35168

JNDl.\N Au I~WULTU.
nAJJ
ItESEAIWH IN~l'[TF'l'E, NEW DELHI.
I, A. R. 1. 6.
lIWIPC-·SJ-IO Alt--2"-(L4U--l ,onH,
TECHNIQUE OF OHGANIC CHEM.ISTHY
AHNOLD WEISSBEHGEH, lldilor
Volum.e IV
DISTILLATION
TECHNIQUE OF ORGANIC CHEMISTRY
ARNOLD WmSSBERGER, Editor
Volume I: Physical Methods of Organic Chemistry

Parts I and II
Second Edition
Volume II: Catalytic, Photochemical, and Electrolytic
Reactions
Yolume III: Heating and Cooling; Mixing; Oentrifug-
ing; Extraction and Distribution; Dialysis
and Electrodialysis; Crystallization and
Recrystallization; Filtration; Solvent
Removal, Evaporation, and Drying
Volwne IV; Distillation
Volume Y: Adsorption and Chromatography
Ilolume Ill: Micro and Semimicl'O Methods
Yolume VII: Organic Solvents
Second Edition
Volume VIII: Investigation of Rate and Mechanism of
Reactions
Volume IV
DISTILLATION
Contributors:
ARTHUR and ELIZABETH HOSE
ARTHUR L. GLASEBROOK
FUEDERICK E. WILLIAMS
CARL S. CARLSON
JOHN n. BO~~\{}N.
u. STUAUT TIPSON
EDMOND S. PERRY
JOHN C. HECKEl{
1951
INTERSCIENCE PUBLISHERS, INC., NEW YORK
INTER SCIENCE PUBLISHERS LTD., LONDON
IN MEMOUY OF
JAMES K. MOORE
Copyright, 19,1'11, by Interscience Puhlishers, Inc.
ALL HIGHTS RESERVED. TMs book or any part thereof ?nl<st not be reproduced in any
form without permission of the publisher in writing. This applies specifically to
photostat arnl rnicrofilm l·cproductions.
INTERSCIENCE PUBLISHERS, INC., 250 Fifth Avenue, N ew York 1, N. Y.

For Great Britain and Northern Ireland:
INTERSCIENCE PUBLISHERS LTD., 2a Southampton Row, London W. C. 1
, ..,
PRINTlm IN THE UNl'rED STA1'ES OF AMERICA DY
MAClt rIUNTI'NG co., IMB'rON, l'ENNi;YLVANIA
.
~.,
.'"" '
TECHNIQUE OF ORGANIC CHEMISTHY
INTRODUCTION
Organic chemistry, from its very beginning, has used speeific tools and
techniques for the synthesis, isolation, and puripcation of compounds, and
physical methods for the determination of their properties. Much of
the success of the organic chemist depends upon a wise selection a.nd a
skillful applieation of thm;e methods, tools, and tcchniques, which, with
the progress of the seienec, ha.ve become numerous and often intricate.
The prei:ient i:iel'ies is devoted to a comprchensive presentation of the
techniques which are used in the orgn.nic laboratOl'y and which arc available
for the investigation of organic compounds. The authors give the theoreti-
Ctll background for an understanding of the various methods and opera-
tions and describe the techniques and tools, their modifications, their
merit!'! and limitations, and their handling. It is hoped tha,t the series will
contribute to a better understanding and a more rational and effective
application of the' respective techniques. Refcl'enee is made to some in-
vestigations in the field of chemical engineering, so that the rmmlts may be
of assistance in the laboratory amI help the laboratory chemist to under-
stand the problems which fLrise when his work ill stepped up to u larger
seH.le.
The field is broad and some of' it is difficult to survey. Authors and
editor hope that the volumes will be found useful and that many of the
reader~ will let them have the benefit of their criticism and of suggestions
for improvements.
A.W.
Re~earch Laboratories
Eastman Kodak Company
Rochester, New York
v
GENERAL PLAN
Volume I (Sct'-oml Edition-in two parts). l>hysicullVlcLhods of Organic Chcm~

1stry. Contributors: J. It. Anderson, E. D. Bailey, W, F. Bale, N. Bauer, J, F.
Bon'ner, Jr., L. n. Brockway, ,T, D. H. Donnay, K. Fajans, I. Fanlmchen, A. L,
Geddes, W. D. Harkill~, W. Heller, E. E, Jelley, T. E. McGoury, H. Mark, L,
Michaulis, D. H, .Moore, O. H. MUller, J. B. Nichols, M. A. Peacock, T. Shed-
lovsky, E. L. SIWll, C, P. Stnyth, D. W. Stewart, J. M, Sturtevlilll" W, Swieto-
~Iaw~ki, G. \V. Thomson, M . .T. VoId,n. D, VoId, n.H. Wagner, II, Wakeham,
and W. We~t
Volume II. Catalytic Ucacliolls, V. 1. KOlllarewsky and C. H, nieszj l)hotochelll~
ieal Reactions, W. A. Noyes, Jr., and V, Boekelheide; Electrolytic Reactions,
S. Swann, Jr.
Volltme III. Heating amI Cooling, R. S. EgIy; Mixing, J. II, Rushton and M. P.
Uofmann; Centrifuging, H. B. Golding; Extractioll alld Distribution,
L. C, Craig and D. Craig; Dialysis tUld Electl'odialysis, H.. E, Stauffer; Crys~
l~~llizatioll and Uccl'ystallization, R. S, Tipson; Filtration, A, B, Cummins;
Solvf.lnt Hemoval, EvaporaLioll, and Drying, G. Broughton
Volume IV. Distillation. Gonirilmiors: ,T, R. Bowman, C. S, Carlson, A, L, Glase"
brook, ,T, C, Hecker, E, S. Pefl'Y, A. Rose, E, G. Rose, R. S, Tipson. and F, E.
Williums
1'olume F. AdsoqHiollaud Chwllllltogl'aphy. H, G, Cassidy
Volume Fl. Micro and Sm:nimicro Methods. N. D, Chcrollis and A. n, Ronzio
Volume FII. Organic Solvents (Second Edition). A. Weissberger, E. S. Pros-
kauer, and.T. A, Riddick
Volume VIII. In'Vestigatioll of Hate !lud lVlechanism of Reactions. Contributors:
G. M, Burnett, B. Chance, S, J, Cristol, n, M. Friess, S. L, Friess, D. E. Green,
E. Grunwald, F, M. HlIclmekens, T, H. James, T. S. Lee, J. E. LuValIe. n.
LiYingst.oll, H, \V, Melville, P. O'COllIlor, W . .T. Priest, and W, D, Walters
TECHNIQUE OF OHGANIC CHEMISTllY
YnlUlllc J \'
DISTILLATION
PREFACE
Distillation is one of the chief operations used for the isolation and puri-
fication of volatile compounds. Much lahar has been devoted to the
understanding of the tll80ry and to the perfection of the techniques of dis-
tillation at ordinary and reduced pressures, but the information thus ob-
tained is not easy of access. The present volume gives a comprehensive
account of the theory and praetige of batch distillation on a laboratory
scale. Its organization owes mnch to Dr. F. D. Rossini, to whom I ex-
press sincere thanks. I am grateful to the anthors, not only for their con-
tributions, but for many suggestions concerning chapters other than their
own. Aut.hors and editor ncknowledge with gratitude Jhe helpful sugges-
tions, loan of materials, and assistance in the editoriEtl work received from
Miss M. W. Graffiin, Messrs. R. L. Bent and 1. S. Bradley, and Drs. C. H.
Bridges, E. M. Crane, M. R. Fenske, R. F. Mnrschner, W. J. Podbielniak,
H. Skolnik, H. M. Spurlin, P. W. Vittum, and J. S. Whitaker. We are
also indebted to Podbielniak, Inc., for assistance in the preparation of a
number of the illustrations in Chapters II and IV. Arthur and Elizabeth
Rose worked out the integmted list of symbol::> (shown on pages xviii-xxvii)
following the approvecllist of the American Institute of Chemical Engineers
as closely as possible. It did not prove feasible to use this for "Distillation
under High Vacuum," and a separate list was made by Dr. J. C. Hecker
for Chapter VI (pages 495 and 496), following the usage in the literature on
high vacua. Dr. E. M. Crane prepared the subject index.
Certain phases of the subject matter of this hook hl1ve been treated in
preceding volumes of the series Technique of Organic Chemistry, particularly
in the following volumes and chapters: VOLUME I, Physical Methods of
Organic Chemistry, Part I, "Tempemture Measurement" by J. M. Sturte-
vant, "Temperature Control" by J. M. Sturtevant, "Determination of
Melting and Freezing Temperatures" by E. L. Skau and H. Wakeham,
"Determination of Boiling and Condensation Temperatures" by W. Swieto-
vii
viii PREFACIU
slawski and J. R. Anderson, "Determination of Vapor Pressure" by G. W.

Thomson, "Determillation of Dem;ity" hy N. Bauer, "Determination of
Solubility" by R. D. Void and M ..J. Void; VOLUMlij I, Part II, "Refraetom-
etry" hy N. Bauer and K. Fajlllls; VOT,TIME: III, "Heating and Cooling"
by R. S. Egly. With fow exeeptiolll:i, the authors of Distillation have not
repeated the information contained ill these chapters. A eumulative
index of the major topics covered in the first. five volumes of this series is
printed at the end of this volume.
During the preparation of Distillation, we suffered the loss of an author,
Dr. J. K. Moore, who died suddenly in July, 1948. We hl1,d the benefit of
his cooperation in the planning of the book as it whole and of Chapter II
in particuhLr. Dr. Moore was a Rcientist of great ability and a fine person
who had become deal' to everybody wit.h whom he collaborated in the
writing of this book.
To the publishers and their staff go thc authors' and my thanks for t.heir
understanding, patient, and generous help in many problems.
A.W.
Research Laboratories
Ea~tmall Kodak Company
Rochester, New York
TECHNIQUE OF OHGANIC CHEMISTHY
Volume IV
DISTILLATION
CONTENTS
l)refacc . ............ , . , ..... , . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . \lii

Symbol Index .......................... , ...... , ............ , ....... , .... xviii
Chapter I. Theory. Ih AR'l'HUR lLud EL1ZABE'l'1! Ro~l']. . . . . . . . . . . . . . . . . . . . . . 1
I. IntroducLion.............. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
. 1. Terminology.................. . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
A. Definitions of General 'Terms and the Apparatus. . . . . . . . . . . 2
B. Definitions of Variables. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
llefiux Ratio, HoldujJ, Vapor Pressure, and Volatility 5
Separat.ing IJower and Theoretical Ph~ies. . . . . . . . . . . . . . . . . {l
2. Helation of Distillatioll to Other Sepa.ration l'l'ocesses-Fl'ac-
tionaticm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . !J
3. Distillation Curves. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
4. Conclusion.............................................. 14
II. Vapor-Liquid Composition Relations........................... 14
Ch.'lses of Liquid Mixtures ...•.......... , ..... , ....... , 15
1. Dinary Systems ......... ' .......... , , ................... , 15
A. Completely Immiscible. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
B. Miscible............................................. 18
Ideal Mixtures ........... , . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
Volatility and Rclat.ive Volat.ility ..................... '" 20
Nonideal Mixtures. Duhem and Other Equations ...... , . 23
Vapor-Liquid Equilihrium Composition Diagrams ..... , . . . 25
Azeotrope~ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
c. Partly Miscible. .. .. . .. . .. .. . . .. .. . .. . .. .. .. .. . . .. .. .. . 29
2. Multicomponent Systems ....................... '" .... . .. . 29
A. Immiscible and Partly Miscible. . . . . . . . . . . . . . . . . . . . . . . . . 29
B. Completely Miscible ........ , . . . . . . . . . . . . . . . . . . . . . . . . . . 30
III. Theoretical Plates, Reflux Ratio, and Column Composit.ions. . . . . . . . 30
1. Plate Efficiency ............. , ............... , . . . . . . . . . . . . 30
2. Theoreticall'lates ....................................... , 31
A. General Procedure for Determining Number of Theoretical
Plates and Height Equivalent to Theoretical Plate. . . . . . . 32
Apparatus for Obtaining Distillate and Still Samples. . . . . . . 33
Choice of Test Mixture and Its Initial Composition.. . . . . . . 35
.IJ. Operation at Total Reflux ... , ..................... " .. . au
'fest Procedure. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
Determination of Plates from Vapor-Liquid Equilibrium
Diagrams. , ..... , .' ... . .. ..... . ...... . ........ .. ... 37
Calculation of Plates by Fenske Equation ...• , ......•..• , 38
ix
x UON'l'EN'l'S
Chapter I. Theory. Continued

Determination of Plates uy Simple Graphic and Tabular
Methods .......................................... . 41
Calculations for Columns Having; More Than 100 Theoretical
Plates ............................................. . 42
o. Operation at Pal'tiallleflux ............................ . 45
Test Procedures ...................................... . 45
Basis for Plate-to-Plate Calculations .................... . 46
Continuous and Steady-State Distillation ................ . 47
Batch Distillation .................................... . 56
3. SWI and Product Composition Relations ................... . 59
A. Under Partial Reflux ....................... . 59
B. Under Total Reflux ................................... . 63
,~. Compositions on Various Plates or at Various Points in a Column 64
5. Summary ....... , ..... ,., .......................... ,., .. . 67
A. Comparison of Methods of Determining Number of Theoreti-
cal Plates ................... , ....... , ............. . 67
B. Precautions in Usc of Height Equivalent to 'l'heoretical Plate
and Theoretical-Plate Data ........ , .. , , , . , ....... , . , . 68
V. Differential Equations of Performance .......... , .. , , ........ , , , 75
1. DiffusioMl Processes in Packed ColumtlH .. , , .... , .......... , 75
2. Packing Factor ........ , ................. , ... , . . . . . . . . . . . . 75
3. Lewis Equations ........................... , , . . . . . . . . . . . . 76
'1. Chilton and Colburn Tl"ansfer Units ......... , . , . , . . . . . . . . . . 77
5. Westhaver's and Kuhn's Equations ..... , ....... , , , .. , . . . . . . 79
6. Effects of Pressure .................•............. , . . . . . . . 84
7. Effects upon Efficiency When Usual Simplifying Assumptions
Are No!; Valid ...... , . , ...... , ... , .. , , ....... , . . . . . . . . . 86
8. Bowman's Equations ............... , ... , ..... , . . . . . . . . . . . 88
g. Composition Relations for Multicomponent Mixtures ... , .... , 91
10. Cohen's Equations ........................... , . . . . . . . . . . . 94
11. Marshall-Pigford Equations ..... , ..... , .. , . , ... , , . . . . . . . . . 96
V. DistiIlation Curves in Batch Distillation ................. , ... , . . 97
1. Curves for Simple Batch Distillation .... , ............. , . . . . . 98
A. Basic Rayleigh Equation ......... , , , . , . , , ........ , , . , . . 99
B. Algebraic Solutions of Rayleigh Equation. . . . . . . . . . . . . . .. 100
c. Solutions of Rayleigh Equation by Graphic Integration. . . .. 102
2. Curves for Fractional Batch Distillation When Holdup Is As-
sumed to Be Negligible ....... , .. ,. , .. , .. , ' . , ...... , . . .. 104
A. B.ayleigh Form of Equation ..... , . , , ... , ... , ...... , . . . .. 104
B. Algebraic Forms Based on Ass\Uuption of Total H.eflux ... " 107
c. Rayleigh Equation for Multicomponent Mixtures. . . . . . . .. 108
3. CUl'VCS Calculated by Stepwise Method. . . . . . . . . . . . . . . . . . . .. 109
4. Curves Calculated from Millimum B.cflux (Infinite Plates) and
Henry's Law Equations .•.................... , . , . . . . . . .. 111
5. Calculation and Determination of Holdup. . . . . . . . . . . . . . . . . .. 112
A. Holdup of More Volatile Component., ............. , ' . . .. 112
B. Total Holdup ............................... , . , , .. , . . .. 117
c. Direct Determination of Holdup of Individual Component.. 119
D. Factors Determining Total Holdup in a Packed Column. . .. 120
CONTENTS xi
Chapter I. Theory. Continued

G. Clll'ves for Fractional Batch Distillation Wh~1l Holdup TH Ap-
preciable. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ] 20
A. General .Equ/itioll by ExtensioTl of Hi1yleigh Equation. 120
B. Differential Equal,ions for Plate Compo~iti()Il~. 125
c. Stepwise Plate-to-Platu Calculations. . . 126
7. Equations for Equilibmtion Time. . . . . . . . . . . . . . . . . . . . . . . . . . 12D
8. CompariHoll of Continuous and Intermittent Take-off Proce-
dures. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 131
(). Curves for Changillg Hcflux lbttio to Mnintltill COJlfltant Distil-
!:tIe Composition. . . . . . . . . . . . . . . . . 131
A. Stepwise Inel'pase in Reflux Ratio. . . . 131
n. Gratiual Increase in Reflux Ratio. . . . . . . . . . . . . . . . . . . . . . .. 132
Equations for Time Required and for Yield Fmctioll When
Holdup Is Negligible. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 133
Equations for Time Required !wci for Yield FmctioTl When
Holdup Is ApIlreciablc. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
10. Curves for Fract.ional Bat,ell Distillation of Multicomponent
IVIixtures. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
VI. Calculated Effect of Process Va.riubles in BlItch Distillation. . . . . .. .. 145
1. Process Varinblcs in Batch Distillation. . . . . . . . . . . . . . . . . . . . . . 14.')
2. Calculated EtIcct of Holdup. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146
A. Etfe~t on Proportion of Charge Distilled. . . . . . . . . . . . . . . .. 147
B. Effect OIl SluupnosH of SeIJI\l·ILtion. . . . . . . . . . . . . . . . . . . . . .. 147
Theoretical Predictions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 147
Experimental Work.... . ... . . . . . . . . . . . . . . . . .. . . . . .. .. .. 152
3. Calculated Effect of Reflux Ratio, Helativc Volatility, and Theo-
retical Plates When Holdup Is Negligible. . . . . . . . . . . . . . . . .. 153
A. Effect of Heflux Ratio. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
Calculations from Log St Equlttio!l. . . . . . . . . . . . . . . . . . .. 155
Methods of Summarizing Effect of Hcfiux Ratio. . . . . . . . . 157
D. :EITeet of Heflux Ratio OIl Effect of Number of Plates. . . . . . . . 158
u. Effect of Hcfiux Ratio-Stedman Procedure. . . . . . . . . . . . .. 160
4 Calculated Effect of Helativc Volatility and Number of Plates
(Total Reflux and Negligible Holdup AS8umed). . . . . . . . . . .. 162
A. Optimum Hange of Over-all Fractionating Factor. . . . . . . . . 164
n. Minimum Plates for a Given Separation.... . . . . . . . . . . .. .. 164
C!. Calculated Position of Break in Distillation Curve ....... " 165
5 Calculated Effect of Initial Composition on Shape and Position
of Distillation Curves. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 165
(). Cakulation of Reflux Ratio and Theoretical Plates by Poh~
Height Equations fol' Tot:,[ Heflux and Negligible Holdup. . .. 1G6
7. Calculation of Theol'oIical PlntoH from Relative Volatilit.ies and
Empirica[ Fractionat.ing ]!'aCLO)'S ..•.•.................. " 168
8. Correlation of Comhined Effects of Helative Volatility, Reflux
Ratio, and Theoretical Plates. . . . . . . . . . . . . . . . . . . . . . . . . . .. 169
A. Separation of Close-Boiling Mixturos ................... , 172
B. Calculation of PlateR Required from Difforence in Boiling
Points ............................... '" ........... '" 172
General References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 173
xii CONTENTS
Chapter II. Ordinary Fractional Distillation. Part I. Apparatus. By A. L.

GLASEBROOK and F. E. WILLIA~fS, . . . . , . . . . . . . . . • . . • . . . . . . • • . . . . • 175
1. Introduction ... , ................ , ' , , , . , , ' , , , , , , ' , , , , . , , , , , , .. 175
II. Column Section .. , ....... , . , , , ' , , , , , , . , , ' . , ' , , ' , , ' , , , , 178
1. Types of Columns .... , ... , ' , , .. , , . . ,' ,,,,.,. 178
A. Wetted-Surface Columns .. , , , , 178
Empty Tube Columns ... , , ' 179
Spiral Columns ..... , . , .. , . , . , " ",.,."." .. ", 183
Wire-Coil Packed ColuIlllls ... ' , , ... ' ........... . 190
Distorted-Wall Columns .... ", ",. , , .. ' " " , , , 193
Concentric-Tube ColumnH, . , , . , 194
ranked Columns .... ,.,., .. ", 108
B. Phte Columns ............ . 210
Bubble-Cap Columns .... ' .. , . , , 220
Perforated-Plate Columns .. ' , ' , , , , . ' , ' , .. , . , , . , , ' , 222
c. Rnctifying Sections Employing Moving Parts. , , , , , ' , , ' , ' , 226
2. Oonstruction of Oolumns ....................... , . , , ...... . 235
A. Column Tubes .............. ' ............ ' , , , . , , , , . , , . 235
B. Paeking SUPPOl'ts ..... , .. , , . ' , .. , . , , , , . , . , , , , , .. , . , , , , 236
c. Packing Procedure ..... ' , ..... , , , .. , . , ,,.,,,,,, 237
D. Column Insulation ..... ' , , ................ , .. , .... , .. . 238
, r. Condenser Section .... , ...... . 249
1. General. .... , .......... , .. 240
2. Total Condensers ...... , , . , . ,' ...... . " 251
3. Partial Condensers ..... " ............. , ........ " .. 250
IV. Still PotH ... , , ............. , .. ' , . ' . , , .. , .. , ' , . , , , , , , . , . , , .. , , 260
V. Heating, Control of Heat Input, and Temperat,ul'e Measurements .. . 262
1. External Heat.ing .. , ............. , . , ... , ... , .. , . , , , , . , ... . 264
2. Internal Heaiflrs .... , .......... , 26ft
3. Control of Heat Input .... , ............ , .............. , 268
4. Reflux-Temperature MeaHlI]'PlllnnL .... , ... ' 273
Measuring Devices .......................... . 274
VI. PJ'eH~ure Cont.rol. .............. , ..... , ........... . 280
1, General....... ' .... , ... , ... ' 280
2. MaIloHtats....... , .................. . 281
A. Effect of Tl'mperature Changes on Manostats .. , 287
B. MisCiellaneous Factors in PresHure Control ..... . 291
VII. Miscellaneous Accessories ........................ , 292
1. Joints and Stllpnocks .. " . " . , ...... , ........ , . 292
2. Lubricants ............ , ........ , , .. , ..... , . , 203
3. General Suggestions for Mounting Bquipment ..... 204
Chapter II. Ordinary Fractional Distillation. Part 2. Procedure. By li'. K
'VILJ.IAMS . . . . . . . . . . . . . . . . . . . . . . . . . ' • . • . . . • . . . . . . . • . . . • • . . • .. 295
I. Int.roduction ..... ' ...... , . . . . . . . . . . . , , ......... , 295
II. PUrpose of Fractional DiRtillatiuli .... ' , ' ... , ........ ' , .... , , . .. 295
III. Choice of Columns .............................. , . . . . . . . . . . .. 296
IV. The DistiUand ...................................... , . . . . . . .. 300
1. Pretreatment ........................ ,................... 300
CON'rENTS Xlll
Chapter II. Ordinary Fractional Distillation. Part 2. Procedure. Continued

2. Size of Batohes. Chasers. Charging ................ ,...... 302
V. Operation ....................... , .... , ...... ' ......... , ... ,. 304
1. Beginning of Operation ................................. " 304
2. Pre flooding ........................................ , ... " 305
3. Establishing Primary Equilibrium ....... , , , , . . . . . . . . . . . . . . . 306
4. Product Collection ................ , , ..... , ... , . . . . . . . . . . . 307
5. Size of Fractions ............. ' ..... , . , ........... , .. , . . .. 308
6. Interpretat,ion of Results ................ , .............. , " 310
General References . .................... , ... , . . . . . . . . . . . . . . . . .. 316
Chapi{'1' III. Extractive and Azeotropic Distillation. By CARL S. CARLSON. . . . . .. 317
I. Intl'Oduction and Definitions ..... , ..... ' , .................... " 317
II. Extractive Distillation ........................ ; . . . . . . . . . . . . . .. 320
1. Scope........................................... . . . . . . .. 320
2. Theory.......................................... . . . . . . .. 322
3. Selection of Solyent ..................................... " 325
A. General Considerations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 325
B. Vapor-Liquid Equilibria and Equilibrium Stills. . . . . . . . . . . 325
c. Estimation of Effectiveness of Solvent. . . . . . . . . . . . . . . . . . . 333
4. Laboratory Apparatus and Procedures .... , . . . . . . . . . . . . . . . .. 341
A. Batch Extractive Distillation without SOIv.ellt Recycle. . . . . . 341
B. Batch Extractive Distillation with Solvent Rccycle. . . . . . . . 344
c. Continuous Extractive Distillation without Solvent Recycle. 347
D. Continuous Extractive Distillation with Solvent Recycle .. " 349
1'. Control of Operation in Extmctive Distillation. . . . . . . . . . . . 350
5. Calculation of Extractive-Distillation Separations. . . . . . . . . . . . 351
6. Equipment Performance. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 354
A. Packed Laborat.ory Columns... . . . . . . . . . . . . . . . . . . . . . . . .. 354
B. Commercittl Bubhle-Plate TowerH. . . . . . . . . . . . . . . . . . . . . .. 354
7. Typical Industrial Application of Extractive DiRtillation. . . . .. 355
TIl. Azeotropic Distillation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 356
1. Scope................................................... 357
2. Theory .............................. ,.................. 35T
3. Selection of an Entrainer ................................ " 359
A. Chemical Considerations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 359
B. Selective and Nonselective Entraining Agents ........... " 363
c. Desirable Characteristics of Azeotrope Formers. . . . . . . . . . . 364
D. Summary............................................ 368
E. Experimental Verification of Effectiveness of au Eutrainer.. 368
4. Laboratory Apparatus and Procedure for Azcotropic Distillation 372
A. Batch Distillation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 372
B. Continuous Distillation .............................. " 378
c. Equipment Performance ............................. " 380
5. Calculation of Azeotropic-Distillation Separatiolls. . . . . . . . . . . . 381
Amount of Entrainer Required ........................ " 382
6. Example of Commercial Application of Azeotropic Distillation. . 385
General ReferCnces. . • . . . . . . . . . . . . . . . • . . . . . . . . . . . . . . • . . . . . • . . . . 385
xiv CONTENTS
Chaptm·[V. Distillation of Liquefied Gases and Low-Boiling Liquids. By

ARTHUR and ELIZABETH HOSJi;. . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 389
1. Fractional Distillation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 394
1. Apparatus and Accessories... . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 394
A. Columns and Packings. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 394
B. Methods of Insulation of Column Proper. . . . . . . . . . . . . . . .. 402
c. Construction of Still Pot. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 406
D. Condenser Design and Refrigerat.ion. . . . . . . . . . . . . . . . . . . .. 408
E. Measurement of Product Boiling Point. . . . . . . . . . . . . . . . . .. 411
F. Removal of l'l'oduct. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 413
n. Collection and Measurcment of Product. . . . . . . . . . . . . . . . .. 415
H. Auxiliary Apparatus for Analysis ....................... , 416
I. Apparatus for Sampling and for Pretren.tment of Sample. . .. 417
2. Procedure. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 419
A. General OperlLting Procedure .......................... , 420
Preliminary Operations ............................... , 420
Entering the Sample .................................. , 420
Establishing Initial ReJlux. Venting of N ollcoudensaLle
G!lsoa ... , ........... ' . . . .. .. . . . . . . . . . . . . . . . . . . . . . .. 421
Scp!lratioll of Methane Fraction.. . . . . . . . . . . . . . . . . . . . . . .. 422
Separation of Intermediate-Boiling Fraetions (C 2 from Ca, C a
from C.). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 423
Separation of Fractions Boiling near Room Temperal.ure. .. 425
Withdrawal and Measurement of Residue. . . . . . . . . . . . . . .. 426
Automatic Operation and Standardized Procedures. . . . . . .. 427
Errors ...................................... , .... "" 427
B. Interpretation of Data. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 428
Calculation of Results. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 42&
Precision and Limitations of tho Method. . . . . . . . . . . . .. 431
c. Principles of Low-Temperature Distillation ..... , . . . . . . . .. 434
3. Special Apparatus and Procedure. . . . . . . . . . . . . . . . . . . . . . . . . .. 441
A. Micro Columns. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 441
B. Total Condenser Columns. . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 441
c. Other Special Columns. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 442
D. Pressure Columns. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 442
II. Simple Distillatioll. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 443
1. Repeated Simple Distillation. . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 443
2. Single Simple Distillation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 445
Chronological BibtiDY'I'I}.phy of Low·'1'emperaturc Dl:stillation A.ppam-
tus . .............................. , .. . . . . . . . . . . . . . . . . . . . .. 457
Clas81:jied Bibliography. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 460
General References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 461
Chapte.r V. Distillation under Moderate Vacuum. By JOHN R. BOWMAN and R.
STUART TrpsoN. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 463
1. Introduction.............. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 463
II. Theory....... . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 467
1 ~ Fluid Flow., j ~ I I I I I I I I I I I I If' I , ' • , •• , , •• I , , I I I ••• , • , , ••• , 467
CONTENTS xv
Chapl.er V. Distillation under Moderate Vacuum. Continued
2. Rectification ........ , ... ,',., ... ,., ... , ....... , ......... . 469
A. Countercurrent Coutant, Hectification, , ....... . 469
B. Batchwise, Countercurrent HedistilJation, Thermal }{~(,.ti-
fieation ................... , , . ' . , . , .. , , , , .... . 4iO
3. VI1porization nnd Condensation ...... , , .. , .. , ...... . 4i4
III. Apparatus ......... , .... , , ........................... . 4i6
1. Stills ............ , ............ , .. ·.· ...... ·, ... · .. 4i6
A. HorizonLal a{,ilb .......... , ... , .................... . 477
n. Vertical Stills, ..... , , ......... , . , ... . 483
2, Pumps ......... " ......... , ............ "., .. , ........ ,. 487
3. Manost!1ts .............. , .... ,· . , , ..... , . ' . , .. , ......... . 488
'1, Traps, .. ,.,., .. , ... , .. , .. , .... , . , , . , . , 488
5. G!1gcs ...... ,., .. ,.,., ...... , .... ,.,"', .... " .. 489
IV. Operation, ....... , , .. . .......... , ......... , .. , ...... . 491
1. Degassing, ....... ,.,., ...... , ......... , ... "., ...... . 491
2. Prevention of Bumping ... , .. . .. ,., ...... ,., ... ', .. " ... . 491
3, Determination of Fractions, ., , .. , .. " , .. ' .' ., ,. " .. 493
4. Control of Pressure ............ , , , ... , , , , ......... , , .. ' , , . 494
General Refel'cnnes . .............. , .. , . , .. , , . ' . ' , , . , ....... ' , .. 494
C'hapter VI. Distillation under High Vacuum. Part 1. Distillation. By E. S.
PERRY .• , , . , ..... ' . , , ... , , .. , . , . , , .•.•. , . , , , , • , . , , ...... , 495
Symbol Index .. , .. , . ' , , . , , , .. , , . , .... , , ..... , ' , , . , . . . 495
1. Introduction .. , ... , . , , .. , .. , ' , , , , ' , . , .... , , .. ' . , ... ' , , .. , . 496
II. General Aspects of High-Vacuum Di::;tilJation. , . , , . , , , .. , , .. , .. " 498
1, Basic Differences in Distillation Pl'ocesses .. , , , , , , , . , , , ..... , 498
2, High-Vacuum Stills-General. , ' , , ....... , " , . , , .. , , . . . . . .. 498
3. Distillation Process ........ , , .. , .. , .. , , .... , , , , .. , , , , , ' , ., 500
A. Rate of Distillation ..... , . , ' , , , , , ... , . , , . , , , ..... , , . , " 500
B. Thermal Hazard ...... , . , , .. , , , .. , , .. , , , , , , , , ... , ... ,. 505
c. Separation or Fractionation .. , , , ... , , ., . ' . , , , .. , . , , , .. " 507
III. High-Vacuum Distillat,ion TechniqucR .. ' , ..... , . ' , ' , , , . , , . , . , , ., 509
1. Simple High-Vacuum Distillation, " , .. , . , .. ' . , , . , ... , , . , . " 509
2. Cyclic Dist.ilJation ... , ., ....... , . , , .. , , , .. , . , , , .... , .. ' .. , 511
3. Analytical Distillation ........ " ""'" " ", ,,' . , ..... ' .. , 517
IV. St.ill Designs for Laboratory Use, ... , , , , .. , , , .. ' . , , , . , . , . , , , , " 527
1. High.Vacuum Pot Stills .. , .. , , . , , . , , . , , .. , . , .. , , ....... , " 527
A. Concent.ric-Tube Types ...... """"""""""""" 527
B. Hickman Type ........ , . , , , , . , , ... , , . , , . , , , , ....... , " 528
c. Matchett and Levine Type, .. , .......... , , , , . , , , , , .. , ,. 529
D. Riegel, Bci$wangcr, and LanzI Type, ... , , .. , , , ' . , .. , , , ,,52\)
J.]. Carothers Type .... , .......... , .... " . ' , , ....... , , , , " 531
F. Strain and Allen Modification. , ...... , .. , , , , . , . , , .. , , , ., 531
G. Gould, Holliman, and Niemn,llll Design,. , , , , . , . , . , , , , , , " .531
H. Almquist Diffusion Type ..... , , . , ....... , , , , , .. , , , . ' , " 532
2. Falling-Film Stills ................ , .... , ... , ...... , ...... ' 532
A. Rickman Design ....... , ..... , , .. , ... , , , , ........ , ' , '. 535
B. Quackenbush and Steenbock Design .. , .. ' , , . , ... , , ... , '. 536
xvi CONTENTS
Clwpicr Fl. Distillation under High Vacuum. Part 1. Distillation. Continued

c. Det.wiler lIud l\Iaddcy Design. . . . . . . . . . . . . . . . . . . . . . . . . .. 537
D. Taylor Dcsign. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 538
N. Breger Design. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 539
3. Centrifugal Stills. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 0540
4. Cascade Fl'IIctionating SUlk. . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 540
Chapter VI. Distillation under High VacuulTI. Part 2. The Vacuum System.
13y J. C. HBCKl'll. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 540
1. Introduction ................................................ , 540
II. Definitions of Terms ........................... '.............. , 541
III. Gas lrlow at Heducetl Pressures ............................... , .545
1. Gellom! Di~c\msi()ll. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 545
2. Flow at. Intcl'mediat.c Pre~Hllres. A Gpllcml Equat.ion. . . . . . . .. 545
3. Free Molecular Flow. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 551
A. The Factor lV .......... ' ............................ 551
D. Aperture............................................. 552
c. Long CyliudriclIl Tubes ............................... , 553
D. Short Tubes of TJlliform Cross Section .................. , 553
B. Ducts of Nonuniform OJ' Irregular Cross Section .......... , 555
F. Bl1ffies, Valves, ete ........... , ........................ , 555
G. Cold Traps. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 555
H. ElbOWH mui Bends. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 557
I. Slits of Heetl1ngula.r Sec tion ........................... , 558
J. Concent.ric Tubes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 558
4. Vi~e()\ls Flow. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 560
A. Lnmiuar Region ..... , .. , .. , ......... , . , ............ , .. 560
ll. Turhulent Region .................................... , 562
IV. Vacuum Pumps. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 563
1. Backing Pumps or Fore-Pumps ........................... , 563
A. Mechanical Rot!Ll'y Pumps ................. , . , . . . . . . . .. 563
B. Steam and Water Ejectors ............................. , 567
2. Oil-Vapol' Pumps.............. ........... .... ............ 568
A. General.............................................. 568
B. Condensation Pumps. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 569
c. Booster Pumps ........................... , . . . . . . . . . . .. 570
D. Oil-Vapor Ejector Pumps ..... , . . . . . . . . . . ... . . . . . . . . . . .. 572
E, Selection of Vapol' Pump-Performance Specifical.iolls. . . . .. 573
F. Liquids for Oil-Vapor Pump Boilers ......... ,........... 574
V. Vacuum Gages.............................................. 576
1. General. ............. , .. , .. ,............................ 576
2. }rore-Pressure Gages. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 577
3. Pine-Pressure Gages .......................... , . . . . . . . . . .. 579
VI. Cold Traps and Baffles.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 582
VII. Leak Detection .................................. , .......... , 586
VIII. Vacuum Fittings ........................... , . . . . . . . . . . . . . . . .. 592
IX. Integrated Systems ......................... , . . . . . . . .. . . .. .. .. 597
General References •• ........... ' ............. , . . . . . . .. . . . .. . .. 601
CON'l'EN'fS xvii
Clwpter VII. Sublimation. By R. S'l't.<Al1:I· TIl'SON . . . . . . . . . . . . . . . . . . . . . . . . . . . Goa

I. Introduction .............. . G03
II. General Considoratiolls ...... . GO!
III. Types of Sublimation .... , . . .............. . GOS
1. Simple Sublimation ....................... , . G08
2. Entrainer Sublinmtion. . . . . . . . . . . ....... . Gon
3. Vacuum Sublimation ....... , ................. . 610
J. Quasisublimution ........................................ , G12
IV. Design and Opel'a,t.ioll of Sublimatol'~. . . . ............. . G13
1. Vertical Su\Jlinmtol's .................................... , . 6B
A. Simple Sublimation ... , ................... , ........... . 614
n. Entmiuer Sublimation ..... , ...................... _... . 61\)
c. Vacuum Sublimation ................................. . 620
D. Elltmincl'-VaCllUm Sublim~1ti()n ........................ . 62i
2. Horizontal Sublimutol's .................................. . 62n
A. Simple Sublimation ................................... . 629
B. Ent.miner Sublimation ................................ . 630
c. Vacuum Sublimation ......... _ ......... , . , .... , ....... . 633
D. Entrainer-Vltcuum SublimlLtion .... , .. , ... , ......... . 639
V. Examples of Sublimation ............ _ . , .... , ... . 641
Gcncml References . ....... , , ........ _ . ,. ' ........ . 645
Index _.......................... " ............................ . G47
Cumulative indexes, Volumes I-V ......................................... . 665
TECHNIQUE OF OHGANIC CHEMISTRY
Volume IV
DISTILLATION
SYMBOL INDEX
The page numbcr~ are those on whioh the symbols are defined or iirst
used. 8ymbols for Chapter VI will be found on pages 41"}5 [end 4\)6.
Symbol Definitions Page

--------~---------
(1) Empirieal eOll:st[Lnt, Vtlpor~pr(1SS\Il'e equat,ion ............... . 1!1

(2) COllveniencc symbol, Van Laal' equation ..... , , , , . , , , . , , .. , 24
(3) Empirical constant, Bownmu"Br);mt, equation ...... , .. , .. , .. 00
(4) Convenience symbol, J"cwis equation ...... , ......... , , . , , . , 77
(5) COIlVlmicnce symbol, disUllation equations including holdup .. 123
(6) Efficiency factor, 'i.e., ratio of throughput. to holdup per plate. , 1'17
(7) IVloles component A in nlUlticomponent mixture, R:1yleigh
equation ...... , .. , .......... , . , ... , ....... , , .. , ....... , JOD
(8) Area of upper bulb of lIllLllostat. , . , .. , ......... , . , ....... , 2SD
(0) Solute, extl'lwt.ivl> dist.ilIation .... " ... , ..... , . , ... , , ...... . 322
.flo (1) Moles component A in IllulticoIllponent mixture in still nt timc
of reference during batch distillation ...................... . 10l)
(1) Moles component A in l11ulticompoucnt mixture in stilll1t dif-
ferent times during batch distillation ......... , ............ . lOS
a (1) Empirical constant, Clark equation, . , , .. , . , , .. , .......... . 25
(2) Convenience symbol, Lewis equation,., ............ , . , .... . 77
(3) Empirical constant, Marshall-Pigford equatiun ............. . 90
(4) Volume of lllercury in height II of manostat, ............. , . , 2S9
(5) COllvenience symbol, empirical sublimation equation ........ , G05
(6) Thermal coefficient of expansion of gases .... , . , ........... . 50!)
a' (1) Empirical constant, Clark equation, .. , ................... . 25
(1) Weight fraction nonvolatile material in charge at various times
during holdup determination ........... , , ........ , , ...... . 118
A. Pt. (1) Aniline point ......... , ........... , , ............ , .. , , , . , 385
B (1) Conveniellcesymbol, Lewis equation ...... ,', .. "., ...... , 77
(2) COllvenienee symbol, ]3o\YlUnu-Briaut equation ...... , , . , ... , flO
(3) Empirical constant, vapor-pressme equation ........... , , .. . I!)
(4) COllvenienee symbol, Van Laar equation ..... , .. , ........ , . 24,
(5) Moles t:OlllpOnellt, B in JI1ulticonlponent mixture, Hayleigh equa-
tion .............. , .. , . , .. , , ............. , , , , , .. , ..... . IOn
(6) Moles "bottoms" taken from stripping section ill unit timc ... . .53
(7) Subscript pertaining to "bot,toms" of stripping section of col-
umn ................. , ........... , .. , ........... , ..... . 53
xviii
snmOL INDEX xix
..._ - - - _ .. _- ...._--
Symbol Definitions P,,~e
._---------
(1) Moles component B iIi Illulticomponent mixture ill still at
various times, Rayleigh equlltion ............ . 108
(1) Empirical conskmt, Clark equation ................. . 25
(2) Convenience symbol, 8J1lOker equation .... . 55
(il) Convenience symbol, Lewis equation ........... . 77
(4) ]i;mpirknJ constant, l'Ibl'shall-Pigford eqlllltion .. no
(5) Convenielll:e symbol, holdup equation ................. . 11(\
(0) Convenienc:e symbol, distilbtion equations including holdup ... 123
(7) Volume of mercury below height h of mftnostat......... . 289
b' (1) Elnpil'ical constant, Clark equation .......... . 2,)
C (1 ) Constant, Carswell equ!1.tion ................... . 7(;
(2) Constant, equation relating n and a . .. , ...... . 171
(:3 ) Convenience symbol, Cohen equations .......... . 9+
(4) Moles componellt C in Illultic:omponent mixtUl'(, .. . lO\}
(5) Constant, empirical sublimation equation .... . 605
G\ (1 ) Symbol, Bowman-Brhnt equation ............. . 90
G\J (}':! (1) Positive constants, Cohen del'iV:J.tions ..................... . 9(\
Cw (1) 'Veight of vohtile (~harge, holdup detel'mina tion. 118
C.B.M. (1) Constant-boiling mixture ............. , . 31\}
C.S.T. (1) Critieltl solution temperature ....................... . 33+
c (1) Convenience symbol, Smoker equltt.ion .......... . 55
(2) Convenience symbol, Cohen equations .... . 9·1
(a) Convenience symbol, Lewis eqUittion ..................... . 77
D (1) Subscript pertaining to di~tillate. . . . .. . ................. . 47
(2) Moles distillate per unit time ............................ . 47
(3) Mo1es distillate .................................... , ... . 100
Dv (1) Molecular vapor diffusion coefficient, Westhaver equation ... . 80
Dv (1) Molecular vapor diffusion coefficient, Docksey-May equation .. 86
(1) Weight of distillate, holdup determination ................. . 11!l
d (1) Sign of differentiation
(2) Effective average molecular diameter, Docksey-May eqnation. 86
(3) Density of liquid, Carswell equut.ion ............... . 68
(1) Density of mercury at T • .......... 289
(1) Density of mercury at 'l's + t:.t .......................... . 289
em (1) Murphree plate efficiency ................................ . 31
F (1) Subscript indicating feed plate ..................... , ..... . 55
fl, f2 (1) Functional notations ... , .. , ............................. . 122
.(h(x, . .. ) (1) Holdup of moles more volatile component, as function of still
composition ........................................... . 121
!,,(x, ... ) (1) General function relating distillate composition x n to still
composition x• .. , . , . , ' , ... , . , .......................... . 121
XX SYMBOL INDEX
Symbol Definitions Page
f' (1) First derivative of a funetion (f) . ...................... '... 121

f~(x, .. . ) (1) Derivative of i)j(xs . .. ) ............... ' , ........ , ..... ' ... 121
J~(x ... . ) (1) Derivative of i.(x •... ) ............... , ,. , .... ' ,., ,., .' ., , 122
!(a) (1) lVleasure of ease of separation, Bowman-Briant equation, . , . ' , 90
G (1) l\1ass velocity of vapor, Chilton-Colhurn equation ...... , . , . , , 78
1I (1) Total moles holdup in entire column ......... , ..... , , , . , , .. 113
He (1) Total moles holdup in condenser ......... , , . , . ' ... , . , , .. , . 121
Hlt H 2,..,Hi (1) Moles holdup per actual pln.te, or ...... ' . , ...... , . , ....... 112
(2) Moles holdup per theoretical plate, as indicated by ~ubscript .. 113
HL (1) Total liquid holdup in column per unit length orsegment ...... 88
lf n - 1 (1) Total holdup in column above plate n - 1 ................. 57
H, (1) Total holdup in top plate ................................. 57
Hv (1) Total holdup as vapor per unit length or segment of column .... 88
H", (1) Total weight of holdup ................................... 118
Hq, (1) Height of separation stage, Bowman-Briant equo.tion ......... 88
H.E.T.P. (1) Height equivalent to It theoretical plate .................... (I
H.T.U. (1) Height of a transfer unit ................................. G
11 (1) Total moles holdup of anyone component .................. 112
(2) Subseript pertaining to holdup ............................ 57
(3) Height of mercury in manostat ............................ 288
ho (1) ]'I'Ioles more volatile component in holdup at the time a particu-
lar boilup rate is commenced ............................. 129
h. (1) Moles more volatile component in holdup when equilibrium is
reached ................................................ 129
hi (1) Holdup of anyone component on an actual plate ............ ll2
(2) Holdup of anyone component on a theoretical plate ....... , .. 113
h. (1) Hcight of mercury at T •. ................................. 289
h, (1) Height of mercury at T. -I- tlt. ......... , ................. 289
I (1) Constant of integration, Clausius,ClapeyroIl equation ........ 19
(2) Intermediate fraction in batch distillation .................. 151
h (1) Convenience symbol, Bowman equation .................... 139
(1) Subscript indicating values applying to It particular point or
plate in column ......................................... 88
(2) Distance up column in terms of H.E.T.P. units ............. 115
(3) Subscript implying values of ith component ................. 92
(4) Component of multicornponent mixture .................... 141
J1 (1) Convenience symbol, Bowman equation .................... 139
K (1) Equilibrium constant in relation of y = K:t .... ............. 30
(2) Constant dependent on stationary vapor film adjaccnt to the
liquid phase, Chilton-Colburn eqU!1tion ..................... 78
(3) Integration constant ..................................... 124
SYMBOL INDEX xxi
Symbol DefinitionH
k (1) Proportionality constant

(2) Constant, Smoker equlltio!l. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
(3) Constant, Cohen equations ............... , ....... '" . . . . . 94
kl' k2 (1) Constants, Diihring equation. . . . . . . . . . . . . . . . . . . . . . . . . . .. . 19
kD (1) Integration constant, integrated Rayleigh equation iu terlllS uf
xp ..•.•.•.......•.....•...•. ......•................. '" 107
J..'H (1) Henry's law consta.nt ........................... ' ... .. ... 111
k1 (1) In tegrn.tion constant. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 111
k. (1) Integration constant, integrated Hayleigh equation, in terms of
x,..................................................... 107
L (1) Liquid reflux flow rate in mol(Js per unit; time .... ' . . . . . . . . .. . 47
(2) Latent heat of sublimation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 604
L" L'-h etc. (1) Liquid rcflux flow rate in moles per unit time from top plate,
etc ........... , .................... , ..... , .... . . . . ..... . 49
(1) H.E.T.P. umler totaJ reflux. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
l' (1) H.E.T.P. under finite reflux ............... , .. " . . . . .. . . . . 82
1lf (1) Molecular weight
(2) Abbreviation, Smoker equation. . . . . . . . . . . . . . . . . . . . . . . . • . . 55
(3) Average molecular weight of two substances, Carswell equation 68
[M] (1) Moles per liter more vobtile component in liquid, Cohen
equations ............. , . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . 94
1U (1) Subscript indicating ?nth plilote in stripping section of GOIUIlm.. • 53
(2) Abbrevia.tion, Smoker equation. . . . . . . . . . . . . . . . . . . . . . . . . . . 511
fm] (1) Moles per lit.er more volatile component in vapor, Cohen
equations ............. , . ... . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 94
m. (1) Slope of bl1tch distillation curve at midheigbt. . . . . . . . . . . . .. 16()
N (1) Number of moles. perfect gl~s law. . . . . . . . . . . . . . . . . . . . . . . .. 16
(2) Number of moles of nth component in multicomponent mixture 10\)
N (1) Number of molecules per milliliter, DockHcy-May equation ... , 86
[NJ (1) Moles per liter less volatile component in liquid, Cohen equa-
tions ................. ' ............................. '. . . . g.}
Nl (1) Number of moles component 1 in perfect gas law. . . . . . . . . . . . 16
N, (1) Number of transfer units, Chilton-Colburn equation. . . . . . . . . . 77
n (1) Number of theoretical plates... .. . . .. . . . . . . . . . . . . . . . . . . . . . 13
(2) Subscript indicating nt.h plate in column ................... , 31
(3) Number of separation steps or stages, Lewis equation. ... . . .. 76
(4) Distance up column measured in transfer units, 'Marshall-Pig-
ford equation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
(5) Constant, empirical sublimation equation... . . ... . . .. .. . . . . . 605
[n] (1) Moles per liter less volatile component in vapor, Cohen equa-
tions ..•..................................... , . . . . . . . . . . 94
nD (1) Refractive index ..... '" . . . . .. . .... . . . . . . . . . .. . . . . . . .. .. .43
xxii SYMBOL INDEX
Symbol Definition. Page

--~~---------------------------
1) (1) Total gaseous pressure. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Hi
(2) Subscript indicating first plate above still, simplified derivation
of Smoker equ!Ltion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
) (1) Vapor pressures of pure subst[\llces
P
PI,
j)A,PU
p~ ! (1) Partial pres[mrr.s of indivicilml componcnts of mixtures
p, (1) Pressure of mereury at '1" ........... ,,.. ,. , . , ,, ,.. ,. , , ' , , 280

p, (1) Pressure of mercury at T, - til .. , ' , ' , , , ..... , .. , .. 289
Q (1) Throughput ............................. , ............. . 181
q (1) Convenience symbol, distillation equations including holdup .. 123
R (1) Gas constant, perfect gas law .................. ' , , ....... , 16
(2) Reference component, II1ulticomponent mixture .. , . , , .. , . , . 142
(:J) Rate of entraillcr flow .... , , , . , .......... , , ... , . , , .. , . , .. 609
(1) Reflux ratio, LID . ................. , ...... , ........... . 1a
(1) Minimum reflux mt.io ........... , ....................... . 51
(1) Reynolds Number ...................... , . , ....... . 467
l' (1) Distance from axis of packed column ..... , ............... . 80
1'0 (1) A particular value of distance from axis of packed column. , .. . 81
8 (1) Surface area of packing, Carswell equation ................ . 76
(2) Cross-sectional area of column, Chilton-Colburn equation .... . 78
(a) Total moles in still in batch distillation at any instant ........ , 99
(4) S = L, Berg·James equations (80), (80a), and (81) ...... , , . , 130
(5) Solvent, extractive distillation ................... , ...... , .. 322
s (1) Fraction of charge remaining in still ........... , .......... . 125
80 (1) Total moles in still in bn.tch distillatioIt at the time when prod-
uct n)Inoval is commenced ................. , .. , .... , , , ... . 122
(1 ) Total moles charge in batch distillation. , . , , .............. . 100
(1) Total moles in still at end of batch distillation with constant
distillate composition ....... , . , .. , ... , ..... , ............ . 132
(1) Total moles in still at time t . ...... , , , ................... . 102
(1) Fraction of charge remaining in pot at midpoint of distillation
curve .. , ........... , ................................ '" 166
s (1) One-half the distance between parallel plates used as distilla-
tion column, Westhaver equation ................. , ...... , . 83
T (1) Absolute temperature ...................... , , .. , , .. , . . . . . 16
(2) Temperature in heating chamber (degrees Centigrade). . ..... 600
1'A, 1'B I
(1) Absolute boiling points
T" 1'2 ~
T. (1) Critical temperature .......... , ...... , , .. , .. , .. , . , , . . . . .. 605
Tic (1) Boiling point, empirical sublimation equation ..... , ..... , . ,'. 604
SYM.llOIJ INDEX XXllI
Symbol Definitions Puge
1", (1) Temperature chosen as standard for mHnostat.. . . . . . . . . . 289

(2) Melting point, empirical sublimat.ion equatioJl. . . . . . . . . . . . .. B04
T, (1) Triple point tem.pemture ..... , .............. '" ....... '" (105
(1) TempcrutUl'c (degrees Centigmtie) ............. , . . . . . .. .. . . 28
(2) MeLm temperature (degrees Centigrade) ..... , ........... " . 44
(:3) Temperature increase in wuter, boilup rate calculation ..... " 271
(4) SUbscript referring to top plate or top of column. . . . . . . . . . . . 47
(15) Time.................................................. \)4
t - 1, ete. (1) Subscript referring to pl:,te next to top, etc. or to point one
theoretical plate removed from top, cte.. . . . . . . . . . . . . . . . . . . . -17
t.1, tIl (1) Boiling points (degrees Celltigmde) of pure liquius at (1 particu-
lar pressurc. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1\)
USA (1) Usual simplifying assumpLions of distillation. . . . . . . . . . . . . . .. ·i7
y (1) GaseOllS volume, perfect gas law. . . . . . . . . . . . . . . . . . . . . . . . . . 1(i
(2) Volume of vnpor space, Carswell cqua,tion. . . . . . . . . . . . . . . . . . 76
(3) Volume of mercury in manostat ............ ' . . .. . . . . . . .. .. 289
(-1) Difference in volume bet.ween solid and its vapor. . . . . . . . . . .. 604.
(5) V npor rate in column in moles vapor per unit, of time. . . . . . . . 47
Va (1) Radially averaged vapor velocity, Westhaver equation....... 81
VB (1) Boilup ra,te. , . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 271
'V" V.- I , etc. (1) V:1por rate at plate indicated. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
v (1) Volatility .............................................. 20,318
(2) Molecular root mean square velocity, Docksey-May equation. . 86
(3) Net rate of transfer of more volatile component into column. . 129
(4) Vapor tension at TOC.. . .. . .. . . . .. .. . . ... . . . . .. . . . ... . . .. 609
(1) Volatilities, various components .......................... 20,318
w (1) Weight of volatile substance, entraincr sublimation equation.. G09

w (1) Weight of nonvolatile material, holdup determination. ... . . .. 118
(2) Thickness of liquid film, holdup equations.. . .. . . . . . . ... . . .. 181
(3) Water flowing through condenser, boil-up rate calculation. . . .. 271
(1) Individual weights of components in mixture. . . . . . . . . . . . . . . . 16
(1) Molar ratio less volatile to moro volatile component in liquid,
Cohen equation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
x (1) Mole fraction more vola.tile component in liquid mixture. . . . . . 21
Xo (1 ) Mole fraction more volatile component in liquid at feed plate. . 92
Xl, X2, ote. (1) Molo fraction components 1 and 2, etc. in liquid mL'-;:turc. . . . . . G
(2) Mole fraction more volatile component in liqui<l at plates 1 and
2 ............... '.... , ',' ................. " .......... " . :38
(3) Mole fraction more volatile component in liquid at times 1 and
2............................................... ....... 110
XI (1) Mole fraction more volatile component in liquid refiux returIl-
ing to still fronl bMB of oolumn. . . . . . . . . . . . . . . . . . . . . . . . . .. 129
xxiv SYMBOL INDEX
S~'lllhvl DcJinitjolls Page
XA (1) Mole fraction component A in liquid, multicomponent mixture 108

XII (1) Mole fraction more volatile component in bottoms from strip-
ping section. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
(2) Mole fraetion component B in liquid, multir,)lllponent mixture. IDS
XAS, XJlS (1) J'vIole fractions A n.nd D in liquid in presence of solvent at equilib-
rium................................................... :ns
x~ (1) Mole fraction more volatile component in charge. . . . . . . . . . .. 166
Xr.i (1 ) J\lole fraction component i in charge of multicomponent mix-
ture ........................ , . . . . . . . . . . . . . . . . . . . . . . .. 141
Xv (1) :'vlolc fraction more volatile component in distillate. . . . . . . . . . 47
:rm (1) Mole fractioll more volatile component at head of column just
before first distillate iR removed. . . . . . . . . . . . . . . . . . . . . . . . . .. 127
XV,[ (1) IVlole fraction component A in distillate. . . . . . . . . . . . . . . . . . .. 142
XDa (1) Mole fntction component B in distillate............... 142
XDc (1) i\fole frl1ctioll more vohLtile component in distiJla.te of constant
eomposi tiolJ. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 133
;,J;Jjlt (I) Mole fraction reference cOlllpolleut B. in distillate. . . . . . . . . . .. 142
:J:p (1) lVlole fradion more vola,tile compOlllmt in feed. . . . . . . . . . . . . . . 55
Xh (1) Mole [metion more volatile component in holdup .......... , .. 130
:1.:/10 ( 1) Mole fractioll more volatile component in holdup at time dis-
tillate removal is commenced. . . . . . . . . . . . . . . . . . . . . . . . . . . .. 127
X/lc (1) Mole fract.ion more volatile component in vapor from charge .. , 130
Xi (1) Mole fraction more volatile component in liquid at a particular
point in column. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
:tin) etc. (1) Mole fraction Inore volatile component on or from plate m, etc.
of stripping section. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
Xu (I) Mole fraction more vola.tilc component at pinch (intersection of
Opel'l\tmg line and equilibrium curve). . . . . . . . . . . . . . . . . . . . .. ;~51
Xf4J Xn-II ute. (1) ]\fole fl'ac:tion of more volatile component in liquid from plate,
'Tt, n - I, etc.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
$p (1) Mole fraction more voln.tile component on plate, simplified deri-
vation of Smoker equation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
(1) Mole fraction more volatile component in still in batch distilla-
tion ................ ,................................... 37
(2) Itatio of mole fraction more volatile component to that of less
volatile component in still, Berg-Ja.mes equation. . . . . . . . . . . .. 130
(1) Mole fraction more volatile component in still when product re.-
moval is cOIDrnenced. . . . . . . . . . . . . . . . . . . . . . . . . . . . • . . . . . . . . 122
(1) Mole fraction of more volatile component ill liquid in still at
times 1 and 2 ..•.....................................• , . 122
(1) Mole fraction component A in still .•............... " ..•.. 142
(1) Mole fraction component B in still. . . . . . . . . . . . . . . . . . . . . . . .. 142
(2) Initial mole fraction second more volatile component, Edge..
wOl'Lh-Johnstone equation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 138
SYMBOL INDEX x.-xv
SymlJOI Dcllnitions Page
X.a (1) Mol" fmetion compOllunt C in st.ill. . . . .H2

:roo (1) IV[ulc fr~(\jioll m.ure vO\:ltile COmpUl(ent in eh~u'ge_ 102
X.F' (1) Mole fraction more vohtile cumponent ill still at end of pJriuti
of constant diHtillate eomposition . , , , , - , , . , . - , , _ la2
J:.R (1) Mole fraction reference component II, in sUll. , _, _, _, _.. , , , _" 142
X.I (1) Mole fraction of more vola.tile component in still at time t, , . " 102
J't, :1',-" ute, (1) Molcl fraction more volatile component in liquid from top pinto,
one plate below top, etc., . , . , , , . _... , . , . , . , - _. , ... , , _. , . _ 47
:fwD (J) -Weight fmotion more volatile component in still after holdup
enters column ..... , ..... , . , ..... , .. , , .............. , , , " 120
x,., (1) \Veight fraction more vola.tile component in charge, . , .. , _. ,', 120
:.,,11 (1) Avpmge weight fraction more vohttile component in holdup_ , " 120
x' (1) Mole fruction leHs volatile component in liquid mixture .. , , , '. 4:3
1'" (I) Malar l'ltt.io less volatile to more volat.ile component in vapor,
Cohen equations ..... , _. , , , , . , , , , ... - ... - _, , , , . , .... , . , . 94
11 (1) Mole fraction more voh.tile component in vapor. , , , .. _,_ . _". 21
1/1,1/2, etc. (1) Mole fraction more volatile component in vapor in equilibrium
with liquid at plate 1, 2, etc .. , .. , .... , , .... , , , , , , ... , .. , . . ::18
(2) Mole fraetioll first and second components, et,c. in vapor, . _ , _. (j
(:I) Mole fraetion more volatile Gomponent in vapor at times 1 tlnd
2 .... , ... ,.,.,',.,' _....... ,. ,." .. ' ....... ,.,.,. ,_., '" 110
1!A, ?lB (1) Mole fractioll components A ami il, etc. in vapor, multicom-
ponent mixture ............. , ........ , ............ , . . . . . 108
1/As,lIHA (1) Mole fractions A (tnd Bin vnpor in presence of solvent Itt equi-
libriUIn .................... , .... _..... , , ........ , _. , . _.. 318
1/1 (1) Intercept of the opera.ting line ...... '.', _" . _... ,.",., '" 50
(2) Mole fraction of more volatile component in vapor at a partic-
ular point in a column ........... , ..... , , , .. , ... , .. , .... _. 88
11m, ?/m-l (1) Mole fraction more volatile component from plate In, m. - 1, etc.
of stripping section .. , . , .. , , , , ... , , _. __ , _, . , ..... , . , .. . . . 58
lIn, lin-I, etc. (1) Malo fraction more volatile component from pl1lte n, n - 1, etc. 31
1h> (1) Mole fraction more volatile component in equilibrium with
liquid on plate p . ... , • , . , , , ... _. _, , ... , . , . , , , ... , , _.. ' . . 54
1/. (1) Mole fraction more volatile component in vapor in equilibrium
with liquid in still dUring batch distillation ... , .. , , , , . , .. , , . 37
11,. (1) Mole fraction mom volatile component in vapor initially
evolved from charge .. _, . , . , .. , . , , . , , . , , , ... , . , , , , .... , ,. 130
y" l1J-I, etc. (1) Mole fraction more volatile component from top plate, one
plate below top, etc .. , . , .. , _. , ... , ... , . , ...... , .. , ..... , . '17
1/e (1) Mole fraction morc volatile component in vapor at feed plate. . 92
(2) Ratio mole fraction more volatile component to mole fraetion-
less volatile component at head of column, Berg-James'equatioll 130
y' (1) Mole fraction less volatile component in Vapor, . , ......... , . 94
XXVI SYMBOL INDEX
cYlllbol Delillitions Pagn
!/" (1) Composition of vapor (mole fraction more volatile component)

in equilibrium with liquid at any specified point in column. . . . . 77
1/.* (1) Composition of vapor in equilibriullllI'ith liquid on phlte n. . . . :31
Z (1) Height of any section of a column .. '. . . . . . . . . . . . . . . . . . . . . . . . :{:3
Z (1) Distance from top of column to unit under consideration. . . . . . 04
Zi (1) Total height of coluIllll. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
0 (1) Subscript assigned to feed plate, Bowman equation. . . . . . . . . . 02
a (1) Relative volatility and vapor pressure ratio (ideal solution) . . . . 6
(2) Coefficient of expansion of mercury ...................... '. 289
"'-1B (1) Relative volatility of components A and B in multicomponent
mixture ....................... , ...................... '. 188
OAn (1) Relative volatility of component A and reference component It 142
fJ!.t (1) RelativA vobtility of component i and !L reference component
in mult,il:ompollllllt mixture ...... , , .................... , '. 1-11
as (1) lHodified relative volatility ................... , ..... , ... '. :318
«7' (1) Relative vol'ltility at average absolute temperature ....... ". 188
all, a n + 1 (1) Enrichment factor or over-all sepamtion factor, , . . . . . . . . . . . . 38
{3 (1) Coefficient of cllbieal expansion of glass .................. '. 289
(2) Ratio of hent of sublimation to heat of liquid vaporization at
triple point ................................... , .... , .. '. 605
"Y, 'Yh 1'~, (1) Activity coefficients ........... , ....................... , .20,322
0, Ii (1) Partial differentiation signs ......... , ........... , ......... 2-1.,94
Ii (2) Parameter, Cohen equations .. , , . , ............. , ....... , . . 96
1] (1) Vapor viscosity .... , ............... , ............ , . . . . . . . 88
'11 (1) Viscosity of liquid ....... , ..... , ........ , .............. '. 181
'1. (1) Viscosity of vapor ..... , ... , ... , . , ... , ... , ......... , . , , '. 181
0 (1) Time ........ , .......................... ,.............. 133
(2) COllvenience symbol, Cohen equations. . . . . . . . . . . . . . . . . . . . . 95
(8) Ratio of temportlt,ure T to triple-point temperature '1.', . .... '. 605
O. (1) Equilibration time, total reflux ........... , .............. '. 129
00 (1) Holdup per transfer unit in vapor phase divided by COl'l'cspond-
jng vapor flow rate, Marshall-Pigford equation. . . . . . . . . . . . . . 96
OL (1) Holdup per transfer unit in liquid phase divided by correspond-
ing liquid flow rate, Marshall-Pigford equat.ion ...... , . . . . . . . 96
A (1) Molar latent heat of vaporization. . . . . . . . . . . . . . . . . . . . . . . . . 19
(2) COllvcnience symbol, extended Rayleigh equation .... , ..... " 123
I\A, A/J (1) Molar latent heat of vaporization of components A and B ... " 178
~ (1) Factor in Crosley equation ............................. '. 142
'Ir (1) Ratio of vapor pressures at temperatures '1' and T, . ..... , .. " (i05
p (1) Density, Docksey-May equation ............ , . . . . . . .. . . . . . 86
SYMBOL INDEX xxvii
Symbol Definitiolls Page
PI (1) Density of lic[1li. I, hol.lup equations ..... . 181

IT (1) Pole height, Bowman equat,ion. . . . . . . . . 16(;
(2) Convenience symbol, Cohen equations ..... ' . 94
IT' (1) :iYIcll.:snre of ease of separat.jol1, BOWl11lUl equ:,tiOIl. \10
,p (1) Measure of driving foree, Bowman equation. . . . . . . . . . . . . . . . . 88
,pJJ (1) Symbol (complex funetion of 11. log "')' Bcrg-J[L)ucS equ!1tiull. .. l:lO
,pIl (1) Convenience symbol, Smoker equation. . . . . . . . . . . . . . . . . . . . . 55
,p,T (1) Yiel.l fraction, Erlgcwol'th-.Tolmstollc cqu:1tion. . . . . . . . . . . . . 1:1:3
,p,TO (1) Yield fraction when H = 0, . . . . . . . . . . . . . . . . . . . . . . . . 1:17
'-'--'-~-"----.' -----
CHAPTER I
THEORY
AnTHUR AND ELIZABETH nOSE, Slale Colle!le, Pennsylvania
1. INTRODUCTION.
"Now I arn come to the arts and I shall begin from distillation, an in-
vent'iou of laicr times, a wondelful thing to be prai8ed beyond the power of
men; riot that which the vulgar and unskilled men use, for they do but
corrupt and desimy what is good; bllt that which is done by skillful
artist8. . . . Let one that loves learning and to search nature's secrets, enter
upon this; for a dull fellow will never attain to this art of distilling."
Porta, 1589. *
Distillation and organic chemistry have been closely associated for cen-
turies. Long before organic chemistry was recognized as a distinct branch
of science, the art of distillation was practiced in the preparation of what
we now know as organic substances. The production of potable forms of
ethyl alcohol is the example best known to the layman, and has in fact
stimulated many improvements in apparatus. The distillation of such
essential oils as turpentine and lemon, and of the many medicinal "waters"
was likewise an early art. The alchemists stressed distillation as one of the
processes necessary in the preparation of the philosopher's stone, and in
their zodiac assigned to it the sign Virgo. By 1500 knowledge of distilla-
tion had become detailed enough to merit the publication of a comprehen-
sive treatise on the subject, Das grosse Distillierbuch, by Brunswig, one of
the earliest books on applied chemistry to appeal' in a language other than
Latin.! This book and other publicu,tions of the period show that the early
experts in the art carried out distillation operations of some complexity.
In the subsequent evolution of distillation apparatus, large-scale and
industrial applications preceded its development in the laboratory. Early
in the nineteenth century, ingenious stills for the manufacture of spirits
were devised in France by Cellier-Blumenthal and Derosne~,3 and in Great
* Della Porta, M agia N aturalis, Naples, 1589; quoted in Egloff and Lowry, Ind. Eng.
Chem., 21,920 (11)29).
1 Brunswig, Das grosse Dislillierbuch, Stm~sburg, 1500; appeared in several editions,
under several titles. Seo Anon., Ind. Eng. Ghem., 28, 677, 1445 (1936).
2 See Ind. Eng. Chern., News Ed., 13, 140 (1935).
3 See also Underwood, 1'rans. Insl. Chern. EngTs. (London), 13,34 (1935).
1
2 A. AND E. HOSE
Britain by Coffey.' These efficiently employed heat in bubble-plate

towers that Were little different from those used today. Later in the nine-
teenth centul''''
,], when orO'anic
0 chemistry emerged as a science, the clevelop-
ment of the stmctural theory and the enormous amount of research on coal-
tal' derivatives were accompanied by marked refinements in luboratory
stills. The l'a,v-material shortages of the first world war and the extensive
chemistry based on petroleum stimulatecl further improvements in appa-
ratus, of which highly efficient packed. columns of all sizes are the most im-
portant.
1. Terminology
Precise definitions of the terms used in this book arc discussed in detail
in the appropriate sections of the text. The purpose of the following sec-
tion is to give the uninitiated reader a worldng knowledge of the termi-
nology.
A. DEFINITIONS m' GENERAL 'L'ERMS AND THE APPARATUS

In broadest terms distillation is a process of separation based on the
difference ill composition between a liquid and the vapor formed from it.
Condensation of the vapor and recovery of the resulting liquid are almost
always implied. Simple distillation involves the application of heat to
vaporize the liquid and the subsequent cooling of the vapor formed until it
condenses to a liquid at a different part of the apparatus. Sucll a process
is effective in the sepamtion of liquids from nonvolatile solids, as in the
distillation of pure water f1'0111 salt water, 01' the recovery of an organic
solvent from mixtures with nonvolatile substances. In such cases the
distinction from evaporation is not sharp.
A simple distillation also can S(ll've as a means of partial separation of liquids of
appreciably different boiling points or volatilities. Simple distillation is sometimes
referred to as differential distillation because in general the composition of both
vu.por and residue changes gradually by differential amounts as the operation pro-
ceeds. In flash distillation or eqUilibrium distillation, a form of. simple distillation,
an appreciable proportion of a liquid is converted to vapor and maintained in con-
tact with the residual liquid until equilibrium is reached.
Distillation of a pure substance yields no separation, ancl is of interest
only as a partial proof that a ~ample is a single, pure substance rather than
a mixture. For this reason discussions 011 distillation almost universally
deal with the behavior of mixtures.
The apparatus required for simple distillation consists of a pot, called a
4 See Thorpe, Dictionary of Applied Chemistry, 4th ed., Vol. I, Longm!1na, Green,
New York, 1937, pp. 176-178.
1. 'l'HEORY 3
still, retort, or kettle, in which the liquid to be distilled (the distill and) is
placed for heating, a condenser, and a receiver in ,,,hich the product, or dis-
tillate, is collected. The term still is also used to refer to complete assem-
blages of apparatus or equipment for distillation. The earliest condensers
were long tubes cooled by the air. The famous Liebig condenser (actually
devised by Weigel" before 1800), in ,vhich the condenser tube is cooled by
a countercurrent of cold water, came into common use about the middle of
the nineteenth century, and must have served in many of the discoveries of
that prolific period of organic ehemistry.
Early in the art, simple distillation was found ineffective in the separation
of liquids with close boiling points, and the first crude fractional distillations
were devised. As originally practiced, fractional distillation meant the
collection of the distillate in successivejractions or cuts, and \vas sometimes
applied to the systematic recombination and redistillation of those cuts.
Young," for example, described in 1922 such u. distillation of a benzene-
toluene mixture, in which the original t.en fractions Were recombined and
redistilled fourteen times, a process taking about. 80 hours, in order to
obtain appreciable proportions of pure benzene and toluene. A similar
procedure was used by Shepherd and Porter7 in 1923 in the analysis of
liquefied natural gas. As it is used today the term fractional distillation
represents the whole process whereby the fractions are made as distinct in
properties as possible. Especially included in this modern concept is the
use of various devices to cause a partial condensation of the vapor over the
still and return of the liquid condensate toward the still. In this sense,
fractional distillation was introduced when the Alexandrian chemists 8
(about the fourth century A. D.) interposed between the still and condenser
tube the first crude still head, thus foreshadowing the modern column or
tower. Into the improvement of these early still heads has been put most
of the effort of the subsequent centuries. The aim of fractional distillation
is the achievement of the closest possible contact between rising vapor and
descending liquid so that equilibrium will be approached and the distillate
contain the highest possible proportion of the more volatile component.
Such purification by the contact of countercurrent streams of liquid and
vapor is generally referred to as enrichment.
The term rectification is closely associated with fractional distillation and
had its origin in the distillation of spirits. Like any concept involved in an
6 See Science, 103,138 (1946).

G Young, Distillation P7intiples and p''I'ocesses. Maomillan, London, 1922, pp. 98-
105.
7 Shepherd and Porter, Ind. Eng. Chem., 15,1143 (1923).
8 Kopp, Beitrttge zur Geschichte der Chemie. Vieweg, Braunschweig, 1869, pp. 217-
259.
4 A. AND E. ROSE
art before it becomes a seiencc, the term is still used with various meanings.
In its general sense, it is almost a synonym for distillation, but in this
chapter rectification will be defined as the enrichment, i.e., the purification,
of the vapor in the head, towel', or column by contact with condensed
liquid returning tow!Lrd the still. Such liquid is known as reflux.
In the early still heads reflux was produced by the natural cooling effect
of the surrounding air, or in some cases by the application of water to the
outside of the still head. Here the reflux and vapor were in contact only on
the innor surface of the head. In modern apparatus intimate contact is
effected in an insulated or nearly adiabatic column or tower, supplied with
packing, plates, or some other device for achieving a large liquid-gas inter-
face. The reflux is produced i1,1 a condenser at the top of the column. The
term still head is now applied [Llso to the condenser and associated appara-
tus for taking off product and mC!Lsllring temperature at the top of a labora-
tory column.
The teml contact rectification is applied to all processes such as that described
above in which rising vapor and descending liquid are brought into intimate contact
and transfer of material takes place from one phase to the other because of the
fundamontal tendency to approach equilibrium compositions. This process is
usually carried out under adiabatic conditions. H.ecently apparatus has been con-
structed in which enrichment is secured in a column by repeated partial vaporization
of liquid or partial condensation of vapor, or both. This has been called thermal
rectification, and the extent of the composition change depends upon the amount of
heat added to the liquid or removed from the vapor. The process has advantages
in vacuum distillation and is discussed in detail in Chapter V.
In continuous distillation (usual in large-scale operations) the material to

be distilled (feed) is introduced continuously into the side of a tower and
product is continuously withdrawn from the heater section (boiler) or still
and from the condenser, and in some cases also from intermediate points
above and below the point at which -the feed entel's.o The portion of
such a column above the point of entry of feed is called the enriching sec-
tion, while the portion below the feed is called the stripping or exhausting
section. In batch distillation the entire sample of the material to be distilled
(the charge) is placed in the still before the distillation is begun and during
the distillation the produet is withdrawn only from the condenser. Both
partial and total condensers are used. In the former, product is removed
as vapor from the top of the condenser and another condenser is required
to liquefy the vapor. P!Lrtial condensers are sometimes called dephleg-
mator8 and the liquid eondensate is e~lled phlegm. In total condensers,
t Excellent scale models of continuous distillation equipment used in the alcohol and
petroleum industries are on exhibit at the Museum of Science and Industry, Chicago.
I. THEOHY 5
all the vapor is condensed and pl'Odud i:-; removed us liquid just I lclow
the condenser. ThiH rmnoval need Hot, he completp and the rem!Lining
liquid supplieH reflux. If l11l the condensate is l'ptllrnet! down the eol-
umn and no product is withdrawn, the column is said to Ill' operating uncleI'
total or infinite I'ejlu:r. SUdl u]lUratiuIl under total reflux is often used as a
preliminary in oreier to bring the (:olllnm as llenr equilibrium as posRihle be-
fore removing produet. If 1he condensate is divided. tlw column is said t(J
be operating unclcr florliol 01' .finit.e 1'1:/111:1'.
B. DJ..:lcIXl'I'IO;\:-; OF Y.\I([.\ llLI';S
Reflux Ratio, Holdup, Vapor Pressure, and Volatility. <hw of Lhe

major factors affeding the Hharpness of separation and the time and eost of
a distillation is t.he proportion of eondenBn.te returned as reflux. This
proportion is measured b~' the ((jlll:c ratio, whieh may be expressed in vari-
ous ways. In this ehapter it is tlefined as the ra,t.io of moles of reflux liquid
per unit time to moles of produet per uuit time. A high reflux mtio means
a large amollnt of reflnx liquid for a given amount of product, and in general
[L better separatioIl hnt it longer time for the distillation of a given charge
than operation with it low reflux ratio. For some purposes it is advanta-
geous to define reflux ratio as the ratio of descending reflux to rising vapor,
both being measured in tlw same units. It is important to be certain
which definition of reflux mtio is being uKed.
During nny time period, a certain proportion of the liquid introduced into
the appanl,tus is aet.ually in the column 0[' towel' as reflux and rising vapor.
This is known ~),s holdup, The proportion of holdup to the total mixture
beiug distilled is of consequence in bateh distillation since it limits the per-
eentage of the charge that. enn be distilled, and also has an effeet OIl the
f;harpness of separation.
In any discussion of separation by distillatioll, vapor pressure is 11 hasie
eoncept. Every liquid (wapomtes into the space around it until the pres-
sure of its vnpor reaches n ehanwterist.ie yalue called its vapor preSSI11'!'.
This is fixed for any liquid at [), givPH tempm'nturc, inereases with temper-
ature, and varies from one liquid to allother. It is llJl equilibrium value in
t.he sense that, if tho aetual pref;Slll'f' of the vapor is gl'caLer Ill' less than the
ehameteristie value, condensation of vapor or cvapomtion of Jil[llid will
oeeur until the equilibrium value is esbhliHhed. This telldeney for vapol'
and liquid to apprO!wh pqllilibl'illlll ('()II~(,itutes t.he aetive dl'ivinp; j'ol'l:e in (dl
(listiUntiull opel'lLtiullS.
A mixture of liquids will ill geneml give a lllix('d vapor t.he eompositioll of
witieh dl'p('ndB upon the values of the partial vapol' pressures of tlle in-
dividual eomponents. The hd.Ol·" c]cdenuining the:'io pal'tiltl vapl)r p1'es-
(i ,\ ..\ND I,. HOSE
slues n]"(~ ('omplex and, in the ease of distillation, are conveniently expressed
in terms of the volatility and relative volatility. A more detailed discus-
sion of the eOHeepts introduced here will be found in Section II. (Although
in practic:c mixtures of several substn,nces are frequently distilled, and
progress has 1>een made in the theory of distillation of liquids of three or
more components, the succeeding rliscus~i6~1 in this chapter will be confined
for convenience and simplicity to binary mixtures unless specifically stated
otherwise.) The volatility of a substance io; roughly proportional to its
vnpor pressure or inversely proportional to its boiling point. It must be
emphasized tIm!; the correspondence is only 11pproximate, and that, partic-
ularly in the case of liquids with boiling points nenr one another, the corre-
spondence mny disappear altogether, and in certain C'Lses the lower-boiling
component may have the smaller volatility. Volatility of substances in a
mixture is usually expressed as the relative valntility, which is assigned the
symbol alpha (a). For materials which form tl,n iden,l solution, this is equal
to the vapor pressure ?'atia, i.e., the ratio of the vapor pressures of the two
pure materials at the same temperature. This may be expressed as:
where PI und P2 nre the vapor pressures (itt. the same temperature) of the
pure substances whose separation is under consideration. 1'he larger
pressure is conventionally pltwed in the numerator. The more general
definitioll of relative volat.ilit.y in terms of the concentration in vnpor anrl
liquid phases is:
,,·here :"lh and Xz are the mole fractions of the two components in the liquid,
nnd Yl [md Y2 arc the corresponding mole fractions in the vapor.
Separating Power and Theoretical Plates. The degree of success in
separating a mixture by distillntion is closely related to the, inherent effi-
eiency or effectiveness of the column and pncking 01' plat.es. The terms
theoretical plate, H.E.T.P. (height equiv~Llent t.o a theoreticnl plnte), platH
efficient:y, H.T.U.ln (height of n transfer unit), and other concepts are used
in expressing this inherent sepamting power of the apparatus. The
H.E.T.P. standard mLS first introduced by Petersll and has been applied
chiefly to packed columns, although it may fLlso be used for plnte columns,
A theoretical plate may be defined itS one which produces the Sitme difference
in composition as f\xists lLt Hquilibl'ium between a liquid mixture a.nd its
vapor.
10 Chiitl>U and Collmrll, Ind. RnU. Chem., 27, 255, \lO.J. (1 \)35).
11 P(~teI'H, Ind. E'nfJ. CIWIIl" 14, 47(j (UJ22).
I. nmORY 7
The magnitude of the change in eompositioll fot' UHn th(~u]'()1i('al pbte

varies with tlw mixture nuder cOIlHidemt.ion, aUfI for ally one mixture it also
varies wit.h thr composition. This may be illus1.l'nted by reference 1.0 thc
following graph of the composition of liquid and vapor plotted again,,!; tIl('
boiling points of lL binary mixtul'P (Fig. 1). Such boiling-point elll'v(,~
V[1ry widely in shape for various mixtures, and for the S11me Inixtum the~'
are modified by pressure. They are most commonly const.ruct.ed for
ta - - -. - - - - - - - - - - - - - - - - - -
w
a::
:::J
~
a::
w
0..
:::;:
~
I-
'b
z
o
0..
(.!)
z
:::::i
o(IJ
b
PURE PURE
A COMPOSITION B
Fig. 1. Boiling point-composition diagram for binary mixtures.
atmospheric pressure. The lower curve represents the boiling points of the
liquid mixture 'with its varying compositions, while the upper curve repre-
sents the composition of the vapor in equilibrium with liquid of any specified
boiling point. For instance, a liquid mixture of composition a would boil at
ta and would be in equilibrium with vapor of composition b. This vapor
composition, b, should also be the composition of the first drop of distillate
from a simple distillation of the liquid wit.h composition a. Such a dis-
8 .\. AND .E. HOSE
t illation is Ilsl1:111y referred til a:; a lWIJ('(;1 .~il/tfllc liistillat£()Il. The term
1JCJject is llspli because sw·h a dis1 ilbtioll iri difficult to achieve in ordinary
apparn,tus, since some eondensatioll aud rectification genendly occur ou
the upper walls and lleck of the distillation fbsk. Specially designed
nquilibrium stills 12 are used to obtain the data for constructing curves suell
as those of the figure. The data nrc llsnnJly plotted as vapor compositioll
versus liquid composition but for the present cliscuHsion the boiling point-
(',omposition curves of Figure 1 are adequate. Refercncc to Figure 1 in-
dicates that. liquid of mmpositioll b would boil at th and would he in equi-
librium with vapor of composition c. A theoretical plate would he one that
produced the same change ill composition as a perfect simple distillation,
Le., from a to b, or from b to c, 01' l),ny other similar change in composition
sueh as d to ('., that is, the composition corresponding to the ends of a
stmight horizontal line betwecn the liquid and vapor eurvcs on tt diagram
of the t.ype shown. Since the liquid and vapor curves lLlways run together
at the two ends, it is obvious that for anyone mixture the composition
differences COlTCRpontiing to a theoretical plate will decrease toward zero as
the liquid eompositions approach those of the pure substances .. It is also
true that the eloser the relative volatility approaches unity, the eloser will
tho liquid and vapor cmves apprmwh one another, and the less will be the
composition difference cOJ'l'esponding to a theoretical plate. A packed
eolumn (or any other rectifying device) that will produce a separation
eorresponding to two consecutive steps or units, such as from a to c, is Raiel
to have the equivalent of two theoretical plates. If the height of such n
prwked column is 10 inches, the H.E.T.P. is 5 inches. Similar reasoning
applies to ally number of plates, ttnd any height of column. Columns haNe
been construet,ed with more than 100 theoretical plates. A given column
and packing may be expected to have about, the same H.E.T.P. wi{,h
IEfferent pairs of liquids if these arc of the i'\ame genoral type as to chemica,l
el1tss, ,'iseosity, nml stu'face tew;iotl. If these chamcteriRtics vary to [L
great extent, the film thickness, nrOt1 of g,Ls-liquid interface, and rates of
ditIllsion tLre presmn!tbly suffieiently changed so that the same eolunm and
packing may giye mm'kedly different lLKT.P. values. Cowiiderable prog-
ress l1a~ been made in expressing eoluIHn performance in terms of the 1'e-
RiRtance to transfer of mnterial fWl'OSS the film between the vapor fmel liquid
B Typical metlwd8 for dct{!l'lIliwttioll of VIlIl(JI'~liquid eqllilibl'ill, Wl soledad by D. F.
Botkin: (1) Differentiu'! distillation nll'thod-%awidzki, Z. physik. Chem., 35, 129 (1900).
(2) Stlttic method-Sage !Lnd Lltcny, 1'/'(1n~. ,1m. Ind. Mining illel.. Enar~., 136, l3G
(H1-10). (3) Mot.hod of I'l'eil'cnlntil)ll I)f cCm<\llIlRed VlIP0I'-Othmel', Ind. Rna. Chem., 35,
IH4 (1943). (4) Method of ei),IIUi:lt,jOll of \':1[101' phase--Joncs, Schoenbol'n, and Col-
burn, hul. Enrt. Chem., 35, (i(lU (1943). (iI) }[et-hod involving simultaneous l'ccircula-
l;ion of liquid phase and cOlHlenRHd-vltpol' Jlhll"e---GilJe~Jlio, Inri. Enf!. Cilmn., llnal. Ed.,
18, ii75 (194{j). See IIlso Fil!;l!r<' -1 of ChlLptf.'l· III.
(~l'd. IV). HmHn'(lI', tht) (:(lIU'npt ()f li1('(II'!'lil'al pbll' and H.E.T.P. h:l.:-'
lioen and still it' of I!:l'entcl' pral·til·a.l 11t.ilitr·
The term theoretical pl:Lle is hasE'll Oil Lhu uSbl1mpLiull thai in n plale
('olumn the liquid on any phtte should, under idml ('Ollliitiol1f;, give l1 VltpOl'
\\'hieh would differ in compm;itioll by the amount indjc:uted h~' HLP \'apor
liquid composition clif~gmUl, ltnd this VlLp0l' Wllltid eon<ieTlf;(' to form \lw
liq11id oIl the pJatp llbo\'c. Thus the 1'00npositioll diffpl'en(:e from 011<' p1:111'
1 () the next \\'ould C01'l'l'spond to that obtainwl by llS(~ of tlw diagram. In
af'tllal nppl'atioll thp ('ompoi'iitioll diff('J'PllceK tIre usually less than this, and
the t(:nll plate Ij!ici(,flC!! if' used to represent dw ratio of the admd COller'll-
I ration eil:mg(' 111 1hat l'xpe(:tpd from II theoretical 01' perfect plate.
Thl! nalu1'(' of the devices llsed tu bring liquid and vapor into contact if;
111c major faetol' dd,cl'mining H.E.T.P. and, in a given diameter column,
alHo cletE'rminpf; the 1l1axinl1un thronghput. 'l'hmughput, 01' vapor vtlodin
(II' boil",p mil', is the rate at which vapol' is passing up tlH'eolumn anr! it'
lls\Lally expressed in termK of the q1HlIltity of liquid equivalent to the vapor

passing up the column pcr nnit time. As long as there is to be How through
the colnmn there must he n. diffpl'(~lH:(, in pressure between the still pot and
the top of the column, and this mllst hI' sufficient to overcome the resistallce
between descending liquid and rising "apor as they pass through the inter-
stices of the packing or between pbtt's, Thi'l'c is n, maximum throughput
[It which the pressure drop (difference in preSSllre between still pot alld re-
('piYer) heeolUCS HO grOltt thttt the velocity of the rising Vllpor i:>suffieient to
prevent normal downward HO\\' of the reflux, Under snch eircumstnn(:es
(1XeeSS liquid fil':3t eoUE'ets in sings [mel then is fOl'eed upward. This is
t1sllally called .floodinrl aml is nut desirable during aetual distillation hut is
rE'g,111arly uscd to give increased efficiency by thoroughly wetting the pack-
ing for a short period prior to the start of a distillation in a packed column.
2, Relation of Distillation to Other Separation Processes-Fractionation

The term fmclionation is often llsE'd to mean fractional distillation or
I'pet.ificntion. Actnally, distillntion is one of a numher of separation proc-
('S'i\es by means of which fnwtionation of mixtures may be aehieved. 111
this bron.der sense fractionation implies n,ny proeeSR for the systematic.
separation of a mixture into relatively purc fractions. Recombination of
the similar fnwtions and repetition of the fundamental selmrat.ion process
m'e nsmdly also implied. The most widcly known example of fract.ionatioll
by a separation proceRS other than distillation is the fractional-el'ystnlliza-
tion proeedul'e. It has frequently been used in the preparation at' certain
nl,re-earth elements. n A more recent important example of fractionation
13 vYepks, ])i,~covcr!J oj' the Elements. 4th Ed., .Jou1'lln,1 of Chemical Educat.ion, 1939,
Chap. 23.
10
is tho flcparation of iRoLopi(~. UJ'alliurn HlloridoR b.v diffllsion barriers.14 In

thiR r.H.RO a vory ingcniollf; syst.em wa:- devised for reeomhining cel't~1in frac-
tiom; and repef1ting; tlw fnndlLmenj,al separntion with fl, minimum of han-
dling. Systematic fractional precipitation of high-polymer solutions is of
current interest as an aid in determining the distribution of molecular sizes.
The washing of impmitics from a solid is a similar much used separation
prOCCHS, [tnd extraction from one liquid to !Ulother has been extensively
discussed and used as a means of separation and fractiollfLtion.
Analogous proeedures involving numerous repetitions of simple distilla-
tion and recombination of clistilhtes are described in the older literature of
distillation. In IlLet, some dcseriptions of "mell fmctional separation are
of such late elate as to empha::;ize how recent. is the development of ef-
ficient laboratory redifien,tion equipment. Distillation through a col-
umn is similar in pl'ineiple t.o pnrifiention of [I, liquid mixture by counter-
current action of another suitn,bly chosen liquid (continuous solvent ex-
traction). A closely related ense is purification of a. gaseous mixture by
countercurrent action of liquid solvent (ordinary absorption-tower action).
These laRt three types of sepamtion operations arc unique in that a multi-
plicity of stages of repetition can be achieved by countercurrent gravity
flow of liquid and 1'1HC of vapor, so that complex handling problems are
entirely avoided. The effeet of morc than a hundred repetitions of the
fundamental sepamtion process has been achieved in a single packed
column.
The principles and theory are essentinlly the same for all the different
kinds of separation processes,15 and more than one process is usually
applicable for anyone separation problem. If the relative volatility is near
unity, consideration should be given to separation by modified distillation
such as extractive or ttzeotl'opie distillation, deseribed in Chapter III, 01' by
methods other than distillation. However, distillation is frequently the
best choice because of the simplicity of the apparatus and procedure.
Absolute purity is not possible by any distillation process, but for all
except u,zeotropie mi.'{tures (Sect. II) such purity can be closely approached
if sufficiently complex [Lud elabora,te processes and appamtus are used.
3. Distillation Curves
Like the apparatus; the theory of distillation has also developed largely
around ol'ganic-chemical applications. The large-scale production of alco-
hols and of coal tar and petroleum derivatives stimulated extensive work on
U Smyth, Atomic Energy f01' llfilital'Y Purposes. Princeton Univ. Press, Princeton.
1945, Chapter 10. .
15 Brncdiet, Trans. Am. Inst. Chem. Engr8., 43, M (1947).
1. THIDORY 11
the theory of continuous distilltLtion. Since this has been discussed ex.-
haustively ill many places,16 this book concerns itself principally \vith ba tcll
distillation and specialized procedures, aml tho section "on theory is written
largely from the point of view of laboratory batch distillation and rectifica-
tion.
The most common method for expressing the results of [L bateh distilla-
tion is by means of H, distillation curve in which the composition, boiling
point, or some other property of the distillate is plotted against the per
cent of the eharge distilled (Fig. 2). A perfect separation would result in
curves with right-angled breaks. Such a situation is impmlsible theoreti-
cally, but is approached in the case of easily separable mixtures or suffi-
0
z
0 0.2
i=
iii
0 0.4
o._
:::;:
0
u 0.6
I-
u
:::J
0 0.8
0
n:
o._
1.0
0 20 40 60
PER CENT DISTILLED
Fig. 2. Batch-fl'uctioll!1tiull eUl'ves.
ciently effective apparatus. The actual sharpness of the break in the curve
gives an approximation of the sharpness of separation att.ained in anyone
caHe, since it is directly related to the purity of the fmctions collected. The
greater portion of the theory section of this book will be a discusi-lion of the
various factors affecting the sharpness of separation as shown by dist.illation
curves. The most important of these factors in fractional dist.illation are:
(1) the difference in composition thnt cnn exist between liquid a,nd vapor at
equilibrium (vapor-liquid equilibrium relation or relative volatility); (2)
the effectiveness of the contact of licnlid and vapor, usually expressed l1fi
plate efficieney or height equivalent to fl theoretical plate; and (8) the
proportion of eondensate directed back into the fraetionating column (refiux
ratio). Although they are interrelated, these factors refer, respectively, to
the mixture under separation, the apparatus, and the method of operation.
I. \Valker, LllWil-l, MeAuams, awl Gilliland, Pl'incipie.' of Clwmiml Enyi-neel'inrf, 31'l1
ed., l\IIeGl'uw-Hill, New York, 1037. Hobill8(JII I1UU Gilliland, Elements of FmctiOllal
iJistillation, 31'd e(L, MeGl'aw-Hill, New York, 1\)39. HlLu~bnmd aut! Tripp, j'rincipieN
and Practice of Industrial Distillation, tI'l1IlHh~L(l11 frnm4Lh (lel'ilU~tl ell., Wiley, :'Ii ml' Y"l'k,
1926. ScetiOll 25 in Science of Pet1'(){enm, Vol. II, Oxford Ulliv. Press, London, 1938.
12 .\ •.\1\1> E. IW,.;!<:
Additional fado!'::! are holdup, thl'oughpilt, aud the amount of excess or net
eondensation or vaP9rizatioll OCCIi rring in the column itself, if it is not prop-
erly immlltted from' its surroundings. Obviously simple distillation cloes
not involve ~Lll these fuctore, and still heads present [tIl intermediate case.
Variation ill the operating pJ'O(~ecl\ll'e ill fmci~iollHtion nllows considerable
control of t,he results. The ehoiee of reflux mtio tDgether with the vapor
velocity rleterlllines the time required i'm' 11 bitt(~h distillation. These
160
140
120
...... _-
~
"'- 100
III
UJ
~a. eo
60
40
20 40 60 80 100 120 140 160 ISO 200

REFLUX RATIO (RD = LID)
lrig, 3. l{!LlIge of calGuluted plat.es lind rdlux ratio.; l"<'quil'ed fol' ~ta.ndal'(l
sep:trutioll, Values of k (L~ on page ITa
vtLl'iables also playa pltrt in determining the holdup and the number of
theoretical plates. Choice of apparatus and packing determines the
minimum and maximum values of the theoreti(~al plll,tes) the vapor velocity
and the holdup, There is even some cualm! over the rela.tive volatility,
since this lllELy be altered by changing Lhe pressure of the distillation
(vacuum distillation) or by addition rtt' a nel" suhstLtnct' to 'the mixture
(azeotropic 01' extractive distillation).
There is a voluminous litcl'l1tuI'C OIl plwkillg anll other contacting devices
for fractiollHting eolumns, in eluding pal'ticubrly the determination of
H.E.T,P. values Itt total reflux. The present discussion will place most
emphasis on the ul:!e of distiIlntioll eW'ves tu represent the end result of
fractionation, and OIl ,the effed of relative voh1tility, reflux ratio, number of
theoretieal piatt'S, nnt! othE'l' radon; lJll nlt~ ~h!tI'Jlll(,KK of the breaks in these
(:Ul'ves.
I. 'J'ln;OHY 1'",','
;\ pl)<lr :';('pal'aLiml iH l"f~l)I'(,~plll<'d !JY ('111'\"('.1 of Fi).!;IIl'!'~' 1';\'('11 IllI' \"f'r."

fin'll; of tIl<' distillate ii-: lit1](, /11'11('1' Ihall nO(,';; )J1l1'I', ;Jnd Ih(' ('III ire firsl half
lwcmges litt k hrttf'l' than 80% jllI]"('. Tlw:-;(' ]1f'r('!!tlIn.).!;ps 1'1'\'('1' i(l Ilw 10\\'81'-
hoiling (!omp()l1fmL The separat ion Ill' tlw higher-hoiling ('mnponent is
even \\'orRe, Rim,e only the lnsf, t.hird (If t Iw dist·illatr, COlli nin,; mnr(' than 90S>(,
of tlml r.:ompoJlrnt, Snch a POtl]' separation O(,(,IIl'S when materinls are of
llf'arl~' the "anw boiling point. Ill' \'obtility, and "'1\('11 eithel' UJ' both the re-
Jinx mtio and the' P(luiYltipnt. llllluber of 1.lw()reti('nl platPH are tuo low.
C'1ll'VC R iudi(,a1.es it reaH(Jfllddy good sPJJtl.ml.ioll.
c\n approximate theol'etiral rPiatiol1Hhip h(,tWl1Cll the c;lml'jm(l::;s of sepal'U-
hOll a,1Hl thl' l'(~htti\'e \,ola,t,i\it,y, tile reflllx nttio, ,\,Uri the cqui,'alcllt lltl111-
hpr of t IlPorpti(':d pl:tteH i" gi VPIl in Fip;lIl'e :3. This is bas(~cl on the value;
Ilf t1wsp fad Il]';-; e:d(·uJat(·d n,,; 1l('('('ssaI'Y to oht.aiu I!. good separation of a
binary mixhm' as indicai('d by the di,;t.ilhttiol1 ('·Iln·f' n of Fig\lm 2. The

1'C'lu,tiollships 1'01' Fi)!;11l'e :{ 1\'C'n' Ilht.aiIH'll hy plotting value,; for t.he reflux
ratio, R D , and t lw llllmh(!l' of UI<'ol'dieaI plnt"s, n, caleulitted to ubtaiu the
desil'0d se.pfLmtioTl (8('(,1. YIR). The ;';!lm(' approximate l'ela.tion is gi.ven
h~':
2.8fi
n = Hil = - - -
log IX
where I'll and 1',4 are the absolllto boiling point.s. These relations indic!Lte
that the reflux ratio should be llumerically about equal to the number of
theoretical plat.es. For Hllhstances boiling near lOOoe. the expression be-
comeS:
250
Rn = n = --"----
TB - 1'. 1
It is, however, not neeessnry to specify exaetly the number of theoretical
plates required in order to achieve the desired separtLtion. A.ny value Of'11
within the range indicated by a solid eurved line on the figure should be
suitable. Within this range the lower values of n may he compensated for
by a corresponding increase in RD , and higher values of n allow use of a
smaller RD , It is clear that it is essentiul that n remain approximl1tel~'
within the desired range since vnlues that are too low cannot he compen-
sated for by any increase in RD1 no mnt,t.er how great, and values that an'
too high do not allow a corresponding decrease in RD' Discrepancies of
eonsiderable magnitude must be expected in the pra,ctical use of these rela-
tions because of the many assumptions involved in the derivations, The
chief of these are the assumptions of constant relative volatility and of
negligible holdup. The limitations, derivations, and extension of these
theoretical relations are discussed ill detail in subsequent seet-ions, Thc
1:1 A. AND E. HOSE
theory of lmteh clistillaJioll has not yet progressed to the stage where useflll
predictions e:m he made rrglll[Lrl~" ;Lnd with confidenee.
4. Conclusion
It is essential to emphasize that the theory of batch distillatioIl is a Ilew

mld umlevcloped field of knowledge. There nre large areas that, are almosl,
unexplored, as, for in::;lltll('p, the f<Ldors tlmt determine the rn,te of transfer
between liquid <\lHi vapor. The whole field is largely unchecked by ex-
periment, and a gl'el1t deal of work [LIang this line remains to be done. The
theory is ehiefly useful as it general f2,11ide rather than as a means of getting
aecumt.e numerical answer:-; to speeifie detailed problems. The number of
vltTil1bles is so great and their interrelations so complex that the theory,
though imperfeet and only partly developed, is essential in order to reduce
ne,cessary confirming experiments to a reasonable number.
The kind of informl:ttio,ll given by the theory may be likened to the view
of a large unexplored region obtained from l:t high mountain top through im-
perfect lenses. One sees only the m!Lin fet1tures--a broad river, an open
valley, a high plni cau l or a forbidding peak and ridge with passes suggested
here Hnu there. It is inspiring, fascinating, and profitable to speculate on
details, but it must be recognized t.hat exact and useful knowledge will come
only from the hard work of mile-by-mile exploration and development,
with many clisltppointments anclnumet'otls difficulties and changes. The
geneml methods of material balances [Lnd calculus [Lre akin to the use of
field glasses and telescope and airplane photography. Painstaking ex-
periment, like the careful work with compass, measuring tape, and plumb
bob, must follow t.he more general approach. Those who venture into or
are by necessity forced int.o snch a region must proceed with EL keen aware-
ness of their situation. They may expeet to discover easy short cuts and
fertile areas for usefulness and achievement; they may also become in-
volved in eonfusing morasses, great barrens, or impt1ss~Lble canyons, In
any case it may confidently be expected that the main features of the
original view will continue to remain as unchanged and reliable landmarks,
and that the details will gradually build themselves into a continuous and
co-ordinated whole.
n. VAPOU-LIQUID COMPOSITION RELATIONS
All separations by distillation depend upon the difference in composition

betweeli a liquid mixture and its vapor. The composition of the vapor is
determined by the actual vapor pressure (partial pressure) of each of the
components under the conditions of the mixture. At any temperature,
I. THEORY
each pure liquid substance has its characteristic equilibrium vapor pressure,
or saturation pressure, which it attains with sufficient time for diffusion and
opportunity for contact of liquid and vapor. In a mixture at equilibrium,
the partial pressure of anyone component depends on the characteristic
vapor pressure of the pure :mbstance, on the composition of the mixture,
and on the effeet of the other components in the mixture.
In actual distillation equilibrium is not attained, since time for diffusion
and opportunity for perfect contact between the countercurrent streams of
liquid and vapor are seldom ndequH,te. Theoreticml analysis ought there-
fore to consider the time, rates, and surface areas involved in the diffusion of
the components from liquid to vapor and vice versa. Such analyses are
complex and still in proeess of development. 17 Despite their promise of
defining relations existing at lLetual operating conditions, the theory of dis-
tillation CtLllllot discard the study of vapor-liquid compositions at equi-
librium, on which all theoretical work is based. For this renson the present
section deals with the vapur pressure of liquid mixtures, HIld the difference
in composition between liquid mixtures and their vapors.
Classes of Liquid Mixtures. It is advantageous to divide !iquirl mix-
tures into several distinct classes for the purposes of this discuflsion.
These are:
L Binary mixtmes.
a. Completely immiscible.
b. Miscible: ideal mixtures and nonidealmixturo:;,
e. Partly miscible,
2. Multicomponent mixtures.
1. Binary Systems
A. COMPLETELY IMMISClBLg
The simplest mnong the above elasses is represented by tL mixture of two

liquids which are entirely insoluble or immiscible in one another and form
two separate liquid layers or phases. In such a case the composition of the
vapor in equilibrium with the liquid mixture is nnaffected by the relative
quantity of the two liquids, but the temperature of the mixture is all im-
portant. Each liquid vaporizes as if the other were not present, and the
composition of the vapor is dependent only on the vapor pressures of the
two pure components. In phase-rule terms, there are two components,
three phases (two liquid, one vapor), and the number of degrees of freedom
11 Chilton and Colbul'll, Ind. Eng. Chern" 27, 255, 904 (1935). Cohen, J, Chern. Pltys. ,
8,588 (11)40). Kulm and Ryffel, Helv. Chim. Acta, 26, 1693 (1\)43); Kuhn and Baert-
schi, ibid., 29, 692 (1946); etc. Westhaver, Ind. Enl/. ChellL, 34, 126 (1942). See also
Section IV.
Hi .\ ..\XD K HO:;E
mllst then 1)(, (}lW. .-\1 allY ow' di:<lillill!( lI'mpt'I'alul'C', Ul(' total pl'eBl:lLll'C or
the system is fixed.
Dalton's Law of Partial Prl:ssw·cs. TIlt' jJlLrtial jJresslll't~S of t.he intlividwd
components in n ,';1,])01' mixture arc related to lhe tutal prcBslirc by Dalton's
law of partial pressures, whi(~h stajps that the Sllm of lhe individutd partial
pressures in H mixtnrp of gLlS0ii Ill' va])ors is ('(tual to the tobl pressure and
that each individllHI gas exerts a pJ'f'~~lIre as if it \\'ere present alone in the
volume occupied b~' tlw mixture. Ex('('pl for disC'J'Ppallcies dne to devi-
:ttiuns from the perfect gas laws, tilt'S!' reitLtions may be cxpressed by the
mathematical fonl1111ml:
lllHl, from the ftllldllITlPlllal gas law,;:
Ji = NH'l'/Y
for the entire mixturc, and:
foJ' a single ['.umponent. Thus tlH~ partial pressure of any component in

the vapor mixture divided by l.hE' total pl'eS8tll'e of the mixture will approx-
imate the mole fnL(~tioll of that (:ompllnent in the vapor:
This simple reln,tioll, pdp = VI, will form the basis for predicting V~ljl()r and
distillate c.ompositions ill must of thc discllssions to full ow. Its Ili'le nH\~' he
illustrated hy estimating the composition of tlw distillate from the tlistilh-
lion of a mixture of benzltldehyde nnd water.
Calculation of Yapor Composition fro/l~ "lIPOI' Pressure. At a temperature of
(l8°e. liquid water is in eqUilibrium with its yapor when the latter has a pressure of
iOi millimeters. The corresponding vapor pJ'!'ssure for benzaldehyde is 56 milli-
meters. Since each liquid is imsoluble ill the other, it thoroughly agitnted mixture of
liquid water and benzaldehyde at 98 0 will be ill equilibrium with a vapor mixture or
'Yater and benzaldehyde at pressuroK uf i07 and 50 !Hill., respectively, Prom the
above formula it follows that the mole fraction of benzaldehyde in the vapur mixturc)
is 55/763 = 0,018, and of water iOi /763 = 0.927. The relative weights of the two
vapors in unit volume of the mixture are obtained by multiplying each partial
pressure by the correspondillg rnolecu1a.r weight eM):
Il'l IhJfl f)G X lOLl 59:3G 1

or about ,_"-_
WZ l~~X~ = 7()7X18 12,'i26 2.15
l. '!'Il]i:OHY 17
Steam iJi8l'ilialioll. Clldi8! iila! ilill. If j he' alnluHpltC'l'i(, Pl'",,:,\IH'P IS

slightly lo\\,pr than the total (7GH mill.) ur tilt' ahlJ\'C' vap()r pl't'SSlIl'eS, the
lirlUicl hf'llZaldehyd('-wal ('I' mi:d 1\ \'(' \\'ill huil ,1wa~' as hHlf!; as I'he tf'mpl'l'-
attul' of 98° is m,lilllaiIlPd, Hlltil 011(' Ill' IlH~ olht~l' l:omprllWlll i~ pxh:tIlsted.
(The llorm:d l)(Jiling poillt of purt' lJ('llzalddl~Td(' is 17~)'[iClC'.) _\ distilbtioll
may thus he e(tl'l'il~d I)ul h~' passing till' \',Lj)Ul'S j,hl'ollgh a ;;11 i ta.bly n, l'\'tlng;ed
l'tllllleHsC'1' thai I\'ill l'ulltll,n:,\(' buth li([\\i<ls. The distillntn \\'ill wntain the
t \\'0 liqllit!:.; ill Illl' :"alll!' proportioJls as in the vapol', eXt:ept for Stich ,;mall
fl'neliollS a~ an' 1l()lliqll!,fil,d hy tk~ l·IJlllleIlS(]l'. In aetlla.1 pmetiee sHch an
uperatioll i" l\,,\\all~' l,;tl'l'ic,d old :lS a stelull didillalioll ill wltieh steam is
Imhhkd tlll'tJ\l~1l til!, lllatl'l'iai j() bt' di"till(:~ll. 'fbi" (.~']Je of opul'lLtiol1 i:'i 11
t'imp\c' ml'ant' of IJ1ll'ifyillg All ]J"l alleec: t.hal. aJ'(\ high builing 0]' Illl:-;tahk at
their boi lillg; puillj" alH! iUlIllic;l'i I)II~ nIH! lI11n~;t('j,i ve wi UI W;Lj,('I·. ~ln,L1()gollA
plll'ilil'u(iulls by eml istillntiml l ' lllay I)(~ C:,lITiu( 1 on t h.v I1Al' uf 01 ht'!' suihLilly
uho8en irnmi::wibll' liqlli(b lJJ' illL'l't g;as·us. Ii; i" 1,0 Ill' IllJt()d that, iW(:l1l1se tho
moleenlar Iypight or mtiPl' (18) is low (~()!llp:U'etl with 11mt of henzrLldeh;\'dn
(l06), the mLiu h~' \\'c'igitt nl' h('llz,dtl(~hyc1e 'to W:t,tet· in t.he Pl'Odllct if) much
gl'e;Lter (1 to 2.1fi I t hall [;11(' I'al.io of the \'UpOl' prCI':ISlll'CK 01' mole l'mctiollS (1
to 12.7). TIlllK it is ,lllv;ml ageO\1S in :m)r pmifiea(.ioll uy c'oc]istiibtiuH to
have the ,Hlded iIlllniKl'ible l:Ompolllm[. of relatively low molecular weight,
sinee this euts dOlm the tulu! material whieh must be distillcfl in Ordf'l' to
o])tn.in the puw clesil'i'([ ("'omponellt. 'When the added eomponent is a gas,
the l'eeeivel' must he kept. c:old enough to reduce the vapor pressure of the
desired material to :t negligible value, 01' the inert gas ,,-ill (",ltrry off ,lIl
appreciable proportion. Ileeircubtion of the gas may be ])rn(~ti(lPd t.o
;woid this difficulty.
]from the \'npor pressures recorded in the literaturel~ it is USllilll? pOf;sible
to pl'ediet appl'oximn,te boiling points and distilla,te eO!'.1positions of mix-
tures of 1,\\'0 completely immiscible substances by tL nwthod illust.rat.ed by
the benzaltlehyde-water calculations. It:8 to ue p:nphn::-;ized that the
18 i\Iorton, Lnboratol'Y Technique 'in Organic Chemi~lr!', l.Vh·CnLw-Hill, Nl'w York,
W3S, pp. H3-144. RI1~sOW I1nd Schultzky, Z. angelO. Chem" 44, 66!) (1931). Othnwr,
Ind. liJng. Clwln., 33, 1106 (1\)41). Schoenburn, Koffolt, and Wit.hl'Ow, T)'ans. fllll.
[nst. Chem. EnOI'8., 37, 997 (HH1). Garber find Lerllll111, ibid., 39, 1l;3 (19·13). "ClIr-
rier-Dist.ilIatioll Analysis of Url1uium," Chem.liJn(f. Ne.ws, 25, 777 (l\hrcI117, 1\)47).
10 International Critical 'l'able,9, McGmw-IIill, N. Y., 192u+. Beilstnill, Handbw:h
del' organischen Chemic, Sprillgnl', Berlin, 1918+. More recent and detailed arc:" G. G.
Drown, Katz, Obm'fell, and Aldell, Natw'al Oasoline and the Volatile Hydrocarbons,
Nat]. Gasoline AssolJ., of America, TulslI, 19,18, 92 pp. Egloff, Physical Conslant~ of
Hydrocarbons, Reinhold, New York, H)3!l. Ktttz lind Rzasa, Bibl£ograpky fOI" PkyS£cal
Bd!aldor of Hydroca1'lJOns U1uler Pr(%~ItTe and llelatetll'lwnomena, Edwards, Ann ArbOl',
1946. Nationl11 BUI'(l1Lll of St,andal'd~, Circular C461, "Selected V1LIllCH of Properties of
Hydrocarbons," November, 1947. Stull, Ind, IEng. Cllem., 39, 517, Hi84 (cnrmctions)
(1 \)47); vapor pl'llssures of more thl11l 1200 organic compounds, 300 inorganic compounds.
18 A. AND E. ROBI,
aecuracy of all sneh calculations depends upon the assumption of sufficient

Lime and intimate contaet so that equilibrium vnpoi' pressures are approx-
imatelyattained. Calculations of steam consumption in steam distillation
are important in industrial applications.~o
B. MISCIBLE
Ideal Mixtures. Raoult's Law. With liquids that are completely

miscible in one another, the V[LP01' pressure of each component depends
upon the c~mposition of the liquid as well as the tempemture. Each com-
ponent affects thc V,lPOl' pressure of the other. There are two components
and two phases, one liquid and one vapo)" so that the number of degrees of'
freedom is two. If the distilling; ternperaLme and the eomposition are fixed
the total pressure is fixed. In the simple:>t case, the ratio of the vapor pres-
SU1'e of a 8ub8tance in such [t mixture to its vapor pressure in the pure state
is given by t.he mole fraction of the :mbsbnce in the liquiu mixture. For
instance, the VlLpO!, pres~mre of pure benzene at 100°C. is 1344 millimeters.
In a liquid mixture of 1 mole benzene and 9 moles toluene :1t the sttme tem-
perature, the vapor pressure. of benzene is 134 millimeters. Similarly the
vapor pressure of toluene in this mixture is 504 mm., about 0.9 the value for
pure toluene at this tempemture (557 mm.). Mixtures in whieh this simple
relation applies are Cltlled -ideal and the rehttion is known as Raonlt's law.
It may be expressed mathematically as:
fit = PIX!
in which PI is the partial vapor pressure of component 1, PI is the vapor

pressure of the pure substance, and Xl is its mole fraction in the liquid.
Ideal mixtures are usually made up of' two or more closely similar com-
pounds, and the law applies most closely to the component with mole
fraction near unity. It also often applies to one component in a mixture
but not to the others.
Henry's Lml). In a limited number of instances the partial pressure ({h)
of a particular component of a mixture can be related to the liquid com-
position by the expression:
ilL = kXI
in which Xl is the mole fraction of the component in the liquid phase and k
is the Henry's law constant for the mixture. This law usually applies for a
component present in low concentration, but applies only rarely at appreci-
a.bIe concentrations. It is obvious that Raoult's law is a special case of
.20 Perry, Chemical Enginee1's' Handbook, 2nd ed., McGraw-Hill, New York, Hl4L, Vp.
1388-1391. Nord! Ind, Enfj. Ch~m,! 371 854 (1945); nomograph.
1. 'I'H80HY l!1
Henry'!,; bl\' in whieh the C~()m;tlLnt k il" rcphwed by the value fur thr' \':1,P01'
pressure of the pure eomponenf. The t.wo hws apply hnst to oppoRite ex-
tremes of c:oneontmtion-Henry\; li1,w for eomponputK luwing mole frae-
tions near zero and Rnoult's law for c:omponents lutving mole fractions near
nnity.
Vapor Press UTeS of Pure S-ubstances at Various Temperatures. Duhrina'8
Rule and Co;r Charts. \Vhile there are some da.ta in the litemtmc OIl the
\"apor pressures of sllbstt111ees nt a, Vt1riety of tempera.tures,t9,2! such in-
formation is still very incomplete, and variolls procedures have been evolved
for making the best use of the available data, and for expressing them C011-
eisely. The 1msic equation for the relation of vapor pressure to temper-
ature is the Clausius-Clapeyron equation:
in which dp/d'1' is the mte of change of vapor pressure with absolute tem-
perature, 1', A is the molar latent hent of vaporization, and Li V it; thB differ-
ence in volume between a mole of liquid and a mole of vapor. If the volume
of the liquid is assumed to be negligible, and the vapor is assumed to be
ideal, so that V = RT/p, and "- is constant with respect to temperature, the
:I,Love equat.ion becomes:
dlnp "-
and In 1) = -
"-
-- +I
. dT = li:1'2 RT
This is the basis for the empirical form in whieh vapor pl'essurefl are fre-
quently recordedin the literature:
log p = (A/T) +B
A graphic equivalent was prepared by Copson and Fl'olieh 22 for the 10\vo1'
hydrocarbons by plottiug log p against (-l/T). The curves ohtained
were pra,ctically straight lines.
Duht/:ng's rule is expressed by the relation:
tIL = kitn + k2
where tIL and tn are boiling points of two different liquids at 11 partieular
pressure and kl and k2 arc constants. Thus values of tA for various pres-
sures give a straight line when plotted !1gainst the corresponding boiling
points of tB' This is equivalent to plotting against one another the tem-
peratures at which a reference substance and another substance have the
21 Miles, Ind, Eng. Ghent., 35, 1052 (1943). Nomograph.
22 Copson and Fl'oJich, Ind. Eng. Ghent., 21, 1116 (1929),
20
:-;nnw vapor prPBHIIJ'O. Olwe the rde)'enm' line· I'm sllbHLmwo A iH (~Htal)
lislwd, only two points arCl needed t.o ddprl\lill(~ Llw posit.ioll of Lhe n lillP
if the l'luhst.anees A. [mel H ILI'n Kimilal' dWlllinnlly,
Another method of expJ'(1KKing vapOl'-pl'(1SKllJ'(' da.l.a iK i.Iw COJ: dUM,t, ~:I
This also gives straight linCK, and oftl'n 1.he entin1 vapOl'-j)rnK;,\ll1'O (:\n'V(~ Intty
be obtained from a single oxperiment.a.l value. The Cox nlml'L 111l1kos 118(1
of a single refel'cnue substanee, such as wH,Lel' or monntry. The vnpot'
pressure-temperature dntn for the reJ:'enmeo sllbstmwo are plotted as Ii
8traight line with slope of nhout 4!'l", This may he ncemnplislwcl by l1Hinga
logarithmic sealo of Pl'CKSlll'e aR oJ'(lilll.te, and m:Lrking tho tLbRuiRK!L with t.he
proper temperature::>, The YB,POl' Ill'esRlll'e-i;0mpnmjw'(\ dat.a for nt.hot'
substanees wm also fOl'm DI~I1J'ly straight liIws wlwll plot,ted ill this WILY, and
groups of related organic compollnds give lilw:i tlmt .i nin in a e0111111011
point. Thus the normal boiling point of a hytll'oc:m\)()u, aleohnl, ete, will
often i:mffico to ostrdllbh the (~()rnplde VttpOI' Pl'()SKlll'(l Lmnpem,!;uJ'o (ml'V(~,
othmer2·! has presented vnIllahle (~()\'l'Dlnti()JIK of vapo!' prORSlll'C with latent
beat and othcl' daLa,
Volatility and Relative Volatility. 'rho terlll voln(.iWll is eomlllonly UHPc]
in t~ brond sonse to refer to the CiLSO OJ' diffieu!l;y (If evaporation or vaporizn-
tion of substmwos. For use in distillation it iR dm;il'able to cIofine the vola-
tility of any substanc.e in lL liquid mixtlll'e lLS 1';K ptu't.ial vnpor PI'(lRRUl'e
divided by its mole fmetioll in the liquid:
VI = jJl/Xl
For a pure liquid, Xl = 1 and VI = Ph that is, the volatility ir-lidentieal wit.h
the vapor pressure, For a subst.ance in an ideal mixture, the volatility is
still equal to the vapor pressure of the pure materittl, sinee the equation
may be rewritten as])I = VIXI and dil'ocL comparison 1'.0 Ru()ult's luw (pI =
P1Xl) indicates that VI must equal PI, Fa)' nonideal mixtures the volatility
will usually vary with compos~tion and be either nbnormtLlly high or low
depending on whether the actual partial pressure of tl eomponent is greater'
or less than the partial pressure calculated by Raoult's law, The volatility
will then be related to the vapor pressure of the pure mat,erial through the
activi(,y coeffic.ient:
where 'YL is defined as PyI!PIXl.

Relaii1!c vola~ilitY:i usually representecl by the symbol ai, is simply one
~3 CO'x,lnd, Eng~Chem" 15, 5H2 (W23), CfIolinga(ll'L Itnrl Davis, 'ibid:, 17, 1~87 (1925), .
Walas,JJhem, &: Met. Eng., 53, N~); 10,124 (1946); nomograph.
24 Othmel', In(l.1!lng, Chern" 34; 1072 (1942),
J. 'I'IllWH.Y 21
yo\:ttility v~tl\\e divided by ttnot,ltel'. The htl'glll' vobWity iH cOllvontioually

pl:wed in the numer:tLor, RO th:tL Lhe mtio iA gl'oater than onr:
The relatioil of the relative volatility Co vapo!' aud liquid l'omposition may
be derived by noting t·hal:
Vj pJ/:r'j I)j '£2
= =
V2 Pz/:t2 pz :r[
But. if Dn,lt;on's htw applies, fit! i)2 = Yl/Y2, Hince .1}1 and 'Y2 m'e the mole fnw"
tions of the two eomponents in t.he vapor phttHI\ so Lhat:
or, for a bin[\,r~r mixture in whieh :02 = 1 - :!!I and 112 = 1- YI:
(t = y
,,_,,-'-', 1 - ;t U/Q, - y)
1 - lJ :1: x/(l - :1')
This relation is knolVn as the relative-volatility eqttation and forms the basis
for much of the current analysis of separation by distillation. It is con-
ventional for x and y t.o refer t.o the more volatile component, unless ot.her-
wise stn.tnd. The equfLtion is often written:
y X
IX - - -
1 - 11 I - :t:
and used in the forms:
aX y
Y = -- ---- and :r = .---.:::.---
I + x(a - 1) a - y(a ~ 1)
When Rao1l1t!s law applies to the mixture, Vj = PI and 112 = P2, so t.hat:
~ = Pt/P2
Under these eircUIXlstances the relative volatility is identical with the vapOl'-
pressure ratfo. Griswold 2fi ,has offered an equation for obtaij1ing the, rela"
I;ive vola Wity of hydrocarbons from vapor preSSlll'e-teinperfLtUl'e va-hiteR.
)';x[Lmination of the formula: ... .
• _,'iJ,
'1'.·....:.. V=
26 Griswold, Ind,. ,Eng. Chrm., ~5j82G (l,948).
22 A. AND E. ROSE
indicntes tbtLt, when rela.tive volttWity cqunls unity, no sepamtion is pos-

sible, and that, when the vttlne is near unity, very little difference in eOJll-
position will exist between liquid and vl1por thnt are in n(lldlilll'ium. 'I'he
relative volatility v!:\,ries gre<Lt.ly for different mixtures, tt often vll1:i(,s
widely for different concentrations of [-,he same mixture. The expreRRiofl:
-Y- = 1)1 X
~-. ~~."-
1-11 P2 1 - x
is stl'ietly true only when Raoult's law npplics, mHl the ideal gus htwK hold
for the vapor mixture, and then only nt conKttmt tempernture. MOi:\(, dis-
tillations are at constant preSSlIl'e, but over n range of temperatures. The
values of Pl and 712 vary -with temperature aecording to the Clnusius-
Clapeyron equation:
so that PI/P2 is constant over H, rnnge of temperature only when the com-
pounds concerned have equal molnr heats of v!1pol'izH,tion,}\. Only adja-
eent compounds in the same homologous serioR, or compounds with iden-
tical critical constants would be expected to meet all the above require-
ments and behave as described by the equations with a constant relative-
volatility value. However, for mixtures of si:m,il~r liquids the variatioIl
with concentration is often not excessive, and use of ~m average vulue of the
relative volatility is common practice when the variation is noL over 10%.
Much distillation is based on the assumption of a constaIlV l1elative vola-
tility for a particular mixture and distillation operation. This greatly
simplifies the mathematical analysis and at the same time gives reasonable
answers.
Relation between Relative Volatility and Baaing PointB. By the use of
Trouton's rule and the approximate Clapeyron equation it is possible to de-
riven a rough l'elation between relative volatility and the H,bsoluto bolling
temperatures of two similar normal liquids:
T2 - Tl
log a
.
= (8.9) --~--
'1'2 + Tl
Fo!' instance, the value of a for n-heptal1e-methylcyclohex'~ne rnixtul'es
(b.p. 98.4 and 100.8°C., respectively) may be calculated by substituting
the corresponding absolute temperatures in the equation:
10 a = ( ) 373,8 - 371:.± _ .
g 8.9 373.8 +
371 ,4 - 0.0286
26 Rose, Ind. $nIO. Chem., 33, 1i9a (UJ41).

1. THEORY
Thus, a = 1.07. The value obtained from the vapor pressure ratio iN
1.074,21
For liquid mixtures in which t.he boiling poi ntH of the componentR lwerugu
about. 100°C. the equation reduces to:
T2 - Tl ' AT
log a = (8 9) - - - - - = (0.012) AT = - ~
. 373 + 373 85
If t.he average boiling point is about O°C., t.his becomes:
log a = (8.9) -=-=-+

'1'
273
'}'1
:,>,
73 = (0.0163) AT =
,
AT
61
Discrepancies must be expected in the application of this derivation, Ull-

less the assumptions inherent in Trouton'8 rule and the approximate
Clapeyron equation m'e justified. The preceding derivation is based on a
Trout.on's constant of 20.5. In spedal C[LSeS a different value of t.his con-
stant is applicable. Thus Edgeworth-Johnstone 28 obtained the relation in
several forms, one of which is:
T2 - '1\
log a = 5 - - -.. .~ or
T
where T is average of T1 and T 2 • Griswold 29 and Melpolder and Heading-
ton30 have proposed, still other equations, and DaNis 31 has constructed a
nomograph baa~d on that of the last two authors: '
log a = At(3.99/1' + 0.00194)

Nonideal Mixtures. Duhem and Other Equations. A great many
mixtures have vapor pressures far different from the ideal values calculated
by Raoult's law. A variet.y of equations 32 have been developed to express
2'/ There has been some disagl't~ement. over this value. The one chosen is an averagod
quantity from tho data of Willingham, l'aylor, Pignocco, !Lnd Rossini, J. Research Nail.
Btt)·. Standani.s, 35, 2H) (19·15).
28 Edgeworth-Johnstone, J. Inst. Petrole:ltfn, 25, 558 (1939); Ind. Rng. Ohem., 35,
826 (1943). The; second reference is [\ comment, on the Ji.rst, and 011 Griswolrl., ibid., 35,
247, 826 (194:3).
2D Griswold, Ind . .Eng. Chern., 35, 247 (1943).
80 MeJpolder and Headington, Ind. Eng. Ghem., 39, 763 (1947); ,
81 DaviS, Chern. Inds., 6~, 872 (1947).. . . Y,
32 Matgules, Sitzber. Akad. }Viss. Wien, Math.-nutul'w. Klasse, AM: lia, 104, ;124:}
(1895) .. Van Laar, Z. physik. (Jhem., 7'1" 723 (1910). Scatohal'd andi.ij'p.fel.'i J. Atn. .
Chem. Soc., 57, 1805 (1935). Cilirl~on and Oolburn, Ind. ErW. Dhem., .3~',i581 (l94li!).
Clark, Trans, FIJI'udall Soc., 41,718 (J945); Wol~l, rLTl'an8;.Ami1~8t; Ohem11$.n(/r8., 12,.
215 (l9~16). . "
21 A. AND Eo HoslD
tIl(] l'eln,tion behvpen vapo!' pre::;s1ll'e a,nel tile COlluentl'at.ioll or llolliLieal mix-
tures. All these l1.J'P r~lated to the Dnhem:l:l equut.ioll:
in which Xl ILnd .r~ are the mole fmetiolls in I,be liquid :tilt! 1'1 and 'Y:~ I.fw
neLivii,y oocffieiC'llts, which nre J'elnJed to tlw vapo]' jll'('SSlll'(1 by Llw (1qlla-
(ions:
Pill 1)1/2
1'1 = nIH!
In these 1) signifies tot.al gas pres:-;lII'C, Jh lLlld /12 lL]'(', (,\10 vapol' Pl'()t;S\lI'P:-; or
the two pure eompOlwnts, and !II :tud ']/2 m'e the nlOk i'mej;ioIlH in the vapor.
The most nseful form of the Duhem nf[u:Ltioll is thnJ dmivl1d hy Vnll
L:I:Lr H and expressed b~' CarlROI1 an!l \'ollHll'll/'; Ufo;:
,"
log 1'1 = -(-------'---'11;;1)---~ ,
1 + --
Ba:2
Wlwu ;1:1 = 0, log 1'1 = A, and whnn ;r2 = OJ lug 1'2 = n, A and B aj'(~
obla.ined for the mixt1ll'C' uucler cOllsidnmtioTl hy using 1111' nqlltttiollS:
Values of I'll I'~, :lh, and ;1:2 lLl'e obtained from experimental meaHtll'Pl)1ouL:-;.
Thus in Figure 4 values of 1'1 nnd 1'2 me plotted tLgaim;t :1:1 for tho mixtlll'Cl
of isopropyl ethel' in isopropyl aleohol. The curves were (laleul:ttod by t.he
Vnn Laal' equation with A = 0.42 and H = 0.60. The points arC) from the
experimenta.l data of Miller lind Bliss.3r. It will lw noted thnt, whon :i:1
approaches zero, 1'1 approximates 2.65 and log "II = 0.42 = 11. Bimilarly,
when X2 = 0, 1'2 approH('heK 4,0 u.r\cllog 1'2 = 0.60 = H. Carlsoll ttlld Col-
burn have indiC!Lted the usc of the Vall LtMLl' and similar equations in tWtL!u-
nting and smoothing vapor-liquid equilibrium data, lLnd also in obtainiuJ,!;
approximatiolls of such relations by lise of n limited nmoullt of experimental
33 DuhOIl1, Compi. rllnd., 102, i,HI) (1~~li).
34 Vall Laitt, Z. ph1Jsik. Cheln., '12, 723 (UIJO).
3D Carlson and Colburll, Ind. Ji)rW eltell/., 34, !i~1 (1\1·12).
sa Miller and Bliss, Ind. Eng. Ghem., 32, 123 (1 li'lO).

J. 'l'Hl~OHY 25
da.ta, or from azeotropie ('.ompoHition::>, iotlLi-pre::>::>lln) (,1lI'ves, U1' uuiling-

point curves.
CluJ'k:l 7 hnH Pl'opoHcd Uw eqllatiull:
y (t;(
"-_.,'--. ==
] - :r
+ lJ
1 - !I
aH a mennH of eorrt'lnting expel'iinental dfiJa 011 the vapol'-liquid equilibrium
('ompoHitioll of nonideal mixturcH. This hm; thc advantage that it, i:.; lineal
ill ,1'/(1 - x) = X and y/(l - y) = Y, so [hut when plotted on graph papCI
the rebtion is a straight line. Clnrk exnmined experimental datu fo], ten
binary mixtures :1nd found t,[m\. vldues ('.:~leull1(,l'd by Che cqnn.tion t\,greed
(·lm;ely with experiment, except 1'01' very :-;mall values of :r anrly. For this
region, Clark PJ'OpOSPR 11w 1'f'('ijJ1'()('td
7r-------------------~
(~( [11 a t i Oil : \)
5
1 y (/'(1 - :t:) + 1/ 4
x '¥ 3-
This iH uIH(~ linoa!' awl a.gree:::; with ex- 2

perimental data ill the region wlwl'fl till)
1.5
fin;t eqllatiou fails. Tho eOIlRtunt::> (I,
h, 0', awl b' I>nllst he cil'termiued ex-
pnl'imcnLnlly fol' eneh given mixture. o I
MOLE FRACTION ISOPROPYL ETHER
The eomposition td, whieh the shift.
Fi~. 4. Va,1I LILliI' equation plot of
must be made from Olle equat.ion t.o the Iu,.t,ivity OO(~Jlilliellt (or) versus mole
(lther iH given by the equation: fl'll<lt.ioll I'm if;opl'opyl td<:(}itol- iRopl'opyl
ether."
In evel'y case of llDlliliealllllxtUl'OS, the vapor pressures of the uomponents

must be obtained by llired. experiments, and for use in clistillntioll all such
results f~l'e hest expressed and correl!tled ItH vo!u,tilities, relat.ive volatilities,
and vapor-liquid cquilihl'illln eompositjol1 diH,gmms.
Vapor-Liquid Equilibrium Composition Diagrams. Tho diseussioll Df
H.ID.T.P. (height equivalent to !L theol'etieal pbte) in Seetion I (see also
Fig. 1) illvolvetlus() of (louhlo curves l'cbting t.he eomllositioll of a biIHtry
liquid mixt.ure to its boiling point nnd vnpor composition. It is ttlso pos-
sible to usc a single curve to relate liquid composition directly to vapor
compoHition. Such a gmph is Cl111ed a vapor-liquid equilibrium composi-
t.ion uiagl'l1In or curve (Fig. 5). These curves form the basis of mOHt dis-
t.illation caleula.tions [1,nd thcoretieall1ll1Llysis since they furnish the funda-
mentnl data of the relationship between nny liqnid u,nd its Vt1por. In order
t,1) be valid, the data should hn detel'lnillCd by din~et experiment), which i"
inl1('l'(~llt.\y diffi.ell11, \)('cnuHP of \1)(\ reet,ifimttion whieh Calms phwe in mOflt
ordinary diAtilhlJ,iollH. Rpe(~ial Il1PthudH or Rpeeial equilibrium Ht;iIlH havn
been (leviHed. as
A:,; illdi(~ut,ed in tlw en.rii(w disellHsion, in some ClLfiCR Lhe vlipor--liqui(\
('.qnil il>l"i111l1 (:llrve ean lie pxprPHHed by tho mhl,iv(}-volnLility O(tltatiOll
(when a: is (~OIlKt,lLnL):
:1'
== (~-.
11 1 :r
ct:ct:
00
0.0- 0.8
~§
~~
<:(w
-z
<!:w 0.6
ON
8
::'z
:l;w
<!ID
<!:z
00 0.4
j:i=
UU
<!«
ct:ct:
IJ..LL
WlLJ
...J...J
00
::;::;;
o 0.2 0.4 0.6 0.8 1.0

MOLE FRACTION AMMONIA IN LIQUID
MOLE FRACTION BENZENE IN LIQUID
Fig. 5. Vapor-liquid equilibrium CUl'ves for ammonia-

wttt(Jr (A) (tlld beuzene-toluene (B) mixtures.
This gives curves clOf;ely resembling those in Figuro 5B. Tho flhnpe of tlHl
vapor-liquid equilibrium enrve ehnnges with increase in rnlative vohtl;ility
as in Figlll'e 6. For allY giycn pail' of liquids tho l'nlaLi\'(' volatilit.y gml-
as Typical methods for det.Ol'JllilllLtioll of vapor-liquid Ilquilibriu, IIH ~OlllCLod hy D. «'.

Botkin: (1) DHferontial ciistilllLtioIl mothoc!-Zilwidzki, Z. physik. Chem., 35, 12\)
(1900). (2) BtLttic method-·Sage ailli Lacey, 'i'rans. Am. In8t. Mining Md. Engrs.,
136, l3u (1940). (3) Method of rlill.lil'culation of comlensou vapor-Othmol', Ind. En(/.
Chern., 35, 614 (1943)~ (4) Method of circulation of vapor phase-..-Jones, Sehoonborn,
and ColbUl'll, ibid., 35,006 (1943). (5) Method involving simultmw(}ul:! rodreulMion of
liquid phase [lilt! ()ol)rleJl~od-vnpor phas(r-Gillespio, Inri. 11n!/. Chern., .. lnal. [iJ(i., 18, 57/i
(19 1H:i). See Itlso Figure .! of Chapter III.
1. THIWRY 27
era,lly deereases with increasing preSfHll'c:J!I aIlcl mny sometimes be in-

ereased by addition of n new ecJmponenj-,. Separa.tion will b€ easier but
other disadvantages may be cneonntered when slIch devices are used.
Vapor-liquid equilibrium curves for nonidcal mixtnrfls may take on almost
any shape. Typieal varieties aJ'e indicatc~cl in Figme 7. Each of t.hese
corresponds to a particular kind of V!1pOr pressme-eomposit.ioll rclation-
ship. Portions of curves below the 45 0 diagonal indicate a reversal of
normal volatilities so that the component with lower boiling point has its
normal vapor pressure reduced so much that its eonecntmtion is leHs in the
vapor than in the liquid. Curves sueh as those for hydl'Ochloric acid -water
0:::
0
(L
<!
>
='=
(.)
>
::;:
z
0
j:::
(.)
<!
0:::
u...
L1J
...J
a
:::;;
MOLE FRACTION MVC IN LIQUID
Fig. 6. Change in shape of VIIPOl'- Fig. 7. Till'en tYl1t1S of VI1POI'--liQllid

liquid equilibrium curve with increase in equilibdum curves: ideal, I1lfLximlllIl boil-
rtlJntive volatility. ing point and minimulll boiling point.""
and aleohol-toluene in Figure 7 !Lre inc1imt.ionH of major deviations from

Raoult's law, and involve marked variation in relative vohLtility wit.h con-
centration. If the deviations are of such magnitude that the curves cross
the 45 0 diagonal on the graph, the system will form it maximum or min-
imum boiling-point mixture. Such azeot?'Ope8 cannot be separated into
their pure components by distillation at the temperature and pressure con-
ditions to which the vapor-liquid diagram applies. By contrast the sepa-
mt.ion of nonazeotropic mixtures by clistilln.t,ion is limited only by the
39 Robinson and Gilliland, Elements of Fractional Disl.iUation, 3rd eel., McGraw-Hill,
1939, Chapt. VII. Cummings, Ind. Eng. ChBII~., 23, 900 (1931). Cummings, Stones,
and Volante, ibid., 25, 728 (1933). Dodge 11nd Newton, ibid., 29,718 (1937).
31'" Walker, Lewis, ILnd McAdams, Pl'inciplc8 of Clwmicl!l Enginee.ring. 2nd (lcl., Mc-
Dmw-Hill, Now York, 1927, p. 593.
2R
dfeetivellC'rii':l or tlppamLUK and pl'()(·('ilul'P. .\JI(!oj !'opie mixLlIl'PK :I.I'(~ off.1'1l

Knpal'atf~d hy di"tilln.Lioll t.L a diffn),('llj PI'PKHIII'(', 1)1' al'1.('1' addition of a j Itil'd
(·.()mponent (H(~n Chap. 1fT). Till! ('()lllposilioll ('.lIl'I·CH]lllllllinl-!; Lo t.hl\ poinl.
at whieh th(! vapm··liqnid nqlliliiJl'il1l11 ('·111'\'(' ('l'OHKPN I hI! ·11i 0 diagollal is till'
azeotropi(~ (~OJnp()Hition.
Azeotropes. A mixtlll'(~ of nJlc)[)tropi(! ('Olll)lOHitioll llnhavns Iikn n pnr('
liquid imlofnl' Ufo! it distilR without ehn,ng('. in ('ompoHitioll 01' hoiling point, lLH
long as the pressUl'l' iR ('ollstanL Chang(! in pl't'f;Hlll'l1 clof)i':I ('Imngp hoth till'
boiling point and j.lw eompositioll oj' L1w :L:t,(~otl'ope as wnll as Uw Hlmpe of
the ViLp0I'-liqllid eqnilihl'illll1 ('UI'V(!. Su(~h ehallgeR an! almoHI, al\YlloYs of It
mino!' llalmo nnleRs the! ]ll'PHHI11'(~ dlHlIgn is eonsideJ'u.bl(!. Tlms, a, f:;ohltioll
of hydrogen ehloride in water eontn,inilll!; 20.2% hy \Voight of hydrogen
chloride boils at llO°C. :LL 7GO-mm. PI'8SS11l'O Co give a diRl.illa,te of Lhe samn
eompm;itioll (Fig. 7). On Iwn,ting a mixj,\ll'o witit any (If,hel' (~()mp()Niti(ln
hot.h compOllcmtH dist.il over in varying allHlImis until j II(! :-;WI (~nlltajns thn
ltzeotl'Opi(: mixf1]l'e, whi<'h thtm di:-:l.iiH al. enll:-:I.alli 1.(!m))(Ir:tflll'n. All mix-
tures eontaining kSK tha1l 20.2(Yo hydrog<'l1 (!ld()l'id(~ n\:l,~r hl! look(I(\Ilj)oll a.s
being composed or IVakl' a,nd the Ho:r.eoj-,I'Opn, \\"a.L(~1' h(~illg Ilw mOl'n volatill'
eompnnent. Mixf IlI'Pi'\ of mo]'{! than 20.2%) hydl'OI-!:e!l dllol'i<io may :-:im-
ilal'ly be t.rmLi;nd aA eompoRed of hydl'og!Hl (:hl()\'icle and the azcoj,l'o)lP,
hydrogen (,I1I01'i<lo i>Pillg the mOl'O \'olnl.ile (~omponellL. "\s with ;;imilal'
systems not involving H.JI('otl'OPC8, diHtillation, if sllffiei{~llf,ly ('ff(~e(.ive, re-
sults in ;;eplLl'atioil into the eomppuonts waf.er alld azeof;rope for :-\VHtems
with lesH thun 20.2% hydrogen chloride, tLIld int.o hycirog;oll (~hl(Jl'id() and
azeotrope for those with more than 20.2%. The water-· hydl'ogoll dtloride
system is typical of t.hose forming maximum hoiling-point. mixtures.
Identical st!d,ements apply t.o systems forming minimnm boiling-poiut
mixtmcs except t.hat in t.hese t.he azeotrope is more volatile thlLn nither (JOUl-
ponellt. TllUR all tolueIle-·aleohol mixtures with l08s than 41 % toluene (~Hll
be separated into nzeoLrope and alcohol and thoKO wit.h ll101'f' than 41 %
toluene int.o azeotrope and toluOTlf'.
The outstanding elmraeterist.ies of binnry aZ(lotropie Hystems lm~y be
summarized as follows:
(1) The vapor-liquid equilibrium Cl1l'ves of sneh systems ('.r08S the 4fi 0
diagonal.
(2) The individual components cannot be separnt.ed from OliO anot.lwl'
by sufficiently refined 01' repeated distillation, as is the case WWI nonttzeo-
t.ropic systems.
(3) Any azeotl'opic mixt.ure may be considered to be composed of t}w
azeotrope and a single component. Except that one component. iR au
azeotrope, such systems behave like nonicleal, nonnzeotropie syst.emR when
sUbject.ed to distillation. Because of the last eirenmstance, further dis-
1. 'l'Hl~OR.Y 20
elll:lHiull on the j~hl)o]'y ur diHtillntioll in this dlUptel' will be eonfined to JlOl1-

H,zeotropic systems, It should, howevel', be mentionell that difficult
separations by distilla.tion are frequently made easier by the Plll'poseful
aliJition of a Hew component thut formH an ILzeotl'ope with II component
already present, Slleh Hzeotropic cliKtillation iK disf:llHHcd in detail in
Chapter III,
These Hytstenu; exiKt ill either OJl(~ or two liquicl layors neeording to
the eompoHition of tlw Kystnm u::; n whole. H onl.v uno liquid phase is
present, each compollent will affect the vapor jl,'el:lHlll'e of the otlier, and the
mixture will exhibit Vltpor pl'm,sures, boiling pointK, lWel vapor compositions
similar to a completPly miscible system. A dist~illate mny or may not
separate into two byers, depemling llpon its pCl'eentagp eompositioll i1nd
1pmpomture. If t.wo liquid layers nrc pl'ol'ent in any mixture, their com-
positiom, llllU-lt always be sneh as I,D prodnc() the same partin,[ vapor pressure
of any given compOlwnL from both phases. Any component for which this
WlLS not trne would of e()urse eontinuc to evnporate from the phase luwing
Lhe highor pressure, nnt! eondellse in the plmso having the lower pl'e:>sure,
unt.il the one phase disappeared or the two pressures did beeome equal.
The partial vapor prt'HStll'e of each eompouent, over Kueh Cwo-phase mix-
tures remaills'nnehnnged for vu,rious over-nll e()mpol:litions, ttK loug as both
liquid phn~es l'euuLin present. Each of t,lw bUm' also remnins c:onstant in
(\ompositioll, so thnt ehtmge in the (lvel'-lLll eornpositiou reKuIts only in in-
el'llfLsing the proportioll of one 01' another of t.h(~ liquid phases, As a result,
Kuch mixt.ures beha vo like totally inllniseible system:>, in that, ttt a fixed
tcmperatnre, the ViLp0l' and distillate eompositions remain constant regard-
less of liquid eompositioll, itS long a,-; both liquid phases remain.
2. Multicomponent Systems
There has been considerahle int,m'L1HL in the steam distillt.tion of liquid

mixtures that are tlwmselveH partly or eOll1pletely immiscible wit.h wate!',
If thero is l~()m.pl()te immhleihility tIll' 1mI'Li:.1 vapor pressure of tlw water is
detnl'mined by its tmnpomt,ul'l', :tIld the purti:Ll vnpor pressure::; 01' the eOll1-
ponenLs in the other phnsl~ nrc determined by their tempe!'nt11l'l~ HIld com-
pOKition as if they Wl'I'P pl'OKPnt. without t.he wate}' htyel'.
II' there iH IHlI't.i:d lllis()ihiliLy, tlw I'dn,tiollS are allalugous to t.hOHH of
parily mil:!eible biulLry mixtures, Tlte illt,l;l'l':-;i, ill sueh systems hUH l)(~en ill
eaSt'S in whieh the presence of tlw w:tlel' (;ffoets lL marked reduction in tlw
\'lLIHJI' pl'eSKlll'e of one of Hovel'u,1 Snbtltallees of nen,l'ly iclentienl boiling point.
ao A. AND E. ROSE
The vapor pressure of a-cresol is thus affected amI the eompound mo,y be
separated easily from its para isomer by steam distillation, although tlw
boiling points are 191.5 and 202.5 0 , respectively. The principles involved
are similar to those discussed in connection with (lxtmct.ive distillation
(Chap. Ill),
B. CO~IPI,m'l'KLY M[SCIHLI~
When Raoult/s la\v is applicable, it is still true that:

Pl = PI;tl rmd 1111112 = a(a:t! x 2)
The vnpol' pressure - eomposition rebtiolls can no longer be shown by
graphs with simple eo-ordinate systems. For ternary mixtures it h..;
possible to use families of curves, triangular eo-ordimLte systems, or three
dimensional structures, but this is rarely worthwhile. Special apprQaches
are usually developed for such situations (Clmp. III). The relation:
y = J(x
in which ]( il'l called the eq'uilibriurn ratio or elJuil:ibriwm con.stant, has been
valuable in muHicomponent (mlculatiolls for petroleum distillation, The
literature contains many references giving v(Llnes of equilibrium eonst.ants
and methods of obtaining them:111
HI. THI~ORETICAL llLATES, IU.<:FLUX nATIO, AND COLUMN

COlVU'OSITION S
1. Plate Efficiency
In Section I a theoretical platB was defined l),S one tha,t produces 11 pcrfe(\(,
simple distillation, tlw,t is, the Sl1me difference in composit.ion as exists Ht;
equilibrium between n liquid mixt.ure and its VlLpor, A general dCfloription
Was given of the derivation of the idea from hoiling point'-COll1POHitiOlI
curves, and the comparison of columns in tcnm; of the number of theol'ei,ical
plntes was indicated, In a plate column, n theoreticnl plate is one in
which the vapor and liquid lcavin!1 tho phtte m'o of sneh compositioll as to bn
in equilibrium with one <Lnot.hcl', Thus etL(',h phtte should theoretie~tlly give
n per£eet. simple distillation. T\l(> sepami.ioll Oil Iwtllal plates is wmnll.v
<0 GILIllSlHl lLnd Watson, NaU, P{,tl'ol~lUn New", 36, IW23 (HJ4·1), .HlI([(itm, 'l'ranN,
...tm. InRt. C/WIII. En(}I'N" 44, 37, l35 (1\1'~8); revil,w,lIolllogmplm. IIllllHlltl ltlld U. U.
Browll, Ind. l!)ng. Clw/f/., 37, 821 (lIJ'l5), lind tlHJ']iu)' :LI'U",11J1l by .B)'OWl! HIId eO-W()1·)WI'N.
MilleI' ltad Hadlly, 'ilJid" 36, 1018 (HN"l). Othuwl', illId" 36, Ulill (104-1). lttmlll(\l',
KOl'pi, tlllge, :Lud L:w(\y, ibid" 39, 201i (H!47), (mel mLl'litll' al'ti()[llK by Sage, Lttt'oy and eo-
wOl'iwl's. Rt\li~h awl WilhOtl, '[',((Ill.', ,.[ III. I It"t. CIII'IIl. 11'11(/1'8., 42, 027 (l\l'1!\). l"lt~(\ :liHO
8tlch. Rtlwdlll'd ohemical engineoring text,H i\H thO~(l ill the f,I;(!Jltmd rOfm'!ill<lO IiRt on p, 17;1,
1. 'J'III';Oln- :31
:'\OJnewbat less, due t,o imperfeet mixing, foaming j,etHimwi('s, enLrn.illlnent,

n.ne! r,he limitations of nwohanieal rksign for va.l'ioll~ c:ohmm diameters.
Numerous tYPflS of pln,tfls hn,vP 11PPlt d!wilwd whic'.h vH,ry widely in plate
efficiency. The latter is a eompnriRoll of the nc:tual sepamtion with the
theoretical, and may be expl'f~ssed in t,wo ways.
Over-all pla,te effieiency is a simple pcn:(mtage l'ehLiol1. Thus if It column
with eight actual pIntes givps separat.ion equivalent to only six theoretical
plates, the plate efficiency is 75(70' The efficiency uf nn individual plate is
usually defined by the 111urphrce plate effieien(T: 11
'1/" - "1,,--1 IOO
(' m = :, ,,~-___:_- (1)
H" - l},,-i
in which y;: is composition of the VttpOl' in equilibrium wiLh liquid on plate n ,

Yn is aetual composition of vapor from pbtc, n, nndlJn-l i[o; aetual composi-
tion of vapor from the plate below. The two kinds of plat.e cffieiency are
not identical, sinee t.he first represents an uvera,ge valne, but they usually
do not diffor gl'e[ttly. It. is obviom; that, plate efficiency if; of great impol'-
tanee in industriltl design, and mnny studies of it luwo beon published. ,12
2. Theoretical Plates
Standards for pa.cked (\olumns have not been obtained so directly. Vari-
OU8 shapes and tYPCK of pueking me HRPcl, as well as varinus sizes of eoilmns,
und by t.rial and enol' the hest, eombiwtt.ion of thefw for vttrious seales of
operat.ion has beon found. These paelwd (:oiumns and many pla,te columns
ttS well are eomm,only eVll,hutted in tormR of t.he height cquivltlent to a theo-
retical plate CH.E.T.P.)4:1 or by the tot.al !lumber of theoretical plates in a
given column. Sinne the ehange in (~OnlpoRition in H, paeked column is
gradual and continllous from top \,0 bottom, whereas the concept of plates
involves stepwise ehange from one plnte La the next, the nse of t.he theo-
retical-plate standard is basimtlly unsound, ltnd other standards avoiding
this anomttly have been worked out (Sod,. IV). It is quite probable that
Kubstitution of a funcln.mental diffusion pietul'e for the theoretical-plat.e
11 Walkor, LowiH, McAdalllH, amI Gillilulu\, Principles of Chemical Engineering, 3rd

otL, McGraw-Ifill, Now York, Wa7, p. 5()!l. :Hul'phr()o, Intl. En!}. Chern., 17,717 (1925).
·12 Itobiuf1()Il and GilIihuul, FJI,CI)umtN (~f F7'Ilctionill Disl:illntinn, 3rd Bel., McGraw-Hili,
New York, 1939, Clmptor Ill. Some recent (!J'liclc.~: Geddos, Trans. Am. Inst. Chem.
EnOl'N., 42,70 (HIM\). Grhwvold !l.nd Rt.!'owal't, InrI. Bny. Chern., 39, 753 (H)·~7). Nord,
Tra,ns. '1m. InNt. Chern, Engr,~., 42, 8li3 (H).1(i). O'Connell, ibid., 42, 741 (1040). Sevel'al
long articles in. French seem. 'Worth notino: rvlltrilllw, BILl.!. assoc. chirn., 59, 445 (1942); 61,
3 (1944). Perard, ibid., 58, 28, 182 (1941); 60, 121 (IIH3). H!l.varit, il>iil" 60, 360
(1M3 ).
,13 Peters, Ind. Eng. Chem., 14, '17U (1922).
32 A. AND lD. HORE
('olleepL willl~veJlLllally l'()S\l11 ill d(!vdo[lIllf'IlL of nil impl'ovnd and ('ompl'(1-

It('nsivc j)l'(leedttrl~ for cvnluntill~ fl'ad.iollaling (·.OIIlIl111S. IIO\\'~'Vf:'I', the
I1ULjoJ'ity of existing pxpcl'imenl.al data al)d literai;mc a.ppliea.bh1 to Lhe diH-
I illation of organic: materials is in terms of thooreti(~al plates. ThiR chapter
I hel'pfore deals with this very useful and convenient Rtnudul'd.
'1'he equations and proeedures developed hore nJ'e nsn[ll! for caleulatill~
the numher of theoretical plates in fractionating columns, and also fol' pl'(~
dieting distillnle 01' produei; ('omposition \"hen still ['ompositioIl, number ()f
plates, rdlux miio, ane! I'elntive volatility al'n fixed. In geneml, the sam(~
methods may b(' liKed to estimn,ic any OIl(> of these \'a.l'i:thle~ when the othel'
four are known. The detailed tlis(:lIs;;ioll in this chapter a.pplios to bin(tr~r
mixtures. M111tieonipouent mi:d1ll'cs involve the same principles, bu1
application i~ of ton complientrd and lahOl'iollS, although new lines oj'
ll,pproach aTe reducing these diffi.(·nlticK.1;1 Howevel', tho methods for bi-
nary mixtures indicate the geneJ'alllatlll'l~ of til(' effed. of reflux rntio, numher
of theol'eticll,l plates, [mel relative volatility ill the distillation of multieom-
poncnt mixtlll'es. The limitnt.ionH llurll'cstrie1ions of the variolls lIwthoil"
rtl'e ('ompnred in the summary ttt the Prld of this dtnp1('1'.
A. GI~NERAL PIWmmUIU] FOlt DWCI';UMINING NUMBmIt Oir '.l'HICormTIC,\L

PLA'!'l']S AND HEIGH'l~ BQUIVAU:KT '1'0 'l'JUmrU;'I'WAI, l'LA'l']<]
The most. common proceci11l'() iH thai, applied 10 ordinary hatch-distillatioJl

appal'atus operating at total rr,!i'H.1:. Tllt' ('olmnn to he evaluated if; t.oHkd
wit.h a suitable binary mix111l'0 for which the vapor-liquid equilibrium (:urV('
is knowIl (page 25 and fool11oi.P :~8). The nceeHHary experimental data an'
t.he (:OInpoHitioIls at the bottom and Lop of the (~olllmll after it is opemtilllJ;
smoothly and compositions an' no longPI' elmllgin~. For the mmal lmtdl-
distillation I1pparatus having U jotal eOllchmsPl', KHlllplos hom the still and
from 111(' (:ondenRer 01' Hw distillat(' take-off lilw an: lIsed. Becmlse of I.Jt(·
]'('ctifieat!(ln that. oeems in It pnl'tial ('OlHkI1Hc'r, top samples from Hl)('h
apparatus must he taken from Ow vapo!' at til(' top of the co]mnn. The
subsequent (liHcl1ssion aSSUllH'S a totnl eOndeIlH(\l' is uHed.
When tIl(' two neer\ssal''y compositiollS have been ohtained, the numlHH' ()J'
theoretical plates is Ktflppe(l off on the ITttPOl'-liqllie] f'(lllilibl'ium diagram
H EdmiHtol', 'i'ralls. Am. In8t. ChclIl. ElI(lI'.,., 42, Hi (1\).10). Gilliland aud Heml, Inri.
Eng. Chem., 34, 5.5! (1\)42). HogmJ, ibid. 33, [132 (1\)11). IlUllllllnl, 'l'rnns. Am.II!.~/.
Chem. En(l1·N., 40, +M (1\)+1). .Teuny, iln'd., 35, (ia5 (loaIJ). ,fr:llUY and Ci(HLieso, Inr/.
Eng. ChCIlI., 37, 9.56 (1945). I\:arnol'Hky, 1:liid., 34, SaD (1!)·l2). Rrm nlso n. G. BI'O\\,Il
:111d Holcomb, Petl'oimun Engl'., 11, No. 13, 27 (11)·10); 12, l\o. j, [-III (1\)·10). Colburn.
Trans. Am. Inst. Chem. En[!1'8., 37, 80.5 (H)41). Gilliland, Inri. Ell!!. Olwln., 32, 1101
(1940). Sclwibel and Mout)'(}ss, iliid., 38, 208 (H)-Hi). Unrlm·\\"o,,,!,.I. flwt. Pr·tro/fUIII,
31,111 (194.5); 32, 5nS, 614 (1\14(\).
1. 'J'HlWHY
(pagn :)7 aml 1i'ig. \l) ur is ealmtlntml by the Few;ke equation (equatioll
;~). TIH~ H.E-T.P. iR then ea,\eu\nted by the fOI'IllU\t1:
H.E.T.P. = Z/n (2)
in which Z is the height; of the packed sedioll (in centimeters or inches)
and n iR the number of t.heoretien,[ phLtes in the column it.self. The divisor
'II, - 1 is sometimes used when n signifies t.he tutal number of thcul'eticnt
plat.es in t.he column and still taken together. The reasoning and equations
lLre applicable bet.ween [LIlY t.wo point:;; in the r~()lnmn, pl'Ovirlecl propel' cnm-
pm,it,iollS and t.he propel' divisor are IIst'd.
The determination of number of plat.es and H.KT.P. at partinlrefiux is a
more complex matter. In both continllous and batch distillation at partial
reflux, 11 materilll-balance study shows that. the composition change from
plate to plate is less than at total reflux because of the unequal quantitieR
of liquid and vapor entering and IO!LVing a plate during any given time
illterval. As n result, different stepping off proeedure or different equations
mnst. be liKed in estimating the number of t.heorct.iml plates at partialrefillx.
l'hcse methods Cpn,ge c~;') r:t seq.) take into aeeollnt the effect of unequal
rates of How of llquirl rmd vapor 011 the material Imlallee for a plate or seu-
Lion of plLeked eolnllln, as \Yell as fnrther complieating factors such as 1,h<'
gl'acina,l el\H.ng(~ in eomposiliull elmm{'tel'i:-;t.i(~ Df 1mtdl distillation ~tt pttl'linl
reflux (pagn Hi).
Actunl distillutions lLre always eal'ried Ollt ttt partiall'efiux oxeept when
(listillate is l'Ulll()ved intermittently 1tftcr periods of operation tlt total re-
flux:!" BVC'll this is probably cqllivalent to operation at partialrefiux un-
less dif:ltillnt(! l'OInov~Ll is at widely slXteed intervals. However, it has been
generally assumecl that the number of theoretical plates of any column at
t oLlLl reflux is indica.tive of the separation that will be obtained in actual
(1istillu.tion at pm'1iall'oflux. Tho experimental procedure and cal(:nlat.ions
for tot.al reflux H1'8 simpler, and it. Ii:! possible to bring the various parts of the
('olullln into eqllilihrium in a more reproducible fashion. In other words,
\ he tobLl·l'eilux method of determining the number of theoretical plates
i'nrnh:llw8 a eOll vpnient, wa~' of eomparing columns, even though the aetual
llnmel'ieal J'0S1ilts mny not hold t l'lIe nt finite reflux.
Apparatus for Obtaining Distillate and Still Samples. Test samples of
llist.ilbt.(· may l){~ taken from the eonclcnsol' in the S~Lme manncr as during a
Ilistillu.Lioll, ()xcept, thnt dead SP:W(1 or holdup in t.he sampling line must be t.
nunmnUl1. It is trull that snuh holdup ean be :Hushed out before a sample
is colloeter1, hut this is the equivalent of operation ltt finite reflux, whereas
t,otft! reflnx is llstmlly desired and assnmed.
For the still samples, the :fractionating column to be tested must be
pqllipped with H deviec fot' removing sttmples of the $t,illliquid for analysis
·Ir, [I'mll'''''', ill 8rimre of ['rlra/mlm, Oxford Uuiv. PreHs, New York, 1938, pp. 1659-]660.
34
without. inLel'l'UpLing r,he dist.illation l).llrl wit.hnuL evn,pon\,tlotl of U1(\ mo\'(~

volnWe eOlnpollPnt from the r;amp]p. It may he ('.ooler! by a small eOJ1-
A B
I ml. INSULIN
SYRINGE
c D
Fig. 8. Still-sltmpling I1rrallgelllcllts: (A) l!'ellske:IDa (B) Mortotl. ~I;
(0) Rose. 41 CD) Prevost. ,8
u. Fenske, ill Science of Petrolenm., Oxford Univ. Pross, Nnw York, 1\):38, p. HHJI,
Fig. 22.
46 Mortoll, Laboratory Technique in Organh; Chemi8trll, McGraw-lJilI, New YOl'k, 1938.
p. 87, Fig. 35.
47 A. Rose, Ind. Eng. Chem., 28, 1210 (1936), Fig. 1.
4l! PrevoHt, !VI. S. Thesis, Ponnsylvania Stato CoJlego, ] 048.
1. 'l'HEOnY 35
denser on the ::;alllple line, or by immersing the sample container in a cooling

bath before opening it to the air'. The sample line should be of as small
vohunens possible. It should be Hushed with the sl;illliqllid just before the
sa,mple is taken. One practice is to incorporate t~ sampling line and valve
drainiug from the luwm;t point of the still (Fig. 8A). POl' ordinary glass
flasks, a siphon Hrmngement may he provided (Fig. 8B-D) if there is
lUi ex.tra opening; at the top I)f the Ihsk. The use of B, pipet to l'effiOV€:
samples from the boiling still liquid is not recommended. In some ca.ses a
small portion of the reflux liquid lea.ving the bottom of the column CH,n be
deflected into a sample contu,iner. Such liquid may be assumed to differ in
composition by one theoreticttl plate from that in the still itself, but errol'
may be illt.rmillced by t.his assumptioIl.'I!! Few experiment.nl data. arc avail-
able as to its validity.
Choice of Test Mixture and Its Initial Composition. The ideal choice
for it test mixture for H.E.T.P. determinations should eome H"l close a.s
possible to the following requirements:
(1) The vapor-liquid equilibrium diagram should be based on well established
(bta. The (l!~ta for commonly used mixtures are given in Tables I to IV.
(2) Puriflcatioll of the itulividtml compoIlents should be simple and certain, and
the mixture Hlloulrl be Hllch that quick und accurate analyses can bc made on small
t:inmples. Filial lJUriiieatiull tlhould be by distillation at 11igh reflux through the
eolUlllll to hu t,e!:ltell, or oue with the SMile or a greater lllullber of theoretical plates.
(8) The mixture lHnst not seriously cOl'l'(J(le tbe packing or other parts of the
l~Pl)t1mtus.
(4,) The lllixtlll'o i:lhould be neither too easy nor too difficult to separate in the
column to be te::;tCI1. (Sce diseu8sioll of Tables I to IV, and the following dismIssion
011 ehoice of initial cOlllp0i:litioll.)
(5) The (JOJllPOllt111tH of thc mixture should be similar chcmic!tlly to those for
whi.eh the column is to he used, ltnd in the samc general range of density, surface
tension, and viscosity.
The initial c:ompoHitioll of the test mixt.1ll'e mllst be such that the sample
from the eOllden~er ellllt.ltins an apprecb.ble u,mount of the higher-boiling
eornponon(;. 'l'ltis is nnOPHH[LI'Y heclLUsc t,he change in eornposition per
thool'!ltieu.l plate beeOIlWS vel'Y Bmnll when the mixture is nearly pure in one
or tho ot.her (·OlllPOli(\llt.. TlmB, if 11 eolumn with five theoretiClLl pln,tes is
t.o be tested wi(,h hellZPIln-toluene rnixt1ll'e, the initial composition should
be neal' 20 llW!P pel' eont benzene mliher tha.n 50 mole per eent. The former
mixture will J.!;ivl\ II ($)lHlellSer i:ll1rnple with ahout 89 mole per cent henzene if
tllOl'O nre [Olll' plnLOH, $5 molo pel' (lent if there arc five plates, and 98 mole
pel' cenL for six plates. 'l'll(~ OOI'l'ospontiillg compositions for the 50 mole per
eellt. init.ial kHt mixt.me wOlllc! he nhouL 97, 99, and 9~) +
mole pel' ceut.
·!U O(,luueJ·,llIti. NIIY. (!111'1II., 35, (j15 (1!H8). t-ltolhnuJI, Can . .T. R8seul'cit, 5,458 (H)31).
:w ,\, AND li~. JWSJ'~
] Il the finit l~n~l~ fill lI11Ct'rtnill (yoI'] (;/;, ill analysi:-; WUll hI noL hn sf1riom:, hll (
iu thl:' Kf'{,Olld ea,He i\ueh an \\l1(,prta.inty would nmke it impossihle to do mon'
than eonducie that there mill'll 1)(' fOlll' or more platt's.
1Vluch more seriolls is nny 1l1ll'l'rlaiIliy or illaN'lIm('~' ill L1w posit.ion of tlw
l~qllilibrjU1n curve jlKPlf. This aJfel'j H t hI' ('oTllJlosi (ion ehange of elwh plat(~
or step and produces a, ('lImulntive error of il1(:J'(~lL"ill!!; magnitllde at, the ox-
tremes of eompositioll \\'Iwn~ tlw eq\lilihri1lIll line alld the 45 0 line I:onw
dose together (Fig, n), Since eVl~Jl IL [) lllol(~ pl~l' eent bOllzelle,-tollH~I)('
mixtnre will give 95 mole per eent henzelw in the (\OlHlenRcr of n. seven- 01'
eight-plnte column, other 1('l';t mixtllres (Tahll':::; I-IV) an: geuomIIy 1Ised
for all exeept the shortest (~()IUnlll::;. It is of eoUl'SU pOF;siblo to use any mix-
ture that approximates the generul requirements already mentioned.
B. OI'l<ntA'l'ION A'1' '1'O'1'A1, H1!JFLUX
(1) Test Procedure. The column to be testf)d should be cleaned and

dried (page aOn) and then eharg()d with enough test mixture to filln.!', least;
half the still. For pn,eked columns hent should be lLpplied until Hooding
oemus throughout the colmnn, nnd the ]l()nt, input then reduced gradually
to t.hn,t, desired for the test. If tJH~ packed soet,ion is equipped with heaters,
these should he [t(ljnstetl so thaI, the eo]umn is operating adiabntieltlly and
the eolumn then iiooded again. It is important t.hat the packing he kept
wet W}WTl heat input. is l'c(hwed nftu)' floodillg, OJ' thp hl'llefieial d'fnct will he
lost. The initial flooding generully mnkl'R sllh;-;cqllPnL behavior more 1'0-
prodneihle, l~nd tdso retimcR the H.K'!'.P. H,]ii.T.P. varieR with thl'OIIp;h-
put so that. this lflW::tj be:' kupt at t,he dwsml value tl1l'Ullg;houL (lll' test.
Operation at. toLal reflllx :;I1OlIld ('on1.illlIe lI11til all parCs of the appnmLuH
nrc opernting smoothly. This may l)()st be t.oHted by tnking small HUlTlple;;
from the condenser at inLervab, until no i'm'HlPl' (',hnngn in composition
oeeurK. All sample:> should be withdrawIl I'll owly and should he lel'ls thaD
, 1% of the chl1rge, in order to minimize disturbHllces in column tOlWelltm-
t,lons und recluee the intervn.ls hetwmm samples. Small columns with few
plates OJ' open paeking willl'et.urn to it sl:t!ndy state qniekly aftel' remuval of
l~ sample, but large columns, and those with twenty 01' more pbks nuty re-
quire nil honr OL' morc. 'Vhen tlw steady stu-to is established, Hamples
should be tn,ken from both still ItlHl l~olldcns(,l', nnd uHed for anttlysis alld :~
computation of the llllInber of ph~t(·s. It iH profprablc to repeat tho J'l'movld
of samples several t,imes, allowing; it ::;(liLable in[,(\)'vlLl tn eltLpHIl betw(wn oaeh
:ltl,mpling oJwl':ttion (,() reuHI·ahli;;h thl' steally >lta!.I" lOlHl {lontilllling; until
ehcek l'esltlt,l'l aro olduill(·ll. A (,lIOl'uug;h t(!~t. :-lhmdd illelndl' :t (:ompint(·,
shutdowlI and n:petil.ioll of I.lw ()xperiuwnL from Hw ]l('gillniug. .A (·est
on n new plteking should intdlldn ('mpi.yiJlg; alld j'('pn('kinp: ()f (itt· I'ohllllll alld
I'epeftting' the test
1. 'l'H1~OJt Y ;)7
(2) Determination of Plates from Vapor-Liquid Equilibrium Diagrams.

When t.he eompm;itiollS of distillate and still !ire known, the vapor-liquid
(~omposition diagram may be used to step off the number of theoretieaJ
plates. Such a stepping-off process was illustrated brieft}· in Seetion I
with the use of tL hoiling point-eompositioll diagmul. It is more common
(,0 use a vapor-liquid composition dillg;rll,ln, since boiling points are not
I kr
0.8
,
I
I
'£-,-------:"" k5
,
il::
oD.. I
I I
~ I
~
0.6 , I
I
u I
> , \
,
;;:;;
6 11
, I
i= I
I
u
~ 0.4
I
I
,
It..
I, I
w
o
....J I
;;:;; \ I
Y5 I I
I I
0:2
I I
\ I
I I
I I
I, \
I
0.6
Fig. n. Oraphie Jllethod of ohtniuing number of tlworetical plates.
(~Hselltiu.l to the determination of the number of theoretical plates. The

('(n've::; Hhould be plotted on 24-in. or 36~ill. sqLHlre sheets of graph paper in
order to J'oduee orrors in plotting and reading.
The stopping-off llroeeils for u typicd vapor-liquid equilibrium diagram
involv(l:-i tlw following reasouing (Fig, B): If t.he twtual still composition
ill Ow tORt iH X:" the V!lPOl' enter.ing the bottolll of the column would be of
uompoHitioll 11.. If nil this vapor Wel'C condensed to complete a single,
38 A. AND E. ROSE
perfect, simple distillation, the distillate would also have the composition y.,.
On the diagram, this value is no", tmmlposed to the liquid-composition co-
0
ordinate XI, which equals 1/", by drawing the line from 1/" [Lcross t.o the 45
line and down at right angles to :rl, This liquid fraction in the lowest
section of tho column will l;e in equilihrium with vapor of eompm;ition 1/1,
and if the vnpor wert: condensed it-, abo would have [,he sILme eompol:lition X2.
The point ;1:2 is obtained oIl the diagram as before, and the proeess repeated
ILI:l many times tLS neeessnry to oht.ain a puint 1}.l, represent,jng distillate as
pure or purc:\r than t.hat nettwlly obtained, 1/", ill the experimental test of the
column. It is of ('OUl'He lIot neeessary to obtain the intermediate points 11"
:1:1, 1Jb ete., since the wnstruetion (or even j,he inspeetion) of Ow points k, kl'
7c 2 , eLe. gives the clesired ellcll'e~mlt. In this example tho system has the
equivalent of between four ltncl five tileoretimd phtes, tlUtt is, Ii = 4, be-
eause four steps are requircd to go from sWI eornposition to uoucienser com-
position. Since the first distillation oceurs between the still Hud the base of
the fmctiormtillg column, the latter itf'lClf has only three tlworetienl plateH.
If the height of the plwked section is 12 inches, the RKT.P. is approx-
imately 4 inches.
It, is equally correct to eany out the stepwise construetion in the reverse
order by shLl'ting with the aetual distillate composition and working toward
the still composition.
(3) Calculation of Plates by Fenske Equation. The li'enskei'li equation
(also derived by Underwood 5l ):
'Un
1 - 1/"
Illay alRo he lIsed to !:nlculate the number of plates when certain test; mix-
[mcs are Ilsnd. It iH lwenssary t.hat the vapor-liquid equilihrium follow the
rehd,ive-volatility relation:
y a .. _..X
_.--.. - (4)
1 - 11 1 - X
and that t.he l'ehLtive volatility be at least approximately COIll:ltant over tho
range of concont.rat.ions and temperatures involved. AUlwemge vnlue of a
is then uRed, and when a doer; !lot vary more than 10(70, reasonable results
may be expected. Methyll:'yc:l()~lexnne-n-heptane,. bcmzcnc-toluene iLnd
benzenc-ethylmw diehloride mixtures meet the requirements.r.2
In the F\~nsk(' (:qwt(,ion :1/" is the mole fmetion of the more volatile eOIIl-
"' l"1:IlHko, Illd. 811Y. Cltell/., 24, ·IH2 (1H32) .
•, lJllri()l'\\'ood, '1'1'01/",. In"l. Chelll. Rnyrs. London, 10, 112 (1 Ha2) .
• 2 A graphie Holutioll of t.hl' 1"(lllHke equation has iJcpu [JllbliHlwd by i\1ejpoldcl' Uoud
1lI!lLdingt,oll, Ind. JiJng. ChenL, 39, 703 (1\)'17).
I. 'rrIlDOHY
pOllenj, in t.he djstillat,(~ of a column opera.ting at total reflux with a totll,l

eondfmser and wit.h 'II. 1h(,oI'0ti(~al plat0s and st,ill composition :r".
II'01' 11llrpnRf'S (If (·:l.I(~11Ia.t,ion t.ho (,({11Il,I,ion may lw \\Tittel):
log ' ~~--. 'JII ') - 1og ( -.- ,r"

.---~- )
( 1 - 1/" 1 -
11- = --.. ----- -----~ . .----------.1,- - 1 (5)
log ex
For example, ill Figll1'c 0, ",h0re it is assumed that!/lI <= 0.9, :r., = 0.1:35,
[tnt! ex = 2.45:
(0.954) - (9.193 - 10)
'Jl. = ~.... --.-~----"---'--'- - -
1
= 3,- ..~J (6)
0.389
This is of eourRc the same value tl11Lt wus obt!1ined by t.he graphic procedure.
The derivation of the Fensk(~ equation is merely the analytical equivalent
of the graphic operations previowlly described for stepping off the number
of theoretical plates. For the purpose of the clerivai,ion, the relative-
volatility relation may be written:
'y, :r.
ex .------. (7)
1 - y. 1 - :1: s
A B
a:::
w f5
<II
If)
Z
z W
w I')
a z
z
8 8
Y3
DISTILLATE
I H.E.T. P.
}
I I H.E.T.'
} I H,E.T.P,
Fig. 10. Derivation of Fenske equation. Composition a.t vl1l'io\lR points in plitt!,
(A) :md pankf,d (B) columnR at tota.l reflux.
A, ANT> ]i;, HOSE
'l'hn <[llH.utity yj(1 - lis) l'(,pJ'(>~(·llt.1:i Che mular raCio uf the ll1O]"(! \'olaLik ltIHI
lpHs vola!.iln (.'(JIllPUllolitH af(l'l' liqllil\ with mobr rat)!) ;1',/(1 _ ;1',,) is 1:111\}-
joctml to n, single, perfect., simpln distilln,t,iOlI. This (',Ol'l'CS}JOIllis ciot;dy to
t.}}(-l degl'(~H of separation between the "till pot Hmi the vapor entering the
base of the eohnnn; thfl.t, is, when still composition is x,, the Y[1pOr entering
the bottom of the eolnmn has the composition lis (Fig. lOA). If the column
i" assumed to bn of t.he pInto t:ypo, Iwd if this vnpor is eompletely eondellsed
on the lowest plate of the coillmn, the resulting liqllid will he in eqllilibrium
with new vapor having compositioll ,Ill, ",hel'o:
1/1 _' =
'-"" i f-_-. ,
=.IX"
, :1:"
IX' " ,- (8)
I - Yl 1 - 'ils
1. - :r"
Complete eondensation of this new vapol' on the ~econd plate from the bot.-
tom would give n liquid of composition lh H.epet,it.ioll of this reasoning
shows that the new Vl1por rising from thn seeollcl pInte wonl(l lutve the
molal' nltio:
1/2 (9)
1 - ,1/2
and, in general, tho V[1pOr from the nth plate woulflluwe molter ratio:
lin
(10)
1 - 1/"
With a, total condenser the di~tillate compoHition would also be Yn' Thus,
if a eolumn has n theoretical plates, n just under total reHux will gi Vp dis-
tillate compositioll Yn when still compositioll is X,I' Conversely, if these
eompositions are obtained in [L test, the eolumn must have n theoret-ieal
pltttes.
A somewhat similar sitllntioll exists in a packed column, Hnd is reprp-
sented approximately in Figure lOB. The end rosult is the same but the
gradual change in eomposition requires a somewhat different analysis. By
definition, the height of eolnmn that produces a sepamtion 1/8 to 1/1 is tlw
H.E.T.P., so th~1t, at. a distance equivalent to ono H.E.T.P. unit above the
base of the C'olumn, the vapor compositioIl would be that corresponding to
the mohr ratio Yl/Cl - Yl). Hepetition of this reasoning shows that, LIt
points distant from the base of the column by two, three, four, etc. times
the H.E.T.P., the vapor eomposition would eorreRpond to molar mtioH of
Y2/(1 - Y2), ya/(l - Ya), 1/4/(1 - Y4), ete" and these are related as follows:
(11)
1/3 /12 = a 1 .."_X. I

= (~ , .... -.. -- (12)
1 - Ya 1 - lh 1 - x"
1. 'l'HIWHY -41
Y,I
(13)
1 - VI
At the distance n H.E.T.P. units the vapor amI distill[Lte compmli1;ion (JIll)
would correspond to eqmLtioIl (8). Thus again, if a packed eolumn has the
equivalent of n theoretict'11 plates, It test under (.otall'rfiux will give prochwt
composition Yn when the still coll1poilitiCH\ is x".
There has been some c:oufusion regarding the lise of equM,ion (3) and of
the equations:
!ill Xl
(ill --_.-_- (14)
1 - ?III 1 - ;'(:1
:l'n ;f[
an- 1 -- -_..--,
(15)
] - .1~n 1 - :l'1
BqmLtion (14) lLpplies 10 the separation in the column itself, or any sect,ion
that ha.s n plates, while eqna.tion (3) with the exponent, n 1 applies to n +
eoillmn and still-pot system in which additional RepartLtion equivalent to
one plate is achieved between the still pot lLnd the base of the column.
The Fenske forlll (15) is m'itten enLirdy in terms of liquid composition (no y
1erms) lWcl hmwG im-()l\'(~s alowcr power of a than (14).
(4) Determination of Plates by Simple Graphic and Tabular Methods.
Simple graphie and l.al>11lnl' methods have also beell developed for rapid
routine calculation of H.RT.P. These are pnrticulal'ly useful when the
number of plates is large. The following example using bellzene-carbon
tetrachloride rnixt.ul'()H will ilhu-;trn,te the method. Figure 11 shows the
vapor-liquid equilihrium dia.gram with theoreticnl plntes stepped off from
the (lomposition a: = 0.05 to:l: = 0.90 (mole fraction ct1rbon tetrachloride).
These st,eps 111'8 lllimbul'cd beginning wit.h zero for the lowest value of x.
Figlll'e 12 shows pnlarl!:ed sections of the same graph. The marleed vltri-
ation in composition (\I!nnge pel' plat.e is clearly brought out. Such a graph
is used by reading off the plate numbcrs corresponding to the still and dis-
tillat.e compositions, and subtracting the smItHer plate number from the
larger. Inclusion of the refractive inuex c:urve in Figure 11 mILlees it
possible to proceed directly from refmctive index to composition to plate
nnmber. Large-solLIe graphs are desirable unless only a rough approx-
imation is needed. Interpolation to obtain fractional plate numbers
(tenths) is common practice, but in must CH,ses the results of a test 1:'L1'e not
r(~produciblc to better than one-half plate.
The entirc operation is facilitated by plotting plate numbers directly
against composition and refractive index, as in Figure 13. Thus if still and
distillate compositions are, respect.ively, 0.70 and 0.20 mole fl'aetion carbon
(rtrachloride, the corresponding pInt.o llumbers H,re approximately 22.2 :tne!
42 A. AND Ji;. nOS.E
Ij.9. The munllm' of plate~ is 15.:1. If t,ite refractive indiees of the samples
from anothel' LnsL ()f n ('o]\lIl1n arc 1.4700 and 1.4900, the eOJ'reHponding
plate Il1l.mbCl'R a.re 21.9 and IUl, giving LS.1l pinks.
Thn lise of graphs iH (mLil'cly (~]imill:LLpd hy Llw La.hul!],]' method, illus-
trated in Table V. Tlw refm.dive indices 1.4700 and 1.4900 give plate
ll1uuhcl's 21.9 nncl6.9, iLS bd'ol'u. Values are given only to the nearest 0.1
0.9 . - - - - - - - - - - - - - - - - - - - - - - - - 1 i t ' - - - - , 1 . 5 0 0 0
35
Q;:
0
CL
:3!
:::; VAPOR-LIQUID
w EQUI LlBRI UM
1.4900
0 CURVE:-.-....._..~
0:::
0 0.6
--' ~....._ x
l: '- 45° DIAGONAL w
u o
« z
0::: 0.5
I- w
w >
I- i=
z 0.4 CJ
0 1.4BOO C2
[I)
lJ_
0:::
« w
u a:
z 0.3
0
i=
r COM POS ITION - REFRACTIVE
u INDEX n~5 CURVE
«
a: 0.2
LL
w 1,4700
...J
0 0.1
::;;
c_-~ _ _...J-_~-_L-_L__ _L_ __...J~_~~_...J_ __...l14640

o 0.1 0.2 0.3 0.4 0.5 0.6 0.7 O.B 0.9 1.0
. MOLE FAACTION CARBON TETRACHLORIDE IN LIQUID
Fig. 11. Dingram for 1.heol'eti(:ld plrLtos at. totall'eIhl:x for bnnzmHl- em·boll tetl'l1ehloririn
mixtuJ'PH.
plnte, so that discrepnneies of this magnitude may oceur in eomparisons be-

tween Table V and tho gmphH of FigureH 11-1:3. There is no need for
gl'c!tter preeision becnnse of the inherent lade of reproducibility in H.lD.T.P.
measurements.
Whenever a uonsidol'ttble number of tests [1),0 to be made with a pal'ticuln,r
mixture, it is advantageous to prepare graphs or tnbles on the above plan.
The constnlCtion of the vapor--liquicl equilibrium diagram, nnd the stepping
off of the plates, ete. can then be done once, with cons.idemble OH!'e, and
subsequent repetitions of these operations are entirely avoided.
(5) Calculations for Columns Having More Than 100 Theoretical
1. 'l'IIIGOHY 43
Plates. Willingham and ROl:;sinif':i suggest the trittI of the two following
mixtures for the full-seale testing of snch collLmns HnJ hiwc worked out
VAPOR-LIQUID EQUILIBRIUM LlNE~
~O~ VAPOR COMPOSIT~_-/-/_;:""SJ-~--2--
,\@)0,®0"0;
o "/ / 0.10
"-...._ LIQUID COMPOSITION \ 0.10/
45" DIAGONAL
PLATE NU MBERS /
~l~ZSZ\ZSlSZSZ\7Sz0s~~---
@/ ',@ @ ® @ ® (BJ @ @ @)/@',
/ , / 090
0.80 0.80 0.90
Fig. 12. Lal'gll-sel~le seetiollS of l"ignre II. Thcordieal platt's stepped-off frO!ll
0.05 to 0.00 mole fmetioll elll'bOIl tetmehloride.
mleflll equ(l,tiom; for them. The first is 2,2,4-trimethylpcntane :1ndmethyl-

cyelohexane, whose refractive index-mole frnetion relation at 2.5°C. is:
nD (mixture) - nl) (2,2,4-trimethylpelll,n,ne) = O.0244x' + O.0072(:r'r
where 'YIp is the l'dmct.ive index and x' is the llilile fnw,tion of methylcydo-
~3 Williugham and Hossini, .T. Be,~e(1rch N aU, Bnr. St(1ndard~, 37, 21 (104(i).
hexane 1ll the mixkn:c. The mnniJcr of thcol"ctienl plal;es-molc fr:wtiOJ
relation i::;:
n +1 = [48.07 + 0.13(t - 1(0)] log [(:1;/X')lw"a/(X/X')110t]
(l>ver the range 97 to lO1.5°C.) where n is the nnmber of pbCcs, tis tJI<
mCI1n temperature in degrees centigrade, fmd x and :1;' nre the mole frnct.ion:
in the liquid phase of the more volatile and less volatile components, respec·
tively. The preparation of the material for this test mixture is also de
36r-r---------------------------------,-----,
32
28
24
cr
w
OJ
::;; 20
::>
z
w
16
S
c.
12 PLATE NUMBER
COMPOSITION
8 CURVE +
PLATE NUMBER
REFRACTIVE INDEX
4 CURVE "'l
O~L-~--~O.72-r~--O~,~4---+--'0~,6~~-,~~--L-~1.0
1.4600 1.4700 1.4800 1.4900 1.5000
MOLE FRACTION CAR80N TETRACHLORIDE
REFRACTIVE INDEX
l"ig, 13. Benzene - C!~rbOll tetl'twh)oride plate Ilumbor

WI'HUS eomposition and refraotive iudex,
scribed by Rossini nnd co-workers in another publieation, The CqUlL "I

t.ions given arc not, n.pplic,a,ble unless t.he test liquids are pUl'iii{ld to the do
gree specified, auti in particular the refmctive indices must differ by til(
same 0.0316 units that arc involved in the derivation of the equntions.
The second rnixtll1'o is 2,:3,4-tl'imetltylpentalle (norIllal b,p. l1:3A()7 O(~,
and 2,3,3-trirnethylpentanc (normal h,p. 114.760"C.), The number 0
theoretieal plates-mole f'mdioll relation is:
11 + 1. = [64,95 - O.:3G(l - 114)] log [(:l;jX')'lfarl/(X/X')poe]
6< li'Ol'ZiHii, Glasgow, WiJlinghn.rn, awl l{OHHilli, .T. ReNeal'ch Natl. Rill'. Slmu/nl'J/8, 36
129 (HJ.1Il).
I. THEOHY Ii)
loyer the mugc 111.5 to 115°C.), where .1: an!l :1;' liro the mole fractions ill
tltl' liquid phase of the 2,:3,4 aud 2,::1,:3 isomers, l'eHpectively. AnldYHiH must
lIe by infrared speet.rometer, and Hnmples are prepared by efficient distilla-
t.ion of commercial mixtures of branched-ehain paraffins.
C. OPEHATION A'l' PAH'l'IAL UEFLUX
For pmposes of design a,nd opemtion, a good understt1nding of thp

1l1<:,thods for predicting the composition from plate to plate Ol' point to point
at partinl reflux is required. Both refiux rat.io and the number of theol'et-
ieal plates have an important bearing on the sharpness ~f separation
TAKE-OFF FOR DISTILLATE

NOT RETURNED TO STILL
I LINE FOR RETURNING

t DISTILLATE TO STILL
Fig. 14. Apparatus [0], J'!ltUl'l1ing diRtilJat(~ to stilL
acdnally resulting in ally distillation. In large-scale continuous distillation,

it; if.! neee"siLry to lllake calculations of the plate compositions at partial
rMlux in order to locate the proper point in the column for introducing feed.
In either continllous or lmteh distillation, the reflux ratio chosen is an im-
portant. faeto)' in determining the cost of operation. FinallY7 sinne reflllx .
mtio influellees the number of pbtes required for a given separation, it is
also an impori!tnt factor ill the eost of construction.
(1) Test Procedures. In a continuom, distillation with constant feed
composition, dnplicate samples may be obtained without difficulty, since
the eompositions of the still and product; liquids should not change with
time after the column is operating smoothly. Samples for analysis can
then emdly be withdrawn from any of the three lines, A related expel'i-
'In A. AND m. HOSID
mental proeedure for studying the portion of l1 eontimlOlls column above

the fm~(l involves the l'Otlll"1l of the distillate eontinu011Hly to the hottom of l1
batch-diHtil1n,jioll still as shown in Figllrn ] 4. This "frDflzes" the nom-
positions a,t, a, particuln,l' set of values so that replicate samples can be
obtained. It is essential to ensure that the product stream is thoroughly
mixed with the still liquid. When an adequate number of samples has been
obtained, the distillate stream may be collected in a receiver for a limited
time, and then again deflected into the still so Lha,t another group of samples
can be obtained, with different still and distillate compositions. This re-
turn to still procedure is not generally applicable to batch distillation.
In bateh distillation still samples of negligible size must be removed by
the procedure already described for total-reflux tests. Distillat.e samples
t1,re obtained directly. Because of the change of compositions as the dis-
tillation progresses, dupliol1te slLlnples cannot be obtained except by repeat-
ing an entire distillation from tile beginning (see page 5G).
(2) Basis for Plate-to-Plate Calculations. Sorel"O was the first to
show that It plate should not he expected to give as much separation under
partial reflux as it does undor totalreflnx, (won though other conditions are
identical. His method of estimating the degree of separation under any
reflux ratio by means of material Itud hea.t halances has been thoroughly
worked out for plate columns and continuous distillation. The same
principles apply to IXL(',ked columns and blLtch distillation, but two addi-
tional factors mnst be inelndecl in a careful theoretical analysis. As a
batch distillation proeeedH, there is it cOlltimu11 change in composition at
[LIlY POillt in the column as the more volatile component is gra(hmlly re-
moved. This adds to the syst.em as wOl'kr~cl out for continuous distillation
a rate of change variable HO tlmt appreeiably dift"erent relations may occur
when hol(lup is not negligihle (15ee p~lge 5(\ bolow). In the ::;e('.o11(l place,
in pnekecl columns tIle (Ji"ad?wL change l)f ('.omposition up the column re-
quires It cliffcrentinl mmlYHis whioh is lWt:miH1Lrily' more eomplex than the
corresponding ease for pInto eolumnR, for \vhieh the theory may be worked
out step by step, corrosponding to iJw distind ehange in eomposition from
pln,te to plate.
A number of different ways of npprmwhing this problem have been pro-
posed (Seet. IV) but there is much evic1ellc:e thnt the relations derived for
plate columns arc reHi:lonably satisfactory for paeked columns. For these
reasons and beeanse the explanations and equatioIlt) arc b118ie and relatively
simple, the next paragraph c:onsiderr-; eontinuous 01' Hteady-state distillation
in plate columns. It is further assumed that the eolumn is adiabatic, thnt
.. Sorel, COTnpt. rend., 1Q8, 1128, 1204-, 1317 (188\); 118,1213 (18\).1). Theso were
summarized and expanded in La Rectification de l'AIr-ool, Gauthier-Villars, Pa.ris, 1808,
and in La Distillation et Rectification Ind'llstriellcs, Gauthier-Villars, Paris, 18m).
1. 'rHEOHY
the heat of mixing in vapor and liquid is negligible, Hnd tlmt the substances
involved have similar thermal properties. These further simplifiClttions are
frequently applicable and me known as r,he uSllal simplifying assumptions
(USA). The reasoning is bttsed un the early work of Sorel, and the simpli-
fications intruduced by Lewis 56 and McCabe and Thiele.
(3) Continuous and Steady-State Distillation. Steady-sttLte distillation
may be defined as any distillation in which thoro is no dmnge in col-
umn compositions with time. Examples are operation at total refiux 01'
ret.urn-to-still operation with a bat.ch column. A more pert.inent example is
ordinary continuous distillation.
Plate-la-Plate Calculations by Method of Sorel and Lewis. With a total
condenser in a continuous distillation at partial refiux it, is legitimate to
assume that the composition of the vapor leaNing the top pinto is the same
as the distillat.e eomposition:
(16)
With [I, partial conuenser the composition of the vapor from the top plate is
determined directly by experiment. In either c[tse it is assumed that the
vapor leaving the top plate is in equilibrium with the liquid on the top plate.
When the relative-volat.ility equation applies, this liquid composition :C t can
be calculated from the relation:
1 l1t
(17)
a 1 - !Jl
For any binary mixture, :C t el1l1 also he [ounu gl'nphically from the vapor--
liquid equilibrium diagram. The composition of vapor coming from plate
t :_ 1 may then be found from the opl'J'ating-line equation (see below):
L D
11/-1 = 11 Xt + Tr :~D (18)
The liquid composition X t - l mlLy now be found as before from the vapor-
liquid diagram ~1nd then used in another uperating-line equation:
L J)
111-2 = V Xl-l + ·v a:D (19)
This process m!1Y be repeated as often as need(~d to obtain Xl-2, Yt-a, X1-:l,
etc.
In order to derive tlw OttuM,ion for tile upel'nbng line, it is assumed that
une Im~ a plnte colulllll lOwl Lhat, Lhe refiux mLiu is LID, in which L l'epre-
~ij Lewis, J. InrI. Eng. C'hem., 1, 522 (J \)0\)); indo En(!. Chem., 14, 'U)2 (1922). Walker,
Lewis, McAdallls, and GilIihtlld, P)'-inciple8 of Chemical 11Jn(!ineel'ing, 8rei eeL, McGraw-
Hill, New York, 1987, p. 553.
,1~
sents moll's l'efiux n.nd ]) the moles distillat.e pel' unit timn. (Reflux ratiu
may he measured by :t variet.y of mnal\~, a" de:·wrihml ill Chapter II, page
251.) The moles of Vl'por pcr unit time l.Ll'e then V = L + /). By tL
TOTAI_
CONDENSER
REFLUX OR
OVERFLOW
TOP PLATE (I)
NEXT TO TOP PLATE (I-I)
(1-2) PLATE
II'jg. ]!"i. Diagrllm of pl:d,e-Lo-plntn e:doulatiollll of COllljlOHitilJll, ok.

by the Illtl\,hod of Sorel, Lowis, am! J.VIeCahe lInd Thie]('..
material balance about the top plate aud eoudellser (a:,; in Fig. 15) it is
obvious that for nny one eomponent:
moles of COlllpOIH'Ilt. mlt,oJ'il)l!; llloll'S of OOllIPOlH'll t. llloks oj' OOIll]lOlJf'llt
tnp pla\,o t in VllJlor frulll plttLo bwing top plat.o ill -I- Inl1villg ;;,YHtOIl1 in diH-
helow, t - .I ovorflow to plate below WlI1te
Expres~etl in symbols:
(20)
where yl_l is the eomposition of the V!LpOI' from pli1t.e t 1., XI the com-
position of liquid from plate t, and :r[) the di::;tillate eomposition. This it'
usually written [)'s equntion (18). A similar material halance may he made
about the two uppermost plat.es, t find t - 1, and the condenser. In thi~
case:
moles of (:olllpOlleut (JIItel'ing lIl()h~H of compunent molos of compouent.
t - 1 plat.o ill VfLpOI' frolll It'llving I - 1 pJatll in + leaving syst(JIU in dil:!-
t- 2 OVI'!'[low to t - 2 tillatu
01', in symbols:
(21)
1. TIUWltr 1!J
III a ::;illlilar nHLllllcr it may be showlI that:

YYt-3 = L:rt-2 + D:"Jj (22)
and in general:
(23)
the opcratina-l1:ne equation, for any purtion of a c,outinuOllH-plute t'.ollllnn
ahove the point of entry of feed, or for any portion whatever of a batth-
plate (\Olllmll operating under steady sinto at finite reflux,
AR RllO\\"ll above, this type of equation mu.kes it possible to ealeulute t,lw
('omposition of the vapor from H,ny partieuln,T plate from thnt of the liquid
from the plat.e abovE', if the distillate .eOluposition and refinx ratio are
knowll. A numerie:Ll example is given heluw in this seetion (page 1)9). The
alJOve derivat.ion assumes t.hat. there is no hoat. loss fnnll the eolUlllll and
tJHtt the tlwrmnl propertiefl of the mixture being cli:,;tilled are such t.hat n. ma-
tel'inllmhuw() will also produce It heat bnhlll:e, ant! mnko L, = [;1-1 = £1-2,
cte. aud TTl = V,. 1, ok. If the t,herrnnl propcrtiuH :-;neh as heat of mixing
an~ not. negligible, and the ht.cnt. heats and Hpeeifie heats of the materials
are not eloHely RimilaJ', t.he derivatioll mllSt. be hased on n hent balance and
the more (~Olnplieated l'elation:,; ()L'iginn,lly derivetl by Sorel 111llHt be uscd,
Lewis pointed out t.hat liquids could be dltssified as either assoeiating or
nonassociating, and that when :1 mixture contained only liquids of one class
the assumption of eqllimola,l overflow (L t = 1.1- 1 = L t - 2 , ete.) was reaSOll-
ablel. This has bc(m subHtlmtin,t.ed by ext.ensiire use of the opm'atillg-line
equnLion:
L D
Y,,-l = 1T X" + ofT Xn (24)
in the design of eontinuous-plate columns. ·When t.he liquids in a mixt.ure

arc no\; of t.he same c:lass, it is often possible to use the above equat.ion by
assigning fictit.ious moleeular weights so that L t = L t - 1 , ete, This requires
recaleulation o{' the vapor-liquid equilibrium diagram in terms of the fic-
titions molecular weights, Graphic procedures have been developed by
POllehon[,7 and Snvarit5S in order to apply the met.hod of Sorel when heat.s of
mixing, smlsible heat.s, and Intent heats VI1l'Y so much t.hat the simpler
methods based on equimolal overflow are not. Sittisfactory. Randall nnc!
LOllgt.in 6U have generalized this procedure.
It is import.ant. to recognize t.hat for tottLl-l'dlux condit.ions D 0 and
LITT = 1, so t.hat. the operating-line equation beeomes:
YlI-l = Xu (25)
67POllcholl, '1'eehldlJ1W nwl1erne, 13, 2U, 55 (11):.l1).
SaVl.Irit., Chiltvie ,t -intiu,stl'ic, 9, Special No., 737 (May, 11):.l3). A lllOJitiCl.L~i()n of
'~ij
Savl,ri1.'s lIluthml Wllfl proposed by SilllnHettlt, Chinvica e i.nrllwlria Mil.an, 27, 157 (1945).
,Q Randl111l.1nd Longtin, Ind. Bng. Chem., 32, 125 (1\),10), and pl'ecelHng papers,
50 A. AND :m. HOSE:
McCabe-Thiele Graphic P1"OCed1.lTe. The following equivalent graphic

method 60 may also be used to calculate p\ttte-to-pbte compositions ~tt pal"-
tiall'eflux (Fig. 1(3). The first step is to draw OIl the va,por-liquid equilib-
rium diagram the so-cnlled operating line, with equation (24). This will
VAPOR ENTERING CONDENSER
Y, -I f - I PLATE
a::
oe.. YI-~ FROM t-2
~
!?;
u
>
::;;
z
o
i=
u
<t
0:::
LL
W ,
-I
w' z
o I-
W
I-
o
::;; <t <t, i=
-I -I, Of)
e.. e.. o
a. e..
... ,
I g ;3:
u
::;:, ::;:
o o
a:: a::
lJ.. lL
Cl
::;
o
::3,
Xi
li'ig. 16. Graphi<J constl'uetion of McC!tbl~ and Thiele. 60
pass through the point d(a.; = Xn, Y ::== Yt '= :t:J)) and will have the slope L/V.
Its intercept on the vertieul Hxis (x = 0) will be:
(26)
so the lille can simply ue dl'UWll Lhl'ough the jJUilltH d aud 'Y I. This line re-
lates the liquid cOlllp<lHition, (;c It ) , from ILllY pinto t.o the composition oj' tho
vapor from the pbte below (y" -1). Thel'ei'ol'o tho value of y oorresponding
eo McCabe alld Thie1e, Ind. Eng. Chem., 17, 005 (HI25).
1. 'I'HEORY 51
to the intersectiun of the operating line with the vertical line :1: = :I;, gives the
composition of the vapor, Yl-J, leaving the t - 1 plate. The nnmposition of
the liquid from this plate may now he found by simply rea(iing ael'I)SS from
Yt-l to t - 1 on the equilibrium curve and then down to J'/_I' In a similar
manner the following are located:
1/t-2 - composition of vapor from t -2 platn
;(:/_, - oomposition of liquid from 1-2 phLte
!/1-'1 - compoHition of vapor frolll t-3 plate
;1"-'1 - oOJl1position of liquid from t -3 plate
ltnd so OIl. Thus, if still composition in a batch plate or packed column is

:1.:/-:1 and distillate composition is Xn, it is obvious that the system as a whole
mUHt have the equivalent of four perfect plates.

In actual practice the number of plates is determined simply by drawing
0
the operating line through the point d on the 45 diagoD!Ll representing dis-
tillate composition, and through the intercept (equation 26) and then
stepping off the eonstruetion of d, t, tl, t - 1, t l - 1, etc. until a value of x
near the Rtill eomposition is reaehed. The equivalent number of theoretical
plateR is the number of stepH less one if separation equivalent to one plate
is assumed to occur in the still itself.
When dealing with composit.ions near zero or unit,y, or with mixtures of
relative volatility near unity, it is usually necessary to plot the x versus y
diagram on an exaggerated scale in order to preserve reasonable aceuracy.
Sometimes mere increase in the scale is suffieient, but in other cases plotting
on logarithmic paper may be helpful. The eqtLilibrium curve may appear
as a straight line, and the operating line as a curve on such graphs. Ran-
dall and Longtin 61 have prepared graphs of y minus x versus x for use with
mixtures having relative volatility very near unit.y.
Minimum Reflux. It can be seen from inspection of Figure Hi that if the
reHux ratio, LID, is made higher ftnd appro!l.ches total reHux, and if the
ratio DIY and the int.ercept Yi approach zero, the operating line will
approach the 45 0 dingonal and the entire operntion approaehes that for
total reflux.
For any particular values of still composition, x., and produet composi-
tion, Xn, there is also a minimum reflux ratio, R Dmin ., corresponding to the
operating line which intersects the equilibrium curve at x.. Fat, example,
such a situation would arise if still composition was Xt in Figure 16. Lower
reflux mtios than correspond to the operating line shown cannot give a
separation as great as from Xp to Xi, because for such lower reflux. ratios the
opemting lines int.ersect the equilibrium curve to the right of ti • It is clear
that the stepping-off process cannot be continued to the left of ti , and that
Gl Randall and Longtin, Ind. Eng. Chem.~ 31, 908 (1939).
52 A. AND E. nOSI~
the l\OlTCtlpouding; still compot>itioll it> tltn minimum possible with the given
vlLpor--liqnid equilibriulll enrve, product eomposition, awl operating; line.
Theoret-ieally, !Ul infinite nmnbet· of plat,(·s is rnquiJ'ml (,0 re:wh sueh:t min-
imum composition. Actunlly those plates operating on material of com-
position neal' the minimum accomplish li(;t]e or no sepnratioll, and inerefLse
in the number of plates does not improve the scparation.
The minimum reflux ratio for the enriching section during <listilbtion of a
binary mixtl1l'e when the usunJ simplifying assumptions l~l'e n,pplieable is
given by:
This ma.y be derived by noting th:lt in Figuro lG, if :c s = :c[, if, = ~Vt, and the
slopo of tho nprraling lin(~ i,.,:
(L/ J))IJI.I'1.
(L/f))mill. + (DID)
which then mny be solved for ilVlI/iit. to obLain the eqm1tion shown above.
Smuker G2 has devised a llomogl'l1,ph fO)' e:,;tim:ttion of minimum reflux and
theoretical plates eorrcRponding to tho (K[wtl,ions of the preceding discus-
sion.
The caleuliLt.ic)ll of minimum l'dhlx fo1' Ilw eontiulIOllS diBtil1ation of
multicOluponent mixtllrm; Ims been Liw snhjm:t of oxl',ensive study beeans(:\ of
its practical impol'tanee in inrlm;1;rinl <listiHation.li~
Procedures "When U8ual 8implzfllinU ..:J:S811'flbJJliOiIS Arc Not Applicable.
The original pl"lleednl'c of Sorel mm be eal'l'iect 0\11, hy It gnLphio eonstrnetioll
analogous to the McCabe-Thiele methocl jm;t descrihed. However, when
the column is not. fldi:Lbatie Ul' tlw (JIltnt' llsual simplifying ItHSlImptiolls arc
not applicahle, thc vnllles of Land Yare lW louger equnl 011 till) diffcll'(mt
plates throughout the f:olumn. Their aetnal values d<~p(md 011 the thermal
qnantities (heats of vlLporizntion, mixing, etc.) and, fiB it r8slIH, the operat-
ing line is curved; otherwise the Htepwii:1C procedure is unehanged. Modi-
fied but. equivalent gr:tphie methods a,l'e lllHch use-d u•l for such (·.~11enhtions.
SlllC)]WI', Ind. Eng. OIwlI!., 34, i30!! (1!)·12).

62
G. G. Brown !llld Holeolllh, 1'(:ll"Oir'/(1II. Buyr., 11, ]\"0.1:3,27 (W·lO)i 12, No.1, 14.[)
63
el ~eq. (1940). Colburn, 'l.'ranN. illll.ln,~t. C'WIII. Engr8., 37, SOli (1041). GillilILIHl, Ind.
Eng. Chem., 32, 1101 (10,10). Helwihl\1 lind M()nj'I'()~B, 'ibid., 38, 2ti8 (lOtiO). lJlIdnl"e
wood, J. InNt. Petroleum, 31, HI (1!N5); 32, lilli', 01·t (l!J4()). Sno niHo: j£l[miRtel',
Trans, Am. Inst. Chem. Rn(Jr.~., 42, Iii (lH40). Hunllllcl, ibid., 40, 445 (1044). .Tenuy,
ibid., 35, 635 (H)39). Gilliland and Hood, Ind. E'n(/. Clwm., 34, 551 (1D,J.2). Hognu,
ibid., 33, 1132 (H141). Jenny and CiC'nioso, ildd., 37, ulin (11)45). Knl.·lwt':-:ky, iiliri.,
34. 839 (1942).
53
]i}'jlwtions jor St,ripping Sec#on. III the sectiull of a continuous column helO\\" the
(,Iitry of the feed, tlin relation het\\"een the cOlll]lo,;itioll of the vapur coming up to a
plate (UH'- ti Rnd tlw h[\uie! (In tt pht<, (xm) is.l!;iven hy the cquation:
L 13
71",-1 = -(; XIII - ,; Xn
where L, V, and B aw, respedivnly, the mole,; IJf overflow (reflux), Vt111I))", and
pror\twt withdrawll :tt, the bott()m ()r tho (:Ollllll11, all POl' uuit tilll(), and :l:IJ i" t]!(~
(·ornpoRjj,ioll (,f 1h,· laHt, FOI' tlw bott.oill platt, 01 :L "oJulllll, thiH k
in which y" awl :rc are the (JolllpositiolJS of the vapor from the still and the liquid in
the Ktill, reHpecti\,ply, ami J,It if< the COlllPORitioll of the liquid on the fil-st l)late. The
valll(~~ (If x" an(l :rH \rill be identical since thr, prodnct is being withllrawn frnm the
~till. The vnlue Ill' II, is obtainable from:
01'from the vapnr--liquiil eqllililll·iulll diagram for the mixture. The only rcn1l1iniug;
unknown if> ;Cl, so its value Can be calclllatf1d ~Llld the proeeRS repeated for the plate",
above to obtain values of X2, :r3, etc, A graphic construction lllay be used, and IL
[arm of the Smoker equation cml also he workc(l out for these operating conditiol\s
ILwl used aH cleR~ribed nn Lite following pages.
Application fo Mllliicomponent Systems. The procedures just discussP(1
a,]'c cntil'(~ly valitl for multieomponent mixtures, hut application is compli-
cated by the difficulty in finding .&,,-L whenYn._l is known. This arises be-
('ause in Illlli!.ieomponcnt mixtlll'os t.he ntille of lin will in gmlCml depenrl
OIl the concentrations in the liquid of' a.ll j,he other components as well as on
x". Thus, a series of equilibrium eurves or equivalent clat~t nre required,
and trial and error, or specially deVeloped methods must be used to find the
eorreet solution. fJ:J.llr.
Smoker arl;ll Other Equations. If the Illllllber of plates is large, or if very
l1umy McCabe-Thiele determinations are necessary, the gra.phic procedure
beeomes quite laborious, so that It number of algebraic methods have been
devised. The maHt straightforward of these is the Smokerl16 equation,
0·1 POW:hOIl, 'l'tchn'ique liIorlC/,/W, 13, 20, fi5 (Ul21). Savnrit, Chimie .. \:; indus/rie, 9,
Special No., 737 (May, 1!l23). Thiel~, Ind. En(/. CltCln., 27, 302 (1935). White,
PetrolculI! Refiner, 24, No.8, 20!li No.9, 357 (1MI») .
.. Robinson nud Gilliland, Ellllnrnt8 of Fractional Distillation, 3rd ed. McGraw-Hill,
New York, 103!), Chapter 15.
60 Smoker, Trans. Am. Inst. (,helli. En(lr,~., 34, 1115, 583 (HI3S).
54 A. AND E. ROSE
which is merely the algebraic 01' analytical equivalent of the McCahe-

Thiele p;mphic pl'ocednrc. As originally ckvelnped hy Smoker it is 1I5cflli
only when relntive vobt.ilit,y may be a.~:;\lnwd t.o he constant. The same
method mn.y he l),pplie(\ to obtain eq1U1tions f()\' any en.se i.n whi.eh an {tlge-
bmic st.u,tement for the equilibrium curve is known [mel can be solved for x
and y. 67 The derivation of the Smoker equation mlty be simply illust.rated
by the example of a column with one plate and a still operating with diH-
t.illttt.e return (Fig. 14). With the same reasoning used for t.ho Lewis pro-
cedure, one obt.ains equation (J 8) or for this instanee:
L D
Y8tiU = V xplate + V Xv
It is also true thilt Xv = Yplatc' Since the vapor from the plate (Yplate) is ill
equilibrium with the liquid on the plate (Xplatc), these two eompositiom; are
related by the relat.ive volatility equation:
= a--X-I'
I - :CIl
so that:
This vallie of ;;t:jJ may be subst.ituted above to give:
Y. = t [a(1 - ::)--+ ;,J + ~ XD
Thus this equaj:ion gives directly the composition of vapor leaving the still
in t.erms of t.he dist.illate composit.ion, XD, the reflux ratio, L/V and D/V, and
the rel:.Ltive volatility a. No graphic operations are necessary. Since the
vapor leaving t.he still, Ys, is in equilibrium with the liquid in the still, x,,
these arc related by:
y. Xs
=a --"'---
1 - 1/s 1 - X.•
so that:
X
1/.
= ---------, -_.
8 a(1 -- Y.) + 1/"
On substituting the value Ys in the above oquation the still and product.
compositions are related by:
67 Olark, Tram. Faraday Soc., 41,718 (1945).
1. 'l'HEORY 55
This final equation gives the still composition directly in terms of the diH-
tillate composition, without graphic operations. If the symbol <PI> is Ul-iCU
to represent the quantity:
the equation appears much less formidable:
<PD
:C. = ---- ---
aCI - <Pn) + CPD
For a continuous column the feed-plate composition replaces the still COIn-
position in the above reasoning so that:
:l:p =
a(I - ¢D) + CPn
In spite of the complexity of the relationship, Smoker has developed the
equation for any number of plates in a rebtively simple form, through the
use of ingenious mathematical transformations. The most useful form for
calculating the number of theoretical plates is:
n - 1 {[XD - kHl - M(x. -

- og(_[x, _ k][1 - M(xn _ k)]
k)l}!logmc
~
2
in which:
"M = ·~!c_:(:__a_-_l)
- a - rnc 2
and Ie is the root between 0 rtnd 1 of:

m(a: - 1)lc 2 + [tit + b(a - 1) - a}le +b= 0
with m = RD/(RD +1), RD = L/D, b = xD/(Rn +
1), and c = 1 (a - +
1)1e. For calculating still composition from product composition when Ot.,
n. and RD are known, it is convenient to use the form:
50 .\ ..\Nfl I':. TlO:-\i':
Bi::;c:si6~ IttL::; fUJ'llisilod 11, llomograph fot· finding 11: in Smoker's equatioll. It
can readily he shown that ILl'. tota.! I'eflux the Smol,er equation reduces to
the Fenske equation.
Other equations simihtl' in objective to the Smoker equation include the
following. UIlllcl'\\'ooc[6u derived relatiolls for the case in which t.he dis-
t.illate is lleftr 10017,) purity, and for other special cases. Thomson and
Beatty 71) t'\,nd Clal'k 71 have also presented equat.ions applicable !lllcIet·
special eircumstanecs. More l'(l(:cnt n.rticles by Underwood give equat.iolts
tha.t are mol'() genet'nlly Use1'Hl. 72 Amundson, Eshaya, Harbert, a.nd Stop-
])e17 3 have eaeh dev(~l()ped a syHtcm of theoretical plate equations by reasoll-
ing along lines similar lio Smoker but using different ma.thematical t.mns-
formatiolls. li'lL:u.;('n 7.\ hus worked 0111, n nomogmph for ealcuhtting number
of theorctil:nl plates and l'etillX mUo, bitt using u different approach from
Smoker. All these relatioIl::; are haspd un the stepwise graphic and nlg(~
braic prOCedlll'e of SOl'el as simplifietl oy Lewis and MeCnbe and Thide.
Various ot}wr methods of eolnJ11n evaluation have also lJeCll derive(\.1f)
(4) Batch Distillation. Bn.1.l'h-dist.ilhLioll apparat.us haH a.lmost always
been tc~ted :,I,t; total reHux ruther tlmn at rml'Lial reflux (page :32) and
there ha.s been but little study of the composition difi'el'cnces to be expeeted
under the hLtcl' circumstances. Collins and Lnntz71i luwe chosen to ex-
prc~8s the composition diffel'l'IlCeS bet;ween still (mel product at partinlrofillx
in terms of plate eq1dvalents, [L standard of eo]umll efIicieney first snggcHt.ed
by Baker, Bnrkenbus, and Roswell."/ Plntn equivalents are a quantij.~'
numerically equal to the number of theoretical plates uperating at tot,n.l
reflux which would give the separation obtained whell the column is opernt-
ing at. a finite reflux mtio. The proeednre iH to Htep off bet.ween the vapor-'
liquid equilibrium curve and the Il5° li11e the number of phLtes bet.ween
still and distillate compositions in eXHeUy the same way as for operation at
total reflux. This simple, direct method of evaluation can serve admirably
as a basis for comparison of colmnn8, but is not. useful for predicting sepul'H.-
Bisesi, Petr'oleWI! Uejiner, 22,236 (1943).
68
Underwood, 7'1'ltn.~. Inst, Chem. En(/1"~. London, 10, 112 (1932); .T. Soc. C/WIII. Tnl/.
60
London, 52, 223 l' (lU33).
70 1'h()lllHOn amI BC!Ltt.y, Ind. Rnll. C/WIII.) 34, 112'1 (1942).
Clal'k, :tran.s. F'al'alil!!I SOG., 41, 738 (JH·15).
71
Underwood,.T. inst. Pctroknlll) 29, 1-1-7 (HHH); il)id., 3D, 225 (1!)44).
72
73 A.mulld~on, 'l'mns. 11m. In.~t. Clwm. Rn(lr,~., 42, \l30 (lH-iU). E~h!tya, ilJid., 43, /iii;'
(1947). Hal'bert, Ind. En(/. ChWI., 37, UU2 (Hl45). St.oppel, ildd., 38, 127l (19,16).
74 Fultsen, Ind, Eng. Chem" 36, U8 (1IH4).
7. Kirschb!LUlU, Z. Vm·. deul. Iny., Vm/(!hrenstech., 1943, No.1, 15. Millul', 0., Bull .
•~(lC. chirn, Bely" 53, 97 (1944), 1I1lt! ullrliul' artiel!IH. Nikola(1VIt, N eftya/uul Klwz., 24, No.
12,33 (1946); C/WIt. AbstmctN, 41, 5758 (UJ47). Be!) also 1:b,Uon IV.
76 Colllllsllud Ll1ute, Ind. gll!{. (.'}WIIl., An!!l. Ed., 18, tJ73 (1940).
71 Bakel', BarkellbuH, and HOHwl~ll, Inti. Itn(j, Chem., final. Ed., 12, ·H18 (1!HO).
r:~
L 'l'lU;OltY dl
Lion umIel' new conditions, '1'0 Lwoid the sa,mpling difficulty inherent ill
batch distillation because of the dmllge of ('.ompositions with time, Collins
and Lantz have used the method of retul'lling the distillate to 1;he still (Fig,
14) to "freeze" the cornposiliolls. However, this method gives results
somewhat different from snap slLmples taken during a regular batch dis-
tillation because the return to still procedure involves steady-state opera-
tion while batch clistillfLtion does not.
Colburn [Lnd Stearns7S have pointed out that the Sorel-Lewis and :Me-
Calle-Thiele procedure cannot be depended upon to estimate the nllmilm' of
theoretical plates or composition differences for ba,tch operation at finite
rl'llux. This is because of the effeet of constantly changing compositiollfl
and holdup on plate-to-plate composit.ions, even though the usual simplify..
ing assumptions are valid. This may be shown by the following reasoning.
For the top plate and condenser of a continuous distillation in which the
1181.lHl simplifying assumptions are valid, the Sorel-Lewis equation is:
VYI-l - LX t - DXD = 0
For the corresponding caso of a bat.ch distillatiun:
VYI-l - Lx, - Dxv = -Il,!:lx,

ill which H t is the tot!d holdup OIl the top plate and ~:!;, is the change in
(~(Jmposition on the top plate during \ he interval over whieh the distillate,
/), if> removed. Similar equations apply to other plates. The additional
term necessari~y results in a curved operating line for batch distillation.
As a result it can be predict.ed that a given number of theoretical plates
will result in n different separation in batch distillat.ion from that in steady-
state or continuous distillation.
In gencral, the eqnation for the opcmting line in batch distillation will be:
VYlI-1 = LXn + Dxv - i1n_l~X"
ill which I{n-l is the total holdup in the part of the column above plate
n - 1, and .6.Xh is the change in composition of this material during the
interval over which D is measured. This ~ffect of changing composition
hus an important bearing on the role of reflux ratio and holdup in batch
distillat.ion. When holdup is negligible, or when the mte of change of com-
position of the holdup is small, the simpler bws of steady-state 01' cOl1t.inu-
ons distillation may be adequate to predict the course of a batch clistilla-
1,ion. The study to date of the much more complicated relationships that
!Lpply in untch distillation with finite reflux anel large holdup leaves little
doubt that in Latch distillation [tppl'eciable holdup Hometimes gives n
78 CoJbul't\ u,ud t:iteu,I'IIS, Trans. A.m.ln,~t. CI!lim. H:ngl's., 37,291 (1941).
58 A. AND E. ROSE
::;hal'per ane! better separation than would be expeeted from the hnvs of
('()ntirnlOlls distillation. This is in addition to the improved separation
Jlml' the bc>,ginning of a batch distillation commenced at total reflux. In
1.0.---------------------------...
H" 57.6%
0.9
\),., 1
,<_'v '0101 I
0.8
~'<J "':!:,,rp / , I~
'V ''(o~ I I 1m
,,'7 ' , I IN
,,0 1 I I":t:
I 1 , I
W
0.7 I I I I
I- I I I I'
<t
_J I 1 J ,,'I
_J
I I I
~ 0.6 I I~ ,~
o J~ '<t /rv
~
I G:? l!t I';;':
u It;; /~ I:; /:t:
~ 0.5 I" ':t: / J
I:t: I / I
z I H"2B.8%
o II II / ,..!-. _.-.-.
1=
u
I / 'j' /' /
;2 0.4 I 'iJ1·~·I-·,L.-.f_.-.
"-
'w
..J I
I I
,/
'f""'" Y/ I
/ Q:-'?-
,,"...0
o
:;;; I /
i"), ' I
I
v
~+
0.3 I / I /1 I </}
I ,I / / / ....'<:'
II I I I r,.\
Ii / / /. "lv r-------------------l
0.2
/ r;-~ LEGEND
/.>.: H" HoLDUP AS MOLE 'I. OF CHARGE.
/ ",Q CHARGE COMPOSITION, MOLE % MVC:
I ",«;
G:> 9.6 TOTAL REFLUX START-UP
0,1 25 TOTAL REFLUX START- UP
50 TOTAL REFLUX START -UP
9.6 FINITE REFLUX START- UP
o 0.02 0.04 0.06 0.08 0.10 0.12 0.14

MOLE FRACTION MVC IN STILL
Fig. 17. Distillate eompositioll versus still composition from eLtlcuhtted bateh·
distillation lIurves.
l\OIltl'Ul:!t thel'(j LL1'll ail'lo combiuatiuns uf opcl'atiug couditiow, thut caust:

larger holdup to give poorer separations in blLtch Jistillution.
The l'ehttion between still and distillate composition for various types of
operation may best be shown by graphs such as FigUl'e 17. This shows
1. 'l'HEOH¥ 59
still composition plotted u,gainst distillate composition as calculat.ed for a

:-;teady-state distillation with reflux ratio of RD = 4/1 and batch distillation
with the same reflux ratio [mel v~nying proportlons of holdup and cha.rge
composition. The limiting case of total reflux is also shown. Entirely
different curves are .obtain(~d with total reflux start-up Hnd finite reflux
start-up in batch distillation. The actual values in batch distillation at
finite reflux depend upon the previous course of the distillation, and fol' the
curves given were calculated by the stepwise plate-to-plate process de-
scribed on page 12(1. These ca,lculations involve numerous simplifying
assumptions but they do take into account the change of composition with
time in a batch distillation. Relations similar to those shown in Figure 17
have also heen observed in experimental curves. 79
3. Still and Product Composition Relations
A. UNDl~R l'AH:rIAL REFLUX
The preceding part of this Chltpter has dealt with met.hods of determining
the number of theoretical plates when still and product compositions are
known. The same methods may be tlsed to estimate product composition
from still composition, or vice versa, when the number of plates is known.
For steady-st.ate dist.illation, or for bateh distillation with negligible holdup,
the McCabe-Thiele and Smoker procedures are the simplest that may be
used for obtaining still and distillate compositions from one another. The
results of a number of calculations of this kind are shown in Figures 18-20.
For example, in Figure 18A the product compositions corresponding to
various ea.lculnted still compositions are summarized for lL binary mixture
having a = 1.25, when distilled with reflux ratio Rn = 19 in columns with 5,
10, 20, 30, 50, and 100 theoretical platoH.
The method for obtaining the curves of li'igures 18 to 20 may be il1mltrated by the
calculation for five plates in Figure 18A. It is first necessary to construct a large-
scale graph for the vapor-liquid equilibrium curve for a binary mixture with a =
1.25. The points on this curve are obtained by calculating values of y correspond-
ing to x = 0.1,0.2, O.B, etc. with the equation:
y X
-_- = ~---
1-y I-x
7" O'Brien, M.S. Thesis, Pennsylvania SttLt.C College, 1948. Prevost, "M.S. Thesis,
Pellnsylvttnia State College, 1948. Rose and Pfeiffer, "Effect of Reflux Ratio 011 the
Sepamtion Achieved in a PI1Cked FmctioulLting Cohnll11 in Continuous and in Batch
Distillation," Division of Indust.rial and Engineering Chemistry, Am. Chem. Soc., Atlan-
tic City Meeting, 1947. Sec also Colburll and St.earns, 'l'rans. lim. lnst. Chem. Engrs.,
37, 291 (1941).
GO A. A:-ID I'~. HO:"\]<:
w
f--
«
..J
..J
i=
(f)
is
~
u
>
::;;: a:~ 1.25 <t" 1.25
z LID = 19 LID" I
0
i=
u
«
a:
u..
w
..J A B
0
:E
0 o
...._------_.. _._ __.. ._-----'
MOl.. E FRACTION MVC IN STILL
100
w
~
...J
...J
~
is
!':
(j
::>
;:l;
~ '" " 1.25
0 LID: 99 Fig. If:\. Ht,ill-prodnet e()Jlljloflit;jnll
i=
0 om·VOH "al<:ulutccj by l\'r(,()ah(~-'l'hi('l(\
«
a::
u.. IH'oeorlu]'(~ for ido!d hinary llIiXt.IIJ'('.
w a = 1.25. Note in 13 t.hat !'ltu Hanw
...J
0 C (,IIl'V(, i" o!ll.:\itled for all V:d\l(~H oj' II,
::li
g'·<:al.i'1" i.han :'i.
0
A series of Olleratillg lines with Hlope L/V = 10/20 are then urawll 011 this graph.
These are so plaeed tJ1l1t their upper endH cro~s the .15 0 diagonal at composition;;
rmob as 0.99, (l.9n, O.~H, 0.02, (l.on, O.8fi, 0.80, 0.70, a.GO, a.50, 0.80, 0.20, 0.15, 0.10,
and 0.05. Additional operating lineR lllay need to be dr[,WIl later in orcinI' to obtain
points for a smooth Rtill-prorlnet oompo:,;itioll (x" ~:j)) Gurve Hueh as in Figm'eR 1H
to 2().
Figure 2 L indicates how point.H arc obtained. Tho point. ~: = 11 = O.90 IIlol<~
fraction of 1l)ore volatile eOlllpOllellt rcprescmt.s an arbitrarily choscn product emn-
position (xu) anrl the operating; line is drawn through this point with slope L/1/ =
19/20. The corresponding still 01' feecl composition (x 8 ) for a column with nn over-'
all separation of five theoretical lllatcs, oporating on the mixture with a = 1.25,
reflux ratio RD ;<= 19, is obtaillecl by stepping off five stepH along the operating line
and equilibrium curve (]lage 50). This gives in Rucccssion the following val!H1R:
I. 'J'IIEOllY til
CompusitiuH nf Value
Liquid from top lllaLo 0.878

y, V!1pOr from ~HCOlld pbt,e (eoulILiug frum top down) 0.878
Liquid from ~riCOlld l'bto 0.8/)4
"'
'/o< Vl1IJ()r ('rum third pltttc 0.851)
t;a Liquid frolll I,hi1'll plate 0.827
1/4 V:ljlO]' from fuurth plate 0.830
X·I Liquid from fourth plate 0.797
II .• = 1/, Vnpor from jifth plate" 0.802
Liquid in still ill J:ive-platel:()hllllll (or liquid from lifth ph,t()
fj'OIU top of long~r coluIllll) O.7(iii
" TIUli, i~, frolU t.IH' Ht.ill ill:t nolUl11n wit.h nvor-:dI8o!xtnttion of fivll thllol'eticlll ph11,",
t,-----~===-------~~~~
n" 39 and 91
LID" I LID = 99
A B
o t 0
ig. Hl. Still-product composition Curves calculated hy Smoker equation for ideal bin-
ary mixture. Ci = 1.05.
oL---------------------------~

"i".. 20. Still-procluc:t COIllj)nsit.itm curves ca.lculated by Smoker eqUlLtiOIl for ide!,] bill-
62 A. AND E. ROSE
ThiH ItlBt vallle i:; the desired ~till eumpo:-litiolL (x.,) corresponding to the chosen
product eompoHitioll (xn "" 0.90) for the five-plate sepamtioll with reflux ratio
l~J) = Ulltm\ the chosen binlu'y mixture with a '" 1.25. ThiH gives olle puiut on the
eorreHponding; X.<1 ;J;}) CUl've (see Fig. 18A, where :e D = 0.90, X,< = 0.765 lieH on the
e\ll've for 11. = 5). Repetit,i(}ll of this procedure with the other opemting lines gives
additic)llnl points. The x"' :l:/J (lurves for ten phtte~, etc. m'e ohtained hy ~i!l\ihrly
stepping otT ten steps along the various operating lille~.
1.0..--------------------_____
0.9
:;
z
0
j:
(Ii
0
I
0-
::t ) 19 UNITS
0 0.8
<J
a:
0
0-
I
<{
::>
_._j".
20 UNITS
0.7 0.8 0,9 1.0

LIQUID COM POSITION (xl
Fig. 21. Calcul:Ltiou by Me:Onbe-Thiele pl'()(:edul'l' of Ht,iil or feotl eWlllj>CJHitioll "'hem
disWhtLC: cumllOHitioll, l'cfiux rtltio, Humber of theOJ'pticJaI plates, liud l'l'llll;ive voll1tiliL.v
/II'{l known and fixed ~qpe Fig, 18).
1'he example chosen has a constant relative volatility, but the S:Ll)10
procedure eoukl be used for any mixture for which the vnpol'_:_liquid equi-
librium curve was known. Whon constallt reIatiive volatility may be
assumed, the Smoker equtLtioll will give results identicnl wit.h the grnphic
method described above, The Fenske eqll!1tion is applieL1ble for toLtd re-
flux u.nd cunstlLnt relative volatility. Similar caleniatiom; call he mtLtle hy
the Ponehon, SHvlnit, and like procedures for caSOt; in whieh the uHIUl.l sim-
plifying assumptions ftre not applicable,
1. THEORY f)3
The cmvct; indicl1te the cOlnplexit.y of thp effeds rlet,ermining the separa-
tion achieved in ciiiiiillntiOI1. They niso aid in ViflllHlizing the eompoiiit.iol1s
dllring ac:t.IHLl (listilhtioll, and, in ('ascs in whip,\) holdup is lllltpprecin,hle,
show how difltillate eomposit:ioll eh:mg('s with t.he decrPHse in st.ill eom-
position as a batch distillation is progressing.
In addition to the specific numerical relationships, tL comparison of the
curves shows the following general l'olntions implicit in McCabe-Thiele
diagrams.
(1) As the number of plat.(:_~s is inertmsed, Im';H :trld ICHH impl'Ovement in
separation results.
(2) No matter how large the number of plntf)S, a high-purity distillate
C1Lnnot be obtained with lo~v still composition and low reflux rat.io.
(S) At very low reflux ratios aJmost the HILme pOOl' separatioll is obtained
rf'gftrdlet;s of t.he number of plates or the still composition.
Curves of the type givon in this chapter C,tn serve to indica,to the general
lUttUl'e of composition relatiolls f)ven for nonideal and l1zeotropic mixtures.
In the latter case the mixture must he considered to be made up of azeo-
trope and one of i1.s pme components, rather than the two pure components.
The general effect of increase in plntes, relative volatility, and reflux ratio
will be that indicated by the calc:ulatccl curves, even though there will be
discrepf111eicH for specific caRes.
B. UND.E1U '['O'l'AL RE1!'LUX
Curves under totall'efiux conditions, such ns those in Figures 22 and 23,

which were calculated from the Fenske equation, show the maximum
difference in composition that ean be achieved with a particular mixture
and column. They also serve as a quick means of determining thc number
of theoretical plates if still and distillate eompositions from a test are
known. It, also appears that in some cases these curves describe the sepa-
ration 80 characteristic of batch dist.illation with appreciable holdup and
finite reflux ratio. This arises from the mn,rked shift in the operating lines
caused by constantly changing compositions in batch distillation with
appreciable holdup. '
Under total-reflux conditions 11 single family of curves may be used to
summarize the still-product separations for all different relative volatilities
and any number of theoretical plates. The path of any eurve is determined
by the value of an rather than the values of a and n. Thus the curve
marked n = 20 011 Figure 22 is obtained for a = 1.05 !~nd 11 = 91 as well as
a = 1.25 and n = 20. The Same curve would be obtained for any other
8() Rose, Johnson, and Williams, "Stepwise Plate to Plate Oalculation of Batch Dis-
tillation Olll'ves," Am. Chem. Soc., Ohicago Meeting, 1948.

\.:omhin.aLi()1l fur which el' is nppl'()Xilllal('I~' 8(Ui. 'rlj(~ qnall! ity n" i:-;
kllO\Yll as the l'/!ric/ulIl'll.t. fw.~t()J', (li' 1I1J('1'-IIU sl'jloJ'({fiou II/dol', :I,lId is a !!:()()d
measure of the maximum slml'plItss of separation that may 1)(\ expr,eLed in
any particul!tr iustmlCc.
W
I-
<:(
_j
_j
j::
CJl
o
z
u
>
::;
z
o
i=
u
<:(
a::
LL.
w
_j
o
::;
I
Fig. 22. SW1-produc:L eOllllloHitioll Fig. 2il. SLill-pl'orhwt t:OItlIHIHi(,ioll

I'llrves (lldcull1t,c([ by tho l<'(mRlm {J(lll:Ltioll eUI'VCH ea.leuIaLo([by ]i'ellslm equtl,t;ioll f<lr
for iti('ld hintlry mixture. «= 1.25. Total idl\!ll binary Illixl:m·('. /' = 2. Total 1'('-
reflux. flux.
4. Compositions on Various Plates or at Various PoInts in a Column

The composition Oll any pln.te or at any point in n, (~Ulnmll may 11<\ (·sl·i-
mated by the same pl'ocechll'os already dis(mSflOd for t:nluulatillf,!; the 1l11l1lbnr
of plates and the still-pl'oduet eompositiull relations. An pxample WUK
indicated in eonnection with the x,,, X]J values rO)' Figl1l'(~ 18A. H1l!~h oalcll-
ItLtions are subject to all the limitations and (lisc)'epuneim; a]n'Li\ly HI('l1-
t.ioneu, !lnd have geneml interest and utility for the K:.tIlW )'(~aK()nN as t.hn
curves relating still and pl'oduet eompositi()Il. They are nBI)(H~ially useflll
for est,imnt,ing the holdup of tho individual (!OlllpouontK. 'J'hi:-1 is dd:(1['~
mined by the coneentration of that eomponont on cuell of the pbtos or
througllOut the packing of [1 column. This varies from plnte to plate or
point to point, and in a batch diHtillation suoh compo:;it.ioils ehange ·gm<l~
nally as the distillation progresses.
Figures 24-26 show the results of sueh calculations. Figure 24 gives the
calculated eompositions under total reflux in a column with twenty theo-
retical plates for an ideal mixture with a: = 1.25, The shape of tho (J1U'ver-;
fmel the values of the composit.ions al'e unehanged if t.he numhcl'ing is from
tbe bott.om up, or from some int.mm(\diat.e plate, aR for t.lw nnaly~is of It
1. TlmoltY Ci5
:;\turter eolnmn. This is true beeau,>l'. ;dl the <:\\l'VCtl have idcntieal slopes
fot' any pal'tienlar eomposition, and i(iPllti(~tt.l nverage slopes between a.llY
two given compositions. This is not i.l'llll of elll'ves at partinl reflux such as
those in Figmes 25-26.
0 Xo o xo
If)
w I
t;c
..J 2 2
n.
..J
<I
u
i=
w
cr 5 5
0
W
I
I-
~
0
w
cr
::J
(/)
<I
w
::< 10 10
z
:::;
::J
..J
0
U
lJ..
0
n.
0
I-
:::;
0
cr
u.
W
U
Z
<I
l- a'1.25
(/)
0 Xs LID' I
20
o 0.2 0.4 0.6 O.S 1.0
Fi!);. 24. Comp(JHitiollS (\Ideulated by the mg. 25. Compositiolls mdculated by

I,'(nu-:k!' nqllLLt.ioll at, vILriou;; lJoints within It Smoker (~qlll1ti()n I1t viLrious po in tH
('O]Ulllll for :Lll ideltl hiwLJ'Y mixture. Twenty within It column lIud various times durin!);
JlltLLn~. Tot.al l'f'flux. . bat(lh disWlations fol' ILll irleal hinary
mixt.ul'e. Twenty plates. P!trti!11 1'' -
flux.
Under ptLrtial-l'eliux conditions t.he lower ends of the curves are fre-
quently straight lines, corresponding to that part of it McCabe-'l'hiele dia-
gram at which operating line and equilibrium curve come together. At
distillate compositions approaching unity a similar phenomenon appears at
the upper ends of the curves. Any of the curves can be used for a smaller
number of theoretical plates than the tot.HI shown in the figures. In such a
case the curves [\'1'e read from t.op down t.o include only the desired numbel'
of plates.
fjG A. ,\ ND 1'~. nOSg
An intcl'c~tilll!: tt]lplimLtioll of thiH type of ()lll've is the )H'ol)()i-\l1l by Pocibiclniak"1

to mmlyze liCllli(h; or liqnefier] gases hy determination of the tCll1pernt,urm; along It
fnwtitlllnting (lOhnllll opnmtillg at totall'(:~flux 011 Ii Hnmple of Rueh Hil'.n that, all of it iR
present ill the COI11l11n itS holdUp. The t,olll]lm'atllJ'(1R aro relntnr\ to the (lolllposition
at variom; pointK in the (lOllll11n, ami the Cllrvn of tmnpemt\ll'c (or composition)
versus length along the column closely rescmhleft the (Jonventiona.J clistil1atioll eurve
relating tempemtlll'c, rcfmctive index, etc. to per (leut of chnrge di;;tilled. It is of
OOlll'fle esscntial that tbe condcnser alld still pot of such :t frnetiollating column be
designee[ to have negligible holduJl AO that all the SIUll)Jlc is diHtrihllted through the
fraetiOlHttilll!: scottOll.
Bowman ItIl(1 Snstry"2 lwve a\Ho diHClISHe(1 t.his llwthnd of lLllalYHis ItS applied to
hinary rnixtlll'CR, and have presentc(l typical experimenttLi <lnta. The experimental
eUl'ves agree with thw;() calculated [or the proee~H, bllt the theory indicates that !t
given eolmnll will give poorer sepamtioll than Wllllll operated in tho usual bateh
(fJ
w
j
n.
_J
«
u
;::::
w
a:
8J:
f--
~
o 40
w
a:
;:l
If)
~ 50 50
::;:
~ '" = 1.25 a.' 1.25
:3 LID = 19 LID = 99
o
u
u.
o
a.
~
::;:
o
a:
LL.
w
U
z
~
A B
~ ~ ~
1000"""".L-lJ':1I1L
. --'--0-'.-4--'--O-'.'-6--,-___jO.'-B--'---'1.0 1000WL-'---0J....--'-OJ-- J --i..- O.B
0 2 2 4 06
Fig. 26. ComposiLions calculated by Smokc!' equatioJl at vlIl'iouR ppints within It
column and variou8 times during hat(lh distillntioll for an ideld biIllI)'y mixture. One
hundred plates. Partial reflux.
Podbielnil1k, U. S. Pat. 2,377,(JOO (Juno 12, HJ.J5).
81
Bowman and Sastry, "Ailltlysis by Continuous RentifictLtioll," Divisil)n of Analyti-
82
cal and Micro Chemistry, Am. Chern. Soc., New York City mcetillg, 1947.
I. THEORY 67
fashion. However, this theoretical comparison neglected the effect of holdup and
finite reflux ratios on batch distillatioll. The chief advantage of the analysis by
temperature measurement at total reflux is the use of simpler applll'atus and pro-
cedure.
5. Summary
This section has preKented the most important methods of determining

the !lumber of plates for dist.illation appamtns. When the number of
theoreticnl plates it> knO\vn 01' fixed, all these methods are a1t:lo useful for
cstimating the composition diffcrcnr:cfl between still and product or at other
points in the column.
A. tJOMPAlUSON OD' MI~'rHODI:i 01" DliJ'l'mUMINING NUMBI~H OF 'l'HEOHE'l'ICAL
PLA'l'ES
If the data are collected when the distillution is proceeding under total
reflux, the number of plates is usually determined by:
(1) The stepping-off process, using the experimental vapor-liquid equi-
librium diagnun. Thif:i if:i applieable to both ideal and nouideal mixtures.
(,Ie) The Fenske equation, which is the algebraic equivalent of the step-
ping-off process, but is applicable only when the average value of the rela-
tive volatility is within n,l)out 5% of the extreme vnlllCR encountered in the
distillation or test.
If the distilln,tion is cll,l'l'ied out under ste[l,dy-state conditions (continuous
distillation) at paTt1:al Tefl'll:!:, the number of theoretical plates may be de-
termined by: .
(1) Sorel's method, which is the basic general procedure for plate-to-
plato calculations. It is not used when the following simpler methods are
applicable, i.e., when the column is adiabatic, and the thermal properties of
the components are similar. Numerous modifications of Sorel's original
method have been published (Ponchon [1nd Sava1'it, Randall and Longtin).
(2) Lewis' method, the basis for the most useful modifimLtions of the
Sorel method. It t),pplies only when the column is adiabatic and the
therml11 properties of the componentis are similar. The original Lewis
equations are less used thlLn the equivalent:
(3) McCabe and Thiele graphic pl'ocedm:e. This is the most frequently
used procedure. It is restricted ill the same way as the Lewis method, but
like S01'el'f:i and Lewis' rfi(~t,hodf:i it is npplieahle to nonideal as well [~S ideal
mixtures.
(4) Smoker equatioll, which is the algebrnic equivalent of the McCabe-
Thiele method but is applicable only to ideal mixtures in which a constant
relative volatility may be assumed.
68
(Ii) PlaLe-equivalen(.H protl'chlr<!. Plaie <'quivnlclltH may hp enl(.'ula.L(~d

for Hlly euse in whinll still nnd product c:ompoe:itiolls nnc! the I'apor-liquid
equilibrium curve are known. They at!' of (:omparative vallI(> only, and
have no genoml utility f01" lJl'edietillg separatiou HIlder new eil'emllstullees.
Tlw Sord-Lewis, M(:Cahe-Thidc, and SmokeI' rdations Hn~ also u.t leu,::;\'
approxim:ttely valid for hut.ch diHt,illation at. fiuit,!) reflux, if huldup is small
and reflux ratio' is not too small. When holdllP iK large, nil the above
methods arc inapplicable, nIHl more emnplex mi:LtiollK tLpply, :LS diRensHcd
in detail in Seetiom, V and VI.
B. PIU~CAU'l'IONS IN us)'; 011' ITl'JWll'[' IGQIHVAI,IIJN'l.' 'ro 'l'HI.;QIl.lD'l'WAl, 1',[,,\'1'1'] AN])
'l'HJ~OlU~'l'ICAr.-I'f,A'l'l'] DA'l'A
In geneml, H.K'.l',P. data lL]'(~ HOt. highly reprochwihle, ptLrtieularly in

packed columnR. H. EV(111 the Ramo tf'Ht mixture lUl!l the saIne column may
give values differing hy lor/~), if the kKt,R [],ro murle at different times,
Different columns of lll(~ sanw dimuIlRiom; awl pa(~ked with the same pack-
ing mELY give clifferell(. l'l'SllltS beelL1lSe of vnrintiolls in the arrangement of
the packing, or heefl.Uf\fl of ~light eorrosioll 01' the IJl'esellee of extmneous
material. Compnmtivn H.E-T.P, vallleH 011 :;nv(~r:.tl colnmm; with n given
test mixL1ll'e arn n l'Plls[)Jwhly t J'llsLw()l'thy haHis for clloice oj' t.he best 001-
11n111, paeking, reflllx l'at.io, or t hl'onghpu!. for a partioulal' sepnmtioll.
Values obtained with standard tt'H!. mixtures are less dep~ndable wh('ll llHnd
to precEd the height, of n column for separation of un entirely different mix-
tnrc. TallIe VI illdi(::Li,es tlw v[tl'in,tio!l found hy P(~kl'HK'1 with different
types of (·,om})ollnds. .
Thel'e is some indiention tlmt eompounds or high HlOleeubl' weight give
higher H,RT.P. vnlueH Hum those of lower moleenlal' weight, C[Lrswell~rl
ascribed this to contaet of rm equHl volume of vapor with a SI1lnUer volmne
of reflux, and gave the furmula:
H.E.T.P. = kMd/T
to indicate the fnctors involved. The i:lymhoI k is it proportionalit.y eOll-.
st.ant, AI iK the average molecular weight of the two subH(;anees being
sepamted, d is the ILvenLgo denKity of the liquid mixture, and l' the absolute
tempemture. Carswell sj,nl,l'H that the inverse temperature relation is dllP
to increase in vapor volume with absolut.e tmnlJcmturo, while the denl-iity
factor arises from the deel'ea::;e in surface area of the liquid at greater densi-
ties, Carswell makeR no mention of viscosity or diffusion propertieH whieh
might also be expo(:kc! to (mtel' iuto this situat;ion.
H3 '['odd, Ind. Ella. Chem., Anal. NrL, 17, 17(j (HWi).
H4 Peters, Inri. JiJng. Chem., 14, ·17() (1022).
R6 Cal"KII'llll, Inll. gng, Chem.. , 18, 2\)4 (1I)21l).
1. THEORY 69
A variety of recent work,SG mostly in terms of transfer units rn.t,her than

theoretical plates, hn.s demonstrated that in many cases H.E.T.P. may be
expected to vary with reflux ratio n.ud composition.
H.E.T.P. is usually greater in a long eolumn than in a short one of ex-
actly the same kind. Thus a low H.E.T.P. obtained in a test on a short
column cannot in general be reproduced in a much longer calumnY It is
even more important to recognize that a low H.E.T.P. in n, small-diameter
column will not even be approximated with larger diameters, even though
packing, test mixture, throughput, and reflux ratio aTe the same. (Chan-
neling, i.e., agglomeration of the liquid into relatively large streams, is be-
lieved to be the cause of this behavior, although good experimental evidence
is lacking.) Reeently developed protruded packing88 is designed to avoid
channeling, by using surface forces to maintain filmlike distribution of
liquid over the packing. There is no record of attempts to accomplish the
same end by addition of a nonvolatile surface-active agent or by changes
in the surface of the packing.
Corrosion or introduction of foreign matter will usually hnve a deleterious
effect out of proportion to the amount of such ma,terial. The H.E.T.P. as
well as the over-all utility of a column or packing is relnted to the pressure
drop for a given throughput. Pressure drop in both packed and plate
columns has been subjected to extensive study. 89
If a column is not insulated to operate approximately adiabatically at all
times, the variation in heat gained or lost through the column walls will
change H.E.T.P. appreciably. This results from variation in the amount
of reflux, which in turn causes changes in reflux ratio, throughput, and
!-I.E.T.P. (Sect. IV). Small-diameter columns are most affected by such
conditions, so that their H.E.T.P. will undergo gross variations with
changes in environment if insulation is inadequate.
The vapor-liquid equilibrium data in Ta.bles I-IV and in most papers
published is for a pressure of 760 millimeters. Little error is involved in
using these data for operation at pressures of 700 to 800 mm., but at other
pressures differences will become appreciable.
66 Duncan, Koffolt, and Withrow, Tmn.s. Am. Inst. ChelT~. En(jl·s., 38, 259 (1942).
GIll'ster, Koffolt, and Withrow, ibid., 39, 37 (1\)43); 41, 393 (1I}45). Griswold and
Stewal·t, Ind. Eng. Chern., 39,753 (1947). Keyes and Byman, Univ. Illinois Bull. Eng.
Expt. Sta. B1l11. Se'!'., 38, No. 328 (1941). Langdon and I{eyes, Ind. Eng. Chern., 35,
464 (1943). Pfeiffer, M.S. Thesis, Pennsylvania State College, 1!)·!7. Richter, Del n.
Kohle, 40, 67 (1944). Schultze and Stage, ibid., 40, 66 (1944). Stage and Schultze,
ibid., 40, 68 (1944). Ston'ow, J. Soc. Chem. Ind. London, 66, <11, 73 (1947).
87 Kirschh!mlU, Angew. Chem., B19, No.1, 13 (1947).
ss Cannon, Ind. Eng. Chem., 41, 1953 (1949).

89 Bcrtet,ti, Tmns. Am. [nst. Chem. Engr8., 38, 1023 (1942). Leva, ibid., 43, 549
(1947). Leva and Grummer, ibid., 43, 633, 713 (1947).
70 A. AND E. ROSE
TABLE I
DATA FOR n-HiCI'TANE-MgTHYLCYCLOUIJJXANE TES'r MIXTURE" (ALL COMPOSITIONS IN
MOLle FRAc'rroNB n-lhn"l'ANB)
~--~--
Vapor-liquid cquilihriluu datu. Datu. for ttnalYHiH of imm})l(~s
VfLIlor Liquill 20
11])
l~o
, ·1
H.p" °e.
C{JHl!)n. (:OlIlIHl. (7!lU lI11n.)
.--------~
0.0000 1.4232 0.7Gn3 100.S0

0.0350 .0310
.0620 .05S0
.0787 1. "1200 .7613 100.55
.1030 .0950
.1-+30 .1330
. Hi38 1.4165 .7535 100.35
.1!l20 .1800
.22\10 .2100
. 24S!i 1.,1135 .NM 100.15
. 28!lO .2715
.3330 .a170
.8372 1. 4100 .7877
.3810 .3630
.'1200 .4010
. 4121i 1.4075 .7312 99.70
..1750 .45UO
.5210 .[JOlO
.518H 1 . 'lO3(l .721S 00.20
.57S0 .5590
. (\1S0 .5\)00
.605G 1.400,1 .7145 no.oo
• (JliGO .6470
.6H03 1.3070 .70GO D8.SS
.72S0 .7090
.7710 .71)(\0
.70·12 1.3042 .6U!)2 1)8.GO
.8100 .70GO
.8535 .8'130
.8\100 .871)0
.9130 .\lOGO
.9400 .9310
.9338 1.3899 .6884, 98.50
.9525 .91';40
.9860 .9S00
1.0000 1.3878 .(\839 98.40
• This mixture is usoful for testing columlls with !tbout. ten to ninoty plal;OA. It is
particularly suitable for columns distilling close-boiling hYllroearbOlIH. AI1ltlysis is by
refractive index or density. GriswohI(I) recommonds mfmetive indllx ItS lending to
(I) Griswold, Ind. EnlJ. Chem., 35, 247 (lD43).

1. THEORY 71
smaller errors. The dntn are from Bromiley and Quiggle. (2) The n-heptallc ant!
methylcyclolwxane used by Bromiley and Quiggle had the following characteristics:
B.p.,OC. d'" n'·D

Compound (70U mIll.) • F.p., cc.
n-Heptane ............ 98.,1 0.G839 1.3878 -'10.8

MethylcyclolUlxane .... 100.8 0.7G93 1.4232
The heptane Wl18 obtained from a commercisJ source !1nd was approved by the National
Bureau of Standm'ds for determination of knoek ratings. The methylcyclohexane was
purified by fractional distillation, washed with concd. H 2SO.I, Na 2C0 3solution, and water,
and thon dried.
Willingham and Hossini(3) suggest the following equations for calculating theoretical
plates with this mixtun,:
(1) refractivll index and mole fraction relation at 25 ac.:
nD(mixture) - nD(n-heptanc) = 0.0306.'1)' + 0.00!8(:C')2

(wlwJ'(\ nIl is tho rcfral;t.ivc index and x' is tho mole fmetion of mothylcyc]ohexi1llc in the
mixf,ure).
(3) theorotical plates and mole fmctioll relation:
n +1= [32.15 - 0.34(1 - 99)] log [(1~-~ -:)

X hw(]
/ (1 ~ X
) 11(1t
]
over the range 96.5 to 101.5°C. (where t is moan temperaturo, in "C., n is llumber of
plates, amI ~r, is molo fl'l1ctioll of the more valatilo compollout). Thc~ equat.ion is not
applicable st.rictly unless tho t.(l~t liquids usod hl1Vo ooon purified so that the I'cfract,iv(l
indices diffel' by 0.035! unit.
TABLE II
DA'l'A FOR MITITHYl~CYOLOHEXAN8-TOLUElNE TEST MIXTURE" (ALL COMPOSITIONS IN
MOLE FRAC'l'IONS MElTfIYLCYCLOHEXANE)
Vapor Compo Liqui,l Compo n~!O

D
0.000 0.000 1.4965

.143 .100 1.4871
.270 .200 1.4782
,378 .300 1.4699
.470 0400 1.4620
.5~0 .500 1.4544
.650 .600 1.4474
.737 .700 1.4'108
.818 .800 1.4345
.n06 .900 1.4286
1.000 1.000 1.4235
7'able continued
(2) Bromiloy ami Quiggle, Ind. Eng. Chem., 25, 1136 (1933).
(3) Willingham and Rossini, J. Reseal'oh N atl. Bur. Standards, 37. 15 (1946).
72 A. AND E. ROSE
a This mixture is usoful for testing columns of IIp to about thirty thooroLical pltttes.
Been,use of the cheapness, ready lIvailfthility, ILTld ease of purificatioll of the eompononts,
it is particularly suitable for tests on large-size columns. Analysis may be nit;hor by re-
fractive indox or density. The data givon are thoso of quigglo tLlHl Fenske. (·1) Tho rola-
tiolls do not follow equation (4) with a constant value of a, HO tim!; gl'l1phic procoduros aro .
necessary to dotfmnine the number of phlteH. Compositioll of tho tos1; mixturo should
be in the mugo 10 to 75 mole per cent methylcyclohm(l1lw to lwoid the coneontmtiollB in
which composition change per theoretical plate is too sma.ll for ILccurntll analysis. The
materials used by quigglo and Fenske had the following CIULt'lwtlll'is!;ics:
B.p.,oe.
Componnd (7(10 llllll.)
dio nfg
Methylcyclo!lexl1uo
Sample J ................ 100.85 O.7IHlfi 1.4234
Sl1mple II .......... , ..... 100.85 .7602 1.4235
Toluene
Sample 1. ...... 110.6 .Sli58 1.49(15
Sllmple II ....... 110.(1 .8603 1.49(j5
The toluene was llitmtioll grade which had beon fractionally distilled through 75 pl:1tes
with high reflux ratio. The mothylcyclohoxlIllo was techllic!tl gmde purified by washing
with coned. ILS04, NI1~C03 solution, and wllter.
TABU': III
DA'rA FOU BEl\IZENE-E'l'IiYLgNE DICHI,ORIDl~ Tms'l' MIXTUl~Ea (AI,r, COMl'OSl'rIONB IN
Moun FUACTIONS BNNZIGNg)
---~---.
--~--. _-- ..--.
Vn.JlOl· Compo Liquid Compo ni~·2
--_._-_.
0.000 0.000 1..14218
.113 .100 1.4,1750
.222 .200 1.45287
.327 .300 1.45825
.429 .400 1.4G30()
.527 .500 1.4(l!)H
.624 .UOO 1. 474G2
.727 .700 1.48020
.818 .800 1.485\)(i
.912 .000 1.4\J187
1.000 1.000 1 ..:Hl71!)
~,- -_--__ ._._ ... -- .... _- .... _----_._,-. -.------~.-.
a This mixture is id0l11 01' IHllIl'ly irIe!d, with tho vu,lu() of c; giVllIl iLS 1.109. It Illay be
used for columns with up to Sixty pit1tes, but therB is some ullcertainty aH to the VI:1P()l'-
liquid equilibriulll values, and the ethylene diehlol'iull mfly bo hydl'oIY7.1111 [IUd clluse
corrosion of metals if water is present. Tho data givon !tl'(l supplied from W!lrd(5) and
taken originally from B.osltnoff and Easloy. (6) .
(4) Quiggle and Fonske, .T. AIIL. Chem. Soc., 59, 1820 (1037).
(6) Ward, U. S. Bur. MineN 'l'ech. Papers, 600 (UJ30). "
(0) Rosanoff lind Basley, J. Am. Uhem. Soc., 31, 053 (lOOn).
r. THEORY 73
TABLE IV
DATA l'OR BENZENE-CARBON TE'l'RACHLORIDE 'l'0~'1' MIXTURE" (ALL COMPOSITIONS IN
MOLT<1 FRACTIONS CARaoN TE'l'IlACHLORIlJE)
Vupor COll1p. Liquid Compo
0.000 0.000 1.4071H

.122 .100 1.40392
.233 .200 1.48997
.337 .300 1.48589
.437 .400 1.48178
.534 .500 1.47772
.028 •(iDa 1. 4736H
.720 .700 1.46963
.812 .800 1.46555
.831 .820
.850 .840
.86B .860
.888 .880
.907 .900 1.461<12
.918 .918
1.000 1.000 1.4,5732
• This mixture has probably been more widely used for test,iug than any otlwr. It is
suitable for columns with up to about 25 plntes. Analysis is almost always by refractive
index, but density mlLy be used. The daLa given are supplied by Ward,(6) and taken
originally froIIllnternational Cril'ical Tables(7) and Rosanoff and Eusley.(6) Initial COIll-
position should be such that still composition at the time of tlLking samples is not much
less than 6 mole pllr llent carbon tetrl1chloride, lLnd the condenser sllmple should not have
more than about 80 molfl per clint carbon tetrachloride. At these concentl'a~iOIlS one
theoretical plate corresponds to :1bout 0.0005 change in refractive iudex. The azeo..
tropic composit.ion of 91.8 mole pel' cent cm'boll tetrachloride must be entirely avoided.
Serious corrosion of aluminum or iroIl packings or columns is likely to occur with any
conceutmtion of this test mixture.
TABLEV
BENZENE-CARBON TETItACHLORIDE REFRA.CTIVE INDEX VERSUS PLA.TE NUMBER
DO 01 02 03 04, ·05 ()O 07 08 Oil

lt~ PI"te nu mLor
1.'1600
10 37.7 37.4 37.1 36.8 3(i.6
20 36.3 3ti.l 35.8 35.(j 35.4 35.2 35.0 34.8 34.5 M.3
30 3<1.0 33.8 33.6 33,3 33.1 32.8 32.6 32.4 32.1 31.9
4() 31.7 3l.5' 31.3 31.1 30.\) 30.6 30.1 30.2 20.1) 29.7
Tabla canlint<cd
(1) International Criticall'aules. Vol. 7, MCGl'ILW..Hill, New York, 1930, p. 77.

74 A. AND E. ROSE
TABLT~ V (Confirmed)
-~-'---'
00 01 02 03 O't 0,1 DO
()7 08 on
.~- .. --~-------------... ---.-.--,---.---.-.-- ..-
n'j; Plate IllIllliJOl'
.._ - - - - , - _ . _ ....- ... _--------
50 29.5 2H.3 2H.l 2S.0 2S.7 28.5 28.:3 28.1 27.0 27.7
GO 27.5 27.3 27.1 27.0 2G.S 2(i.(i 2ti. 5 2(i.3 2ti.2 2fi.l
70 25.n 25.7 25.5 25.'1 25.2 25.0 24.0 24.7 24.(i 2-1.5
SO 24.3 24.2 2·1.0 23.0 23.S 23.7 23.5 23.4 23.3 23.2
no 23.1 23.0 22.9 22.7 22.fi 22.5 22.4 22.3 22.2 22.0
1.4700 21.9 21.S 21.7 21.(i 21.5 21.,1 21.:3 21.2 21.0 20.\)
10 20.S 20.7 20.li 20.5 20.'1 20.3 20.2 20.1 20.0 10.n
20 19.5 H)'S H).7 In.7 10.li 10.5 lilA 10.3 19.2 10.1
30 11).0 IS.\} IS.S IS.S 18.7 IS.li 18.5 18.'1 18.3 18.2
40 18.1 IS.O I7.ll 17.S 17.7 17.7 17.G 17.5 17.4 17.3
50 17.3 17.2 17.1 17.0 In.n lO.\) HI.8 Hi.7 Hl.7 W.H
60 W.5 ](i.5 Ui.4 Hi. 3 l(j.2 10.2 ](i. 1 W.O 15. f) I5.!}
70 15.8 1.5.8 15.7 15.(i 15.5 15.5 1.5.4 15.3 15.2 15.1
SO 15.1 15.0 14.0 H.B 14.S U.7 14,. G 14.5 H.5 H.4
no 1<1.4 14.3 14.2 H.I 14.11 1'1.0 13. !J 13.8 13.S 13.7
1,£1800 13.7 13.0 13.0 13.5 13.'1 1:3.4 13.3 13.2 13.1 1a.1
10 '13.0 12.n 12.n 12.S 12.S 12.7 12.7 12.H 12.5 12.4
20 12A 12.3 12.2 12.2 12.1 12.0 12.0 11.0 11. \J 11.8
30 11. 7 11. 7 lUi lUi U.5 11.".1 11. ,1 11.3 11.2 11.1
40 11.0 11.0 10.\) 10.n ]0.8 IO.S 10.7 10.0 10.n 10.5
5Cl 10.4 ](J.3 10.3 ]0.2 10. ] 10.0 10.0 n.n n.!l !l.S
00 n.s 0.7 O.G n.5 n.5 0.4 G.a n.3 !J.2 0.1
70 !).1 n.o 8.0 8.0 8.8 8.8 S.7 s.n S.5 8.5
80 8.'1, 8.3 8.2 8.2 8.1 8.0 8.0 7.!) 7.8 7.8
gO 7.7 7.7 7.(\ 7.5 7.4 7.3 7 ']
.~ 7.1 7.1 7.0
1.4900 (J.n 6.9 6.8 !l.S n.7 (Ui Q .U~ H.<1 U.S 0.2
10 H.l a.I 0.0 5.0 5.S 5.7 5.6 5.5 5A 5.3
20 5.2 5.1 5.0 4.n '1.8 4.7 4.0 4.5 4,4 4.3
30 4.2 4.1 4.0 a.n 3.8 3.7 3.11 3.5 3.3 3.2
4.0 3.1 2.0 2.8 2.7 2.5 2,4 2. :J 2.2 2.0 U)
50 1.0 1.6 1,4 1.2 1.0 o.n 0.7 0.5 0.3 0.2
60
TABLE VI
VAUIA'l'ION OF H.E-T.P. WI'l'H DWFJilRJolN'l' nlNAHY MIX'l'UUBS
----~----- ..-- ..
.~.,.-- ------.--~
Mixtu,'''" H.B.T.P., in.

---~--.--- ... - . --,------~-.-~
BOllzene-toluOIlO ............................... 10
Acetic acid -WtLtlll' .............................. 10
Nit.ric acid -wn!;er ............................... 8
Aoetone-ethyl ILleohol .......................... 8
Ethyl alcohol-witter ............................ <1
Methyl alcohol-water ........................... 3
I. THEORY 75
IV. DIFFERENTIAL EQUATIONS O'F PERFORMANCE
1. Diffusional Processes in Packed Columns
The separation achieved in fractionating columns in terms of theoretical

plates was discussed in Section III. The Fenske and Smoker equations
were treated in detail because they are typical equations of performance
that relate tIle separation achieved to the properties of the mixture being
distilled, the length and character of the column, [md the operating condi-
tions. It was pointed out that the theoretic!11-plate concept is useful but
essentially unsound for packed columns. Even for plate columns it has
been necessary to use the concept of plate efficiency to explain the difference
between calculated and experimental results. A number of concepts other
than theoretical plates have been used in attempts to gn,in a clearer picture
of the fractionating process, and of these the present section discusses thc
following:
(1) W. 1(, Lewis equation for differential change in composition from plate to
plate or point to point;
(2) Chilton and Colburn transfer-unit equations;
(3) Westhaver, Kuhn, Docksey and May, and Bowman equations;
(.4·) Cohen equations for the rate of approach to equilibrium;
(5) Marshall-Pigford equations.
All these methods use a differential approach n.ncl so are particularly appli-
cable to packed columns.
2. Packing Factor
The irregularity of the interstices in n. packed column continues to be a

complicating factor, so that empirical or experimental values are necessary
in evaluating and comparing packings. The most useful approximations
concerning packing materials themselves are the em}Jil'ieal relations of
Fenske and co-workersDO and of Carswell. 91 The former noted that the
surface area per cubic inch of packing multiplied by the percentage of free
space gave numerical ratings which p!Lralleled the usefulness of various
packing materials. Carswell suggested the relation:
H.E.T.P. = aVIS
where V is the volume of vapor space, S the surface area of packing, and C
is a proportionality constant.
9Q Fenske, Tongbel'g, and Quiggle, Ind. EnU. Chem., 26, 1169 (1934).
91 Carswell, Ind. Eng. Chem., 18,294 (1926).
76 A. AND E. ROSE
The relations to be discussed in detail in the present section pertain not

to the characteristics of different types of p:wkillg, but mther to the be-
havior of a given packing or contacting device under various circumstances.
3. Lewis Equations
An approach toward a differential treatment of the composition change
with height in a column was suggested by Lewis. 92 He arrived M, an ex-
pression for the change in liquid composition from plate to phtte by writing
the basic material-balance equation as:
VYn - DX/l
X1l +1 = ---------L- --
and then subtrtwted Xn from each side to obtain:
d~; D
-1--'
X,,+1 - X" = (,n = y" - x" - -1·.·.:
,
(xv - YII)
When inverted, this becomes:

dn 1
= --~'---'-'---'-
d:c
Integration gives:
l.
'~D dx
n = --------.---..-.----... -..-
y" - x" - iD (Xn - YII)
Graphic integration may always be used to obtain numerical values, but

algebraic integration oa is also possible when the relative-volatility equation:
Y x
a--
1- y 1 - x
is applicable. In this case:
92 Lewis, Ind. Eng. Chem., 14, 492 (1022). Slle also Rohinsollltlld Gilliland, ElemcntN
nf Fractional Di8t7:11at-ion, 3l'd eel., MeGrnw-Hill, New York, 1939, PP. lUi-lIS.
03 Dodge lind Huffmnn, Ind. Eng. Chern., 29, 1434 (H137). Dodge, Tl'an.~. Am. Inst.
Chern. Engrs., 34, 585 (1938).
I. THEOHY 77
so that:
(zn d:~
n = J:Va ----O'-X-n~-----D-:--;:::[-------O'x-.n---]
-l-+-x-'(O: -
n 1) - Xn - L XD - -l-+-x-',,~---l-)
and integration gives:
n = (1 + (bj2) In [(XD + A)(X. + B)] _ 72 In [(ax~ + bXD + c)]

Vb 2 - 4ac :);. + A Xv +B (ax, + bx. + c)
where
a = (1 - a),
b = (0: - 1) (1 - f xv) + D
0' _.
T./
and B
2a
It is to be emphasi.zed that this method does not assume the stepwise

changes characteristic of plate columns, but provides for a differential
change from plate to plate or point to point in the column. The equations
are particularly applicable to difficult separations for which a large number
of plates are used and the change in composition from plate to plate is smItH.
4. Chilton and Colburn Transfer Units

Chilton and Colburn94 have devised a distinctive differential method of
design for packed columns which em'relates distillation phenomena with
other diffusional processes such as absorption. 96 Columns are compared on
the basis of their H.T.U., or height of a transfer unit, in correspondence
with the H.E.'l'.P. concept. The number of transfer units is determined by
the rela ti on :
vt'=XD dy
Nt = l 'II,
-*c-..:;:___
Y - Y
in which y* is t.he eomposition of vapor that would be in equilibrium with
liquid at any specified point in the column and y the actual vapor composi-
tion at that point. The limits of the integration YI and Ys are the vapor
M Chilton and Colburn, Ind. Eng. Chern., 27, 255, 904 (1935).
96 Deed, Schutz, and Drew, Ind. En(/. Chem., 39, 766 (1947), and IUany others:
78 A. AND E. ROSE
composit.ions at t.he t.op of the colulllll and in t.he still, respect.ively.

Graphic int.egration is usually necessary, but Bakel' and White U6 have pub-
lished shorter graphic constructions which are of assist;ELllce. Oolburn,u7
Scheibel and Othmer,98 and Underwood o9 have also proposed modified
met.hods of calculating t.he number of t.ransfer units.
The Ohilton and Colburn equation for transfer units is derived from con-
sideration of the diffusion and flow phenomena occurring in a packed
column. The differential rate of increase of the moles of low-boiling com-
ponent in the vapor, dw, due to its upward flow may be expressed as:
dw = dpGS/Vllf
where dp is the pressure differential of the more volatile component over
the small height of the column, dZ, that is under consideration, G is the
mass velocity of the vapor, S is the cross-sectional area, p the total pressUJ:e,
and J.11 the average molecular weight. -
The moles transferred by diffusion must balance the above, and may be
written as:
dw = J{ flpSdZ
where K. is a constant dependent on the stationary vapor film adjacent to
the liquid phase, and flfj is the rliffercIlt:e between the actual partial pressure
of the more volatile nompOllont and the pl'OSSlll'e this component would
have if in. equilibrium with the liquid at the point in the column undor con-
sideration.
The above equations may be combined to give:
or:
ii
[ , dpjlljJ = KplVIZ/O
}I,
so that the height of eolul11n mquil'ed £01' a given separntioll is:
Z = (G/KpM) I )J
I
P
' (dpj AP)
The nse of this relation has hocm limited beca\lse valnos of the eOll:-;iallt K
DO Balwr, Ind. EnrJ. Clwm., 27,077 (1035). WhiJ,n, 'l'mIt8. Am. Insl. elWin. }lJn(fI'.~., 36,
359 (1()40).
97 Colburn, Ind. Eng. Chem., 33, 159 (19'U).
08 Scheibel and Otluunr, 'l'mnN. Am. In.~t. Chern. Enol'.~., 38, 33!l (1042); Ind. liJno.
Chern., 34, 1200 (lU42).
00 Underwood, J. Inst. Petroleurn, 29, 147 (1943); 30,225 (1944).
I. 'l'HEORY 79
are difficult to obtain. loo The transfer-unit concept has been most useful
when the equation is expressed in the form:
or the equivalent:
v, (I"
Nt = i u. (dyj /:1y) = )u., dy/(y* - y)
The H.T.U. and H.E.T.P. for givcn column and operating conditions will
frequently have almost identical values, depending on how nearly parallel
the operating and equilibrium lines happAll to be. lOl A theoretical plate
wjU involve less concentration change than·[L transfer unit if the equilibrium
curve has a steeper slope than the operating line, and vice versa.
It has been shown by Colburn 102 that the J-LT.U. may be expressed either
in terms of the resistanoe to tnLnsfer of material through the liquid or gas
film at the interface between liquid and vapor during distillation, 01' to the
over-all resistance of hath films. The theoretical relations between these
three varieties of H.T .U. are such that it was predicted that the value of the
over-all RT.U. would depend upon the slope of the equilibrium curve and
upon the reflux ratio. Such dependence was noted in several experimental
investigations. 103 Unless otherwise indicated it is the over-all H.T.U.
which is referred to when this quantity is mentioned.
5. Westhaver's and Kuhn's Equations

Westhaver 104 has shown agreement between actual top and bottom com-
position differences in a column and the same values calculated by a formula
based on the diffusion and flow characteristics of the countercurrent. streams
of liquid and vapor. Kuhnl05 has carried through the same type of reason-
ing independently and arrived at the same conclusions. !Ie has also ex-
100 Simon and Hau, Ind. Eng. Chern., 40, 93 (1048), have developed an equation for
expressing column height directly in terms of the mass-transfer coefficients of th(l gas and
liquid films between ph!Ls(ls in distillation.
101 Clark, Tmns. Fnrada1J Soc., 41, 749 (11)'15).
102 Colburn, Trans. Am. In8t. Chem. Engrs., 35, 211,587 (1939).
103 FUl'llas and Taylor, Trans . .c1m. Inst. Chem. Engrs., 36,135 (1040). Surowiec ;tud
Furnas, ibid., 38, 53 (1942). DUnC!lll, Koffol1" nnd Withrow, ibid., 38, 259 (1942).
Gerstel', Koffolt, and Wit.hrow, ibid., 39, 37 (1943); 41,393 (1945). Johnstone and Pig-
ford, ibid., 38, 25 (19'12).
101 Westhaver, Ind. Eng. Chern., 34, 126 (1942).
105 Kuhn, lIelv. Chim. Acta, 25, 252 (1942). Kuhn and Ryffel, ibid., 26, 1693 (1943).
80 A. AND E. ROSE
tended tl1is reasoning to predict the course of batch distillation. lOr. West-
haver's formula applies only to distillation through unpacked tubes in
which liquid reflux exists only as a film on the wall of the tuhe, and the rising
vapor is merely a continuous column. Extension to a conventional packcd
column requh-es the use of arbitrary const!1nts.
Westhaver sets up the equation:
Dv. d(rdy/(b') -I- D ,9j~ - v ~}L 0

r dr I OZ2 oZ
(a) (b) (c)
to represent the dynamic equilibrium of the convention and diffusion proc-

esses that result in constant composition at any point in a column operating
under total reflux on a binary mixture. 'The distillation operation is as-
sumed to be carried out in an unpacked column under adiabatic conditions.
The symbol DI' is the molecular vapor diffusion coefficient, rand Z indicate
distance from the axis and base of the column, respectively, and V is the
vapor velocity. (Some of the original symbols have been changed to con-
form to those used in this book.) It is assumed that t.here is laminar or
viscous flow of vapor, and also that the descending reflux film evenly wets
the entire column wallfl, is of uniform composition throughout its radial·
thickness, and has negligible surface-trlLnsfer resistance. 'The last assump-
tion is justified bee!Luse kinetic-theory caleulations show thnt the V[Lpor
bombarding a liquid surface is of the order of grams per second, most of
which is condensed ancll'eplnced in the vapor by material evaporating at
the same rate. Since this mte of transfer is very great compared with the
net transfer due t.o the distillation opemtionJ thore is every reason to be-
lieve that there is equilibrium between the liquid surface and the film of
vapor adjacent to the liquid. The assumption of uniform radial eompoRi-
tion through the liquid film is (lquivulcnt to assuming thl1t liquid diffusion
is not a transfer-limiting factor.
Term (~t) of Westhaver's basic equation describes the clULnge in eom-
position of the vapor at a pnrtieular point due to diffusion in n. radial dirce-
tio11. This is relat,ed to the difference in (]ompositions at the axis and the
wall of the column, and the rate of change of composition along 1'1, radius.
The term (b) describes the eh:lUge in composition due to diffusion in a ver-
tical direct.ion, and term (c) the ehange due to the motion of the vapor
stream.
At. tot.al reflux there is no net flow of eit.her component because the
amount of the more volatile (Jomponent carried upward by the vapor stream
lOG Kuhn, Ilelv. CMm. Acta, 29, 26 (1946). Kuhn and Baertschi, ibz·d., 29, 692 (Hl46).
1. THEORY 81
is equal to that carried downward by the reflux and vertical vapor diffusion.
This may be expressed as:
(ro
)0 [V(y - x) - D v(oy/oZ)]2m'dr = 0
This equation involves the previously mentioned assumption that liquid

composition is independent of r, and that vertical diffusion in the liquid is
negligible. The equation:
log [ ( -Y -)
1 - Y Z=Zt
/ (- -
1 - 11 z=o
y) ] = (Z/log a) / (11-48. Va1'~
- - +-
Dv
,Dv)
V.
is then derived by use of Poiseuille's law for laminar flow, and the further
assumptions that downward vapor velocity near the liquid surface is
negligible, that change in composition is the same along the axis as along the
wall, and that the relative volatility may be expressed as:
log a =
(y - X)r=/'o
_~_---=-_:c__
Yr=l'o(l - Y)r-ro
In these equations Zt is the total height of the column, Va is the radially
averaged vapor velocity (flow in cubic centimeters per se<iOncl divided by
7Tr~). The symbol 1'0 is the radius from column axis to the downward-mov-
ing liquid surface. Since a column of height equal to the H.E.T.P. has:
l = H.E.T.P. = (11/48)Va(r~/Dv) + (Dv/Va)

and for taller columns:
Zt
n = -----~---------
(11/48)(Var~/Dv) + (Dv(V a)
The minimum value for the H.E.T.P. can be obtained by taking the ,de-
dvative of the equation: '
l = (1l/48)(V arVDv) + (Dl·/V.)
and setting it equal to zero, whereupon:
lmin. = 0.96ro
when:
Va = 2.1Dv/ro cm./sec.
82 A. AND B. llOSE
The latter is of the order of 0.5 em. per second, which is so low as to be un-
likely to be att;ained experimentally, and is about one-tenth of the values
llsed in practice. The term Dv/V a arises from the back-diffusion term,
Dv(oy/oZ) , in the bfLSic total-reflux equation:
1 )
"0 [V(y - x) - Dv(oy/oZ) ]2-JTTdr = 0
At vapor velocitie:-; eornpamble with those in aetun'! use, ba.ck diffusion is

negligible and the term Dl'/Va mny he dropped, so that:
l = 11 VA/48D v
Thus smnller vnlnes of l or H.liJ.T.P. should be attainable by using smaller
vapor velocities, smaller-bore tubes, t1ncl by increasing the diffusion co-
efficient- All these eonulusions are in a,greement with experimental ex-
perienee, and vELlues of l calculated by the last eqnation are in reasonable
agreement with those of experiments, as shown in Figures 1 and 2 of West-
haver's original paper.
For the (l[Lse of finite reflux Westhnver assumes l1 small upward flow of
vapor (n V,,) beyond that equivalent to the reflux liquid. The equation:
(I"u [V(y _
}o
k) - Dv(oy/oZ) ]27l"1'dr =0
is replaccd by;
(1'~ -
1 "0 [
V(U - x) -
oy
Dv-::;.~Z~
u"
- ZA-V I, --2---
1'0
1'2) ]
27frdl' = 0
and the derivation then gives:

l' = Zll - [(V,,/nV a )(1 - y)(log 0:)]-1)-1
wh8re l' is the H.E.T.P. under finite reflux, Val nVa = RIJ 1, and l is the +
H.E.T.P. under total reflux. This equat.ion indiCELtes t.hat u high reflux
ratio at finite reflux is needed to obtain a product approximating the com~
position of that secured under total reflux. Westhl1ver gives the illustra-
tion t.hat RD must be at least 200 for n mixture with ex = 1.1 in order that l'
be 'within 5% of t. The equation for l' may also be used to compute totnl-
reflux H.E.T.P. when a vuluc at finite reflux is known. No examples of
such calculations have been published.
Westhaver assumes maximnm economy of produetion occurs when l'Ru
is a minimum, and for this condit,ion shows that l' must eqlULI 2l. This
means operation with a reflux ratio that will ha,lve the number of pli1tes
obtained at total reflux. POl' this RD must be 2/ [(1 - y) log (Xl.
r. 'l'HEORY 83.
He has also derived the similar equations:

l' = (17/35)(Vas2/Dv) + (Dv/V a)
and:
for operation between elongated parallel plates a space of 28 apart with,

respectively, both sides and one side only wetted by reflux. A derivation
for a packed column was not attempted but the available H.E.T.P. data
for packed and empty columns indicate that the turbulence and reduction
of dimensions produce at least a tenfold decrease in H.E.T .P. A decrease
in pressure should increase the diffusion coefficient and thereby also reduce
H.E.T.P., according to these equations.
The expression:
t"T dT
-=.--=
is given by Westhaver for the total temperature gradient required to obtain

a stable H.E.T.P. equal to l a,t total reflux. R is the molar gas constant
(not the reflux ratio here), X the heat of vaporization, T absolute temper-
ature; p is column pressure, l'J the viscosity of the vapor, and M the average
molecular weight. The instability of l arises from the fact that the velocity
of vapor flow in an open-tube column depends on the longitudinal temper-
ature gradient, which is the sum of the flow-producing gradient and that
necessary to sustain the weight of the vapor even in the absence of flow.
Since the weight-sustaining gradient is usually relatively much 'larger, a
small change in total temperature gradient will cause a large change in the
flow-producing gradient, and thereby in the va.por velocity. Thus for a
mixture of 95% benzene and 5% ethylene chloride the tempera.tul'e require-
ments to hold H.E-T.P. in the range 1 :!: 0.1 cm. are shown in Table VII.
As a consequence of the apparent severe limitation on temperature gra-
dient, Westhaver suggests the use of a forced stabilization of the temperature
gradient by an auxiliary column outside and independent of the working
column. Independent corroborative evidence along these lines comes from
Podbielniak'slo7 discovery of the need for a controlled temperature gradient,
in precise analytical fractionating, and in the observations of Rose 10S that
very small but markedly unstable H.E.T.P. values are obtained in a well
insulated open~tube column operating at very low vapor velocities.
107 Podbielniak, Analytical Distillation and Its Application to the Petroleum Indust1·Y.
Podbielniak, Inc., Chicago, p. 8.
84 A. AND E. HOSE
TABLE VII
CJl.LGULA'l'Jo;J) TJ"i\ll·gRA'rumc-GItADI~]N'l' l{lli(~uLHlliMEN'l':-;'o"
.._-_
'rubo rEt(_lills 1:J 1'essHt'o, dt/dZ. PermiSHihlo varln.tion

('-o).OIll, atm. °O./e111. of dl/dZ, (i~
0,95 1.0 !).O X 10-0 ='=0,0017

0,47 1.0 n,o X 10-5 0,027
0.47 0,1 8,'1 X 10-· 1.1
0,20 0,1 3,5 X 10-.1 7.0
0.10 0,1 4,5 X 10-'3 9,8
Westhaver has summarized the fLssnmptions involved in his hasie equa-

tion:
as follows:
(1) Mixture is close-hailing, i,e., Ci = 1.0 to 1.1. (2) Operation at
equilibriuIllundel'tot!Lll'cfiux. (3) Vapor-flow luminu,r. (4·) Vapor veloe-
ity constl1nt and independent of height (uniform vertical pressure gradient
and no condensation in column). (5) Vnpor veloeity In.l'ge compared with
liquid vcloeity. (6) Liquid filIll flows uniformly over entire wall surface
and has neither surface-tmnsfer resistance nor radial-diffusion resistance.
(7) H.1D,T.P. is large compamd with tube radius. (8) The diffusion eo-
effieient is independent of vu,por eompositiun. (9) BlLek diffusiun is
negligihle.
It is clear that the equations and c(lUcilu;ions of West}uwer woulll noi;
apply if relatively thick liquid films were involved, nor would they apply if
there was a slow-moving vapor 01' liquid filIll with apprceiable diffusion re-
sistance, near the surfaee. The applic:at.ion of the equatiolll:! to conditions
in which column composition is c:hanging, e.(/., initial separation period of
any distillation, or entire period of batch operat.ion at finit.e reflux, would
involve equating the original equations to the c:hange in composition with
time.
6. Effects of Pressure
Superfieial examination of the equations of Westhavcr indicates IL two-
fold effect of deel'eased pressure. Both of these result from the inerease in
the diffusion coeffici.ent. This increases the rate of mixing in a horizontal
direction, which aids diffusion bet, ween phases and thus tends to improve
separation. At the same time mixing in a vertieal direction oJso beeomes
more rapid, so that the vapor stream cannot sllpport a vertical concentra-
tion gradient. There is also a reduction in the eatOe of establishment of
108 Rose, Ind, Eng. Chem., 28, 1210 (1936).
I. THJ~ORY 85
equilibrium between phases due to reduction in absolute vaporization and

condensation rates at low pressures. The practical operating difficulties at
low pressure are well known, and include the high vapor velocities and
appreciable pressure drops, as well as the effects of hydrostatic head and
surface tension on vaporization. Byron, Bowman, and Coull 109 have ex-
pressed the theoretical relations by the graph of Figure 27. The reduction
of the vaporization and condensation rates as the pressure approaches zero
more than counterbalances the high transverse diffusion since the latter in-
W
.J
m
<t
a:
o
>
Lt
Fig. 27. Hebt,ive importllllce of severlll factors in determining effcct of pres-

sure on separation in fractionation. loo
fIuences the vapor phase only. At high pressures separation also ap-
proaches zero chiefly because of poor transverse diffusion and at the critical
pressure separation must be zero. It is to be recalled that the above dis-
cussion applies to contact rectification, and that thermal rectification is in-
fluenced only by improved longitudimtl diffusion as pressure decreases. It
is for this reason that thermal rectification is advantageous for vacuum-
distillation operations.
Docksey and MayllO derived expressions for the form of the equation
relating column height for a given separat.ion to column diameter, vapor
velocity, pressure, and presence of a diluent. Different expressions were
obtained for streamline and for turbulent flow. It was assumed that the
interchange of material between liquid and vapor depends on the rate of
diffusion through the layer of vapor in which flow is streamline. Experi-
100 Byron, Bowman, and Coull, "New Thermal Rectifying Colunm," High Vacuum
Symposium, Camb.ridge, Mass., 1947.
110 Docksey and May, J. Inst. Petroleum 'l'echnol., 21, 176 (1935).
86 A. AND E. ROSE
ments with benzene-toluene and with aniline-nitrobenzene mixt,ures sup-

ported the theory. The conclusions are summarized by the relations:
Z ex: (V1'2/D v ), or Z 0: (V1' 2f.id 2/v)
for streamline flow, or for turbulent flow !1t eonstant Reynolds number, and:
vr2
Z 0:: -----<
D.(vrp/ TJ) 0.8
for turbulent flow in general. In these equ[1tions Z is the height of (Jolumn

for a pal'ti(Jular separation, V is vapor velollity, p itl denflity, 'IJ is viseosity,
l' is column radius, Dv is the diffusion cooffieicnt, N is number of molecules
pel' milliliter, d is effective average molocuIn.r diameter, and I) is molcculf11'
root moan square velocity. An empty-tube column is assumed, and sim-
plifying assumptions are made to obtain values of several of the above
quantities for the mixtures tha;t are necessarily involved. No further con-
firmation or study of this !1pproach has been reported since the work of
Docksey and M!1Y.
7. Effects upon Efficiency When Usua! Simplifying Assumptions

Are Not Valid
All the del'iv!ttions and equations described here for calculating the num-
ber of theorel;illal plates 01' otherwise expressing separating power of a
column or pl1cking ass.umc t1111t the moles of reflux liquid flowing arc uni-
fOl'Ill throughout the eolumn. The equations of Scction III are worked out
with the reflux liquid rate, L, expressed in moles pel' unit time, and assumed
to be the same at all parts of the column. The equations could be derived
equally well with L on a weight basis. However, it is rather likely that it
is the volume rate of flow of reflux that is uniform throughout the column.
Fortunately the densities of many organic compounds are in the range 0.8
to 1.0, and with such mixtures weight, t"1ud volume will be approximately
proportionall'egardless of composition. This is noL the case for mixtures
in which OIle component has a high clensii,y, suell as benzene-carbon tetra-
chloride. The literature records 110 methods of dealing with this situation,
although Colburn and Stearns111 have pointed out that the basic material~
balance equations may be developed in terms of volume anel. volume pOl'
cent if the volume changes on mixing ul'!lucgligiblo.
In a column which is losing an appreciable amount of heat iihrough its
walls, the reflux ratio is greater at the bottom thu,n at the top. The
throughput at the bottom will be limited by the amount of vapor that can
pass the relatively large amount of reflux at the bottom of the oolumn.
111 Colburn and Stearns, Trans. Am. Inst. Ohern. Engl's" 37, 29l (1941).
1. 'l'HEORY 87
The upper part of the column will have greater capacity for vapor, but even
less vapor will be passing than at the bottom, becanRe of the partial con-
densation of the vapor in ri'3ing through the column. The effeetiveness of
separation will be different at different parts of the column and will be very
sensitive to slight changes in the rate of heating the st.i.ll, and to drafts or
other influences in the transfer of heat through the column walls. The
column mayor may not show greater difference tha,n an adiabatic column
in composition between top and hottom. This will depend on the com-
bined effects of the different reflux ratios and the diffuBionalresistances at
various points in the column. The throughput will almost certainly be
less than for an adiabatic column giving the same composition differences.
z
o
f=
«
Il::
~
lJ.J
(f)
Fig. 28. Qu:tlitative expression of results of llonadhtbatic operation. 100

(I) COIllhill:J,t,ioll of thermal and contact rectification. (II) Thermal rectifica-
tion only. (III) Contact rectificl1tioIl only. (Q, -Q) Heat added or sub-
tracted.
The usual reasoning is that there is a particular vapor rate that gives the
best separation, (lonsidcring the diffusional resistances, the reflux ratio,
compositions, and rate of production of distillate. If a column is not
ndiabatic, only one part of it can be operating at the optimum vapor rate
and the rest must be under less favorahle conditions, so over-all effective-
ness must be less. There [Ire weaknesses in this reasoning. One is that an
appreciable range of vapor rates might give the same result, because of a
combination of changes in reflux ratio and diffusional resistance. In this
case nonadiabatic operation would not affect separation as long as the vapor
88 A. AND E. ROSE
rates romained within the specified range. Another instance would arise
when a combination of vapor rates at various points in the column gave
better results than any single rate, because of the changing effect of concen-
tration, surface tension, etc. on the diffusional transfer between liquid and
vapor.
Byron, Bowman, and Coullion have analyzed the effects of nonacliabatic
operation by means of a graph such as Figure 28, in which the degree of
separation or rectification is plotted as ordinate and the quantity of heat
added to or lost from the column is plotted as abseissa. If separation due
to contact rectification alone is considered, the reRulting curve is be11-
shaped with a maximum for zero heat transfer to or from the column, and a
graduoJ decrease in separation for either addition (Q) or abstraction (- (J)
of increasing quantities of heat. However, for any eondition of heat addi-
tion or removal a eomposition ehange occurs due to partial evaporation of
liquid or condensation of vapor, i.e., thermal rectifiell,tioll will occur. The
curve for the resulting change of composition will be zero for zero heat
transfer, but with a parabolic rise as heat transferred is increased. The
net effect of the contact rectifieatioll and thermal rectification taken to-
gether is shown by the curve with a maximum at zero heat transfer, slight
minima at either side, and rapid rise for greater heat transfer.
Thus small deviations from adiabatic operation should theoretically re-
duce efficiency, but large OIles should improve it. In practice it, is difIicult
to operate with highly Ilonadiabatic conditions because reflux becomes vcry
large ~1lld vapor flow very small, as deseribed earlier. This diffieulty is
avoided by the speeially designed thermal rectifying column of Byron,
Bowman, and Coull. Simonetta ll2 h[18 also reported results on the opera-
tion of nonadiahatic eolumns.
8. Bowman's Equations
Bowman and 131'i:1nt 1l3 have presented a comprehensive dis(lUs:,lion of the
theory of packed columns, following and extending the general reasoning of
Chilton and Colburn, Westhaver, and Kuhn. The hasic mltterial-balance
equations for transfer in and out., and aecmhulation by, a segment of liquid
and vapor are, respectively:
and:
112 Simonetta, Chimica einduslria Milan, 28, 114 (1946).

l1S Bowman and Briant, Ind. Eng. Chem., 39,745 (1947).
r. THEORY 89
in which L is the liquid-flow rate down the colnmn, V the vapor-flow rate
up the column, and H Land flv are the total holdup in the liquid and vapor,
respectively, i.n the segment under discussion. 'The subscript i indicates
that the various quantities apply to it particular point in the column.
There are as many pairs of these equations as there are components. The
first terms represent transfer into the segment under consideration as the
result of the downward flow of liquid or upward flow of vapor. The second
Yt
y I
I
I
I
I I
"'_~-l
I I
I I
I- X .x* _I
I I
I
o x
Fig. 29. Determinlttion of Bowman's 1> all X,11 graph. ll3
term represents llet accumulation or depletion as the result of the transfer

due to the first and third terms. The third term represents rate of transfer
into one phase from the other. The variables Z and t represent height of
column and time, respecti.vely. 1{J> is a quantity analogous to H.E.T.P. or
H.T.D., and cp is 'a measure of the driving force causing transfer from one
phase to the other. This rate is ~ssumed to be proportional to the amount
of the component which would have to be transferred from one phase to the
other to bring the two phases into equilibrium, assuming equal amounts of
the two phases. The last requirement is reasonable because the transfer
rates are influenced principally by the concentrations at or near the inter·
face, and not by the depth of material behind the interface. The value of 1>
may be represented on the X,y graph for a binary system as shown on
90 A. AND E. ROSE
Figure 29. * It is proportional to the distance along a line that has a slope of
-1 between the operating point and the equilihrium line. For eases to
which Henry's l!1W is applicable:
4> = (lex - y)/(k + 1)
where Ie is the Henry'S law constant. When lhoult.'s law applies:
11+4> ~!:-4>
------- = a --~
l-Y-4> 1-:!:+(/)
or:
2¢ = x - 11 + ~~±_2 - ~I~~ + 1J)2-~ 2(x ~- y)~(-~;-+-!:)~

a-I a-I
It is possible to represent the analogous relation for nonidenl mixtures by
equations of the type:
24> :::: x - Y + ((T'B/A)(A - B)(l - B)
where lJ = (x + 7/)/2, IT' = f(a) and is n measure of ease ofsoparation, andA.
is related to the extent of nonidenJity.
For steady-state eonditions as in continuous distillation the basic
material-balance equations re~lueo to:
and:
l' (011t/OZ) - (2cjJt/ II "') = 0
These may be combined to give:
This on integration gives:

Vy - Lx = C1
which is the conventional operating-line equation with:
C1 = (V - L)lIn or
depending on reference tiO the enriching or stripping section. The mate-
rial-balance equations also give:
2Z/I1", = l' f (dyjcp) = L f (dx/4»
These integrals may be evaluated graphically by plotting the equilibrium
and operating lines of an x,y diagram, then measuring tihe values of cjJ, llud
* Figures 29-32,50-52,55-58,63, and 64 conrtosy of Industrial and Enaineerlna Olwm·
istry.
I. THEORY 91
plotting 1/4> against Y or x as desired. The resulting area is 2ZjVH", or

2Z/LHq,. When Henry's law applies, and operation is at total reflux in
an enriching column the integration may be performed algebraically to
give:
~=~
VII,,, Z a-I
(a + 1) log y.
?f..:_
This compare,'! with:

11 = Zjl = (l/log 0/.) log (ye/YB)
for conventional plate columns where l is H.E.T.P. For all values of 01.
ordinarily encountered, Ijlog a is substantially equal to (a + l)j2(a - 1).
When Raoult's law and total reflux are applicable, a similar correspondence
is obtained, that is:
_£ = ~ (01. + 1) log YIC! - y,)
VII", 2 a-I ys(l - YI)
whereas conventional plate theory gives:
1 I y,(l - y,)
n = - - og
log IX YB(l - YI)
Similar correspondence occurs for the analogous partial-reflux expressions.
For the case of appreciable holdup in batch distillation, nonsteady-state
conditions are encountered and the problem is more complex. However,
the general procedure has been worked out in detail by Bowman and
Briant for the case in which Henry's law governs the vapor-liquid equilib-
rium compositions. Although the mathematical problems are involved,
t.here is reason to expect that the method can be developed so as to prove
useful for calculations of initial equilibration time in batch distillation,
effect of intermittent talce-off and other irregularities, as well as normal
batch-distillation operations.
The most useful result of the Bowman and Briant analysis has been the
agreement with numerical values obtained by analysis based on transfer
rates in plate columns. Thus the conventional plate equations may be
used wit.h additional confidence for packed columns, even though they are
without theoretical basis.
9. Composition Relations for Multicompol1ent Mixtures

Bowman114 has devised the scheme of representing composition by a
function x( a), in which the mole fraction with relative volatility in the
1U Bowman, Ind. Eng. Chern., 41, 2004 (1949).
92 A. AND E. ROSE
range a to a + da is xda. This is equivalent to using a ItS n, variable that

dOflignntes the several cornpononts. In such a scheme it is most convenient
to take the relative volatilities as the ratios of the vapor pressures of the
cOlnponents to the total pressure at some fixed temperatul'{). Fig\ll'(J 30
shows curves expressing the compositions of two typical mixtures. Curve I
eorresponds to a mixture containing equal amounts of an infinite num her of
components uniformly distl'ibuted with respect to their values of a. Curve
II corresponds to a mixture rich in intermedia.te-boiling materinJ, and with
smaller nmounts of high- and low-hoiling components. Those eomposition-
distributioll curves are analogous to true hoiling-point CUl'V(JS. The area
under all composition-distribution curves
must be unity, i.e., mnthematically:
1'" x da = 1
X I
This is merdy equiv[Llent to the requirement
t.hat the sum of the eOllcentrations of all COlll-
ponents must he unity. A simple example
of the use of the composition-distribution
concept is the derivation of the relation be- 2
tween the compositions on various plates in Fig. 30. Composition exprcRsed

a column under steady~st[Lte conditions with in terlllH of relative volatility.llol
reflux ratio high enough so that separation
is determined by the number of theoretical pla.tos. The eOllvelltiounll'ela-
tion for a mnltico mponent mixture is:
C~j:rn(i)
Yn(i) = -----
r:(~iXn(t)
i
where y" and X" are the eompositions on the nth plate, and the subseripts i
imply the values for the ith component. Tmnsbted into composition-
distribution relations, 1;he above becomes:
Ynda = (aa.~nda)/(.fo'" ax"dec)
from which the dec factors on the left and in the numemtol' divide out at
once. If the feed plate is assigned the subscript 0, and plates 11bove are
numbered in sequence:
yu = (axo)/(,fo'" ec;t;oda) = Xl
the latter following because Xl = yo for the i1ssmned cOllditiollf; of high re-
flux. l3y substitution of this expression for Xl in:
X2 = YI = (axl)/(..!o'" ecx1dec)
I. THEORY 93
there is obtained:
X2 = Ca 2a.:o)!C';;"' a 2x oda)
and by continuation of the process:
Xn = (anxo)!C';;'" anxoda)
Application of this procedure gives Figures 31 and 32 for the composition
distributions on several plates in a column in which the feed-plate com-
positions are those represented by Curves I and II of Figure 30.
The same type of reasoning gives the equation:
XD = [CRD + 1) - CRDa-l!fo'" ysa- 1 dOl) ]Y8
for steady-state operation in an enriching column in which separation is
determined solely by reflux ratio because the number of theoretical plates is
Fig. 31. Composition distributions Fig. 32. Composition distribution

on four-plate column at total reflux on fourt,h and eighth platea at. totnl re-
with feed or charge composition I of flux with charge composition II of
Figure 30. 114 Figure 30. 114
high. The equation is applicable only for the special case in which there is a
pinch point for all components at the bottom of the column. Equations
for more general cases arc also possible in principle, but detailed equations
have not been derived. Applications to simple cases of batch distillation
are presented in Section V.
94, A. AND El. ROSEl
Kuhn and Baertschi llIi have used n, different method to arrive at the
change of disti.llate composition with time in the distillation of [t multicom~
ponent mixtUl'c.
10. Cohen'S Equations

The treatments of Westhaver, Kuhn, and Bowman deal wit.h t.he int()r~
ehange bet.ween vapor and liquid streams after a eolumn has reached equi~
librium at total reflux, and after a steady state is reached in continuous
distillation at finite reflux. Cohen llO has considered the changes occurring
during the initial period of operat.ion I1S the column approaches a steady
state. '1'hi8 is important because of the length of such a period when iso~
topes 01' other very close boiling mixtures are being separated. Coulson,
and Berg and James 1l7 have also derivecl expressions lor tho time required
to reach equilibrium; these are discussed in Scction V.
Cohen equated the transfer in and out of a sm::tll nnit length of packed
eolumn to obtain the equations:
H,/)x'lot = (-Lux'loZ) -+ (1I,/Jx'IBt)
and:
I-lvoy' lot = (11oy'102) - (Ih'ax' jot)
for the liquid and vapor phases, respcctively. In these x' is the mole frfLc~
tion of the less volatile componcnt in the liquid, H L the total liquid holdup
per unit length, L the liquid reHux flow rate, while y', III', and V are the
corresponding quantities in the vapor. 'l'he length of the column is Zt,
while Z is the distance from the top of the column to the unit length under
consideration; t is time. Cohen considers only continuous distillation or
its equivnlent, and for total reHux obtnins:
')
(_]j__
1 - y' tal)
=
(')
_JL__
1 - y' feea
e2q (1-Z/z,l
wherc:
(J' = (lccC/2L)(1 - cx)Zt
and:
a
[N][m]
= -------.-
[M][n]
= (
Xi I y' ) X'
1 - xi 1 - y' = Y'
116 I{uhlll1ml Baol'tschi, Ilelv. CMm. Acta, 29, 692 (1946).
110 Cohen, J. Glwm. Phys., 8,588 (1940).
117 Coulson, J. Soc. Chern. Ind. London, 64, 101 (1045). Berg and James, Trans. Am.
Inst. Chern. Eng,/,s., 44, 307-314 (1948).
I. THEORY 95
[N] and [n] are the concentrations (moles/liter) of the less volatile com-
ponent in liquid and vapor, respectively; cis [m] + [n], and Cis [M] +
[N], where [M] and [m] are the concentrations (moles per liter) of the more
volatile component in liquid and vapor, respectively. The constant k is
defined by the relation, based on analogy with lL bimolecular reversible
reaction:
HLox'/ot = -k([N][ml - a[Ml[nl)
The difference in composition between the bottom and top of the column
is given by:
2a
Y'(O)top/Y'(Zthouom = e
where Y'(O) signifies the value of Y' when Z = 0, and Y'(Zt) the corre-
sponding value when Z = Zj'
E.e-examination of t.he expression for rr shows that for values of a near
unity e2<Tis neal'ly equal to (l/a)kCaz/L. Thus:
Y'(O) (y,) / ( y' )

Y'(Zt) = 1 - y' t01Jo/colltrnn 1 - y' leed
is nearly equal to (1/ a)kcaz/L. Recalling that 1 - 1/' refers to the more
volatile component, one recognizes the above as the Fenske equation with
kcCZt/L representing tIle number of plates.
For the morc general case of finite reflux, Cohen obtains:
Y'(O)
---- -
(1 -FE) + ()
----~--~-------
(1 - y)(e- + 8 2a )1+ao
where D/L is the fraction of flnw drawn off (the inverse of ordinary reflux
ratio) and:
fJ = (D/L)/Cl.- a)
As an example Cohen calculates data for the case a ::= 0.98 and when (J' has a
series of reasonable values. Even a very small withdrawal of product such
as 0.02% (8 = 10-2) effects a marked reduction in the ratio Y'(O)/Y'(Z!).
The above relations are only approximate solutions of the basic differential
equl1tions of Cohen. When the quantity Y' e2a is greater than 0.5, it is
necessary to use more complex, exact solutions of the equations.
Cohen has :11so derived the expression:
Y'(O)t=t = Y'(O)t"'oe 2a - Cle-Iodt - C2e- 1B'lt - ..•
96 A. AND E. nOSE
in which t refers to time, CI , C2 , .,. are positive conl:itants, and 01, Ih., .. are
pal'tLmet.ers. Because of relations among the constants and paramet.ers the
time of coming to equilibrium is rlet.onnined by the parameter Ih. rrhis is
dependent in a complex manner upon (x, L, Ih, Zt, and IJ". In an example
Cohen obtains:
Y'(O)t~t C' ()- J ('\ -II,I)U~21 I) 18n-2,l1t () ()h)" -(l,Ulit _ OI)'Je-12t

·Y'CO)t~O = l. i) - ,to ) .. c - ~." - .. c, .~ D - •••
and a similar equation for the transport of hmwy material from vupor to
reflux strettm, ltnd fimLlly the value of X'(Z)/Y'(Z). FigureH 5 and 6 of
Cohen'H original pnper llS describe the vttl'intion of ull these wit.h time.
11. Marshall-Pigford Equations
Marshall and PigfordllQhave also derived the basic equations for ml1.terial
transfer in packed and plate distilht.ioll columns, as well a.s ot.her [).un.IogmlH
countercurrent. diffusional opemtions, l~or a rmcked column the bl1sie
relationship is:
OGOL(CP'y/ot 2) + COL - flG) (Q2 y/O nuy) - ('02 Y/O n 2) +
0
( [,
+ ~:!~
L
OG) ?Ji -I- (~.!_.1 -
at L
1) 91t'On : ; ; 0
in which fla and 0[, are, respectively, the holdup per transfer unit in vapor
and liquid phascs divided by the corresponding flow ratcs, V mHI L, n is t.he
dist.ance up the column (mol1Hured in transfer units), t i~ time, 11 is the mole
fraction of more volatile component in the vapor, and (~ is a consttl.nt de-
fined by the simplified vapor liquid equilibrium rebt.ion 1/ = ax b. +
FOl' a plu.t.e column the corresponding eqlmtion is:
(L/V)(h(oxn/ot) + Oa(01/,,/ot) = (Yn-l - y,,) + (L/V)(X n-!-l - x,,)
where Xn , 1/", etc. refer to liquid and vapor concentrat.ions from plates incli-
oated by the subscript.
Marshall and Pigforcl make use of tho si.mplifIed equilibrium relation 11 =
ax + b and the I,aplaco iimnsformatinn in order to obtain solutions to these
equations. Thus the first of the two preeeding equn.t.iOlll-l is made the hftsis
for estimating the rn.te of (lpproaeh to equilibrium of a column opemting at
total reflux. They n.ssurne that the column is initia.lly filled with liquid
ident.ical in composition to t.he vapor entering the column, and the latter is
assumed to remain const.ant. during the approach to equilibrium. The
!l8Cohen, J. Chem; l'hlJ.~" 8, 588 (1940),
11UMmshllll and Pigford, Applicat-ion of Dij)'crmtlial Equations to Chemical Enainllering
pj·ablclIUl. Univ, Delawa.re, Newttrk, Dell1ware, 19·17, p. 144.
I. THEORY 97
reflux returning from the condenser is assumed to have the same com-
position as the vapor leaving the top of the column. These conditions con-
stitute the boundary conditions required to evaluate the constants intro-
duced in the course of the application of the Laplace transformation and the
subsequent solution. The mathematic:al operationtl are somewhat ex-
tended but the end result is a graph in which the number of transfer units
in the column is plotted against the number of times total column holdup
must be changed to achieve a particular degree of approach to equilibrium,
say 95 or 99%. With log-log scales the relation is nearly a straight line.
This relation involves the further approximation that the equilibrium line is
parallel to the 45° line. When the relative volatility is near unity the
number of times that the column holdup must be replaced is approximately
equal to the number of transfer units.
Pigford, Tepe, and Garrahan 120 have used the preceding plate column
'equation as the basis for predicting the effect of holdup in batch distil-
lation (page 150).
V. DISTILLATION CURVES IN BATCH DISTILLATION
In Section III the separation achieved in a distillation process was ex-
pressed by still-product composition curves. Such curves can be used for
relating still and product compositions, relative volatility, number of theo-
retical plates, and reflux ratio. They are applicable to continuous distil-
lation if certain simplifying assumptions are valid. Their usefulness in
batch distillation is limited because the change of composition with time
distorts the steady-state relations.
The results of a batch distillation may be more clearly and concisely pre-
sented by a curve plotting the quantity of distillate versus its composition.
If desired, the curve for distillate quantity versus still composition may also
be included on the same graph. Such batch-distillation or collection curves
show the purity of the product at all times during the distillation, lLnd from
them the average composition of any fraction of the distillate may be ob-
tained. Thc sharpness of the breaks and the flatness of the plateaus in the
distillation curve are a measure of the effectiveness of the operation.
Such curvcs call also take into consideration the factors of holdup and
initial composition.
Boiling-point curves are of this general nature, and have long been used
to follow the progress of a distillation. Refractive index and density are
often a more satisfactory means of measuring distillate composition, be-
cause boiling points are often inaccurate when obtained directly from the
thennometer at the head of a column.
120'Pigfofd, Tope, and Garrahan, "The Effect of Column Holdup in Batch Distilla-
tion," Am. Chern. Sao., Philadelphia Meeting, 1949.
98 A. AND JoJ. ROAlD
The literature and tradition of batch distillation inelude mueh qualitative

information regarding the effect of reflux ratio, relative volatility, number
of plates, and other process varil1bles on the shape of the (list,ilbtion cmve.
It is remarkable that there are few direct quaIltitl1tive dl1tu, on these l'oln,-
tions. Batch distillations in particular seem almm,t, always to have been
carried out in order to get something distilled or analyzed, rnther than as
part of ttIl investigation whose primary object was to improve understand-
ing of the process of distillation. Most of the latter type of work is quite
recent and experimental data are very limited in seope.
In the present Hectioll equ[1tions are dmived and methoels desGribed for
calculating distillation curves with whatever values of the reflux ratio,
number of plates, etc. are of interest. A series of curves may then be
worked out to ostim(1te the effeet of changes in the proeess vtLriablotl.
Numet·ous speeifie examples of the resultfl of H1Wh ealllulatioll ltrO given in
Seetion VI. The oqlH),tiolls and caleuh1tiollS vary in emnplcxity according
to the simplifying aSHumptions used in the dorivatio1l8. The Himplcsi; ea.se
is the gl'aphie pl'Oecciure for s'imple disi,illatiotl. This l1ssumes only tlmt
thero is no holdup nnd that the vapor-liquid equilibrium eurve gives the
true relation between still and produeij eompositinn. The algebmie solution
for tIm R.ayleigh Himplo distillation {l(lllations includes the additional as-
Rumptjioll that the relative volatility equa.tion eorreetly expresses vt1por-
liquid equilibrium ClompositiollS. The most convenient methods for ealen-
iating eurves for fractional distillnt,ion in n, eolumn use the Hayleigh j)ro-
eedul'e but as:'Hnne that holdup is negligihle, the eolulllll is adiabatie, heats
of mixing are negligible, the thel'lllL11 propertieH of the components are
simibr, and that the procedure of Sorel (page 4G) is applieahln in bl1teh
as well as continuous distillatioll, in Hpitc of the gru.dlml ehange in Clom-
position at all points in 11, ba,teh column (the time ef'feet). Any or all of
these simplifying assumptions ean be avoided by use of mOl'f) (lOluplioatr,d
calculatiolls, but little h11,S been done along this lino. All the met,llOds are
illustmted in the following pageH. The equations and methodA of Cl!11cnla-
tiOll are considered in fLpproximate his1;orienl order, which iH also the order
of increasing eomplexity. '
1. Curves for Simple Batch Distillation
An equation for the curve relating distillate composition to amOlUlt of

distilln,te in [1 simple distilll1tion was first derived by YOUl1g,121 This in-
cluded systems of two 01' more eomponents rmel [t!1Pl1rently denlt only with
those mixtures having low rnlntive volatilitios. Hnyleigh 122 derived the
m Young, Distillation Principlos and P'I'OCCSNCS, Mnnmillan, Loudon, 1B22, p. 117'-
m Rayleigh, Phil. Mag., 4, 521 (11)02).
1. THEORY 99
general equation for simple distillation, and applied it to the special case of
very dilute solutions. At a much later date it was pointed out 123 that the
same reasoning and essentially the same equations apply to fractional dis-
tillation with a column, if holdup is negligible. If the holdup is appreciable,
the same reasoning can be extended to derive analogouK equations and
curve;;, but complicating factors enter the situation. 12 '1
A. BASIC RAYLEIGH EQUA'l'ION
The fundamental material-balance procedure of Rayleigh is illustrated

by the derivations and cnlculations for the simple distillation of a binary
mixture. Consider any instant during the distillation when the mole frac-
tion of the more volatile component in the still is Xs and there are S total
moles of material remaining in the still. At this instant assume that the
vapor in equilibrium with the mixture in the still will have composition
Y = XD, and in a perfect simple diHtillation will give distillate of this same
composition. After a very small amount of this distillH.te CdS moles) is
collect.ecl, the total moles remaining in the still will be S - dS, ancl the com-
position in the still will have chH.ngecl slightly to Xs - dx It follows that:
N•
8x s = (8 - dS) (x, - dx s ) + xDdS (1)
or, in words, in terms of the more volatile component:

Moles in Rtill before moles in still after re- moles in small quall-
removal of SIllttll must equal moval of small quall- plus I;hy of distilla.te
quantity of distillate tity of distillate
The above equation may be multiplied out and simplified, dropping out the
negligible second-order differential d8dx s• After rearranging terms, this
beeomes:
Sdxs = d8(XD - x s) (2)
or:
d8/S = dX,/(XD - 3:,) (3)
On integration:
1n (l
i) = f dx,
XD - Xs
(4)
This is the basic form of the Rayleigh equation. In this form, gr!;1,phic
integration is necessary to obtain the numerieal values needed to plot a
particular distillation curve. An example of such a procedure is given on
page 102.
123 Rose and Welshans, Ind. Eng. Chellt., 32, 668 (19-10).
124 Colburn and Stearns, Trans. Am.lnst. Chern. En(JTs., 37, 291 ( 1941).
100 A. AND E. ROSE
n. ALGmmAIC SOLU'l'IONFl OF UAYLEIGII ]i:QUA'l'ION
If Xl) and :", are rclfttcu by the relative-volat;iIjt.v equnt,ioll, then: 12fi
XJ)
x - - ILncl - --~,----- 1 ]
• 8 - a - XD(a: - 1) a - X])(a - 1)
(5)
and:
(5)
so that:
(7)
This may he integrated to give:

1 "
logS = --1 (log Xl)
a-
- alog (1 - XD) + (a - 1) log [a - xn(a - l)]l
+ integration constant (8)

or:
S = lc(Xn)l/(a-l)[Q! - ~t;])(a - 1)]/(1 - XD)"'/(a-l) (0)
Since Xl) represents the mole fraction of the more vo!a,t;i!e component in the
dh;tillt1 te und B represents total moles remaining in the slm, the above
equations represent the theoretical shape of the batch-distillation CUl'VO.
If it is desired to plot quantity of distillate, this may be found by the relu-
tion:
(10)
where D is the moles of distillate nnd Be the moles of charge.
It is also possible to eliminate :i~D instead of x, from the basic equa.tion ('1)
and obtain:
log S = _!____l [log ::1:, - a log (1 - x,,)]

Q!-
+integration wIlRtant (11)
This relates ,~lill composition t.o toiJI,\l moles remaining in the still.
Tho actual use of such equatiollR may he illustrated by ealculatillg the curVe for
the simple distillation of 100 moles of an equimolar binary mixture with eonRtant
relative volatility, IX := 1.25. It is first necessary to find the value of the integration
constant by substituting the initial conditions of the distillation in the equation for
126 Rose and Welshans, Ind. Eng, Chem., 32, 660 (1940).
I. THEOItY 101
log S. The equation for S could be uscd and would give the same end result, .but
the computations are more difficult. Any number of moles of starting material
may be considered, but for convenience in this example 100 moles is used. Thus
log S = 2. Since the initial still composition of the equimolar mixture is x, = 0.5,
the composition of the 'first portion of the distillate is:
XD = aX, = (1.25) (0.5) = 0,555

1 + x,(a - 1) 1 + (0.5)(1.25 - 1)
Using these values of log S and Xv in equation (8), the integration constant is 1.2149.
This constant is used in all subsequent calculations for this example, so that the
equation to be used for finding values of XD and S during the distillation is:
1
log S = 0 2 {log XD
. 5
- (1.25) log (1 - xn) +
(0.25) log [1.25 - Xl)(O.25)]} + 1.2149
0.60
w
I-
<t 0.50
..J
..J
i=
II)
2i 0.40
!!:
u
>
::E
0.30
z
0
i=
u
<t 0.20
Il:
LI..
W
..J
0
::E 0.10
a 20 40 60 80 100
PER CENT OF CHARGE DISTILLED
Fig. 33. Points on calculated batch-distillation curve,
Thus if Xl) = 0.550 (rather than the initial value 0.555), S = 90.5, which represents
a point on the distillation curve for this instance. Repetition of the calculation for
XD = 0.40 gives S = 6.21, and it becomes possible to sketch a rough outline of the
distillation curve (Fig. 33). It is clear that little separation will be achieved and
there is no use in calCUlating further points in this case.
The values of XD used in calculations such as the above are chosen at random
from the range of possible values. Thus in the above case, no value of Xv can be
greater than the initial value of 0.555, because the concentration of the more vola-
tile component in the distillate necessarily decreases as the distillation proceeds.
102 A. AND ]]. ROSE
The first random choice of Xn flhollld always be somewhat less than the value of Xn
corresponding to initial st.ill composition.
It is often less work to use equation (ll) to calculate distillation CUl'ves. The
calcuitttioll for the initial point on the above CUI've (:c s = 0.5) gives 1..(1900 as the
integration constant. From the vapor-liqlli(! equilibrium relation, when XJ) =
0.55, :t" will be D,4,95. Substituting this last value of x" awl U)9HO u.s the integra-
tion const,ant in equation (11) above gives S =' DO.Ii, a value that is identical
with that obtained above. The smaller number of terms in this (\(lImtion makes it,
:-;impler to usc, particularly when corresponding values of x., aIld Xl) IlU1y be read
from a vapor-liquid equilibrium curve.
C. SOLUrrIONS OIl' RAYI,lDIGI-I EQUArl'ION BY GHAPHIC IN'l'WWA'l'lON
If the relative-voln.tility equation does not eOl'l'eetly indicate the relation

between x., and XJ), equations (8) and (11) mtn no longol' be llsed as illus-
trated above. On the other hanel, the fundamenta.! equation (4) can be
used whether or not t.he rolative-vol::l.tilit.y relation i::l vH.lid.. Tn order to
make the (mieulations in sneh t1 ease, the equation must be expressed with
definite limits and signs, and with logal'ithmH to the base 10. The result i:,;:
log 8, = log 8, - -------

- I
1
0 2.:30~1
1'""·"'a'
dx"
-------
XlJ - ;'f,s
(12)
Here 8 t represents the total moles initially present., X.Ie the (:ol'responding
mole fradion of the more volatile component, and 8, and XSI the moles and
composition at a pn,rtieula.r time during the distillation.
For purposes of illustration the above calculations will bo repeated. Tho problem
was to obtain the curve for the simple distillation of 100 moles of all equimolal' bi-
nary mixture with O! constant and equal to 1.25. In this case S. = 100 and X HC =
0.5. Values of ':ClJ and Xs were obtained by solution of the equation:
1- XD
= O!
1 - Xs
(18)
TABLE VIn
POIN'l'S ON Dn,'I'HoLATION CUHYFJ U'lWM H.AYLIWlH EQlTA'I'ION
__ ..
,~_ "._m._~_
:C,' xn - X8 1/(.,]) - :c.) Al'PU,!-i
0.555 0.500 0.055 18.19

.550 ,495 .055 18.19 0.09095
(;f,J) '" O. ,1)51.) to O. 550)
.500 .445 .055 18.19
.450 .395 .055 18.l!l
.'100 .347 .053 18.85 2.80
(XI) = 0.lifi5 to 0.40)
---~ .. ~----~~------------- .. - ------
1. '['HBORY 103
In order to find the value of f~~"d"-~.J(xJ) - x,,), it is necessary to plot l/(x D - x,,)
versus Xs and determine the areas nnder the resulting curve and bctweell x" and
X.,t· Over the range of values of XD (0.555 to 0.40) needed to repeat the above
calculations, the values are shown in Table VIII and plotted in Figure 3,L For
the value of S wlten XD = 0.55:
log 8 1 = 2 - (0.09095/2.303) 1.9(i03
81 = 91.3
For the value of S when a:D = 0.40:
log 8 2 = 2 - (2.80/2.30;) 0.7842
040 0550555
I I I
I I I
I I I
I I I
I I I
I I I
20 I I I
I I I
r-- I I
18
16
AV. I/(Xo-X s ) =18.52
I AV. I/(xo-xs ) =18.19
14
,,<VI Ax =0.048 tu =0.10
I
",'" 12 AREA = 0.89 AREA' 1.82
::::: 10
AREA: (\8.19)( 0.005) =0.09095 __,._
o
0.34 0.36 0.38 0.40 0.42 0.44 0.46 0.48 0.50
MOLE FRACTION MVC IN STILL tXsl
Fig. 34. Graphic integl'l1tion of f [dX,/(XD - x,,) J. Simple distillation.
The use of equation (12) with values of x, and XD from an experimental

vapor-liquid equilibrium curve does not depend upon any simplifying
assumptions except that there is no appreciable holdup or fractionation be-
tween still and receiver, so that still-product composition relations are
those of the vapor-liquid equilibrium curve. Although there is undoubt.-
edly some slight holdup on the upper walls of t.he still and in the condenser
104 A. AND E. Rosr~
during ordinary simple distillation, and slight fractionation may occur,

these effeets are probahly small. Sueh ealculations have bccn regularly
made in eonnm:tioll with the design and operation of Birnple distillation
npparntlls. 12fJ N omngmplts h:wo been worked out for the solution of the
Rayleigh equati()n.l~!7
2. Curves for Fractional Batch Distillation When Holdup

Is Assumed to Be Negligible
A. RA.YLmIGII ))'OHM OF l~QUA'l'roN
The derivation of tile equation for the distillation eurve for this case is
almost identieal with that for (,limp Ie distillation. Consider any instant
MOLE FRACTION MVC IN DISTILLATE (xol

G 0.20.4 06 08 0.905
!w 1.0
~AREAO"5
I-
« 5
_J
_J
0.6
j:
(/)
0
~
u
>
::!;
4
3
:V
:
: I
AREA 0.1869
/
AREA 0.3097
AREA 0.456
'
: :'ri- I~
z
0 2
i= I I I I I
(.)
«
a: 0.2
I I I I AV. HEIGHT I
lJ..
I I I I 2.23 I
I I I I
W
.J I I' I fJ. x = 0.205 I
0 I' I I I
::!;
0 0 0.2
0.2 0.4 0.6 0.9 1.0 0.1 0.3 0.4 0.5 0.6
MOLE FRACTION MVC IN STILL (xsl
Fig. 35. Still-product composition CUl've. Fig. 36. Gmphic integratiou of
a = 1.25, n = 10. Total reflux. f [dx./(xn - :1:.,) J. Fractional distil-
lation. Ordinate, l/(xD - :l:s).
during the distillation when the mole fraction of the more volatile com-
ponent in the still is Xs and there arc S total moles of material in the still.
At this instant, assume thnt the distillate composition is XD' After dS
moles of distillato passes off, Ule total moles remaining in the still will be
S - dS, and the composition in the still will he x. - dx.. As before, equa-
tion (1) is obtl1ined, which on solution and integration gives the fundamen-
tal equation (4). This IMY he expressed in the form suitable for caleula-
126 Robinson anu Gilliland, Elements of Fracl'ionltt Di8tilllltion. ard ed., McGraw·lIill,
New York, W3n, pp. 47-50.
121 Nord, Ind. Eng. Chem., 39, 232 {1!}4'l). Stanton, ib'id., 39, 1042 (1947).
I. TH1<JORY 105
TABLE IX
POINTS ON DISTILLA.TION CURVE FROM RAYLEIGH EQUATION
'"81 XD - X" 1(XD - ",.J Areas
0.905 0.500 0.405 2.466

1.n8 0.456
.800 .295 .505
2.174 .310
.600 .140 .460
3.015 .187
.400 .068 .332
5.65 .195
.200 ',023 .177
-._._-----_._
tion (equation 12) and evaluated graphically by use of still-product com-

position curves (x" Xv curves) such as }"igures 18-20, 22, and 23. The
method is essentially the same as that given above and may be illustrated
by calculation of the curve for fractional distillation in the case when Se =
100, X.e = 0.5, and the still-product relations are those given in Figure 35.
The details of the calculation are given in Table IX and Figure 36. The
resulting distillation curve is plotted in Figure 37.
1.0r--------------------------.
LJ.J
l-
e:(
:I 0.8
i=
en
2i
~ 0.6
u
>
:::E
z 0.4
o
i=
u
e: ,0.2
e:(
LJ.J
...l
o
:::E 0~-~--~2~0--~-~4~0--~-~6~O---L-~~8~0---L--~IOO
Fig. 37. Distillation curve calCUlated for So .. 100 and !V. t = 0.5. Still-product
" composition as in Figure 35.
As with the graphic procedure for simple distillation, this method involves
the simplifying assumptions that holdup is negligible and that the still-
product composition curve is correct. The latter depends on the validity
of the methods given in Section III for calculating distillate composition
from still composition (or vice versa) in terms of reflux ratio, number of
plates, and vapor-liquid equilibrium data. Experiments with benzene-
106 A. AND E. ROSm
toluene and benzene-ethylene dichloride mixtures gave good agreement

(Fig. 38) between the eX}Jerimental batch-distillation curves and those
calculated by the a,hove gra.phic method. Since the validity of the methods
of ealcnlating di~ti1late compositions from ~till composition has been well
tested in the deHign and operation of continuous columns, such agreement
is not surprising when holdup is but a few pel' oent of the ohm·ge. The
important result~ from oalcuhtionH of batch-distillation curves of the pre-
eeding types are their approximate predictions of behavior in actual dis-
tillation, and their general information as to the probable effect of the
A B
~
, \
\
\
\
"- ....
FRACTION DISTILLED
:Fig. 38. Cnlculuted (- - -) am] expcJ'iment~tl (~) eUrves for batch fmetioIlll.l dis-
tillation of bcnzmlO -ethylene dichloride (A) [lnd benzene-toluene (B).m
various process v[1rlables. 128 Complete agreement between experiment

and the simplified theory cannot be expected for the various reasons
previously mentioned. These may be summarized as follows:
(1) Even at low holdup there is some effect of the gm,dUftl change of
composition which is ignored hy the simplified theory. (2) 1.'he1'e is u,n
inherent variability and difficulty in controlling batch-distillation opera-
thms. (3) The simplified theory a.'lsumes finite reflux conditions from the
very start, whereas most actual batch distillations are at total reflux just
before the first portion of distillate is withdrawn. (4) The simplified the~
ory neglects the effeet of reflux ratio on H.E.T.P., and also the similar effect
of composition (slope of equilibrium curve). (5) The usual simplifying
assumptions are always a possible source of discrepancies.
12\1 Smoker and Rose, Trans. Am.lnst. Chem.itngl's., 36, 285 (1940). Rose, Ind. Eng.
Chem., 33, 594 (1941). ROSG and Long, ibid., 33, 684 (1941).
1. THEORY 107
II. ALGEBRAIC FORMS BASED ON ASSUMPTION OF TO'l'AL REFLUX
In the theoretical analysis of fractional batch distillation, it is not possible

to reduce equation (4) to a completely integrated algebraic expression as
was done for simple distillation. This is due to the fact that the equations
relating Xn and x, at partial reflux, such as the Smoker equation, are so
complicated that their use to eliminate x, from the equation leads to ex~
pressions that could not be integrated. Similar diffieulties are encountered
with the analogous formulas that might be obtained from the relations
discussed in Section IV. By making the restrictive simplifying assumption
that still and distillate compositions are related by the :Fenske equation for
total reflux (14, page 41) it is possible to reduce equation (4) to completely
integrated algcbraic equations:
'( 1
log,s = - - - Ilog XD
an -1
- an log (1 - XD) +
(an - 1) log [an - Xv (an - 1) l} + leD (14)
1
log /) = ---1 [log
cl~ -
Xs - an log (1 - xJ 1 + k, (15)
h;,i!/a'-l) [an - XD(et n- 1) 1

(1 G)
(1 - XD)""/(<<"-l)
where kD and k, are integration constants. The method of deriving these,

and also their use for calculating actual CUl'ves, are an[l.logous to the corre-
sponding case for simple distillation, since the Fenske equation and the
relative-volatility equation are similar in form.
Typical curves represented by the above equations are given in Figure 39.
These have been calculated for the fractional distillation of 100 moles of an
equimola!' mixture under circumstances in which XD and Xs are related by
the Fenske equation with 01" equal to 1.25 and 9.313, The shape of thfl
break in a curve is entirely dependent on the value of an and not on partic-
nll1!' values of a and n themselves. Thus curve B of Figure 39 is obtained
for any values of a and n such that an is 9.313. Such combinl1tions as a =
1.25 and n = 10, a = 1.28 and n = 9, a = 1.32 and n = 8, etc., would give
the same curve B. It must be strongly emphasized that these relations are
based on the simplifying assumptions of negligible holdup and the validity
of the Fenske equation. The latter is strictly applicable only under total
reflux, and to mixtUl'es in which the variation in relative voll1tility is not
excessive. The curves of Figure 39 and other curves calculated by the
equations on which it is b[l,sed (see page 153) there10re represent not
actual sharpness of separation,. but only the limiting case of sharpest pos-
sible separation, for mixtures in which relative-volatility relations are ideal.
108 A. AND E. ROSE
This sort of information is URefU! when a sharp separation is desired, since

it leads either to the imI1;lediltte and definite rejection of the column, or the
knowledge that the desired result may be approximated by sufficiently high
reflux ''lith the apparatus at hand. The equations and curves are also URe~
ful in suggesting the general nature of the relations between relative vola,-
1.0
1JJ
~
..J
..J 0.8
i=
(/)
5
;g: 0.6
0
> A (co' '1.25)
::!O
Z
0 0.4
j::
0
<t
rc
ll.. O,~
1JJ
..J
0
:2il
0 10 20 30 40 50 60 BO 90 100
Fig. 39. Distillation curves calculated for S, = 100, x.,, = 0.5 and: (A.) an = 1.25;
(B) an = 9.313.
tility, maximum sharpness of sepaTation, a11(l minimum number of plates.

Kuhll129 has proposed a different method for predicting the course of a
distillatjon when reflux ratio is very large and holdup is negligible.
C. RA"YLJ]IGH EQUATION FOR MUI.TICOMPONEN'l' MIXTURES
The Rayleigh equation may also be written:

log (A 1/A 2) = a log (B1/ B 2) (17)
where A1 and A2 are the moles of component A in the still at two different
times during the batch distillation, and Bl and B2 have a similar meaning
for a second component. In this form the equation is applicable to any
pail' of components hi a multicomponent mixture. The validity of the
equation depends upon Dalton's law of partial pressures, as well as on con-
stant relative volat.ility, as indicated in the following derivation. The rela-
tive volatility, Ct, may be expressed
0: = PAjPB = PYA.jP1J1I (18)

XA XB :VA Xn
129 Kuhn, Helv. CMm. Acta, 29, 26](1946).

l. 'l'liEOHY 109
where PA and Pn are the partial preil:;ures of eomponcnts .A and B, respec-

tively, when the corresponding mole fractions are X.1 and ;tn) and 1) is total
pressure, while YA and YB are the mole fractions of A anci B in the vapor.
This relation will apply to the formation of a differential quantity of vapor
from a complex liquid mixtuI'C whose composition may be expressed in part
as:
XA = A/(A +B+ + .. .N)
C (19)
:Cn = B/(A + B + C + .. .N) (20)
where A, B, 0, ... N refer to the molc::; of the various (lornponents present.

The composition of the differential quantity of vapor is:
dA
(21)
PYA = P dA + dB + dC + ... rIN
dB
(22)
PYB = P dA + dB + dC + " .riN
Substitution of thef:ie values in the relaj,ive-vol!1tility expre:,;sion give:;:
cr=W;/~~=~~/~ (23)
or:
rlA./A aeriE/B) (24)
Thi:; may be integrated to give:

log A = a log B + (log All - a lug Bo) (25)
where AD, Eo, are moles uf A and B at a time of rofe1'onl.'e. The more fre-
quently used form of this relation is:
A1 13 1
log - = Cf. lug -" (2G)
A2 B~
3. Curves Calculated by Stepwise Method

An approximate batch-distillation eurve may also be obtained without
use of calculus by employing a stepwise method of ealeulation. 180 This
may be illustrated by carrying out the procedure for fractional distillation
of 100 moles of all equimolar mixture with relative volatility, ex, constant
and equal to 1.25. These are the same conditions as for the example given
130 Rose and Welshans, Ind. Eng. Chem., 32,671,672 (1940) ..
DO A. AND E. ROSE
on page 104, and comparison of the methods and resulting curves may be
made.
It is assumed at the start that the distillation is to be carried out by allowing dis-
tillate to flow through a cup in or just below the condenser, and that the contents of
this cup are removed from time to time. 131 If the cup holds 5 moles of material,
then at the time of the first removal of distillate, a material balance of the more
volatile component gives:
(100) (0.5) = g,)Xl + 5Yl
If thc Fenske equation is applicable:
111 =
1 + xl(a" - 1)
and therefore:
a"Xl
50 = 95xl + 5 ------
+
1 xl(a" - 1)
This gives Xl = 0..179 and thereforc Yl = 0.897, the latter being the composition of
the first portion of distillate removecl. This is plotted in Figure 40.
0
w - CONTINUOUS REMOVAL
I-
<I: ----- INTERMITTENT REMOVAL IN
..J
..J 0.2 10 MOLE PORTIONS
i=
II) - - INTERMITTENT REMOVAL IN
is 5 MOLE PORTIONS
~ 0.4
<..l
>
:!;
z 0.6
0
i=
u
<t
0: 0.8
lJ.. I"'~"'''
W
..J
0
::E 1.0
0 20 40 60 80 100
Fig. 40. CI11cubtcd batch-distillation eurves for continuous ::Ind intermittent
removal of product.l!O
After removal of the first portion of distillate the cup will again fill up, and a
second portion of distillate will be removed after the appropriate time interval. A
matarial balance now gives:
(.l!"X2
95xl = gOX2 + 5Y2 = g,OX2 + 5 - - - -7t - -
1 +
X2(a - 1)
131 Fenske, in EC'imce of Pei1'01eUIn. Oxford Ulliv. Press, New York, 1938, pp. 1659
!lJld 1660.
1. TnEORY 111
so that the composition Y2 of the second 5% of distillate may be calculated and
plotted. This is repeated step by step to obtain a graph as in Figure 40.
The method is similar to that used in the derivation of the Rayleigh
equation, except that in the present case tl finite quantity of distillate is re-
moved. If these finite quantities of distillate are made very small, the
stepwise curve approaches the Rayleigh curve. The labor of the Rtepwise
calculat.ions is considerable, and enOl'S are cumulative and may become
serious after a number of steps. Nevertheless the method is not always at a
. disadvantage compared with the graphic Rayleigh pl'Ocedure. Compara-
t;ive calculations are an excellent way of building up confidence and appi'e-
ciation of the power of calculus as an ttnalytical tool.
4. Curves Calculated from Minimum Reflux (Infinite Plates) and Henry's

Law Equations
When distillation is at finite reflux and with such a large number of plat,es
that the difference between still and distillate composit.ion is independent
of the number of plat.es:
(XD - Ys)/(XD - x,) = L/V (27)
(see also Sect. 1II2C3). Colburn and Stearns 132 substituted this relation
in equation (4) and obtained:
In S= (1 - ~vL)f ~ =
1ls - :t.,
1 - (L/V) [In :t:.
a - 1
- a In (1 - x,) 1+ kJ (28)
where:
1], = 1 + Ca - l)x,
and k1 is the integration constant. They ~llso derived the corresponding
expression for the case of dilute solutions in which y., = kHx. (Henry's law)
to obtain:
In S = ~J, - L/ Y ! dx, = 1k- L/; In x, + leI (29)
"H - 1 x. H -
which may be written:

l-(L/V)
klI-l (30)
SdS2 = (:C./x.,)
For the case in which Henry's law is applicable but infinite plate conditions
are not satisfied:
XD = (7c H )nx. (31)
102 Colburn and Stearns, Trans. A:m. !nat . Chem. Engrs., 37, 291 (1941).
112 A. AND E. )108]1]
Hnd:
or;
log l',
100'
",.
8 = ------
(kH)1Z - 1
+ /.;1 (33)
5. Calculation and Determination of Holdup

The preceding derivations and equations have all assumed negligible
holdup. The results are of interest in many connections because, in actual
distillation, holdup is often so small that it may be neglected. 133 It is how-
ever desirable to analyze the batch-dist.illation process for cases in which
holdup must be taken into consideration. The general method of approach
is simihtr to that already discussed, but the basic material-balance equa-
tions must be modified to include the quantity of a particular component
present in the column as holdup. It is also desirable to know the t.otal
holdup. Very little experimental information along these lines is avail-
able, but some general observations and ealculations are of interest. 134
A. HOLDUP OF MORE VOLATILE COMPONENT
The simplest situation is that in a plate column in which the total holdup
per plate, H I, is the moles of material actually on and over any plate. The
holdup in moles of one component, hi, is then given in terms of its mole
fl'aet,ion, Xt:
(34)
In many cases the volnrne of holdup pel' plate will be constant from plate to
plate, and from time to time throughout a batch distillation. The mass of
holdup and the tota.l ?nole.g of holdup are subject to variations due to density
and molecular-weight differences, but it is convenient to ignore these for
the present in order to keep this preliminary discussion as simple as possible.
On this basis the total moles of the more volatile component of a binary
mixture present as holdup in a plate column during a batch distillation
would be !Lpproximately:
i~n
h = H1xl + H + H x3 + ... H,~XII = 2: Htxj

2X 2 3 (35)
'l== 1
133 Smoker and Rose, Trans. Am. Inst. Chern. Engl's., 36, 285 (1940).
114Rose, Welshans, and Long, Ind. Eng. Chern., 32, 673 (1940). Edgeworth-John.
stone, ibid., 35, 407 (1943). Edgeworth-Johnstone, ibid., 36, 1068 (1944).
I. TIIEORY 113
or, if III = H2 = Fla, ... = IIi moles of total holdup per plate:
i=n
h = Hi I: Xi (36)
i=1
If H is the moles of total holdup in the entire column:

i=n
Ht = Hln and h = Hln i=1
z:::: Xi
This reasoning is given in diagrammatic form in Figure 41. The com~
positions of the liquid on the several plates (:lh, X2, Xa, etc.) will change as
the batch distillation progresses, so that the aetual values in any case of
PLATE NUMBER COMPOSITION
COUNTING FROM ON
BOTTOM PLATE
H5(~ I} 5 x5
H4(~ I) 4
X4
H3(~ I} 3 x3
H2(~ I} 2
x2
H, (~ I}t x,
Fig. 41. Holdup of more volatile component in plate column. Total

size of block respresents t.otal holdup on plates. Shaded portion repre-
sents holdup of more volatile (Jomponent.
interest must be obtained by the equations or graphs (Sect. III) showing

composition from plate to plate.
In the case of a packed column, a somewhat similar situation exists ex-
cept that the change in composition of the liquid is gradual instead of
making sudden jumps from one plate to the next. In this case IIi and hi
refer to the holdup per theoretical plate. The situation is represented
graphically in Figure 42, which leads to the approximate equations:
hI = H1Xo.6,
or, in general:
(37)
114 A. AND E. ROSE
where XCi-o.,,) signifies the composition one-half theoretical plate below i

plates from the still. Then:
i;:::lln
h = H/n
i=d
L: X(i-O.5) (38)
where E is total holdup of all components, n is the number of theoretical

plates cOlTesponding to the difference in composition between the still and
the liquid at the top of the packing, and x is the mole fraction of the more
volatile component in the holdup at various distances from the base of the
column, as indicated in the subscript.
DISTANCE LIQUID COMPOSITION
LIQUID VS.
~~?L~ COMPOSITION POSITION IN
COLUMN
Hs
X4 - - - -
X 3.5 --- H4
xz.s--- H3
Xz - --- -
xI.s --- Hz
I times H.E.T.P. XI - - --
xo.s -.- - ... HI
Xs 0
Xi
Fig. 42. Holdup of more volatile component in packed column. Total

blocks and shading as in Figure 41.
The preceding expression for h is slightly in errol' for several reasons.

These are: (1) The total holdup, HI, in the lowest section of the column is
probably always appreciably less than that in the other sections higher up,
since the lowest section includes the upper portion of the still. The total
holdup per theoretical plate may also vary in the remainder of the column.
Specific experimental data are required in order to correct for these errors.
(2) The expression assumes that the vapor at any point has the same com-
position as the liquid. The vapor is a minor proportion of the total holdup,
and an expression for it similar to the above may be derived only with con-
siderable increase in complexity. (3) The holdup in the condenser is not
included. (4) The values XCi-0.5) used in the above expression are not the
I. THEORY 115
correct average compositions unless the change in composition with dis-

tance up the column is a straight-line function. This is generally not true.
l'he first three of these sources of error arc neglected in the suhsequent
discussion. The fourth is eliminated by deriving an expression in terms of
very short units of height.' Thus if the sections are made 0.1 of the
H.E.T.P. in height, the expression for h becomes:
i=n
h = 71 0 ,1 + lIo.? + h + ." = I: h,
O,3
i=O.l
(hy tenths)
where H j and hi now signify the holdup in 0.1 H.E,T.P, Since:
h, = IIix(i-o,oo)
h = H O,lxo.05 + HO,2XO.15 + H O.3xo.26 +

If, as before, the total holdup is uniformly distributed:
H O,1 = N O•2 = H o,3 = (H/n)(I/lO)

then:
i=n
h = (H/n)(I/lO) I: XCi-O.OS)
i=O.l
(by tenths)
which is nearly equal to:

i=n
(Hjn) (1/10) E
i~O.l
Xi
(by tenths)
If instead of tenths the sections are made stilI smaller, the difference be-
tween the two forms in the above equation becomes negligible. When the
sections are made of a very small height, ili, where 1: represents distance up
the column in terms of H.E.T.P. units, the last expression becomes:
... n
h = (H/n) M 2: Xi
i=O
= (Hln) ;:~~n xidi (39)
In most instances it is necessary to obtain the value of the above expres b
sian by gr:1phic means. An algebraic solution is possible in. the case of

dhltillation of an ideal binary mixture nnder total reflux. In fluch a case
the composition Xj of the liquid l1t any point in the column is relat.ed to the
still composition by the Fenske equation:
(40)
116 A. AND E. ROSE
or:
(cbs)/(l - x. + ch.) (41)
If for convenience:
b = (1 - x.,)lx s
then:
e:::l
CJ
::i
!1:
u
>
::IE
z
o
i=
u
<l:
0::
U.
W
...J
o
::IE
O.5 L - - - - - - - - - - - - : ! 3 0
0
DISTANCE FROM BASE OF COLUMN
IN THEORETICAL PLATES (i)
Fig. 43. Composition of liquid holdup at various
distances from hase of <)o\umn.
Equntion (39) for the holdup of the' more volatile component l36 now be-
comes:
h = (Hln) (i=n _~ di = (Hln) log [1 + xs(an - 1) 1 (42)

Ji=O b+a i
log a
01':
(I-Xs
HI og ---;:- + an.
I! = ;; )/1 og a + .mtegratlOn
. constant (43)
I. 'l'HEOll.Y 117
The use of both the graphic method and the equations may be illustrated llS
follows: The graph of Figure 43 represents the composition of the liquid in various
parts of a thirty-plate column during distillation of a mixture with ~ = 1.25, and at
a time when the still composition is x" = 0.5. The values of XI were calculated by
equation (41), where i was given the successive valucs 1, 2, 3, etc., up to 30.
By counting squares to the right ofthe CUl've in Figure 43, values for the holdup of the
more voh,tile component in that pal·t of the colu\l\ll.correspondiIl.g to any pbte or
plates may be determined. For the lowest ten plates the area is 7.348. Thus the
holdup of the more volatile component for these ten plates is, from equfltioll (39),
(HI10)7.348. If the total holdup lor this ten-plate column, H, is 10 moles, h =
7.348 moles. If equation (43) is then solved with the values for the base of the
column (x, = 0.5, n = 0, ex" = 1, and Hln = 1), the integration constant is - log
+
(1 1)/log 1.25 = -3.11. When n = 10:
h -- log (1 + 1.25
log 1.25
10
) _ 3 11 - 7 G 1
. - .3 mo es
The graphic information on variation of cumposition from point to point

in a column, as given in Section III, may be used in a similar manner to
estimate the holdup of a given component at either total reflux or finite
reflux. The algebraic solutions given above are not applicable at finite re-
flux because of the use of the Fenske equation. By substituting the Smoker
equation, it might be possible to derive equations for calculating the holdup
of one component at finite reflmc. Such equations would be complex and
subject to many simplifying assumptions. It is preferable to employ
graphic methods for this purpose.
B. '£CY£AL HOLDUP
All the preceding discussion has dealt with methods of obtaining the
holdup of one component. These methods required the amount of the total
holdup as well as the assumption that it was uniformly distributed. No
experimental work has been published on the uniformity of distribution of
holdup in packed columns, but total holdup may be estim!1tecl in several
ways. A very l'ough estimate may be obtained by allowing the mixture to
be distilled to drip into the packed section of the column from its top at the
rate to be u~ed during the distillation. Wb.en the packing seems to he
uniformly saturated, the process is stopped and the p!.1Jcked section is
drained into a measuring flask. The liquid 80 obtai.ned is called the holdup.
This method gives serious errors because temperature and therefore vis-
cosity and surface tension are different from those of actual distillation, and
also because an appreciable amount of liquid remains on the packing (the
136 Rose, Welshans, and Long,1nd. Eng. Chem., 32, 673 (1940).
118 A. AND E. ROSE
statjc holdup) and is not measured. In addition there is an errol' due to

absence of the countercurrent gas stream.
Some of these disadvantages may be avoided by measuring the quantity
of material in the column during actual distillation by one of the following
methods. Distillate may be removed until the still pot is just dry, when
distillation is stopped and the still heat cut off. The bulk of the holdup
then drains into the still pot where its amount is measured. Corrections
should be made for the vapor in the still pot if this is large, and for the static
(nondrainable) holdup. The former may be estimated from its volume,
and the latter may be removed and collected by passing a stream of warm
inert gas through the column. An alternative method fa)' measUl'ing the
liquid static holdup is to pour a known quantity of test liquid into the top
of the column and note the amount of this that fails to drain up to the time
the rate of drainage becomes negligible.
A calibrated collection tube and valve may be placed in a liquid return
liue from base of column into the still pot and vapor passed through a second
separate line also containing a valve. Total holdllp may then bc deter-
mined by simultaneously closing the valves in the liquid and vapor lines,
measuring the quantity of liquid that collects in the collection tube, and
adding to it the static holdup of the column.
The most practical way to determine total holdup is to introduce with
the charge a small amount of inert, soluble, nonvolatile material whose con-
centration can be easily determined. 136 When heat is applied to the charge
and the system brought to equilibrium under total reflux, the concentration
of the nonvolatile material in the still pot will be increased because some of
the volatile material has entered the column as reflux or holdup. The
following equations may be set up. The original concentration of the non-
volatile material is:
w/(Cw + w) = al or (44)
where w is the weight of nonvolatile material introduced into the charge,
Cw is the weight of the volatile charge, and al the weight fraction of the
nonvolatile material in the charge. The concentration of nonvolatile
materia 1after equilibrium is reached at total reflux is:
or (45)
"
where a2 is the concentration of nonvobtile material in th~ still after equi-

librium is reached and H", the weight of volatile material present as holdup.
Since w is the same in both cases:
aleC.. + w) = a2(C + 1V -
w H",) (46)
136 Tongberg, Quiggle, and Fenske, Ind. Eng. Ghem., 26, 1213 (1934).
I. THEORY 119
(47)
(48)
The same type of calculation may be made after a weight, DUl , of distillate
is removed, because the new concentration of nonvoln,tile material in the
still is then:
(49)
and:
(50)
Thus:
H1JJ = (C,o + w) (1 - Cl3~) - D1JJ (51)
By taking samples from the still at intervals during the distillation, the
holdup may be followed throughout a distillation. A good choice fot, an
inert, soluble, nonvolatile material for hydrocarbon-distillation studies is
stearic acid, which can be easily determined by evaporation of the volatile
components or by titration with a base, provided that rust or similar mate-
rial docs not react with part of the stearic acid. A nonvolatile fraction of
rosin oil has also been used successfully, analysis in this case being either by
means of optical rotation or refractive index. Since the composition of
rosin oil varies from batch to batch, it is important that the relation be-
tween composition and optical rotation or refractive index be confirmed for
each batch.
When optical rotation or refractive index is used for the analysis, it is
necessary that the volatile material in the pot consist of a single component,
but analysis by titration or evaporation and direct weighing permits the use
of a mixture. The latter is the only condition corresponding to actnal
operation. Whichever procedure is used, a material balance should be
made on both the volatile and nonvolatile components, to assuee absence of
errors from leaks or other losses. Examples of the last are tile reaction of
stearic acid with rust in an iron pot, and the carbonizing of the nonvolatile
material due to overheating a dry section of the still pot.
C. DIRECT DETERMINATION OF HOLDUP OF INDIVIDUAL COMPONENT
This may be determined for a binary mixture by analysis both before and
after holdup enters the column, and when nonvolatile additive is absent,
120 A. AND E. nOSE
but, total holdup is known. A matcrial babnce of t.he more volatile com~
ponen!; gives:
(/)2)
in which XWI. is the average weight fract.ion of the more volatile component
in t.he holdup. C,e, t.he weight of the charge, X,." its composition, and XwO,
the composition of the still contents after holdup has entered the column,
can be determined experimentally. There are no published applications of
this procedure, but it is an excellent tool for checking other calculated
values of holdup and the concepts on which they depend. Colburn and
Stearns l37 have used a closely relat.ed pl'ocednre.
D. F.AC'roRs DETERMINING TOTAL HOI,DUP IN A PACKED COLUMN
The experiments of .lesser and Elgin 138 showed that in packed towers the
liquid holdup is an exponential function of liquid velocity and the exponent
is essentially the same for anyone type of packing. Change of holdup with
variation in speeific surface tension, specific gravity, and relative viscosity
of the liquid was also an exponential function. This was the same for all
types of packing, and for constant specific gravity and relative viscosity,
it was independent of liquid veloeity.
6, Curves for Fractional Batch Distillation When Holdup Is Appreciable

When holdup is appreciable, the theoretical relations of batch distillation
are sufficiently complex so that no simple methods of prediction have been
developed. Three different. a.pproaches have been used. The first is ex-
tension of the Rayleigh equation with inclusion of terms to represent
column holdup. The form of the general equation may be obtained as
shown later, but numerical solution is impossible because there is no
method of predicting the still-product composition curves. A second
approach involves setting up diffe;'ential equations for composition of the
distillate, still liquid, and liquid on each plate of the column, as functions of
the fraction distilled. The algebraic solution of such equations is probably
impossible, and even approximate numerical solutions a1'e very laborious.
Some progress has been ma.de through use of a differential analyzer. The
third approach is by stepwise plate~to~plate calculations, also exceedingly
laborious. Some details of each of these methods are discussed in turn.
A. GENlmAI. EQUATION BY EXTENSION OF RAYLEIGH EQUATION
The form of a general equation for batch fractional distillation may be

derived by extension of the material-balance procedure introduced by
137Colburn and Stearns, Trans. Am. Inst. Chem. Enyrs., 37, 291 (1941).
ua Jesser and Elgin, Trans. Am.lnst. Chem. Engrs., 39,277 (1943).
T. THEORY 121
Young and by Rayleigh. 139 Since complcte experimental dlLta are lacking
on the variation of H.E.T.P. values along the length of a column, and also
on the distribution of holdup, and in order to take care of all possible vari~
ations in these factors, the generalized expression:
h = fn(x s • .• ) (53)
can be used to represent the holdup of the more volatile component in the
column. This is analogous to the equation:
h = H log [1 + x,(a n - 1)]
(54)
n log a
used for expressing holdup of the more volatile component in the derivation
of equation (43). Similarly, the relation bctween product and still com-
position will also be used in the generalized form:
Xv = !,,(x., ... ) (55)
so that the derivation will not be limited by the simplifying assumptions
used in cn,lculating distillate and column compositions from still composi-
tion. This latter general relation is analogous to the Fenske equation:
(56)
The derivation of the general equation for a batch-fractionating curve
involves a material ballmce just before and after the removal of a very small
quantity, flD = dB, of distillate, 139 similar to that llsed in the simpler cases
already discussed. This giveH:
8x s + !h(X .... ) + HCXD = (x" - dxs)(S - dB) +
hex, - dx s • •• ) + xnrlS + (xv - d.1:D)H c (57)
In this equation Ex. represents moles ·of more volatile component present in
the still just before removal of the small quantity of distillate, and (x, -
dx.) (8 - dB) represents the moleR of the same component just after the re-
moval. The quantities fh(X, ... ) represent the moles of the more volatile
component present as holdup in the column, while Hex D and H C(XD - dxv)
are the moles of the more volatile component in the condenser. When
terms ai.'e transposed and multiplied out, and the second differential dx,dS
is dropped out as being negligible, the equation becoIlws:
!h(X. - dx•.. ) - !h(X s ••• )
-
dS . S
= x, -- + - dB
Xv -
+ Elc -dXD (1':8)
0
dx. dx s dx, dx.
The quantity on the left side is by definition equal to -f~(X8"') where
139 Rose, Ind. Eng. Chem., 32, 67(194.0).
122 A. AND E. nOSE
prime signifies the derivative of hex"~. ,.). Graphically, this is the slope of
the curve of equation (53) at whatever value of Xs is involved in the specific
numerical calculations. Also, by deii.nition, the quantity dXD/dx s is the
derivative of fn(x s • •• ):
d:t:D/riX, = lJx s ... ) (.59)

Therefore:
x,(dSjdx,) +.') - xv(dB/dx s ) + IlcJ":,(:c s ... ) = -j':'(x s ••• ) (GO)
and:
elF) '" = -Hc/;'C:e s ' ' ' )
~.-.---.,
+ j'~(x" ... ) (G1)
This is a linear differential equation which can be partly solved to give:
dxs
S = k exp
J _._'-
XD - X,
+
f [-I1cf:'(x,,· .. )+f:'(~:s, .. )ldx8
exp
f dxs
---
XD - :c,
. '--
(XD - xs) exp J dX../(XD - xs)
(62)
The indicated integratiolls may be performed graphically whenever definite

graphs for f;, and j'~ are available. The equation is essentially a relation of
the form:
(63)
lmd is thus the desired relation hetween produet. composition and quantity
of product distilled, i.e., it represents a distillation curve.
Colburn and Stef1l'lls140 expressed this equation in the form:
In ..8-1
82
= iX" '~'2 .T]) -
dx.•
x, - (Hrlxlt/rlB)
(64)
where Hdx,JdS is the rate of removal of the more volatile component from
the holdup.
The actual Use of equation (62) requires that it he expressed in a form
with definite limits, such as: 1H
S = Eo. _ fx,of~(Xs ... )dx s - H C£:,(x 8 •.• )dxs (G5)

& "\I.O
I\J~!I ;I::;t
( )'x.
XD - Xs I\XS!
140 Colburn and Stem'as, Trans. Am. Inst. Chem. Engrs., 37, 291 (1941).
HI Rose, Bailey, and Bertram, "The Problem of the Holdup Effect in Batch Fl'ac-
t,ioliation," Am. Chem. Soc., Atlantic City Meeting, 1942.
I. THEORY 123
and
"x.""" __expo. j,r, ~.
dx s.
.1'D - .1-"
By assuming condenser holdup to be negligible so tha.t He = 0, the equation
becomes:
(66)
In actual calculations wHh this equa,tion, all the quantities are obtained
from the curves of equations (53) and (55). Thus the distillation curves
calculated with this eqnation are as correct or as ineorrcct. as the equations
for XD and h. Unfortunately all three of these equations are dependent
upon one another, and their form is so complex that combination into a
single expression has not been achieved.
The use of the above equations may be illustrated by considering a case for which
the Fenske equation may be used to describe the relatioJl between XD and x,. This
assumption limits applications to systems inwhich reflux ratios are so large or other
circumstances are such that separations approximate those under total reflux.
The equation for moles of holdup of the more volatile component then takes the
form of equation (54). This represents the holdup of the more volatile COlll-
ponent in the column at every instant during the distillation.
By making a material balance in the same manner as for the preceding cases, and
including tIle terms for holdllp,142 there is obtained the expression:
SXs + [II/(n log a)] log [1 + x,(a" _ 1) 1= (x. _ dxs)(S _ dS) +
[Il/(n log a)] log [1 + (xs - cZxs){al!; _ 1)1 + xDdB (G7)
This may be solved as follows, using A = H/(nlog a), b = an _ I, andq = 1j(an _
1) = lib in order to simplify the writing of the various expressions. The first step
is to eliminate XJ) by using the Fenske relation, whereupon:
x. dS +S_ a"Xs dS _- A log [1 + b(a:" _ &"c,,) 1- log (1 + bx,) = 0

dx. 1 + bxs dx s dx .•
(G8)
Since by the definition of a derivative:
log [1 + b(x" _ dX s 21-= log (l + b~~~ :::: d log (1 + In:.,,) b

+ bxs
-~--
([:r;., d:rs 1
dS dS
then: Xs -
dx,
+S -
1
a"X
+ s -
bx s dx.
+ 1 +Abbx. = 0 (G9)
142 Rose, Welshans, and Long, Ind. Eng. Chern., 32,673 (1940).
124 A. AND E. ROSE
which is a linear differential equation that can be partly solved to give:
(70)
q = l/(a n - 1) (71)
The remaining integral may be evaluated as nearly exactly as desired by expanding

the numerator by the binomial theorem and then integrating and retaining as many
of the terms aR is neeesoary:
!_ (~ -
2b b
1)(__
1)
2 - q
(x )2-a
.,
+ '"
]
The series converges rapidly for all cases of interest and often all terms beyond the
first are negligible, so that the equation reduces to:
S = I{ __(~_)q_ _ Ab (72)
(1 - xs)a+l (1 - x s)a+l
The Fenske equation may be used to express this relation in tenns of distillate COIll-
position:
(73)
For purposes of computation it is lJ~ually more convenient to use equation (72) and
the ]'enske equation separately.
The integration cOllotant, K., is calculated 143 from the initial conditions of the
(listillation. The value of x, llse(l in this calculation is not that of the original
mixture. The original value of X.I (hereafter referred to as x.,,) decreases somewhat
even before any product is removed, due to the passage of some of the charge into
the column proper. The value of the still composition X.I (hereafter called X SlI )
when equilibriuIll has been reached, but before any product is removed, is given by
the solution of the expression:
S· eX"e -- (OJ _
'')c
H)
11. Xso + H log (11og+aT!.bx 80)
(74)
Trial and error is. the only method available for solving t.his relation, and the value
of x.o so obtained is used in:
(75)
l<8 Hose, Welshans, and Long, Ind. Eng. Chem., 32, 6N (1940).
1. THEORY 125
or its more complex forms in order to obtain a value of J( for use in subsequent cal-
culations. These equations contain the factors £:1:, n, and II. By llSiiLtlning suitable
values of these and for the iuitial cumposition of the mixtlll'c, values of Sane[ Xv
may be obtained and plotted as in Figure 44.
~
<{
0
..J
..J
i=
(f)
C
~
u
>
::;:
z
a
1=
u
cr
a:
LL
~ 1.0ol--===~=:=:~:::::-_---'----~IOO
PER CENT DISTILLED
Fig. 44. Calculated distillation curves for 1, 4, and 10% holdup (dis-
tillation at total reflux).
B. DIFFEREWrIAL EQUATIONS FOR PLATE COMPOSITIONS
In the case of a five-plq,te column the following equations l44 describe the
rate of change of composition of the liquid Oil the several plateR, if the usual
simplifying aRsumptiolls are made:
d:~6IdS = - (SIR.) [(RD + I)Y4 - RDx" - xn]
dX4/dS = -(SIH4)[(RD + l)(ya - Y4) + Rn(xs - :r'l)]
dxt/dS = -(SIHI)[(RD + l)(U. - !/I) + RIJ(xz - Xl)]

dxsldS = -(liS) [RDXl + X., - (RI) + 1)Y8]
In these S represents the fraction of the charge remaining in the still and
H6 = I-I~ = I-I3 etc., i.e., the total holdup per plate is uniform. Since such
equations descrillc the c.omposition on the various plates they may be com-
bined to prochwe the general operating-line equation derived hy Colburn
and Steal'ns. 145 The solution of the equation by algebraic means is ex-
144 n.. C. Johnson, unpublished work. Marshall Md Pigford, Application of Differ-
ential Equations to Chemical Engineering Problems, Univ. Delawt),l'e, Newtlrk, Delaware,
1947, p. 144. Pigford, Tepe, and Garrahan, "Thl1 lijffect of Column Holilup in Bat.r.h
Distillation," Am. Chern. Soc., Philadelphia Meeting, 1949.
m Colburn and Stearns, Trans. Am. In.~t. Chem. Engl's., 37, 291 (1941).
126 A. AND E. nOSE
t,l'emely lengthy if not impos:-;ible. Numerical solutions by approximation

methods1oj.J arc also lengthy, but not impossible in the simpler cases. In
the few cases in which such cah~ulations have been completed, the results
are in agreement with those of the stepwise plate-t.o-plate procedure.
Pigford, Tepe, and Garrahan j.jij have used a differential analyzer to ob ..
Lain numcrical solutions to such differential equations for batch distillation.
The results arc indicated in Section VI.
C. Sl'EPWISE PLATE-'l'O-PLA'fE CALCULATIONS
This pl'ocedure t47 necessitates est!1hlishment of the compositions of vapor

and liquid leaving each plate of the column after initial total- (or finite-)
reflux conditions have been established but before any distillate is with-
TABLE X
DlS'l'lLI,A.'l'B, STILL, AND COLUMN COMPOSITIONS. S'l'EPWISg PLATE-'l'O-PLNl'E
CALCULATIONS FOR BATCH li'UAC'l'IONAI, DIS'l'ILLATION
(0) (1) C~) (3)

After total
reflux
conditions
estl1bli"hed After reulOvitl After removal After rcJUOvn.
hut prior to of slnall of s""ond of third
ramo,ralof initial porHon small portion smull portion
first product of distill a!,,, of distillate of distill"t"
Mol~'S distillate ........ none 0.00318 0.00636 0.00954

Distillate composition. .0873 .H867 .0858
]jiqllid Vapor Liquid Vapor Liquid Vapor Liquid Vupor
Top Plate
#5 ........... 0.9736 0.9880 0.9722 0.9873 0.9707 0.9867 0.9690 0.9859
#.1. .. , ......... 9431 .9736 .94.00 .9722 .9367 .9706 .9332 .9689
113 .... ......... 8813 .9431 .8751 .9399 .8685 .936,1 .8611 .9326
#2 ............. 7690 .8813 .7578 .8747 .7455 .8672 .7358 .8613
If) ............. .5989 .7690 .5819 .7563 .1)728 .7494 .5647 .7432
Still ........... 4010 .5989 .3994 .5973 .3972 .5951 .3945 .5927
Moles in still ..... 0 .3926 0.381)42 0.38624 0.38306
Total moles holdup .1590 .1500 .15DO .1590
Charge .......... .5516
Molesmve"in still .15743 .15553 .15343 .15126
in holdup ...... .13248 .13124 .13020 .12923
in distillate ..... .00314 .00628 .00042
Total moles mvc·.0.28fl91 0.28991 0.28H!n 0.28991
a Marc volatile component.
1<6 Pigford, Tepe, and Garrahan, "The Effect, of Column Holdup ill Batch Distilla-
tion," Am. Chern. Soc., Philadelphia Meeting, 1949.
147 Rose, "Agreement between Calculated and Experimental Batch Distillation Curves
with Appreci:tbJe Holdup," Chemical Engineering Symposium on DisWjatioll, Pitts-

burgh, 1946.
I. 'l'HEonY 127
drawn. These may be obtained by trial and errol' enJcubtions with the
equation:
(7n)
in which Be is the total moles charged, x"c is the mole fraetion of more vola-
tile component in the charge, :C,o is the still composition just before the first
distillate is removed, H the total moles of holdup, and X'IO the average (:0111-
position of the material in the column just before the first distillate is
1. 0 1--_ _ _ _ _ _...:.Y-"-5_ _ __
,_,-,-,_,-,-,-, ,,
0.9
r-)-'-'_'_,_,--
J Y4 and Xs
Y3 an d x4 '
: £I Y5
I
I £I Y4
)1--): _ _
x--x_~_iC.~ : + X5
X3
x-, : £I Y3
I
O.B 1+.1'4 :z:
I 0
U x---~ __ x__ i=
> x-";;;;x __ Y, : .:. Y2 <!
::;; I ..J
X2. iC.---X _ _ t. ..J
2: I
0 0.7 I
1+ X3
~
i= 0
U Y,
I £I
<t I III
ex:: ::l
lJ.. I 0
IlJ Y. I ::l
..J 0.6' It; z
Ys ;:
0
;,:
'_,--)(- ~ + x2
.... Z
XI-"-x_% I 0
u
I
I
0.5
I
I + x,
I
I
\
0.4 '-'_'_,-,_,_,-. : + x.
x. I
I
I
0.3L-~ __ ~-L-~ __ ~ __ k-~~ ______ ~
\ 234561
PORTiONS OF DISTILLATE REMOVED
(BATCH DISTILLATION)
Fig. 45. Effect of changing composition on separation achieved in
batch distillation. Total reflux conditions at start.
obtained. By trinl and error it is possible to find the value of X.o that satis-
fies the preceding equation. Trial values of X"o are chosen to be somewhat
less than X.a and vaLues for X/to lUay then be obtained frolU graphs of still
composition versus column average composition. An alternative procedure
is to choose a trial value of XnO and obtain coluIlln compositions (and thus
Xho and x. o) by the McCabe-Thiele procedure. Repeated trials lead to a
set of values such as in column 0 of Table X. If distillate is withdrawn,
128 A. AND E. ROSTIl
for examplc, at the rate of 0,00318 mole per unit time with a reflux ratio
L/D = 4/1, the following gain and loss of tho morc volatile component oc-
curs on the top plate during a first interval of time:
gain: VY4 + J"XD = 5(0,00318)(0.9736) + 4(0.00318)(0,988)
loss: VY5 + Lxu = 5(0.00318)(0.988) + 4(0,00318)(0.9736)
1.0
>'5
, Y4
)(5
)'3
,--,--,--, )(4
0.9
>'2
,--,--,--, )(3
>'1
K _ . ' I - - K _ _ J.
U X2
>
::! 0,8
z
0
i=
---------
u
_'__
«
Cl:
lJ.
w Ys
.J
0
::!
,-x , XI
0,7
0.6 ' .............-.........

'............_,
~xs
,
o \ 2 :3
PORTIONS OF DISTILLATE REMOVED
:Fig, 46. Effect Ot changing composition on sepal'ation
achieved in batch fractioIlntion. Finite reflux conditions
at swrt,
The resulting net loss of the more volatile component from the top plate,
. together with the total holdup on this plato (0,0315 mole) and its original
composition (X5 = 0.9736), gives a new top plate composition (X5 == 0.9722)
1. 'rHEORY 12!)
for the end of the short time interval. The results of a series of such calcu-
lations are given in columns 1, 2, an(13 of Table X and arc plotted in Figure
45. For purposes of comparison the values of various plate compositions
have been computed by the ordinary l\lcCabe-Thiele procedure for the
still compm;ition at the end of the fifth time interval. Thnse are also
given on Figure 45.
Solid lines are vapor compositions and dotted lines mftrketl X :-\,!'e liquid
compositions, both for batch distillation with total-reflux conditions at
start of distillation, as calculated by stepwise platc~to-plate material bal-
ances. Points marked Ll are vapor compositions and those marked + are
liquid compositions cn,lculated by MeCabe~Thiele procedure for continu-
ous distillation conditions, with still composition the snme as that at the
end of the fifth interval of batch caleulations.
The lesser over-all separation in the continuUlul-distillation calculations
and the different. relationship between the pairs of vapOl' and. liquid com-
positions is immediately obvious. As the distillation proceeds the com-
position of t11e vapor entering the bottom plate rather quickly lJecomes
greater than that of the liquid on the bottom plate, m; required by lVIcCabe-
Thiele considerations. The effect also begins to appear 011 the upper
plates, one after another, but the relationship continues to be much closer
to that for total reflux, than for the finite reflux as predicted by the Mc-
Cabe-Thiele procedure. The same situation is indicated in a different way
in Figure 46. This gives the results of Htepwh;e plate-to-plate ealeulations
showing the eifeet of shifting to ordinary batch distillation Itt finite reflux
after equilibrium operation at t.he same reflux with return of distilhlte to the
still. A more extensive series of such CtLlculated curves is ill Section VI.
7. Equations for Equilibration Time

Coulson has expressed the time for a column to come to equilibrium H,t
l48
total reflux as:

Oe = (ho - he)/v (77)
where ho is the moles of the more volatile component in the column at the
time a partieular boilup rate is eommenced, he is the corresponding value
when equilibrium is reaehed, and v is the net rate of transfer of the more
volatile component into the column. The value of v is easily seen to be
approximately
(78)
where Ys is the composition of vapor entering the base of the colunm, 11nd Xl
is the composition of material entering the pot from the base of the column.
148 Coulson, J. Soc. Chem. Ind. London, 64, 101 (1945).
130 A. AND E. ROSE
The value of he is in general Iix/to, where Ii is the total holdup and X/to is
the average composition of the holdup when equilibrium is reached.
Methods of obtaining both Ii and Xlio are indicated in connection with the
determination of holdup (page 112). The value of ho is also of the form
HXh. Coulson assumed II to be the same before and after equilibrium was
reached, which mayor may not be true, but is a reasonable first approx-
imation. For the value of ~;il Coulson assumed that the holdup was of the
same composition as the vapor initially evolved from the charge (Xii =
Xhc = 1/so) • On this basis:
[(
lOg [11++ C~: =~~::"cJ) _ aX,c ]
n log a 1 + (a - l)x,e
Coulson gives the approximate relation:
log xsc)Jj [xsc(a-l)]

8e~ [(VH) (l+nloga (79)
fOl' cases in which x" and a are smaJI and n is greater than 50.
Berg and Jarnes 149 have modified the derivations and equations of Cohen
to obtain:
Yo/X" = (n8/Il)(a - 1)(a In a)O (80)
1_~ x s a n +1
= a expo _ n8_ (a -
H an
1) 8 (80a)
and:
1 - (Yo/x s an +1) (n~\ Ca - 1)
1 _ (1/ an) = a expo - (CPll) II} an 8 (81)
Equations (80) and (80a) are simplified versions of equation (81), which
aSSllmes that holdup in the condenser and return system is negligible, that
(1 +
y)/(l + x) = a, and that the vapor holdup is negligible. The first
equation is applicable when a and n are large so that an becomes very large.
The second equation assumes CPB = 1 which is nearly true when el' ap-
proaches 400. CPR is a complex function of n log a, and depends alHo on the
system and the eolmlln used. Two experiments gave confirmation of the
calculated times predicted by the equations,
Hi Berg and James, Trans. Am. Ins!, Chem. Engl's" 44, 307 (1948).
1. 'l'f-II'lOllY 131
8. Comparison of Continuous and Intermittent Take-off Procedures

The theoretical analysis of batch-distillation processes has not yet pro-
ceeded to the point where such a comparison can he made. The changes in
composition that occur with time in [til intermittent t::tke-off proceHS are
dependent upon the rate of approach to equilibrium during the total-reflllx
period, and on the rate at which the disturb:1l1ce due to total take-off pro-
ceeds down the column. Little is kno\\"n of these rate processes, either
from a theoretical or experimcnt::tl approach; even the direct experimental
evidence regarding the relative merits of continLlolls and intermittent, take-
off is conflicting. Thus Oldroyd and GoldblatP&O reported a favorable
effect for intermittent operation but O'Brien l5l found no appreciable effeet
in a column of more than 100 plates.
9. Curves for Changing Reflux Ratio to Maintain Constant Distillate

Composition
All the preceding discussions have assumed constant reflux ratio through-
out a distillation, and the equations and calculations have dealt with the
change in distillate composition as the batch distillation proceeded. It is,
however, common practice to increase reflux ratio, when the amount of a
component in the still beeomes depleted, and thereby inerease the propor-
tion of that component in the distillate. The equations already obtained
can serve to estimate the results of sueh distillations with more than one
reflux ratio.
A. STEPWISE INCREASE IN REFLUX RATIO
For instance, during the distillation of 100 moles of equimohlI' binary

mixture with ex = 1.25, through fifty theol'etieal plates with negligible hold-
up, reflux ratio of 19, calcula,tions by equation (12) show that the distillate
composition remains near unity until the still content is reduced to 70 moles
(Figure 47A). When it reaches 65 moles the distillate composition has
dropped to 0.98 and after distillation of five I110re moles would drop to 0.87.
At the time still content is 65 moles, the still composition is 0.21. If at this
point the reflux ratio is increased to 49, the distillate composition will again
rise practically to unity andl'emain there until still composition drops to
approximately 0.1. The distillation cutve fat' the portion of the distillation
with reflux ratio 49 can also be ealculated by equation (12) using St = 65
and values of Xn and Xs from the still-produClt composition Clurve for O! =
150 Oldroyd and Goldblatt, Ind. Eng. Chem., Anal. Ed., 18, 7nl (1046) ..
161 O'Brien, M.S. Thesis, Pennsylvania State College, 1948.
132 A. AND E. ROSE
1.25, n = ,50, and RJ) = 49 (Fig. 47B). The calculations indicate that the
increase ill reflux ratio will result in nearly pllI"e distilbte nntil about 58
moles l'elll:~in in the l'till.
1.0 1.0
0.9
DISTILLATE
COMPOSITION
LLJ
I-
<[
_J
(
_J
O.B INCREASE IN Ro FROM
u f= 0.9
If)
> 19 TO 49
~ Ci
01
z i!: <t
0 u !!?
f= > O.B "
u ::;; r:r.'"
<[ r:r.""
0::
lJ.. 0.5
z
0
w f=
_J 0.4 u
0.:, <[ 0.7
::f :'iO.3 0::
lJ..
0.2 w
_J
0.1
0
~ 0.6 I
100 90 80 70 60 50 0.1 0.2 0.3
MOLES REMAINING IN STILL (5,) MOLE FRACTION MVC IN STILL (~s)
A B
Fig. 47. (A) Effect of increase in reflux ratio when amount of morfJ volat.ile compon-
ent in still has been depleted. (B) St.ilI-product composition c1lI'ves on which A is based.
a = 1.25, n = 50, Rn = 19 and 49.
B. GRAnUAI, INCm,;Ai'\J,; IN m,;FJ,UX UNL'IO
If l'eflux ratio is continually increa.sed at the proper rate from the be-
ginning of a distillation, 11 limited amount of distillate of constant com-
position will be obtained. The distillation curve in such a case is merely a
straight line that extends to the point where total reflux is required to main-
tain constant product composition. Any distillate recovered beyond this
point will be of lower purity, depending on a new choice of reflux ratio.
For instance, if 100 moleB of equimolar mixture with a = 1.25 is distilled
through ten plates with negligible holdup, calculations by the Smoker equa-
tion indicate that an initial reflux ratio (Rv) of 19 will be required to produce
initial distillate composition of 0.86. This can be maintained during the
distillation by gn,dual increase in reflux ratio. Calculations by the Fenske
equation indicate that with a = 1.25 and n = 10, when still composition
reaches 0.385 infinite reflux will be required to maintain distillate com-
position at 0.86. The moles remaining in the still when this point is reached,
SF, can be calculated hy the material halance:
(82)
in which Be = So ;= 1.00 moles (the charge), XatJ = 0;5. (the initial composi-
~i'( ':"i';"
I. THEORY 133
tion), XsF = 0.385, the still composition at the end of the period of const.ant
distillate composition (xnc = 0.86):
50 - 0.385SF = (100 - SF)0.80
SF = 75.8
Thus, after an initial fraction of 24.2 moles, no further eLstillate of purity
0.86 can be obtained.
Equations for Time Required and for Yield Fraction When Holdup Is
Negligible. BogartI 52 developed the equation:
o= So(xnc =xsn] I"F("'o dX.,/(Rn + 1

V
1 ) (xnc - x s}2 (83)
for calculating the time required for such a distillation or for calculating the
vapor velocity (V) if time (0) is fixed. In this equation So and X 80 have
their usual meaning, Xnc and XsF are the constf1nt distillate composition and
final still composition, Rn is the reflux ratio, LID, and Xs is the still com-
position corresponding to product composition, Xn, and any chosen reflux
ratio. The equation is used by calculating a series of values of:
by choosing random values of Rn , using the McCabe-Thiele procedure to

calculate corresponding values of X s, and then obta,ining the value of the
integral by graphic means.
Edgeworth-Johnstone 153 hlts developed formulas for calculating the
amount of distillate that can be recovered in such distillations. This is
done by means of the yield fraction (cpJ)) which is the ratio of the distillate
actually obtained to the maximum amount that could have been obtained
from the charge. When holdup can be neglected:
So - SF (So - SF) Pap
rpJ = = - - - - Xnc = - (84)
SoxoolxDc SOx'l) . FaJ
in whieh So is the total moles of charge, SF is moles remaining in still when

total reflux is rea.ehed, X80 is charge composition, and XDC is distillate com-
position. P, F, aPJ and af are the symbols used by Edgeworth-Johnstone
and correspond to (So - S~.), So, XVo, and x~o, respeetively. Thus in the
preceding example:
¢J = 24.2/(50/0.86) = 0.416
162 Bogart, Trans. Am)rzst. Chern. Engl's., 33, 139 (1937).
163 Edgewol'th-Johnstj~,.!nd. Eng. Chem., 35, 407 (1943); 36,1068 (Hl44).
134 A. AND E. nOSE
Edgeworth-J ohnstone uses the material-balance equations (82) and:

(So - SF') + SF = So (85)
and the Fenske equation to obtain:
(86)
Then, with the original equation for <PJ:
(
1-
Xvc
XDc
)(1 - X'D
X so )
CPJ = ----- (87)

an+! - 1
This allows calculation of ¢J in terms of ex, n, x.o and any desired distillate
composition. The expression neglects holdup, and assumes distillation
is cll,rried out with gradually increasing reflux (so that XDc is constant) until
total reflux is 1'6l1ched.
TABLE XI
CALCUl>A'l'IONS OF AVERAGE REFLUX RA'l'IO
Fill") liD Filial xa x/)c - X,Of Sp 100 - SF "'.r Avemge liD
0.385 0.475 75.8 24.2 0.416

""
99 ..113 .447 80.5 19.5 .335 38
49 .435 .425 84.8 15.2 .261 28
29 .463 .397 90.8 9.2 .158 24
19 .500 .360 100 0 0
Since it may not be economical in many cases to continue increasing reflux until
total reflux is reached, it is desirable to make the same calculations for specified re-
flux ratios at which it may be desired to stop the constant distillate composition
distillation. As an example, the value of cf>J will be calculated for the same initial
conditions as on page 132 but for a final reflux ratio of RD = 99. Calculations by
the McCabe-Thiele procedure with a = 1.25 and n = 10 indicate that, when still
composition reaches 0.413, the reflux ratio will be Rn = 99. Then, substituting in
equation (82):
50 - 0.41.3SF = (100 - 8 F )0.86
SF = 80.5
CPJ = 19.5/(50/0.86) = 0.335
Such calculations may be repeated for a series of different choices of final reflux ratio
1. THEORY 135
(Table XI), and a graph prepared of refiux ratio versus yield fraction, or against
moles of distillate (Fig. 48). From the graph it is possible to calculate the average
reflux ratio corresponding to any of the final reflux ratios, and thus the relative time
of distillation.
100....------,------,
80
a:;'"
0"
~a:
x
::::J
_j
lL.
W
a:
_j
<t
z
ii:
o 10 20
MOLES DISTILLATE COLLECTED
Fig. 48. Final reflux ratio versus moles distillate in con-
stant distillate composition distillation.163
Equations for Time Required and for Yield Fraction When Holdup Is
Appreciable. Holdup is taken into account by modifying the basic equa-
tion as follows:
(88)
in which H is total holdup.
As an illustratioll, calculations may be made for the same example used above for
the case of negligible holdup (100 moles equimolar binary mixture with O! = 1.25,
and ten plates), but with the assumption of a total holdup of 9 moles. The still
composition will decrease to x,,, = 0.482 before any distillate is obtained, due to
passage of part of the charge into the column. This value of x" is found by trial
and error solution of the material balance:
S.x" = SoxBo +h (89)
in which Se = 100, X,," = 0.5, So = S. - H = 91, and h is the moles of holdup of

the more volatile component when still composition is x" and distillate composition
is the desired value, XDc = 0.86. With trial reflux ratios RD = 19 and RD = 29,
the McCabe-Thiele operations give the compositions indicated in Table XII. It
will be noted that only nine theoretical plates are indicated. This is because the
130 A. AND E. ROSE
column is assumed to have an over-all sepamtillg power of ten plates and nille in the
column. The values of the compositions are those of the liquid phase.
Since the .total holdup is 9 moles, the holdup per plate will be 1 mole (equal dis-
tl'ibution being assumed in the ah8ence of any experimental values). The com-
position values given in Table XII for plates Nos. 2 to 10 are therefore also the actual
TABLE XII
PLA'i'E COMPOSITIONS WI'I'H VARIOUS REFLUX RATlOS
l'l'iall'cfiux ratios (interpolated)

Theoretical platcs
RD = 11l
Distillate ................. 0.86 0.86 0.8{j 0.86

Top pll1t.e #10 ........... " .83 .83 .833
Plate #B ................... 7B .78 .792
#8 ................... 75 .74 .753
#7 .......... ' ........ 715 .70 .708
if6.. ................. 688 .65 .657
#5 ................... 650 .62 .610
#1.·.................. 610 .58 .560
#3 ................... 570 .54 . .502
#2 ................... 535 .50 .445
Still #1.................. .500 .465 .483 .385
Sum #2 to ~10 ....... ..... 6.14 5.92 6.03 5.86
moles of holdup per plate of the more volatile component, and their sum is the total
holdup of this component, i.e., when Rn = HI, h = 6.14 moles. Substituting in
equation (89):
50 rf (91)(0.5) + 6.138 = 45.5 + 6.138 = 51.638
A s~cond trial with Rp = 29 gives h = 5.92 and:
50 ;;& (91)(0.465) + 5.92 = 42.3 + 5.92 = 48.22

A rough interpolation for RD = 24 gives h = 6.03 and:
50 = (91)(0.483) + 6.03 = 43.95 + G.03 = 49.98

This indicates that an initiall'efiu'X ratio (R D ) of 2-1 will be required to obtain the
desired distillate composition (XD' = 0.86). It is assumed in this procedure that
th~ distillation is started at this finite refiux ratio, rather than at total reflux, as is
more usuaL
If distillation is cOlnmenced with distillate composition kept constant at 0.86 by
gradual increase in reflux ratio, total reflux will eventually be reached and, as be-
fore, still composi.tion will be 0.385. The moles remaining in the still are obtained
from the material balance:
Sox... - SFX 8 F - h = (Sc - II - SF)XDC (90)
50 - SF(0.385) - 5.86 = (91 - S1")0.86
SF = 71.8
I. THEORY 137
Therefore:
rPJ = (100 - 71.8 - 9)/58.1 = 19.2}58.1 = 0.3:30
In this instance the huldup hal; all appreciable effect Oil the results.
IBdgeworth-J ohustone uses the m~Ltet'ial-balance equations:
(Sc - SF - H) + SF = Se - H (91)
(Sc - Sp - H )XDC + Sp;r,,~, = S,;c se h (92)
lLnd the Fenske equation to expre:-;s rPJ:
(93)
This is the algebraic equivalent of the operations in the above example, and
assumes that the distillation is continued to total reflux. The correspond-
ing equation for a finite reflux ratio is:
rl-. _ ;rDC[(x,O - xsp) - (hiS,) + (Hx.~,/Sc)]

(94)
~'J -
X8(J(XDe - x.,/,,)
In this case the value of X.P must be obtained from XDc by a McCabe-Thiele
or Smoker calculation, using the desired final reflux ratio. The same pro-
cedure also gives the figures for estimating h by graphic means.
As an example of the use of equation (94) the distillation of mixtures of chloro-
and bromobenzenlj is given by Edgeworth-Johnstone. Young's data W are used
with a = 1.8896, X'c = 0.4, n = 10, H} Se = 0.1, and XDC = 0.98. The resulting
values of rPJ are given in Table XIII, and in Figure 49. The corresponding values
(tPJ.) when H = 0 are also included. The negative values for Rl) = 4 indicate this
reflux ratio is too low to achieve the desired separation. These negative values can
however be used in making a graph, and by its means the correct initial reflux ratio
can be obtained.
TABLE XIII
YIl!lLD FRACTION AND REFLUX: RATIO
Rl) ". 4>J <PJ,
4 0.443 -0.365 -0.196

7 .200 +O.31!J +0.476
12 .157 .5133 .723
20 .117 .55!J .803
to< Young, J. Dhem. Soc., 81, 768 (1902).

138 A. AND E. ROSE
Equation (93) is tnodified by Edgeworth-Johnstone for application to complex

mixtures, to obtain:
a,,+l - :rnc ) [Sc(:rsc + a;"n) - H] - 1]

(----
c/>J = [~cX"c}:_J[ 1 - XDc Sc:!;,," - h .
S,:"Csc an+! - 1
In this X,D represents the initial mole fraction of the second most volatile component
and it is assumed that the distillate and column contain only the two most volatile
components.
20~------------------------~---r---.
: I
18
4>J> HOLDUP H/Sc • 0.1
I '
16 <PJo> HOLDUP H' 0 I I
I I
o 14 I I
~ I /
~ 12 I I
~ / I
_J 10 / /
lJ.. / /
~ 8 / .,., ...
_J 4> .,.,,,, "t'JC)
<>: 6 ;. "" .......... , ....
~ _....-"- _",-
4r----""'II&-"'::I'..-:::..-::..-.--
2
I I
-0.5 o 0.5 1.0
YIELD FRACTION
Fig. 49. Yield fraction at finite reflux ratios in distillation of ch10ro-
- benzene Imd bromobenzene. 153
As an example, 1000 moles of a ternary mixture of A, B, and 0, with X.1e = X.JJ =

0.21 and aAB = 1.5, is to be distilled under constant distillate-composition condi-
tions through ten theoretical plates in a column with holdup of 2 moles per theo-
retical plate. The product is to contain 90 mole per cent of the most volatile com-
ponent, A. The holdup, h, Was calculated to be 9.94. Substitution in the above
equation gave CPJ = 0.863.
As another example, it is required to choose between a plate column with ten
theoretical plates and holdup of 15 moles per plate, and a packed column with five
theoretical plates and a holdup of 3 moles per plate. Distillation at atmospheric
pressure, with constant distillate composition, is to be llsed to produce an 80-140°F,
fractio]l in 95% pUl'ity from a material containing 4 mole per cent of this lowest
boiling fraction, and 5 mole POl' cent of the 140-160°F. fraction. These fractions
are considered as the two lowest-boiling components, and it is assumed no others
enter the column during the distillation. Values of the relative volatility were
obtained by lise of the equation:
rl'~ - T~
log aT = 2.5 T2
1. THEORY 139
where aT is the relative volatility at the average absolute temperature, T, of the

hailing points 'PA and TB of the two components. The average boiling points, 110
and 150°F., were used, and average temperature at the bottom of column was
estimated at 120 of., that at the top as 140 of. The corresponding values of a were
2.24 at top, 1.74 at bottom, resulting in an over-all average of 1.99. Using S, =
2500 moles, x," = 0.04, X.IR = 0.05, and :tDc "" 0,95, the McCabe-Thiele construc-
tion gives h = 9.09 for the packed column and h = 58.0 for thc plate column. The
corresponding values for CPJ are CPJ = 0.55 for the packed column and CPJ = 0.415
for the plate columIl, so that the use of the packed column is indicated.
10. Curves for Fractional Batch Distillation of Multicomponent Mixtures

Bowman 155 has used the composition-distribution concept (Sect. IV) to
show how curves may be calculated to predict the progress of a batch dis-
tillation involving any number of components. This is done for the case of
assumed negligible holdup and total reflux by combining the Rayleigh
equation in the form:
C%S) (xsS) = Xp
with the composition-distribution equation relating pot and distillate com-
positions:
It is assumed that the column has a total of n theoretical plates. This

gives:
(b/oS)(xsS) :::: aUX,/ fa'" anx,da
The v[triables are separated by the substitution:
II/S = fa'" a"x,da
to give:
or:
o(x,S)/XsS = 0/."013/11
and, between the limits S = 1 and S = S, integration gives:
In (x,S/x BO ) = an;;_s (dS/Il )
This gives:
where:
166 Bowman, Ind. Eng. Chem., 41, 2004-, 2608 (1949).

140 A. AND E. ROSE
Multiplicll,tion by da, and integration with the expression:

fo'" xda = 1
gives:
Solution requires trial of v[l,riou8 arbitrarily chosen values of J 1 (8). Thes(~

are t.hen plotted against the cOl'responllillg values of S from t.he last equ~1~
tion, a.nd this graph is used to supply values for calculating :l:" verSL1S S,
according to the relation:
The equation:
is derived and used in a similar manner to obtain the corresponding diH-

tillate compositions. Figure 50 shows plots of distillate eomposition diR-
f\
-~
~---
-~
-- --- - iT-~
\ W'Q.IO--J
I
1
rFw.I.O
II_ W'O.9
iT /W.O.B
----
W'OJ-
~
VI
II W·Q.7 w'Q'~J l;:z:
J I\- 1\ W'0.6 WI 0.3 r\
I 1/\ \- I
I'..\W 'O.5 IJ
I-I J
v .) ~B:A 1'\
If
17 "7 1\ I ~ KI r!I 1\
I
-jI 1 7 / YIi) 1\ l~ IX VA ~\
II f7 V
iJ 1\ \
lt\ II V~ K 1\ \
/ / / X Z k1 \ \ 1\
17/17/
~ ~ f/ \ 1\ \
:....--v- ~
po \
'\
0 r-.... f"-
05 1.0 1.5
a. w
Fig. 50. Distillate-composition dis- Fig. 51. Distillation curves for
t.ribution during ba,tch distilla.tion of fiul. a four-component, mixture. 165 W =
ticomponent mixture.l5; W = S "" moles S; i refers to r:omponent numbers.
remaining in still.
1. 'l'HEOHl" 141
TABLE XIV
INl'l'IAI, CHAlWE COMPOSITION I'OH HYPOTHIDTICAL BATCH DIi:iTILLATION
OF FIGURES 51 AND 52
1 1.00 0.3
2 1.50 0.3
3 1.75 0.1
4 2.00 0.3
tl'ibution for various values of S during batch distillation of a mixture with

initial composition equal to that of curve I of Figure 30 (Sect. IV). Figure
51 shows distillate compositions versus S for a four~component mixture as
specified in Table XIV, using a fifteen~plate column. Figure 52 is the
--"[-"
I
I
I
- "
., 1 /
,---
.
o
/
.~ I
/
1.0
LJ w
o
Fig. 52. Distillation curve for four-component mixture,

but in terms of boiling point of distillate. 165 W = S.
eorresponding batch-distillation CUl've in terms of temperature versus per

cent of charge distilled. Similar equations have been derived by Bowman
for the case of minimum reflux.
Crosley166 has also presented methods for calculating distillate-collection
156 Crosley, "Batch Reetific[Ltion of Complex Mixtures," Am. Ohern, Soo. Chicago
Meeting, 1946.
142 A. AND E. ROSE
curves for multicomponent mixtures in simplified cases. Thus if the

separation between still and distillate follows the Fenske relation:
XDA = (aAR)"x.'A/~
where R represents a reference component, A, B, etc. represent the other
components, and:
~ = (aAR)"x SJl + (aBR)"x,B + (acR)nXSO + .. .:'D'R
Similar equations can be written for each component so that:
XDB = (allR)nXsB/~
and:
and:
xDJdxDR = (aARnX'A/~)/(XSR/~) = (aAn"XSA)/XSR
If holdup is negligible:
DXDA = :._ EX'A; DXDB = ~ EX.B; DXDR = - SXSR
or:
etc. Combination gives:

-d(SXSA)/ -d(SXSR) = aAR"Sx,,[/SXSR
and similar expressions for other components. Integration then yields:
log [(SXBA.h/(SX S A)2] = aA/I," log [(SX.n)1/(Ex'R)2]
The amount of the reference component remaining in the still at various
stages of the distillation is taken as an independent variable, and the corre-
sponding residual quantity of each of the other components is calculated as a
function of the residual quantity of the reference component. The sum of
these residual quantities comprises the total charge remaining, and com-
positions may then also be computed. Table XV gives the results of cal-
culations on the distille,tion of a ternary mixture of 333 moles of each of
three components, A, Band C, through a column with a total of ten theo-
retical plates. The relative volatilities assumed were aAG = 3 and aBO =
1.6. These give Oil~ ;:; 59,000 and a~~ :;; 110. The initial values of
SXaA, Ex.s, and Exso are each 333 and the corresponding logarithm is 2.5224.
These values are listed on line a of the table, in columns 2 through 7.
Oolumn 8 Hsts the total moles in the still, columns 9, 10, and 11 the still
compositions for the three components, and columns 12 and 13 the:
log ~SXDA = log S(aAR)nXsA and log'~SxDB = log E(aBR)"X 8B
1. 'l'HEORY 143
TABLE XV
l\1UL'l'ICOMPONEN'r BA'rol! DrS'l'ILLATION CALCULA'l'IONB
2 3 4 5 6 7 8 10 11
Log Log Log
SX,A. SX,B SX,a BXIIA SX!B S"'.o ZSxs lOOXoA
So . -1
10°".8
C;~ B
10°"'<0
Sfl C
0 2.52244 2.52244 2.52244 333 333 333 01)9 33.3 33.3 33.3
-0.295 0.00055 0.000005
2.22744 2.52189 2.522435 168.8 332.6 333 8M 20.2 39.0 30.9
2 1. 93244 2.02134 2.52243 85.6 332.2 333 751 11.4 41.3 44.4
3 1. 63744 2.52079 2.522425 43.4 331. 7 333 708 6.1 46.8 47.1
8D 2.51254 2.52235 325.5 333 058.5
0.0011 0.00001
8.E 2.51244 2.52234 324.7 333 658 '19.3 50.7
8F 2.51034 2.52233 323.8 333 657
12 1.63694 2.51430 43.34 326.0 370 11.6 88.4
0.22 0.002
13 1.4169'1 2.51230 26.12 325.4 351 6.3 92.7
.:_
0.4-1 0.004
14 0.97G04 2.50830 0.48 322.4 332 3.0 07.0
19 2.48839 307.fl 307.9
0.004
20 2.48439 305.1 305.1
21 2.48039 302.3 302.3
12 13 14 15 10 17 18 19
Log Log Sum
~SxDA tSXDB ~SxD.4. ~B:r.DB W 100xDA lOlltDB 100xDG
7.293644.56364 19,660,000 36,610 19,697,000 99.8 0.2 0

-0.295 0.00055
6.998644.56300 9,n09,OOO 36,570 10,0013,000 9fl.13 0,37 1
6.703644.56254 5,054,000 36,520 5,091,000 99.3 2
6.408644.li6199 2,562,000 36,470 2,599,000 98.6 1.37 3
1.986344. 55a74 90.3 35,790 36,123 0.27 9S.8 0.92 8D
0.59 0.0011
1.393644.55264 24.7 35,700 36,053 0.07 99.0 0.93 8E
4.5515'1 6.3 35,610 35,943 99.1 0.93 SF
0.80364 3.67814 4,766 5,093 93.6 6.42 12
0.22
3.45814 2,872 3,107 89.S 10.2 13
0.44
3.01814 1,043 1,365 70.4 23.6 14
0.81814 6.58 314 2.1 97.9 19
0.44
0.37814 2.39 307.5 0.78 99.2 20
9.93814-10 0.87 303.1 0.29 99.7 21
respectively, as obtained from values in previous columns. The corre-

sponding value of log I;SxDa is already tabulated in column 4 as log Sx.a.
Columns 14, 15, and 7 then contain values of I;SxDA, I;SxDB, and ~SXDC'
Column 16 contains the sum of columns 14, 15, and 7, and is therefore ~D.
From this the values of 100xDA, 100xDs, 100xDa are calculated and tabulated
in columns 17, 18, and 19. The first line of the table (line 0) corresponds
to the initial instant of the distillation, 01' 0% distilled. Corresponding
values for the other lines are obtained as follows:
144 A. AND E. nOSE
Three smnl! numbers in the mtio of 59,000 to 110 to 1 are selected more
or less arbitl'nrily and subtracted, respectively, from the values in line ()
of columns 2, 3, and 4. For the present illustration the numhers seleett~,)
are 0.29,5, 0.00055 and 0.000005. When the snbtraetion is performed the
1.0
z 0.8
0
i=
iii
0
a.. 0.6
::!'
0
(..)
LLJ
f- 004
<t
_J
_J
ti
i5 0.2
XDC
0 60
20 40
1.0
0.8
z
0
i=
iii
0
a..
::!'
0
C>
..J
..J
i=
U')
o 20 40 60
MOI..E PER CENT DISTILLED
Fig. 53. Calculated collection and still composition curves for three-
component distillation. 166
values on line 1 of columns 2,3, and 4 are obtained. The same subtraction
operation is used to obtain values for line 1 of columns 12 and 13. All
other values on line 1 are derivable from those in columns 2,3,4, 12, and IS.
The key values (those in columns 2, 3, 4, 12, and 13) for line BE are oh-
tained in this example by subtraction of 0.59, 0.0011, and 0.00001 from the
proper values in columns 2,3,4, 12, and 13. Tue results of the calculations
are plotted in Figure 53 which plots the mole fraction in the distillate of
1. THEORY 145
each component against mole pCI' cent distilled. In prn,ctical reetificatioll

at finite renux and with appreeiable holdup the steep fronts of the collec-
tion curves will disappear. The calculated curves do explain the practical
observation that the more volatile component is eliminated from succeeding
fractions more effectively tha,n a less volatile component is eliminuted from
preceding fractions.
The extension of the Crosley procedure to cases of finiterefiux or uppreci-
able holdup is limited by the complex nature of the relations between still
and distillate composition.
VlI. CALCULATED EFFECT OF PROCESS VARIABLES IN

BATCH DISTILLATION
1. Process Variables in Batch Distillation

The more important variables of batch distillation are: (1) reflux ratio,
(2) number of theoretical plates, (3) ratio of holdup to clu1rge, (4) vapor
velocity or throughput, (5) relative volatility, and (6) initial composition.
The first four of these are factors dependent upon the appamtus and
method of operation. The last two are characteristic of the mixture being
distilled. All the variables are subject to some choice and control, but in
general they are also dependent upon one another, on the physical proper-
ties of the components, and on the nature of the column and its plates or
packing. Between them these variables determine the time required to
complete a batch distillation, and the sharpness of separation of the com-
ponents in the mixture. The minimum time required to complete a given
distillation is predictable from the actual, operating vapor velocity, the
average reflux ratio, and the total quantity to be distilled over.167 Such
calculations of the time involved are straightforward except that they ig-
nore initial equilibration time (Sect. V), which is appreciable fol' the more
efficient columns. Calculations of the effects of the process variables on the
sharpness of separation are much more complex. The sharpness of separa-
tion achie~ed in any speeific case may be measured by the difference be-
tween still and product compositions (x" XD curves) at any instant, or bet-
ter, by the shape of distillation curves (8, XD curves) or by the composition
of successive distilhtte fractions. The derivations and procedures for cal-
culation of these curves are set forth in Sections III and V, respectively.
The present section dettls chiefly with the results of these calculations and
cites the limited experiments on the sub.ieet.
157 Rose, "Agreement between Calcuhtted and Experimental Batch Distillation Curve_s
with Appreciable Holdup," Ohemical Engineering Symposium on Distillation, Pitts-
burgh, 1946.
146 A. AND E. ROSE
Because each of the process variables affects some of the others, it is

difficult to deal experimentally with the problems of batch distillation.
Thus a change in refiux ratio or vapor velocity may cause a change in hold-
up, and also a change in H:E.T.P. or plate efficiency. The gradual change
in composition with time, characteristic of batch distillation, causes changes
in moles or mass of holdup as well as in H.E.T.P. and probably in the rapid-
ity with which equilibrium is established. These in turn affect all the other
variables of the process, and the shape of the distillation curve. There has
been, and is still, an almost complete lack of precise experimental data on
the above interrelations. A conclusive and rigorous investigation will re-
quire a multitude of careful experiments.
Because of the lack of experimental data, calculations of the effects of the
process variables can be most useful as an approximate basis for design and
operation. They will also aid in planning experimental work and in de-
creasing the number of experiments required. In order to reduce mathe-
matical complexities, theoretical analysis has used simplifying assumptions
and has considered the effect of only one varia,ble at a time.
In evaluating and using the methods [md conclusions presented, it should
be recalled that in ordinary batch-distillation practice it is not usually con-
venient, in fact probably difficult, to get accurate measurements of reflux
ratio, l'elative volatility, holdup, and equivalent number of theoretical
plates. The exact agreement between actual and calculated distillation
curves is less important than an approximate general knowledge of how
various conditions will affect the results of a particular distillation. In-
cluded in the discussion are: (1) calculated effect of holdup; (2) c.alculatecl
effect of reflux ratio, varying the number of plates and relative volatility,
but with negligible holdup; (3) calculated effect of relative volatility and
number of plates, under total reflux and negligible holdup (giving limiting
conditions); (4) calculated effect of initial composition; and (5) calculated
correlation of the combined effect of relative volatility, reflux ratio, and
number of plates, when holdup is negligible. Each of the methods is based
on a restrictive assumption, and therefore gives only limiting values or the
general nature of certain relations. None of the methocls developed so far
can be expected to give accurate agreement with experiment. The correla-
tion given last in the above list is the best available procedure for quickly
estimating the conditions required for a given separation by batch distilla-
tion.
2. Calculated Effect of Holdup
A large amount of data has been collected on the H.E.T.P. or H.T.U. of

various types of packings and contacting devices. This information is
I. THEORY 147
necessarily the starting point in the practical application of distillation

theory. While a low H.E.T.P. is certainly desirable, it is only one of the
required characteristics of a good column, since produetiQll of distillate in
adequate quantities is often as important as sharp separation. Low holdup
or low holdup per plate is usually mentioned as equally important, and the
Same may be said of high throughput per unit cross section. The ratio of
thl'Oughpl\t to holdup per plate (efficiency factor, factor A) has u,lso been
cited as a desirable criterion for evaluating packing materials. Yet both
the slight experimental evidence and the calculated curves do not justify
the assumption thl1t apprcciable holdup in a column is a[1.vays unfavorable.
A. EFFECT ON PROPORTION OF CHARGE DIS'l'ILLED
Holdup has a dual effect in batch distillation, since it limits the propor-
tion of the charge that can be distilled, and also affects the sharpness of
separation of any two components. The disadvantages of a large holdup
ean sometimes be decreased by use of a larger eharge, or by the use of chas-
ers (Chapter II), but this is not always possible or advantageous. The
effect of large holdup in limiting the proportion of the charge that can be
distilled is of major importance when the highest-boiling component or com-
ponents are present in small amount or when the available sample is
limited. The magnitude and seriousness of the effect is directly propor-
tional to the ratio between the quantity of holdup and the quantity of the
highest-boiling components. If the qual;ltity of any of the latter is less
than the holdup, their presence may not even be detected. This is more
likely with sharp separation than with poor.
B. EFFECT ON SHARPNESS OF SEPARATION
While the effect of a large holdup in limiting the proportion of the ehal'ge
that can be distilled may be simply determined, the effect of holdup on
sharpness of separation is much more complex. It would seem logical to
keep holdup Pel,' plate at a minimum in order to avoid concentrating an
appreciable portion of the sample within a few theoretical plates, and thus
prevent its effective separation. As stated before, the evidence, both ex-
perimental and calculated, does not completely confirm this generally
accepted opinion. It is interesting to speculate that, if it turns out that
total holdup and holdup per plate have no important effect on sharpness of
separation, and if chasers can be used successfully, the only l'ema,ining fac-
tors of importance in batch distillation are those that determine distillate
rate and equivalent theoret.ical plates.
Theoretical Predictions. Figures 54A, B, and C show the results of
148 A. AND E. ROSE
1.0 0,6
--- .... -. - .. /H' 0 TOTAL REFLUX,
",'
RAYLEIGH • ,/H =0 McCABE THIELE
W
~
...J
0,5
'/'
,
. ,
RAYLEIGH
H = 28,8
...J
i= ,,
Vl
0 ,
0.4
~
)-*28,8
'S!
~ )---" / \, I
z H'57,6j"" , 0.3 H=576
0 0.4 ,
i=
u
<:! H·Q McCABE
a:
.....
.
THIELE
w 0.2
...J
RAYLEIGH , 0.2
0 H= 7.2
~
0.1
0 5 10 15 0 5 '10 15
MOLE PER CENT OF CHARGE DISTILLED
A B
_ri= 0% TOTAL REFLUX

RAYLEIGH
0.8
w
I-
o;!
...J 0.1 H =0% McCABE,
...J
i= THIELE,
til RAYLEIGH
15 0.6
~
u
>
::;; 0.5
z
0
Ei
o;!
0.4
a:
LL.
0.3
~
0
::;;
0.2
0.1
oL------75------7.0~----~1~5----~2fO~----~2~5----~3~0------~35
MOLE PER CENT DISTILLED
c
0.80
0.60
o 4 8 12 16 0 4 8 12 16
WEIGHT PER CENT OF CHARGE DISTILLED
D E
1.0 r - - - - - - - - - - - - - - - - - - - - - - - - - - - ,
W
I- ~/3% HOLDUP
«
..J
..J 0.7
i=
'!?
0
0.6
~ ~',o% 6< 20')'.
u
> 0.5
:;;
z
a
i= 0.4
u
«
0:
lJ._
w 0.3
..J
a
:;;
0.2
0.1
0 I 2
0 5. 10 15 20 25 30 35 40 45 50 55
WEIGHT PER CENT OF CHARGE: DISTILLEO
F
Fig. 54. Curves for batch distillation of ethylene chloride -toluene mixtures with
appreciable holdup. 158 (A) Calculated; five Lheoretical plates and still i 9.6 mole per
cent ethylene chloride in ch!Lrge; total reflux startup. CB) Calculnted; five theoretical
plates and still; 9.6 mole pOI' cent ethylene chloride in charge; finite reflux startup.
(C) Calculated; five theoretical plates and still; 25 mole per cent ethylene chloride
in charge i total reflux startup. (D) Experimental; four theoretical plates and still;
9.6 per cent by weight ethylene chloride in charge; totall'efiux startup. (E) Experi-
mental; four theoretical plates and still; 9.6 per cent by weight ethylene chloride in
charge; finite reflux startup, (F) Experimental; four theoretical plates and still; 25
per cellt by weight ethylene chlOIide in charge; total reflux startup.
Reflux l'atio, RD , is 4/1 except where otherwise noted.
150 A.. A.ND E. ROSE
calculations hy Rose, .Johnson, and Williams/ 58 using the stepwise plate-

to-plate method described on page 126. Figure 44 shows similar CUl'ves
calcula.ted with the assumptioll of total reflux, and using t.he general equa-
tion for a batch-distillation cllrve as descl'ibed on page 123. All calcula-
tions by the latter procedure indicate a poorer separation as the ratio of
holdup to charge becomes larger, regardless of the specific values of initial
composition, relative volatility, number of theoretical plates, and reflux
ratio. The stepwise calculations for finite reflux (Figures 54A-C) indica,te
a much more complex situation, with two new factors entering the situation.
These are the charge composition and the initial conditions of the distilla-
tion. Thus in Figure 54A the curves were calculated for distillation of a
mixtme contnining 9.6 mole per cent ethylene chloride in toluene, through a
five-plate column with reflux ratio 4/1 and holdup of 2.88, 7.2, 14.4, 28.8,
and 57.6% of tlie charge. In these calculations it was assumed that the
column was at equilibrium at total reflux before the start oithe distillation.
The curves of Figure 54B were calculated for the same distillation condi-
tions except that holdup was assumed as 28.8 and 57.6% and the column at
equilibrium at 4/1 reflux (with return of product to still) before the batch
opcl'ation commenced. For the curves of Fig, MA it can only be said that
the effect of increasing ratio of holdup to charge is complex. As far as they
go, the curves of Figure MB indicate an effect similar to that at total reflux
such as in Figure 44. While batch distillations are seldom started under
finite reflux conditions, it is quite possible that the second and succeeding
breaks in distillation curves for multicomponent mixtures might be related
as are those of Figure 44. Similar considerations might apply to mixtures
where the break in the distillation curve does not occur so near the begin-
ning as is the case in Figure MA. Figure 540 shows such curves, calculated
for conditions similar to those of Figure MA, except that charge composi-
tion of 25 mole per cent ethylene chloride was used instead of 9.6 mole per
cent. In Figure 54C it appears that the curve for 28.8% ratio of holdup to
charge givea a slightly sharper separation than either smaller or larger
ratios. The smallest holdup gives the best initial separation but this situa-
tion does not persist. There is however but little difference between the
curves for 2.88, 7.2, 14.4, and 28.8% holdup ratio. The corresponding eX-
nerimental curves are shown in Figures MD-F.
Pigford, Tepe, and Garrahan 159 (see pages 96, 126) have al::10 cal cu·
latecl bat-eh-distillation curves for CtLSes involving appreciable holdup.
lli8 Rose, Johnson, and Williams, "Stepwme PII~to to Plate Calculation of Batch Distil-
lation Curves," Am. Chern. Soc. Chicago Meeting, 1948 (to bo published in Ind. Eng.
Chern),
159 Pigford, Tepo, and Garrahan, "The Effeot of Column Holdup iu Batch Distillation,"
Am. Chern. Soc. Philadelphi!1 Meeting, 1949.
1. THEORY 151
Their calculations were made by using n differentiall1nalyzel' to solve the

basic differential equations previously derived by Marshall and Pigford. llio
The vapor holdup was assumed negligible, and vapor-liquid equilibrium
was assumed to be of the form y = a'x + (a' - 1)x~, where a' = (3a - 1)/
(a + 1). The usual relative-volatility relation waS avoided because it
reduces the capacity of the differential analyzer. Even so, the existing
analyzers cannot perform solutions for cases involving more than five to
ten plates, depending upon the exact design of the analyzer. Some dis-
crepancies in material balances also occur when this procedure is used.
CalClJlations were made for eighteen cases, as summarized in Table XVI,
TABLE XVI
EFFECT Ol!' HOLDur ON SIZE OF INTEHl\mDIATE FRACTIOJ>,'69
Ill} II/Bo I Xo x, (check)
2.23 8 0.024 0.269 0.470 0.451

2.23 8 .024 .279 .470 .456
2.23 8 .029 .294 .482 .475
2.23 3.2 .0'15 .490 .473 .466
2.23 4 .045 .431 .473 .468
2.23 8 .045 .35() .473 .463
2 10 .02 .31 .502 .478
2 10 .04 .38 .502 .486
3 10 .02 .125 .510 .504
3 10 .04 .170 .518 .495
3 10 .08 .310 .526 .506
3 3.2 .04 .248 .518 .515
1.5 10 .02 .834 .535 .519
1.15 10 .02 .500 .484
but five were discarded because material balance errors were more than 3
mole per cent. All cases were for a seven-plate column, with initial equi-
librium at total reflux. The charge composition was taken as approx-
imately 0.5 mole fraction of the more volatile component, as indicated in
Table XVI. This also gives the size of the intermediate fractions having
composition limits XD = 0.95 and XD = 0.05. The last column of the table
gives the mole fraction of more volatile component that would be obtained
if all the dist.illate and the residue in the column were combined. These
values should be identical with charge composition; the discrepancies are
apparently due to computation errors of some kind.
Based on these calculations, Pigford, Tepe, and Garrahan concluded that
there was a general tendency for the size of the intermediate fraction to in-
160 Marshall and Pigford, Application of Differential Equations to Chemical Engineering
Problems. Univ. Delaware, Newark, Delaware, 1947.
152 A. AND E. ROSE
crease with increasing holdup. In one case involving high relative vola-
tility the size of the intermediate fraction went through a minimum with
increasing holdup, the optimum being for holdup corresponding to 2% of
the charge. These authors make a preliminary correlation of the results in
terms of (HiISo)/(a - 1) and I/(a - I)(RD). They conclude that the
presence of holdup in a column results in an "inertia" effect that makes
sharp separations sharper, but fails to affect relatively poor separations.
This effect is stated to be more noticeable at the lower reflux ratios. Some
of the calculations are for cases corresponding approximately to the experi-
ments of Colburn and Stearns,161 and, while direct comparisons were not
made, it was noted that the calculated intermediate fractions approximated
those obtained experimentally.
Experimental Work. Podbielniak 162 observed that difficult low-temper-
ature analytical separations were improved by usc of a higher holdup to
charge ratio, and this has been taken into account in the design and oper-
ation of the Podbielnink low-temperature columns. Colburn and Stearns 161
showed by expcriments with ethylene chloride-toluene mixtures that the
McCabe-Thiele procedures could not be depended upon to predict still-dis-
tillate composition relations in hatch distillation when holdup was appreci-
able. Their results showed that separation was better than predicted by
the McCabe-Thiele calculations, rather than poorer as might luwe been
expected. In other words, holdup was beneficial rather than detrimental.
This was explained in terms of the effect of holdup and changing composi-
tions with time in batch distillation, as discussed in Sections III and V.
Stepwise plate-to-plate calculations and curves of Figures 54A and C in-
volve the same considerations, and also show the beneficial effect of holdup
as compared with the McCabe-Thiele calculations for no holdup. This
work leaves little doubt on what to expect when experimental distillation
curves for appreciable holdup are compared with those calculated by the
McCabe-Thiele procedure either with or without holdup. A point of even
greater interest is the comparison of experimental curves with one another
to establish a beneficial or detrimental effect of increasing ratio of holdup to
charge.
Direct experimental work on the effect of the ratio of holdup to charge in
batch distillation has been done by Rose and Houston, O'Brien, and
Prevost.163 Rose and Houston showed that in a 21-plate packed column,
151Colburn and Stearns, Trans. Am. Inst. Chem. Enurs., 3'7,291 (1941).
182Podbielniak, Analytical Distillation and Its A.pplication to the Petroleum Industry,
Podbielniak, Inc., Chicago, p. 7.
m Rose and Houston, "Experimental Determination of the Effect of Ratio of Charge
to Holdup.in Batch Distillation," Am. Chern. Soc. Chicago Meeting, 1948. O'Brien,
M.S. Thesis, Pennsylvania State College, 1948. Rose, Williams, and Prevost, Ind. Eng.
Chem., 42, 1876;(1950),
1. THEORY 153
[I,small charge gave D, sharper scp:1ration than :t larger one, cxeept for the
ease of 40/1 reflux, where the same separation was obtained regardkss of
the size of the charge. Similar compal'n.tive disWltttiollS in two different
thirteen-plate eolumns showed neither flworable nor unfavorable effor't of
increasing the mtio of charge to holdup. All the experiments were on
binary mixtures with charge composition ranging up to 40% of the more
vol::ttile eomponcnt.
Thus these experiments indicated a favomble effect of inereasing ratio of
holdup to charge in some cases, and no effect in other cases. The work of
O'Brien indieated an unfavorable effect of increasing the holdup ratio for
ha1eh distillations which were started at finite reflux. The experiments
of Prevost were similar to those of Rose and Houston, and gave similar
results. In general the experimental curves resemble those calculated by
the stepwise plttte-to-plate procedure, in that relative positions are the
same in the crossing and recrossing of curves for different holdup, and in the
shape of the x,, XD curves.
Consideration of all the experimenta,l and calculated curves suggests that
for each mixture and column there is a critic!11 reflux ratio above which
there will be an unfavorable effect of increasing the holdup, and below
which there will be a favorable effect. ,The critical reflux ratio increases
with the number of theoretical plates in the column, and so for a given
mixture a favorable effect may be obtained with a column having few
plates, and an unfavorable effect for another column with a larger number
of plates. Neither the experiments nor theory are developed to the ext.ent
that specific predictions can be made for llew cases.
3. Calculated Effect of Reflux Ratio, Relative Volatility, and Theoretical

Plates When Holdup Is N egligihle
Curves to show the effect of these variables on the sharpness of separation
have been calculated 164 by the equation:
log St = log So _ _ 1_1"'"

2.303 x. j
dx"
XD - x,
The details of the derivations and methods of calculation are those
described on page 102. The procedure always depends upon the
nnmerical solution of the log St equation. Graphs of per cellt distilled
(Be ~o B X 100 =
t per cent distilled) versus distillate composition (XD) may'
he constructed as in Figures 37 and 55. For example, the C'.lll've marked
RD = 1 in Figure 55A represents the distillation of 100 moles of an equi-
,
164 Rose and Long, Ind. Eng. Chem., 33, 684 (1941).
154 A. AND E. ROSE
A
n=l1
.2
0<'2
:I
B
n = 6.5
.8
1.0
~
~.2
i:::
~.4
c:i C
~ n = 47
~
~~
l!S
~
~L
I;,J
D
n = 91
Fig. 55. Calculated effcct of refiux l'atio when holdup is assumed

to be negJigible,16l Ol'dinm'y distillation curves, Pel' oent of eharge
distilled = 100 - moles remaining in still, Numbers on curves indi-
cate reflux ratio, RD.
I. THEORY 155
molal' binary mixture with a = 1.5, n (number of plates) = 11, reflux ratio
(RD ) = 1, and negligible holdup. The series of values of Xv and Xs used in
evaluating the integral in the log Be equation were obtained from still-
product composition curves such as those of Section III. These were
calculated by using the McCabe-Thiele or Smoker procedures.
The choice of a charge other than 100 moles in the preceding; calculation
would make no difference in the final curve, if values of S I1re converted to
per cent distilled. The chief effect of an initial composition other than
x, = 0.5 would be to displace the curve in a horizontal direction, but there
.would also be an increase in the sharpness of the break as higher values of S
were u~ed. The succeeding curves in this discussion are calculated on the
basis of 100 moles and initial composition (x s) of 0.5. A change in the
relative volatility, the number of plates, or the reflux ratio used in this
calculation would cause a change in the still-product composition curve,
and therefore in the series of values Xs and Xl) and finally in values of log 8,
and thus change the shape of the distillation curve. The McCabe~Thiele
procedure or the Smoker equation were used in all the examples of Figure
55, but any other suitable method (Sects. III and IV) could be used if
simplifying assumptions are not valid. 'l'hus,.if holdup is negligible, the
equation above may be used to estimate effect on bl1tch~distillation curves
of reflux ratio, relative volatility, t,heoretical plates, or any other variable
that can affect still-product relations. Over the restricted range that has
been investigated, curves calculated by tlse of this equation coincided closely
with experimental curves,lG6 within the limits of experimental error in reflux
ratio and H.E.T.P. determinations.
A. EFFECT OF REFLUX RATIO
Calculations from Log St Equation. The calculated curves of Figure 55

give a definite idea of the relative advantage of successive increases in reflux
ratio when holdup is negligible. Figure 55A shows the effect when relative
volatility is 1.5 and there are eleven pJatesj Figures 55B-D give similar
sets of curves for mixtures with higher and lower relative volatilities. Each
of these sets of curves has distinct characteristics in spite of the fact that
the over-all fractionating factor, an, is approximately equal in all cases.
Thus in Figures 55A and B, when IX. = 1.5 and 2, respectively, a reflux ratio
of 9 is sufficient for a fair separation, and increase of Rl) beyond about 30
results in little further improvement in sharpness of separation. This is
emphasiriled by comparison with the curves for the sharpest possible separa-
tion calculated on the assumption of distillation under total reflux (RD =
00 ). The curves of Figures 55C and D were calculated for smaller values of
166 Smoker and Rose, Trans. Am. Inst. Chem. Engrs., 36, 285 (1940).
156 A. AND E. HOSE
---- ~:~;;
,
/
20 30 40 50 GO ?O ,sO .90 /OO---;?O/}----O<J

REFLUX RATIO
o
20 40 1i0 dO
REFLUX RATIO OR
RELA7IJlE TIME RlQ(I/,flO
D
Fig. 56. CulcuhLted effect of reHux ratio when holdup is assumed to be neg-
ligible. 161 (A) On size of intermedifLte fraction. (D) On slope of break in .distilJa-
tion curve; the figure plots the angle of the distillation curve with the horizontal,
rather than the actual slope. (0) On composition of 40 to 45% fraction. (D)
On composition of first 40% distilled.
I. THEORY 157
a (1.1 and 1.(5) but hrgcr 12 so that an is approximately the R~mc us for
Figures 55A and B. In these c:J,scs a reflux ratio of \) results in pOllr
separation, and high reflux ratios arc nneessnry to aehieve shnrp sPp:tratioll.
Increase of Rn to 99 or U)9 results in all appreeinhle impt'ovmnent ill these
examples.
Methods of Summarizing Effect of Reflux Ratio. In ol'd(~r to show
den,rly the effect of reflux ratio and other process variables under different
conditions of distillation, it is de8irable to have some method of combining
and snmm:1l'izing all results from studies such as those of Figure 55. This
may be done in a variety of WEtyS, some of which are illustrntecl in }i'igure
56. Yield fmction (page 133), could similarly be used, but published
curves are lacking.
In Figure 5GA the size of the intermediate fraction (1) is plotted against
reflux ratio for each of the curves of Figure 55, as well as for Some addi-
tional similar studies. The intermediate fraction includes all the distillate
intermediute in composition between the cut purest in the more volatile
component and that purest in the other component. In the present case
the limits of purity have been arbitrarily chosen as 0.9 and 0.1 (mole fmc-
tion of more volatile component). The numerical value for such an inter-
mediate fraction can be read from Figure 55A as SXD=O.9 - S"D ~O.l. For the
RD = 9 curve, this is approximately 33. Such curves are advantageous in
that they show the relation between the reflux ratio and the results directly,
the lowest possible values of I being the most desirable. The value of 1 may
also be calculated by formulas snch as the following. For the simple case
in which both total reflux and negligible holdup may be assumed:
I = S"n=O.9 - Sxn=O.l
= k~(0.9)1/(""-1) [a" - (0.9)(0!" - 1) ]/(1 - 0.9)""/(0:"-1)

- 1c~(O.l)l/(an-l) [a" - (O.l)(a n - 1) ]/(1 - 0.1)"0/(,,,0- 1)
This is a form of the solution of t.he Rayloigh equntion for fractional dis-
tillation hy means of the Fenske equation (Sect. V, equation 16). The
size of the interme<iinte fractiun, th!1t is, the sharpness of separation, de-
pends only on (Xn and k~ in this casco The integration constant k~ in tUrn
depends upon charge composition and on a" so that these are the ultimate
variables for the case in which total reflux is assnmed. The choiee of Xn =
0.9 and XD = 0.1 as the limits of the intermediate fraction is arbitrary.
Other. choices may be made as convenience dictates. Similar equations
may be derived for cases in which total reflux and negligible holdup cannot
he assumed. In such cases the size of the intermediate fraction depenrlR
upon the actual reflux ratio and holdup.
A second method of summarizing the effect of reflux ratio is given in
158 A, AND E, nOSE
Figure 56B in which the slope of the steepest portion of the break in the
variOLHi curves of Figul'C 55 is used as a measure of the sharpness of separa-
tion and is plotted against the reflux ratio, This method would be useful in
cases such as totalrefiux, in which an algebraic solution exists for the log
8 t because the value of the slope at the point of inflection where it is at n,
maximum can be found directly by mathematical methods and without
resort to the plotting of graphs,
Probably the simplest and most useful criterion in batch fractionation is
the purity of an arbitrarily chosen fraction collected at a specified point just
prior to the break in the distillation curve, In Figure 56C the purity (as to
the more volatile component) of the fraction coming over between 40 and
45% distilled (8 = 60 to 55) is plotted against the reflux ratio. This set of
curves probably shows most clearly the effect of reflux ratio and the vari-
ation of this effect with a, an, and n, The larger values of XD at 40 to 45%
distilled represent sharper fractionation, Curves A, B, and E show the
marked effect of reflux ratio when a and an are both largc and Rn is smallj
curves C and D show the relatively slight effect when a is small, even though
an is just as large as for curves A, B, and E, Comparison of curves C and G
shows the effect of increasing n when a is small.
Since the time required for distillation of a specified amount of product is
directly proportional to the reflux ratio used, curves such as those of Figure
56 give a relation between the time required and the results of a distillation.
In Figure 56D, the average composition of the first 40% distilled is plotted
against the reflux ratio for various typical cases of batch distillation, Such
relations are of great practical interest since the proportion of the charge
that is recovered and the purity of the products obtained in a distillation
determine the retul'Il on the operation, while the reflux ratio is directly
related to the time, heat input, and cooling water consumed. A complete
study must of course include the variation of prices with purity, and con-
struction costs per theoretical plate,
D, EFFECT OF REFLUX RA'l'IO ON EFFECT OF NUMBER OF PLATES
Figure 57A gives calculated curves for the distillation of an ideal equi-
molar binary mixture with a = 1.25 and reflux ratio RD = 9 in apparatus
with the equivalent of ten, twenty, thirty, and forty plates, Figure 57B
gives similar curves for the identical case when RD = 49, Figure 58 sum-
marizes these and a number of other cases by the method of Figure 56C
(40 to 45% fraction), The heavy lines on Figure 58 refer to Figures 57A
and Band similar'unpublished curves and show the large effect of n when
RD is large and the slight effect when RD is small, They also show that in-
crease in reflux ratio beyond 49 gives little improvement in separation when
a = 1.25 and that there is little to be gained by using a large number of
I. THEQRY 159
Fig. 57. Calculated effect of number of theoretical plntes; (A) at low

reflux l'atio (R D = 9); (B) n,t high reflux ratio (RD = 49).164
EFFECT OF ,PLATES AND REFLUX /i'ATlO

~1.0
~ 'r---t-~~--r---r-~~~~~~~
~
It
~ •.9
~
c:s
~.o
~
~ l~4-~~~__~~__+-__~1
~
~ .6'I--~:__-1-~I---l--+--l---c"'+""'~
~
~.S~-4~~__~__~~~~__~~
u 5 m ~ ~
THEORET/CAL. f'L;1 T~5
Fig. 58, Summary of calculated effects of theoretical phtes and
reflux ratio (R = RD).I04
160 A. AJ'!D E. HOSE
plates when RD has us low u value us 9. The other curves of Figure 58

indicate the changes in these rebtiol1s when (lC has smaller or lurger values.
In euch case, as ltD is incl'Cftsed, the sharpness of separation approaches
that of distillation uuder tota,l reflux. When a is large (l.5 or 2.0) the
limiting sharpness of separation is nearly reached at low values of RD , and
further increase in RD can have but little effect. With the smaller values
of D! (1.1 or l.05) the limiting sharpneBs is approached more gradually and
regularly as RD is increased. This npproach is often so gradual that it is
of considerable value to know the limiting sharpness of separation for a
specific case. Thus by ma.king calculations for total reflux it is possible
to show definitely that curve D of Figure 56C approaches a value of XD =
0.94 as a limit, while curve G will approach only XD = 0.7 as its limit. The
limiting sharpness of separation is determined only by the value of a".
Thus all the curves of Figure 55 approach almost the same limiting eurve
as RD is inel'eaHed, sinr:e an is nearly the Rame for all cases.
C. El"F1GC'l' Ole HJ<iFLUX: ItA'l'lO-S'l'EDMAN 1'1WCgDUm~
Steelman 166 suggested tlmt the expression:

(kn + 1) -'hV ID
be used as u, measure of the over-all effectiveness of a column and procedme.

This formula expresses the relative time required pel' mole per cent per
theoretical plMe, in terms of n (theoretical plates) and V ID (the vapor-
distilhLte ratio).
5 THEORETICAL PLATES
w
..
> METHYL ALCOHOL OBTAINED IN ONE DISTILLATION
~...J
'0"/0
I
W
a:
0.5 0.6 - DIV} REFLUX

III 0.67 -- LID RATIO
Fig. 59. Stedmall calculations of distillate composition, rell1tive time, and reflux
ratio. 1GG
Stedman used this expression to obtain tho curves of Figure 59, in which
relative times for various separations of ethyl and met,hyl alcohol are plotted
laG Stedman, Can. J. Research, 5, 455 (1931).
1. THEORY 161
TABLE XVII
FAC'l'OR BY WIIICH CRl'l'lCAL VALUg OF Dry' l'on INFINl'l'E COLUMN SHOULD BE Murm-
PLmD l'OR MOST ECONOMICAL WOHK16G
.~.------
Charge composition Adequate lvlnderatcll'

x colulllIl adequate COiUlllll a Inadcqultte 001U11111
o.2 or higher 0.5 to 1 0.25 to 0.5 or 1.5 to 2 . 5 or more "as the column
1 to 1.5 becomes less efficient
0.05 to 0.2 0.5 to I 0.25 to 0.5 or 2 to 3 or more as above
1 to 2
Less thun 0 . 05 vi;; vi;; Vllx
a For modemtoly adeqUl1te column t!1ke t,]m top figure if the column will jus~ give the
desired ruslIlt in OIW distilbtiuIl; if two are nerecled take the lower figure; if more than
two, considl~r tho column inadequate.
against the distillat.e-vapor ratio. The gnl,ph also shows the more fre-
quently used reflux rat.io (LID), and the purity of the product corresponrl-
ing to the various reflux ratios and relative times. Considera,tion is given
to columns with 5, 10, 20, 40, and an infinite number of theoretical plates,
the last being obtained by [L modified method of calculation.
LID DIV b.t
67/ 1 .......... 1 . 0 , . . - - - - - - - - - - - - - - - - - - - - - . ,
1.5/1 ......... 0.4
0.2
9/ 1 ......... 0.1
30/1 ........ 0.04
99/ 1 ........ 0.01

o
~
3 cr:
10'I"X'" 0.001
~
LJJ
.1071 .. ~••• 0.0001
0.001 0.01 0.1 0.2 0.4 0.6 1.0

MOLE FRACTION MVC IN CHARGE
Fig. 60. Stedman calculat,ions of boiling point difference, still composition, and
. optimum reflux mtio. 'G6
It is to be noteu that the f01'ty- and twenty-plate columns (adequate

columns) show [1, minimum of time required at about DIV = 0.04: (LID =
24/1), and that this minimum is about half the DIV value (the critical
162 A. AND E. ROSE
value) corresponding to zero time on the curve for infinite plates. For
smaller numbers of theoretical plates (inadequate columns) the minimum
time is displaced to much higher distillate-vapor ratios (lower reflux-
distillate ratios). Stedman genemlized his recommendations (see Table
XVII and Fig. 60) showing the general effect of still composition and rela-
tive volatility on the proper choice of reflux ratio.
Stedman's reasoning involves all the usual simplifying assumptions, i.e.,
ideal mixtures, normal liquids, and applicability of the McCabe-Thiele type
of material balance. The latter necessarily restricts the recommendations
to steady-state distillations, or to instantaneous conditions during n batch
distillation with inappreciable holdup.
4. Calculated Effect of Relative Volatility and Number of Plates (Total

Reflux and Negligible Holdup Assumed)
Thc curves represented by equation (14) (Sect. V) give the relation be-
tween distillate composition a11d pel' cent distilled on the basis of total reflux
and negligible holdup. Curves such as those in Figure 61A are obtained,
which vary in shape according to the values of relative volatility (a) and
number of plates (n) substituted in the equation. Thus D (Fig. 61A) repre-
sents the curve calculated for distillation under substantially totall'efiux of
100 moles of an equimolar mixture with a = 1.25 and initial composition
(xs) of 0.5 in a column with the equivalent of ten theoretical plates and no
appreciable holdUp. Curves Band C represent the 'systems with twenty
and thirty plates, respectively, with all other circumstances identical.
Curve F is a portion of the corresponding curve for forty plates. Because
of the assumption of total reflux and no holdup, the curves obtained repre-
sent only the limiting case of the sharpest possible separation with the
specified mixture and column. In this sense they are analogOlls to product
concentrations of a chemical reaction as calculated by means of equilibrium
constants-they represent a limiting condition that may not be attained,
but that probably cannot be exceeded. This sort of information is of great
value when a sharp separation is desired. Even casual study of Figure 61A
indicates that the increase from thirty to forty plates has a relatively small
effect on the shape of the distillation curve and the maximum sharpness of
separation when a = 1.25, and that further increases in plates can have
even less effect. In order to obtain a definite measure of the sharpness of
separation, one may arbitrarily designate as intermediate fmction (1) all
that distillate having a composition bet\veen 10 and 90% of the more
volatile component. In Figure 61B the moles of such intermediate frac-
tions (mole per cent of charge) are plotted against the corresponding values
of n. A law of diminishing returns is obviously operating, and there is an
I. THEORY 163
optimum number of plates for u given Hepartition. Further increase 1I1

plates fails t.o effect any great improvement in separation.
w ",n=807 ,,'=86
!;t n = 30 C B n =20 o
....J 0
....J
f=
",n =9.3
<f) n = 10
Ci 0.2
£
u
>
::;: 0.4
z
0
f= 0.6
u
«
a:
L1.
w
....J
0 "n = 7523
::;: F n = 40
10 20 30 40 50 60 70 80 90 100
PER CENT DISTILLED
Fig. filA.. Calculated effect of relative volatility and number of

theoretical plates at total reflux. lu7
w
r.!)
a:
«
:I:
u
....J
g
l-
lL.
o
~
...;
n
",n
Fig, 6lB. Calculated relation between size of intermediate fraction

and number of I;heoretical plates. Limits of intermediate fraction: G,
10-90%; J, 5-95%; K, 1-99%.107
167 Rose and Welshans, Ind. EnO. Ohem., 32, 669 (1940).
164 A. AND E. ROSE
A. OP'TBfUM RANGE OF OVER-AI,L FRACTIONATING FACTOR
The preceding discussions indicate the greater signifiennce of the over-all

fn1Ctionating factor, a:", as compared with either the relative volatility, Ci,
or the number of theoretical plates by themselves. Examination of Figure
G2A shows thnt only certain rehttively nalTOW ranges of values of an (those
near the origin) are of prl1.ctieal interest. Decrease of an below a minirnum
value results in poor separation (a large a.nd rapidly increasing intermediate
fraction); increase of an beyond a certain maximum can result in but little
improvement in fractionation but does involve increased cost of construc-
tion. In genoml only intermediate fractions of less than 10 a.nd morc than
60m---------------------, 30r-----------------~
50 25
40 20
,_, 30
2.0
w~
o 1000 2000 4000 6000
an
A B
Fig. 62. (A) Calculated relation between size of intermediate fraction and value
of overall fractionating factol'.167 (B) Graph of A near origin. G, J, and J( have same
values as for Figure 6IB.
0.5 mole per cent and the corresponding values of an are of interest. This
portion of Figure 62A is plotted on a larger scale in Figure 62B. Figurcs
62A and B apply only to cases of substantially total reflux and negligible
holdup, and show the calculation of the limit.ing cases of sharpest possible
separation and minimum plates.
B, MINIMUM PLATES FOR A GIVEN SEPARATION
By use of Figures 62A and B it is possible to calculate the over-all frac-

tionating fa.ctor necessary and the minimum plates required for a given
separation of a specified mixture. For example, suppose it is desired to
I. THlWRY 16,5
make the intermediate fraction containing 90 to 10 mole per t:ent of the

more vobtile component us low as 5% of the charge. (The remaining 95%
wOllIcI constitute the prttc:ticaIIy pmc [radians of the tlVO components.)
In order to achieve this, with a = 1.2,5, 'II must be ahout 25. A column of
at least this number of plates would be necessary to attain the desired
seplLl'ation.
C. CALCULATI~D POSITION OF BIUDA-I( IN D[S'l'ILL:l..TION CURVE
It is common practice in a,nalytict11 fractionation to usc the break in dis-

tillation curves as a basis for establishing it cut point an'd thereby determin-
ing the composition of the mixture being distilled. Examination of curves
Band C (Fig. 61A) shows th[Lt the most nearly vertical portion of the curves
does not coillcide with 50 mole per cent distilled, as might be cxpected.
Instead it occurs at a slightly greater perccntage distilled. If in C (Fig.
61A) the cut point were placed to correspond either to the steepest portion
of the curve or to distill[1te composition of 50 mole pel' cent, the. original
sample composition would comc out nearly 51 %. This displacement of the
steepest portion of the curve becomes greater with smaller values of an.
Thus with Ct." = 86 (curve B) a sharp break is obtaincd but is displaced
several per cent. Similar effects may also be noted in the curves of Figure
55, which were calcubted for finite reflux and negligible holdup. In actual
practicc it. is often not mole per ccnt product distilled that is plotted as
ahscissa, but more probably weight per cent. Furthermore the ordinates
are often not composition of product in mole per cent but boiling point or
some other property not linearly related to molar composition. These
factors lead to distortion of various kinds in a.ctual distillation curves that
have llot been closely ex[uuinecl.
5. Calculated Effect of Initial CompOSition on Shape and Position of
Distillation Curves
It was stated above (page 153) that the maj or effect of a change in initial
composition was to displacc the calculated distillation curve along the
horizontal axis. This statement applies to log St curves (page 102) caJou-
liLted for cases in which holdup is assumed to be zero. If the initial com-
position (x,) is 0.5 in one instance Hnd 0.7 in another case, the corresponding
equations are:
1
log S!o., = log 8, - 2 3
•
1°·5
"'"
_d_,x_:G:__
Xn - x.
and:
log StO.7 = log; Se _ 213 (0,7 dx •
. )""t XD - Xs
106 A. AND l'~. nOSI>;
in which S/o., and St•. 7 represent values of S eOlTesponding (;0 the same still
composiLion, X,'I, for each of the two cases. The charge Se and the XD versus
x, relations are the same for the two cases, fiO that subtraction of the first
equation from the second gives:
log StO.7 - log S/o., = -

1
2.3
1°·7 0,5
dx,
--'-,-.
XD - .t"
= a constant
The distillation curve for initial composition (x,e) of 0.7, i.e., the series of
values of log S for this case, can be obtained from the corresponding values
for the ease x." = 0.5 merely by adding the constant:
1 (0.7 dx,_
2.3 )0.5 Xn - x,
Thus any value of log Stu" differs from the corresponding value of log S/o.,
by a eonstant factor, as long as Se is the same and the same series of XD and
x. values are used. Choice of X,1e values other than 0.5 and 0.7 does not
alter the reasoning. Because the corresponding logarithms differ by a con-
stant, the values of S themselves differ by an exponentially increasing
amount, and so the break in the distillation curve will be sharper for higher
initial compositions.
6, Calculation of Reflux Ratio and Theoretical Plates by Pole-Height

Equations for Total Reflux and Negligible Holdup
The pole height, 0', of a binary batch-distillation curve is defined 168 as the
product of the slope of the curve at midheight, m s, and the fraction of the
charge remaining in the pot, S,,:
(J = m"S"
When holdup is negligible, the pole height has the property of being inde-
pendent of charge composition, as exemplified in Figure 63. For usefully
sharp separations the fraction of the charge remaining in the pot at the
midpoint of the distillation curve is approximately equal to the fraction of
the heavy component in the original charge, so that.: .
S" = 1 - Xc
This gives:
or m, = 0/(1 - xc)
Since II is identical for a given set of values of Ct, n, and RD , it follows that
t.he sharpness of separation is inversely proportional to t.he concent.ration
ISS Bowman and Cichel1i, Ind. Eng. Chem., 41, 1985 (1949).
I. THEORY 167
of the heavy component in the charge. Better sepamtion is obtained on

rieh mixtures thall on Iran ones, as has already been notcd in the preceding
section all the effect of initial composition. The dependence of the pole
height, CJ, on a and n at infinite reflux is sho,,"n by Bm\'man aml Cichelli to
be:
u = (a 2n - 1)/8a"
which rednees to u = a n/8 when
separation is sharp, so that a 2n is ltLl'ge
compn,red with unity. This is derived
t'rom the Rayleigh equation in the
form:
S«(h,,/d,s) = :r.:D - :r,
by multiplying both sides by dXD/dx s :
S(rb:D/dS) = (clxn/dxs)(:CD - xs)
and combining with the derivation of
the still-product composition relation
for total reflux:
XD
~:s = ---.----
an - (an - l)x])
(dXD/dx s) = (1/ an) [an - (an - l)xD)2
The result is:
S(dxddS) = (1/ a")xn(l - XD)
(an - 1) [an - (an - l)xn]
For Xn = 1/2 (midheigM of distillation Fig. 63. Uniform pole heights of biltch-
curve) : distillation curves (TV = 8).168
S(clXD/clS) = S ..ms = if = (a 21' - 1)/8an

The derivation and result are applicable only fOT total reflux, negligible
holdup, constant relative volatility, and the other usual simplifying assump-
tions.
For these circumstances the corresponding slope of the distillation curve
at Xp = liz becomes:
rns = a"/8(1 - :r.c)
if the approximation Sd = 1 - Xc is used. On this basis the number of
plates required for a given slope or pole height is:
n = log 8(1 - xc)m,,/log a = log 8o/log a
168 A. AND E. nosJo;
When the number of plates is very large and separation is controlled by re-
flux ratio, a derivation similar to the above gives:
rr = 1/2[(a - I)R v + al
and:
2cr - a 2(1 - xc)rns - a
Bv = --- = --------
a-I a-I
For very sharp separations these become:
20" 2(1 - xc)m,
R - ~- = --'----'---
D-
a - I a-I
and the quantity RD(a - 1) is thus a measure of the sharpness of separation
for the conditions of the derivation.
Bowman and Ciche11i recommend doubling the values of reflux ratio and
plates calculated by the above formulas when they are used to estimate the
conditions for a particular separation. The formulas are also useful in
estimating the lower reflux ratio required for a long column which is to
make the same separation as a short one at higher reflux. Laboratory dis-
tillations are often run at relatively high reflux ratios, but efficient columns
and low reflux are prefernble for large-scale operations. Another applica-
tion is the use of the formula:
n = log 80"/log a
to estimate the number of plates in a column from a distillation curve ob-
tained at high refiux. The equation:
S = (1 - Xc)/ [1 - - 1
(a - l)R D
]
can serve as a means of estimating the propel' reflux ratio for cases in which
refiux is controlling. The proper value is the minimum that will give sub-
stantially pure more volatile component as distillate. Higher reflux will
improve purity but little, while a decrease in reflux will decrease purity con-
siderably.
'1. Calculation of Theoretical Plates from Relative Volatilities and

Empirical Fractionating Factors
On the basis of actual experience with a variety of columns COOk 169 has
suggested a fractionating factor (an) of 1000 for ordinary distillations, and
of 10,000 for preeision distillation. The first of these figures is obtained on
the assumption that ordinary distillation will be slttisfactory if the distillate
16~ Cook, personLLI communication.
I. THEORY 169
is 99% pure when the Rtill r:ontnins 10% of thH most volatile component,
and 90% pure when the fitill composition in 1(;~J of the most voh,tile com-
ponent, Substitution of the:,;c valllPs ill the expression;
gives nppruximately WOO ill ('nth C'H.KP, For pl"t'eibi()u diHtillatioll the di~
tillate plll'itieH are ilWn'W;;(,d tu 99.9 a.ll(i 99.0%, !'PKpeeti\'eiy, and all is
ac-cordingly l'lLi:;ed by to facto!' of Ipl1. The theoreticfLI plntes needed for
TABLE XVIII
TUI';OHlG'l'ICAI. l'LH'Es PIW1f E.\IPLI!ICAr. FRAC'l'IONA'I'ING FAU'l'OIlS16"
Plates needed
Di1ft.~~enCl· in, J~o~lillg Ordinal')' Precision
pomts, :'7 C. di~tillutiou di"tillation
- - - - - - . -.. -~-.
3.00 30 5 8
2.00 20, Il 13
1.50 10 17 22
1.30 7 25 35
1,2fi 6 30 42
1.20 5 38 50
1.15 4 50 li5
1.12 3 70 80
2,,) 100
1.07 2 100 150+
various cases may then he obtl1inwl from '1'[111le XVIII. It, is of interest t.o
note that these l'eL1tiollS me apPl'oxinmteiy the S!1me us those predicted by
use of the equat.ions:
2.85 .
n = .--_ = RD and
log a
which are (liscm,sed in the succeeding paragraphs.
8. Correlation of Combined Effects of Relative Volatility, Reflux Ratio,

and Theoretical Piates 170
A convenient correlation hl1S been aehievE;d through the concept of a

"standard separation." It depends upon calculation of the plates and re-
flux required for separation of n pair of similar liquids with normal vapol'-
liquid equilibrillm relat.ions, WhCHl the uS\lal simplifying assumptions of dis-
tillation are justifieu and holdup is negligible.
17(1 Ro.le, Ind. ll'n!l. Clwm., 33, 5M. (1!J41).

170 A. AND E. ROSE
In balch fractionation it iR confusing to usc the difference between still

and product compositions as a measure of the effectiveness of the process.
This is because the compositions and differences in composition are chang-
ing continually throughout the distillation. The best procedure is to use
the shape of the actual Cllrve of product composition versus per cent dis-
tilled (8, XD curves) as the measure of effectiveness and thus relate the con-
ditions (0:, n, liD, etc.) of the distillation directly to the final results. This
has been done by choosing u standard separation, as follows:
Each eUl've of Figure 2 (p. 11) represents a different separation of two
componcnts. Kwh may be satisfactory for some purpose, and each separa-
tion requires differcnt eonditions. Thus for an ideal binu,ry equimolar
mixture with a: = 1.25 it can be calculated that ten plates are necessary to
obtain curve A, twenty plates for eurve B, and thirty plates for curve C,
provided t.he proper refiux ratio is used in eaeh ease and holdup is negli-
gible. Before attempting to devise any means for calculating the number
of plates required for a sepnr[Ltioll of the two components, it is essential to
choose one batch-fractionation curve as a standard of satisfactory Repara-
tion of the two components.
To make such a choice it has been assumed that for most practical pur-
poses a curve such as B of Figure 2 will be considered H sntisfactory separa-
tion. The first 40% distilled will have an average purity great.er than 95%.
This is therefore designated the standard separation; the following para-
graphs deal with the methods for caleulating approximately the number of
plates needed to achieve a separation of this standard sharpness. The
reasoning assumes charge compositions of 50 mole per cent, but is approx-
imately applicable to other charge eompositions.
The calculated values of relative volatility, reflux ratio, and equivalent
theoretical plates that will result in such a standard distillation curve have
been cOl'l'clated in Figure 3 (p. 12). The individual points for each curve
for different values of a: were obtained from graphs such as those of Figures
55 and 57. For instanee, Figure 57B shows that wit.h a: = 1.25 and RD =
49, about twenty plates are neeessary to obtain a, separation as sharp as the
standard (B of Fig. 2). This is expressed in Figure 3 by a point A on the
curve for a: = 1.25, A similar set of calculated curves for a: = 1.25 with
refiux ratio RD = 29 and a series of values of n shows that about 25 plates
are required in this case to give the desired separation. This is expressed as
point B in Figure 3. Other points on the curve for a: = 1.25 on Figure 3
were obtained by H similar procedure, as were the values needed to obtain
the curves shown for 0: = 1.05,1.1, and 2.
This correlation illdieates that it should not be necessary to fix the value
of n in order to achieve the desired standard sepa.ration, since any value
within a considerable range will be adequate. Within this range a lower
I, THEORY 171
valuc of n may hH eompnnsated hy a eOJ'l'csponding inCl'C'asc in RD , and

higher values of n may allow usc of t1 slUallc'l' R D , It is ossential for n to
remain within the desirable range since values of n that. arc too low cannot
be compensated by incrense in reflux and those that are too high do not
allow a corresponding docrease in reflux, To achieve the standrll'cl separa-
tion with the most effective use of the plates in n, column (and of the time
consumed) the reflux ratio should not be less than 2/3 nor mOTe than 3/2of
the number of plates.
The diagonal straight line:
RlJ = n = 2.S5/1og O!
was arrived at by inspection, and by analogy with the similar equation:
n = log aNlog a
in the following manner: It can he shown (see curve B, Fig. 61A) that the
standard separation of a normal mixture could be achieved at total reflux
by using a value of n large enough so that the over-all fra.ctionating factor,
an, is about 100. Thus the number of plates could be calculated by the
equation:
n = log lOO/log a
This is of limited usefulness because it applies only for total reflux or its
equivalent, but it seems reasonable to assume that the same form of equa-
tion might be used for calculation of plates required at finite reflux, i.e.,
that:
n = C/log a
.The diagonal line in Figure 3 with C = 2.S5 is the desired relation, The
line representing the lower limit of the desirable range of n is given approx-
imately by:
n = 2.3/log a
and the line representing the upper limit by:
n = 3.6/log a
The values of the constants were obtained entirely by inspection, that is,
it can be seen by examination of Figure 3 that use of a combination of re-
flux ratios and plates outside these limits involves either the use of too high
a reflux ratio or too many plates. Thus for O! = 1.25, the use of twenty
plates and reflux ratio of RD = SO gives substantially the same separation as
twenty plates and RD = 40,
172 A. AND E. nOSE
A. SEPATlA'L'ION OF CJJOSE-llOILING Aux'rUJn;s
The great diffieulty of slwrp separation of dose-boiling mixtures with

slmtlll'eln.tive vo\n.tility is lndlen.tecl by the eurvc for C/. = 1.05 in Figure 3.
Not only nre a very lnrge Humber of phtes required, but the minimum re-
flux mtios are nlso large. No doubt the timc required to reach equilibrium
becomes n faetor in such cnsf's. In Figure 64 are plotted the values of n
100 I
I W- _-
RANIJE OF PLATES _
I
140 "-
SUITABLE FOR
STANDARD S£PARA770N
/20 IJf'P£R CtJRV£ C' 3. /)
Cc,vT£R CL//i'//£ C' ?8S
LOWt:!? CtlRVe C'.?3
~/OO /},--k._
I-Oo.lOOC
!'l
~ 80
'1;;
~ 60 l\
40 ~
20 "~ ~ r-
}.O lJ t2 /.3 /,4 /.5 /.IJ /.1 /.8 /..9 ZtJ

VAPOR PRESSURE RATIO (cc)
Fig. 64. Calculated l'a,nges of theoretical phtes (and rcf!lL\: ratios)

suitable for stl1nrit1rd sepamtjoll nt different relative vol:Jtilities (n ""
C/log ",)yo
(or Rv) required to obtain the standard separation with various relative
volatilities. This indicates the exponential incrc!1R6 in the plates required
as relative volatility H,pproaches unity. The inclusion of the minimum and
maximum curVCR emphasizes the l'el1ttiYcly narrolV range over which nand
Rn may vary in order to achieve the desired separation efficiently, or at all.
B. CALCULA,!'ION OF PLA'l'F;S HEQUIRED FIWM DIFFERI'~NCE IN BOII,ING POINTS
A method for calculating plates required directly from boiling points

would obviously be of great. use. It must be renlized at the start that all
such methods arc subject to serious limitations and cannot be used indis-
criminately. At best t.hey can indicate only the general neighborhood of
the value of 13, bnt even this is so important that it seems desirable to point
out the basis for suoh caleulations. This basis lies in the combination of
the equation n = C/log a with Tronton's rule and the approximate Clapey~
r. 'l'HEOltY 173
run equation. TllU~ for n pair of liquids A and B (B is higher hoiling),

Troutol1'S rule states:
20.5(1'0 - '(J = h - },.t
in whieh I'll and 7'.1 nre the boiliug points at ab~olute tempC'wtnre, and Ao
and A,I the corresponding heats of vaporization, The appl'Oximate Clapey-
ron equation for Ow same two liquids is:
'All - 'A. t = RT In P.l - Ill' In PIl
in ,,'hich R is tlw constant in I) V = ,V NI' nne! P"l and ]111 are the ,-,apor pt·cs-
smos of the two liquids ~Lt '1' = ('1',1 +
'1'll )/2. The last equat.ion may be
written as approximately:
if TA and Tn are not widely different. Combining this with the Trouton
expression gives:
20.5(1'0 - '1'.1) = 2.3(7'B + T.{) log a
and this with n = Cflog a gives:
11 = 2.3C('l'B + T A )f20.5(1'B - TAl
which is of the form:
When the canst.nnt C in n = Cflog a has its optimuDl vI1lue 2.85, this be-
comes:
n = ('1.'Il +T .{ )/3(7'13 - T.)
.1.
At about room temperature (27°C.) the above equation becomes:
n = 200/I1T
n = 233/ ~7'
It is obvious th[Lt. the same equat.ions and the relation Rp = n = 2.85/log IX

may be used for cl1luulnting the required rejl'ux mtia. (Compare with page
22 for similar relation for a.)
General References
Badger and McCabe, Elements of Chemical Engineering. 2nd ed., McGraw-
Hill, New York, 1936, Chapter IX, "Distillation."
174 A. AND ]~. ROSE
Dodge, Chemical Engineering Thermodynamics. McGraw~Hi1l, New York, 1944,

Chapter XIII, "Distillation Processes."
Fenske, in Science of Petroleum. Vol. H, Oxford Univ. Press, London, 1938,
Section 25, "Laboratory and Blllall Scale DistilhLtion."
Klit, Studiel' over Fraklionel't Destillation. 1'llaning &; Appels, Copenhagen, 19;13.
With a SUmIllal'Y in English.
Perry, Chemical Enginecr8' Iia,ndbool". 3rd ed., McGraw-Hill, New York, 1950,
Section 9, iiDistillation and Sublimation."
Robinson and Gilliland, Elements of Fractional Distillation. 4th ed., McGraw-
Hill, New York, 1950.
Hose and Rose, Distillation Literature, Inde;); and Abstracts, 1941-1945. State
C()llege, Pa., 19'18 (bibliography).
Stage and Schultze, Theory, Ap]Jw'atu8, as Well as Procedures oj Distill(!tion and
Rectification. lIobart Publishing Co., WashIngton, D. C., 11)47 (bibliography,
1920-W44).
Underwood, Trans. lust. Chern. Engrs. London, 10, 112-158 (1932). (Historical
review as well as original derivations.)
Vilbrandt et al., Distillation Bililiography. Bull. Virgini(! Poly tech. Inst., Eng.
Ex})t. Sta. Se1'ies, No. 62, 194(l.
Walker, Lewis, McAdams, and Gilliland, Principles oj Chemical Engineering.
3rd ed., McGraw-Hill, New York, 1937, Chapter XVI, "Distillation."
Ward, U. S. Bw·. Mines, Tech. Pape?'s, 6'00 (1939). Review of the literature on
the construction, testing, and operation of laboratory fractionating columns.
Young, Distillation Principles and Processes. Macmillan, London, 1922.
CHAPTER II
ORDINARY FRACTIONAL DISTILLATION
Part 1. Appamtlls. A. I~. GLASEBROOK and F. E. WILLIAMS ... page 175

Pllrt 2. Procedure. F. E. WILLIAMS . . . . . . . , .. , . . . . . . . . . . . . . page 295
Hercules Powder" Company
Pa1't 1. APPARATUS
I. INTRODUCTION
A batch rectification unit consists of a still pot, rectifying section, and

still head as major components. The still pot holds the distilland and
transmits the heat required to boil the cha,rge and to supply vapor to the
bottom of the fractionating section. In the rectifying section, or column,
the upfiowing vapors and downflowing reflux liquid are intimately con-
tacted to obtain interaction leading to the exchange of heat and material
between the phases to produce fractionation. Since the actual separation
of the individual components making up the still-pot charge is effect.ed in
the rectifying section, it is the most important part of the entire unit.
Most of the increased efficiency of modern laboratory rectification equip-
ment has been achieved by improvement of the rectifying section.
The devices used for vapor-reflux contacting are of two general types:
the film type and the plate type. The film-type rectifying section can be
further subdivided into the simple and complex varieties. In the simple
or "wetted-surface" variety of film-type rectifying section the counter-
current vapor and liquid streams follow predetermined simple paths
through the column, and diffusional processes are relied upon for interac-
tion of the two phases. A vertical open tube is an elementary example of
this form of column. Numerous modifications exist in which devices are'
inserted in the tube to change the shape and length of the passageway or
modify the liquid-film surface, but in all cases the two phases follow pre-
scribed paths. The complex form of film-type rectifying column is gener-
ally called a ((packed" column and consists of a vertical tube filled with
solid material or "packing" to break up the direct path through the tube.
The packing material is usually of a regular size and shape and is packed at
random in the distilling tube. The vapor and reflux follow a tortuous,
175
170 A. L. GI,ASICnROOK AND F. E. WILLIAMS
unpredictable path through tho packed sodion, awl the two phases thor-
oughly interact bY' diffusion at the smface of the packing. Since packed
columns nrc easy to const.ruct and have good opemting; characteristics,
they are more widely used in the laboratory thnIl any other type.
A plate rectifying column c:onsists of a tube contnining a vertical series
of equl111y spaced horizontal plates which impede the downward flow of the
reflux. Liquid gathers on the upper surface of the plate until it reaehes [1,
predetermined level maintained by lLll overflow pipe. This pipe extends
into the liqnid 1nyo1' on the next lower plate and furnishes it with reflux.
Various methods of plate construction I1re employed to diHpersc the ascend-
ing vapor to allow it to bubble through and intemet, with the reflux liquid
on the plate.
Other considemtiollH being equal, the rectifying section showing the
greatest enrielunent per unit length is the most desirable. However, de-
pending upon the partieular fractionation problem involved, other fact on;
may have a more important bearing on the choice of the rectifying section.
The throughput or capacity of tho column is one of the fantors which de-
termines the time required to [:arry out the fraetionation. Plate and packed
columns generally have much higher throughputs than the simple wetted-
surface columns for a given fnwtionating efficiency. The operating holdup
is another important factor in the choice of a rectifying section. Gener-
ally, it is dmlirable to have the holdup as small as possible, particularly
when fmetionuting charges containing some constituents in small amount.
Simple wetted-surface columns have low holdup compared with plate or
packed columns. Pressure drop aCl'OSH the rectifying section is an impor-
tant consideration particularly in vaC',llum fraetionations. The pressure
drop of plate columns is comi(krably higher than thai; of the other i;ype~
for a given throughput.
The still head which is ahove the rectifying Imetion serves the dnal func-
tion of furnishing the eoilimn with reflux and of rlelivering; fractionated
product to the distillate receiver. As a consequenee of these functions,
the still head also eontrols the reflux ratios. The vapors from the eolumn
are either partially or totally condensed in the still-hend condenser. If a
partial condenser is used, all the condensed vapor is returned to the column
as reflux ancI" the uncondenHed portion paSRes to an auxiliary condenser.
If a total condenser is employed, t.he still head contains a partitioning de-
vice for withdrawing a portion of ihe condensate as product and returning
the remainder to the column as reflux. Partial condensers improve the
fractionation by effecting fllrther enrichment of the vapors; however, the
enrichment achieved is very slight and since the operation of a partial
condenser is difficult to control, most laboratory fractionating units em-
ploy total condensers. The proper functioning of the still pot, rectifying
II. ORIlINAHY I<'IL\C'l'IONAI, DISTILI,A'l'ION 177
section, and still head requires considerable auxiliary equipment. Many

well construetecl and potentially highly efficient fractionating units give
unexpectedly poor results in actual operation due to the inadequacy of the
auxiliary control equipment.
As shown in Chapter I, the rectifying section must operate under adia-
batic conditions in order to obt.ain the maximum amount of separation.
None of the latent heat of the vapors entering the rectifying section should
be used to eornpensate for heat lost to the surroundings through the column
walls. Neither must heat flow through the column walls and vaporize a
portion of the reflux. Vacuum jaekets or compensating heat jackets are
the method" generally used to obtain adiabatic conditions.
Since the column effiL:iency v:.tries with throughput, it is essential for
proper interpretat.ion of the distillation curves that the throughput be
maintained con~tant. To accomplish this, vapors must be generated in the
still pot at a steady, uniform, and controllable rate. If the still pot is
properly designcd and is readily responsive to chttnges in the heat supplied
by the still-pot. heater, a bade-pressure manometer can be eonnected be-
tween the still pot and condenser section to indicate and control the
throughput.
In order to follow the course of the fractionation it is customary to meas-
ure the temperature of the reflux. Sevel'lL1 types of t.emperature-measur-
ing devices are used [01' this purpose. The proper placing of the measuring
element in the still head to ensure accurate and reproducible readings is
the most important problem connected with the measurement of reflux
temperatures. Aceurate determination of reflux temperature, particu-
larly in vacuum fractionations, requires a constant pressure at the head of
the column. In addition to affecting the reflux temperature, variations
in pressure also affect the efficiency of the rectifying section. It is almost
imperative that the pressure be controlled automatically; a number of
manostat systems have been devised for this purpose. When the column
is operating under reduced pressure, it is essential that t,he fractionating
unit be equipped with a distillate receiver to permit withdra.wal of frac-
tions without disturbing the operation of the column.
Many precise fractionations require more than a month of eontinlious
operation for completion; therefore, in order to prevent excessive distilla-
tion losses, the number of joints must be kept to a minimum. In particu-
lar, ground-glass joints in contact with hot vapors are to be avoided. All
stopcocks should be precision ground and lubricated with a grease which
will not react with or be dissolved by the material being distilled.
A properly constructed fractionating unit combines all the elements
mentioned into a good workable unit. Construction should be such that
the assembly is mechanically strong, easily cleaned, versatile as to operat-
178 A. L. GLASEBROOK AND F. E. WILLIAMS
ing temperatures and pressures, and the construction material should be

corrosion resistant toward the materials to be fractionated.
In the laboratory, fractionating columns are used for a wide variety of
problems involving the purification or analysis of mat.erials. The equip-
ment requirements relative to theoretical plates, throughput, holdup, and
similar factors for optimum efficiency in these various applications may not
be identical. Furthermore, in cases in which laboratory-scale fractional
distillation is used extensively as a primary tool for analysis 01' sample prep-
aration, expensive intricate equipment is justified; whereas in cases in
which fractional distillation has infrequent application, only the simpler
and less expensive types of apparatus are justified. For these reasons, no
attempt will be made in the ensuing discussion to designat.e any rectifica-
tion assembly as the best; instead, the advantages and disadvantages of
the various designs in common use will be presented to aid the reader in the
selection of equipment for particular' rectification problems.
II. COLUMN SECTION

1. Types of Columns
A. WE'1''l'ED-SURFACE COLUMNS
In the wetted-surface variety of film-type column, the vapor and reflux:

streams follow prescribed paths through regularly shaped passageways,
generally of simple structure, such as an open tube, concentric tube, or
spiral tube. The downward flowing reflux covers the surface forming the
passageway with a thin film of liquid which interacts with the ascending
vapor stream to effect the separation of the constituents. The over-all
transfer process consists of three parts; mixing across the liquid film,
vapor-liquid transfer at the phltse interface, and radial mixing of the vapor.
Usually the first two processes are quite rapid, and the vapor diffusion
process controls the over-all rate of transfer. In most wetted~surface
columns the vapor flow is streamlined, and under these conditions the vapor
diffusion process is slow so that the throughput is. low when the column
is operating with reasonable fractionation efficiency. It is not unusual to
operate wetted~surface columns at, take-off rates as low as 0.1 m!. per hour;
at these rates the rectification efficiency may be very high.
Owing t,D the open, uncomplicated construction, the holdup and pree~
sure drop of wetted-surface columns is generally very much lower than can
be obtained in packed columns or plate columns of equivalent rectification
efficiency. In order to obtain the low throughputs required for efficient
operation, it is essential that the jacket surrounding the rectification section
be extremely efficient.
II. ORDINARY FIUCTlONAL DlSTlI,LNl'ION 179
The combination of high rectification efficiency and low holdup make

wetted-surface columns very attraetive for Lwalytical fraetiOl1tll;iollR, par-
ticularly those involving small quantities of materiaJ.
Empty Tube Columns. The simplest type of rectifying section con-
sists of a straight, vertical empty tube of circular cross section. This
design is amenable to mathematical treatment itS is shown in Chapter I
(Section IV). In spite of tllPir simplicity, empty-tube eolumns have not
been e)..'tenflively studied; however, the literature gives a fow referenees. l - J
Rosel studied I-ft. sectiom; of empty tubes, 0.:3 cm. and O.Li em. in diameter;
the results are shown in Table 1.
TABLE I
EMl"l'y-'!'uml COLUMN (30.3-(;:11. LONO) RlcCTIF[Cn'ION EF]o'[(~[illNCY AT 'l\.l'rAL Rli;FLliX
ColllInn diameter
0.3 em. 0.6 em.

n.E.T.p .. em.a
Through- Vapor II.E.T.P., em/' Throulo!:h- Vapor Vllcllum
put velocity, put velocity, jacket
liquiu 1 clTl./sec. Vacuum Air liquiJ. CIll./set'. Vncuum anJ
nll.jmill. (cak) jnd::ct j[l.ckct nll./min. (calc,) jacket heater
8.0 144 GO.Ei

3.00 216 30.3 23.8 6.2 112 60.6
5.0 00.0 60.6 30.8
2.00 1-14 15.1 20.2 2.5 45.0 80.3 30.3
1.60 11.5 7.6 13.8 1.0 34,2 1.5.1
1.30 03.6 10,1 7.(j 1.4 25.2 15.1
1.00 72 5.05 4.33 o.n 16.2 6.1 15. I
b
0.80 57.fi 4.33 0.8 1'1.4 4.33 7.6
0.56 40.3 3.ID 0.45 8.1 5.05
0.22 3.06 2.33
0.17 3.06 1.73
a Determined with benzene-carbon tetrachloride test mixture.
b No refiux at top of column.
As illustrated in Figure 1, Westhavcr 5 showed that calculated values of

H.E.T.P. were in good agreement with Rose's) experimental results for the
(Ui-cm. column insulated with a vacuum jacket and compensating heater.
Using only the vacuum jaeltet, the O.G-cm. column showed greater efficiency
than when the heater was also used. Since throughputs were measured
only at the base of the columns, the results can probably be eXDlained on
1 Rose, A., Ind. Eng. Chem., 28, 1210 (1936).
2 Craig, L. C., Ind. Eng. Chem., Anal. Ed., 9,441 (1937).
8 Bragg, L. B., Ind. Eng. Chern., Anal. Ed., 11,283 (1939).
4 Kuhn, W., and Ryffel, Helv. Chim. Acla, 26, ]693 (1943).
6 WesthaveI', Ind. Ena. Chern .• 34, 126 (1942).
180 A. L. GLASIWROOK AND F. J~. WILLIAMS
the assumption t,hat the vapor velocities in the upper sections of the un-
heated vacuum-jacketed column wcre less than ealeubtecl from the through-
put nt the base. This is borne out by the fact that Rose was unable to ob-
tain reflux in the unheated column tit very low throughputf:>. A comparison
of the results obtained in the 0.3- and O.G-cm. columns does not indicate a
greater €fficiency for the smaller column as predicted by the theory. Here
BENZENE + CCl4
0
60 (Open tuba, 6 mm. 1.0.)
50
E
L>
40
a.: -..... Yo' 0.3 em.
f-' 0, 0.050 cm~/sec.
tJ 30 0 0
:i
20
0 0
10
0
0
o 20 40 60 80 100 120 140

VAPOR VELOCITY, Va cm.lsec.
Fig. 1. H.E.T.P. valullH of ~IIl elllllty-tube COhUlU1,5 Expcri-

Hlcntlll YS. theoretical.
again the explanation probably lies in the insulation. In this regard Rose
says:l
"Operation of columns becomes more and more difficult as the rate of boiling be-
comes very low because the slightest variation in the insulation of the col"unn heat or
input to the pot destroys the equilibrium in the column. For successful operation
at very low ratcs of boiling thc columns should be insulated almost as well as a
calorimeter. Under such conditions efficiencies as high as 3D plates pel' foot seem
possible."
Kuhn and Ryffe1 6 investigated the rectification efficiency of 11- and

lOO-cm. open-tube columns 1.0 em. in diameter. The columns were vac-
uum jacketed and of similar design to that used by Rose. To further mini-
mize heat losses the columns were operated at 60 mm. of mercury. At this
pressure the carbon tetrachloride- benzene test mixture boiled at about
15°0. The experimental and calculated R.E.T.P. values given in Table
II show good agreement.
6 Kuhn, W., and R.yffe l , Helv. Chim. A.cta, 26, 1(lfJ3 (1943).
II. ORDINARY FRACTIONAL DIS'l'ILI,,\'l'ION 181
TABLE II
E~rr"ry-Tl;BI' COUT.\lN (l.O-CM. DJ.UlETlm) HECTIFWA'fION EFFJGIl':""CY
AT TO'r,\L IlEl'!.T1X
Column Inn~th
-----
11 CIIl. IOU em.
V"por H.E.T.P., em. Vapor Il.E.T,P" em.

velocity, velucity,
C1l1./SCl!. Dlm. Tlwor. rL em./see. Dba. T1Jcor. U
4 a 1.1 0.0:) 101 20 22

2.5 0.112 0.5G 31 7.1 G.D
----- __ __ _--------::
. ... _._._ .....
" Calcui:Ltuti from Westh,wur's equat.ioll using [), = 0.26 cm. /see.
24
2
5. !I .S 3
Unfortunately, experiments on empty-tube columns have not included

studies of the pressure drop and holdup. However, theoretical equations
are available which em.ble bot.h these factors to be est.imated wit.h good ac-
curacy. To calculate the pressure drop for an empty-tube column, West-
haver' developed the equation:
.6.p -;r + 1.2 X 10- TNIP] l

= [ 81]"Va 5 (1)
where .6.p is the total pre~tmre drop, p is the lwerage column pressure
(dynes/em. 2), 1]v is the vnpor coefficient of viscosity, Vais the rauially aver-
aged vapor velocity, M is the average molecular weight of vapor in the
tube, T is the lwerage column temperature (DK), and I is B.E.T.P. in
centimeters.
The holdup can be ealeubted from the equation relating throughput to
the thickness of the liquid film :6,7
(:3)
where w is thc thickneHs of the film, Q is the throughput, 7)1 is the viscosity
of the liquid, PI is the density of the liquid, and g is the accelemtion due to
gravity. For [1 tulle 3 in. in dit1metel', holdup values caleubted by equa-
tion (2) lmve been found to be in good agreement 'with holdup vnlues de-
termined experimentally.8
In Table III by means of equations (1) and (2), the pressure drop and
holdup pel' phLtc have heen calculated from Hose's data for the column
0.6 cm. in diameter. The calculations arc based on a 50:50 benzene-
carbon tetrachloride test mixture and est.imated values of 1.5 X 10-4 for
7 Friedman and Miller, C. 0., Ind. Eng. Chem., 33, 885 (1941).
B Willingham, Sedlak, Westhaver. and Rossini. Ind, Eng. Chem., 39,706 (1947).
182 A. L. GLASEDROOK AND F. E. WILLIAMS
TABLE III
EMP'rY-TuBE COJ.UAIN (0.6-eM. DIAMETER)PRESSURE DROP, HOLDUP,
AND EFFICIENCY FACTOR
'l'hroughput Pressura Liquid Efficiency

liquid, H.E-T.P., d)'op, mm, holdup, facto!',
ml./min. cm. Rg/plate ml./plate pl"te8/hr.
8.0 60.0 l,[) 1.0.5 460

6.2 60.6 1.8 0.96 390
5.0 30.3 0.94 0.45 670
2.5 30.3 0.\)2 0.3Ci 420
1.9 15.1 0..16 o.W 710
D.\) 15.1 0.46 0.13 430
0.8 7.6 0.23 0.061 7!l0
0.'15 5.05 0.15 0.031 880
0.22 2.33 0.060 0.012 1000
0.17 1.73 0.052 0.0083 n 1230
" At [1, low but unmeasured. tlu'oughput, Podbielniak (Table VI) fo~nd the holdup of
a lO1.6-cm. lOllg, O.54-em. diameter empty tube to be 0.374 m1., whieh compares to a
holdup of 0.0064 mI. for a tube 1.73 em. long. This agreement is close enough to indi-
cate that the cl.lculated values in this table are fairly accurate.
and 4.4 X 10- 3 for '1')1' Table III also includes the "efficiency factor"
'Y]"
defined by the relation:
ffi . f t throughput (liquid ml./hr.)
e Clency ac or = holdup (m1./theoretical plate)
In reciprocal form, this factor was first used by Bragg. 9 Podbielniak 1o

introduced its use in the present form. Since it is a measure of the number
of theOl'etical plates through which the material passes per unit time, the
efficiency factor is a valuable aid for evaluating and comparing columns.
The data presented in Table III show that empty~tube rectifying sec-
tions combine high fractionation efficiency with low holdup and low pres~
sure drop, provided the throughput is kept extremely low. Design theory
is available so that empty-tube columns can be built to meet specific frac-
tionation problems. The construction is very simple requiring only a
straight, smooth, empty tube mounted absolutely vertically so as to obtain
uniform wetting,
Chiefly because of their low throughputs, empty-tube rectifying sections
require the finest type of insulation, such as an extremely efficient vacuum
jacket combined with an auxiliary compensating heater. In addition,
columns of this type are unduly sensitive to very slight changes in the stilI-
pot heat input, and consequently are difficult to operate. Best still-pot
operation is obtained with a bare-wire immersion heater and vacuum mantle
u13rl1gg, Trans. Am. Ins!. Chem, EngTs' j 37, 19 (194:1).
Podbielniak, Ind. Eng. Chem., Anal. Ed., 13, 639 (1941).
10
U. OltDLNARY FRACTIONAL DISTILLATION l83
insulation. Owing to their operating difficulties empty-tube columns an)

not recommended for general use.
Spiral Columns. In the wire-eoil p[Lcked and distorted-wall columns,
the reflux path is lengthened but the vapor path is essentially the same
as it would be in an empty-tUbe column of equivalent length. In spirul-
\__ V
l_'-
",.
",.
-
~
"-
""'"
"'=
"---,-=
~/
I! HOLES 5 MM.
2 HOLES 4 MM. 01
~~~ ~1Q UIO OUTLET?i5 MM. 10.
Fig. 2. Spiral-type column,l6
type columns (Fig. 2), made by winding tubing in the shape of. an elon-
gated spiral, the vapor and liquid paths are equal but considerably longer
than they would be in an empty-tube column of the same height. This
type of column was first described by Warren l l in 1864, and in various modi-
fications is still very popular.
II Warren, C. M., Mem. Am. Acad., N. S. 9, 12111864); Ann., 4 (Suppl.), 5l (1865).

184 A. L. GLAS.EnJWOK AND F. E. WILLIAMS
Despite the numuer of descriptions of spiral eohlmnH in the literature, 12 -10

very felY data are available to nllow an aeeurate appraisal of their rectifica-
tion characteristics for comparison wit.h ot.her column types. The apparent,
advantage of the spiral tube compared with the straight tube is the great
saving in length whieh simplifies the construction and insulation problems.
As disadvantages, it would appear that the spiraling should retard the flow
of reflux and thus increase the holdup and decrease the maximum through-
put.
y oung17 states that the spiral columns have lower holdup than straight
empty-tube columns; however, other data indicate the opposite1 2 ; a 4-
m.long, 0.5-cm. I.D. tube, coiled to give a length of 101.6 cm., had a holdup
of 2.51 Ill!. at low tht'Oughput. The maximum throughput was 7.02 ill!.
per minute.. Although 0.5-cm. I.D. empt.y tubes were not investigated,
data for 0.38- and 0.54-cm. LD. empty tubes indicated holdups of 0.57
and 1.5 m!. and maximum throughputs of 7.01 and 24.0 ml. per minute,
respectively, for 4-m. lengths.
Shephel'd 16 used 3-m. lengths of 0.5- to a.G-cm. I.D. tubing to make a
spiral 40 cm. long. The spiral was encased in a vacuum jacket. At an
undisclosed throughput this column measured 30 theoretical plates. The
inside of the spiral was coated with 40-mesh carboruridum fused to the
glass. This roughening of the surface undoubtedly increases the rect.ifica-
tion efficiency. The spiral column has a low throughput and' requires very
efficient insula.'tion for satisfactory operation. Modern construction favors
vacuum jackets. 15 • 1S In winding; the spiral, cafe must be exercised to ob-
tain a uniform pitch.
An obvious construction modification of the spiral column consists in
creating a spiral passagew~.y in the annulus formed by two concentric
tubes. This method of construction is used in the well-known Widmer and
spiral-screen columns ~hown in Figures 3 and 4. Coiling a metal wire
around the inner tube is the easiest method for making this modified spiral
column. 18-22 Todd 22 wound a gO-cm. long, 0.15-cm. diameter MoneI" wire
center core with Gturns pel' inch of O.15-cm. diameter Moncl wire and in-
12 Podbielniuk, Ind. Eng. Chem., Anal. Ed., 5, 110 (1933).
13 Berlemont, J. Soc. Chern. Ind. London, 14,821 (1895).
14 Brown, F. D., J. Chem. Soc., 37, 59 (1880) i 39, 517 (1881).
Ii Doris,.;lnd. Eng. Chem., Anal. Ed., 1, 61 (1920).

16 Shepherd, J. Research Nail. Bur. Standards, 26, 227 (1941),
17 Young, S., DistiUalian Prin,:iples and Processes, London, 1922, p. 131.
18 McMillan, J.Inst. Petroleum Technol., 22, 616 (1936).
19 Fuucar, Brit, Pat. 19,999 (1908).
20 Dufton, J. Soc. Chem. indo London, 38, 45T (1919).
n Booth'and Bozarth, Ind. Eng. Chern., 29,470 (1937).
22 Todd, Ind. Eng. Chern., Anal. Ed., 17, 175 (1945).
185
serted the assembly ill [L n.li-ern. diameter glass tube. This fractionating
section had a maximum throughput of ~~.:) Illl. per minute. Operating ai,
a throughput of 1.(,5 ml. per minute, the fraotionating section had an
H.E.T.P. of 1.S em., a holdup of 0.4 mL per theoretieal plate, and an cf-
Fig. a. Widmer enlumn. Fig. 4. Spiml-scl'een uolulUn.
ficieney factor of 2475 plates per hour. Widmer 23 substituted a small-

diameter glass rod for the metal ·wire. Stranded wires have also been used
for the spiral. 24 Complete directions for fabricating the glass spiral for
the Widmer column are given by Fahlandt. 25 Midgl ey 26 improved the
construction by wrapping a metal wire wit,h a square strip of metal and
silver-soldering the spiral to the wire to eliminate the interstices and to
23 Widmer, Helv. Chim. Acta, 7, 59 (HI24).

24 Booth and lVlcMabrey, Ind. Eng. Chem., Anal. Ed., 16, 131 (1944).
25 Fahlandt, Chem. Analyst, 25, 28 (1936). '
26 Midgley, Ind. Eng. Chem., Anal. Ed., 1,86 (1929).
186 A. L. GLASEIlROO:K AND F. E. WILLIAMS
reduce the holdup. A refinement of Midgley's method is to cut a thread

in It metal rod.
A column described receatly was made by cutting a rectangular thread,
5 turns per inch, in a 2.5-cm. diameter stainless steel rod. The cut-out por-
tion of the thread was 11/82 in. deep and 3/32 in. wide. The thread was
ground to fit closely the inner tube of a vacuum iacket. 27 The rectifica-
tion characteristics of this column, as determined in a 320-cm. long sec-
TABLE IV
RECTIFICATION Cn,illAC'fERlS1'IeS OF S~roOTH-SPIRAL COLUMN"
Pressure drop,
Throughput mIll. Hg per Efficiency
liquid, H.E.T.P., theoretical Holdup, C factor,
mi./min. cm.b plate ml./plate plates/hr.
1.42 2.64 0.012 0.33 260

1.83 2.77 0.017 0.35 320
3.1 2.95 0.029 0.37 500
3.8 3.12 0.039 0.39 590
5.3 3.35 0.063 0.42 755
7.0 3.32 0.092 0.42 1000
7.9 3.35 0.11 0.42 1160
8.6 3.40 0.12 0.43 1220
9.2 3.58 0.14 0.45 1230
10.0 3.88 0.17 0.49 1230
. a 2.5-cm. diameter tube fitted with a 5-turns-per-inch spiral, 11/32 in. deep and a/!2
in. wide.
b Determined with n-heptane-methylcyclohcxane test mixture.
, At medium throughput the holdup was determined to be 35--40 ml. Holdup pel'
plate was calculated on assumption that totl1l holdup was constant at 40 ml.
tion, are given in Table IV and plotted in Figure 5. Inspection of Figure

5 shows that the rectification efficiency is constant over a considerable
range of throughputs, This characteristic is a very desirable one since it
leads to stable column operation and smooth, easily interpreted distillation
curves, An actual distillation curve obtained with this column is given
in Figure 6.
Increasing the pitch and enlarging the cross-sectional area of the spiral
opening reduces the pressure drop and makes the column suitable for oper-
ation at very low pressures. Williams 28 carried out fractionations at head
pressures as low as 1 mm. of mercury in a 120-cm. long, 4-cm. diameter
tube fitted with a 2-turns-per~inch aluminum spiral with an 'opening 1.4
em. deep and 1.05 em. wide. The rectification efficiency of the spiml col-
umn is increased by coating the surface of the spiral passage'way with fine-
27 Smith, GJasebrook, Begeman, and Lovell,lnd. Eng. Chem., Anal. Ed 17,47 (1945).
28 Williams, Ind. Eng. ahem" 39, 779 (1947).
II. ORDINAHY I1RAC'rIONAL DISTILLA'l'ION 187
mesh c!1l'bol'undum. 26 This modification also would be expected to in-

crease the holdup.
Lecky and Ewe1l 29 used a metal-gauze spiral and obtained increased ef-
1:30r-------------------------,1.6
.:
1.4 .-.
n:
o f-'
If) W
~ 110 1.2 :x:
<t
...J
0..
-.J
;3 100 1.0
i=
lU
a::
o
lU 90
~
80
20 40 60 80 100 120 140 160 180

PRESSURE DROP, mm. dibutyl phthalate
100 200 300 400 500 600

APPROXIMATE THROUGHPUT, cc.lhr.
Fig. 5. Performance of [L smooth-spiri1.1 column. 21 n-Heptane-mcthylcyclohcxane
test mixture. (GIl) TlwQl'eticttl phtes. (0) H.E.T.P.
70
;i cJ
w·
0::
::>
tra~. 1.4100 '2
\i0:: 65 c·c'/:·c·c !Ii
)(
W
a.. ILl
:::; a
w 60
I- 104000 ~
Z
0 TEMP. >
~
-'
~
cr:
-'
t; 1.3900 f::1
cr:
Ci
1.392.0
o 100 200 300 400 500 600 700 800 900 1000 1100
DISTILLATE, co.
Fig. 6. Fractionation of IL mIxture of single"branched hoxenes in a smooth-

spira,} Ciolumn. 21
29 Leeky and Ewell, Incl. Eng. Chem., A nal. Ed., 12, 544 (1940).
188 A. L. GLASEBIWOK AND P. E. WILLIAMS
ficiencies due to t.he roughened sUl·faee. They also introduced the further
refinement of cupping the :-;piml :-;0 thai; the edges were turned up at an
angle of about L15°. ny making the diameter of the spiral slightly larger
than that of the tube, this construction ensured a clORe fit when the spiral
was inserted in the tube. In addition, the cupping caused the reflux to
run down and over the gauze at the points of eontact with the retaining
'fABLE va
RlWTlI'ICATION CUARACTlmIS'l'ICS 0[" SPIRAL-SCwmN PACKINGS
Thr011'p;hput Uqllirl Efficiency

liClUid, H.8.T.P. holdllP, factor,
nl1./min. em,l; ml./plnte plates/hr.
Cupped spiral 17 ill. long, outside diameter 0.75 in., root dill!l1eter 0.25 ill., 67 turns pel'
..'. foot e
0.25 1.45 0.23 65
0.83 1.62 0.24 208
1.07 1.68 0.26 385
3.33 1. 72 0.30 6G7
5.00 1. 72 0.35 800
6.67 l.73 0.42 950
8.33 1. 77 0.52 960
10.00 l.88 0.68 880
11.67 2.10 1.0 700
12.5 2.20 1.44 520
Cupped spiml 17 in. long, outside diameter 0.5 in., root dbmeter 0.125 in., 67 turns per
foote
0.42 1.71 0.14 180
1.67 1. 71 0.16 6215
3.33 1. 71 0.18 1110
5.00 1. 74 0.26 1150
5.83 1. 78 0.34 1030
II. 67 1.8(j 0.,50 800
----------------------------_._--_
20
_... -_. ..
a Data taken from the graphs of Lecky tlnd Ewell.
b Determined with n-heptane-methylcyclohexane test mixtul'e.
o Pressure drop approximately 1 mm. of mercury per foot.
tube rather than through it. The rectification characteristics of two sizes
of spiral-screen packing arc given in Table V and presented gmphically in
Figure 7. The spiral-scrccn column resembles the smooth-spiral column
(Fig. 5) in that the H.E.T.P. remn.ins constant over a considel'able range of
throughputs, a condition conducive to smooth operation. However, the
gauze spiral is considerably more effective than the smooth spirn.I. In the
portion of the curves where the efficiencies are const.ant the gauze spiral
has one theoretical plate for every 3 314 turns, whereas the smooth spiral
requires 61/2 turns pel' theoretical plate. The efficiency of the gauze spiral
was reduced. when the holes were filled with collodion. 29 The data showed
n. ORDINARY l~RAC:'l'ION,\L llIS'l'ILLA'l'ION 189
that the throughput und H.E.T.P. nrc proportional to the spacing between
the turns, but that the !lumber of turns per theoretic-al plate and the holdup
per theoretical plate are independent of the spac·iIlg between tnl'lls. In
If!<
I6
\
~ 18
>-
'--
XO
o )(
l
l,L'
« L':::::
0" Z
,-I
0..
,-116 L 0
«
~
/ ~l
>- l,..-"" "\
- fe=tt-r,
uJ ~ ...........
:5 14
u.J
::t:
.. ')0
x
>--
12
100 200 300 40J 500 600 700
REFLUX RAiE ML. PER HOUR
(a) 0.75 in. diameter packing
2 I. 6
w
5 ~
e Ii
0::1 8" I. 2 0::
w u
a.. --~ w
0..
~ ~
<fl
w
1;:;:1 6 8
_J
CL.
;i
!dl 4 L o.
tJ ..-' l..--"j
§
:r:
1-1 2-r- 100 200 300 400 500
REFLUX RATE ML. PE. R HOUR
(b) 0.5 in. diameter packing
Fig. 7. Performallee of sel'l'en-spiral packing:!9 (0) effiniencies for 17-in. long
seetiou; (X) etlieicllcies for two 17-iu, long sections ill serills; (.) holdup data
determined for 17-in. long section.
contrast to many types of packed columns, the H.E.T.P. did not iuerease
when the colmnn was lengthened.
The Lecky-Ewell columns were made with GO- by GO- or 80- by 80-mesh
nickel gauze. For larger-diameter spiral-screen columns (3.7 em.), 50- by
50-mesh gauze gave better results. 30
80 Stallcup, Fuguitt, and Hawkins, Ind. Eng. Chcnn., Anal. Ed., 14, 503 (1942),
100 A. L. GLASEBHOOK AND Ii'. E. WILLIAMS
The construction of the gauze spiral column is more difficult than other
types of spiral columns. Dies were used to stamp split cupped washers
from the gauze. These were then spot-welded
Spacing from together to form a long continuous spiral. A
inner wall glass or metal rod was used to fill the center hole
Wire coil and provide support. In order to insert the
fractionating fragile spiral into the outer tube, a solid spiral
tube was screwed into the gauze spiral, the entire as-
sembly inserted in the tube, and the solid spiral
Fig. 8. Single wire-coil removed by unscrewing.
rectifying section. Wire-Coil Packed Columns. The empty-tube
TABLE VP2
COMPARISON m' Howup AND MAXIMUM THROUGHPUTS OF EMPTy-TuBl" COLUMNS AND
WIRE-COIL PACKED COLUMNS
LiJUid
Tube hal up,u Maximum
diam., m!. per throughput
em. I01.0-crn. liquid,
(inside) Paoking length ml./min.
0.26 Empty 0.113 0

0.26 Single-wire coil No. 26 B & S 0.19 2.75
7.9 turns/in.
0.3 Empty 0.094 3.5
5.2 turns/in.
0.38 Empty 0.145 7.01
7 turns/in.
0.38 Double-wire coil No. 20 B & S 0.695 3.1'1
outer coil 7 turns/in.
iuncr coil 4 turns/in.
0.64 Empty 0.374 24.0
5.5 tUl'ns/in.
0.64 Double-wire coil No. 18 B & S 1.60 10.8
outer coil 5.5 turns/in.
inner coil 4.5 turns/in.
0.78 Empty 0.518 36.3
0.78 Single-wire coil No. 18 B & S 1. 68 26.5
4.5 turns/in.
0.78 Double-wire coil No. 18 B & S 2.14: 21.2
outer coil 4.5 tUl'lls/in.
inner coil 3.5 tlll'ns/in.
" Determined at a low but unmeasured throughput.

u. ORDIN.mY FRAC'frONAL DISTILLA'fION 191
column has been modified by inserting a continuous spiral coil of wire. 31-38
The coil fits the tube snugly, has a pitch of 5 to 8 turns per inch, and is
made from wire with a diameter much smaller than that of the tube (Fig. 8).
Columns using "'ire-coil rectifying sections have heen lIsed widely for
low-temperature fractionations.
Aside from t.he effect of the reduction in cross-sectional are:1, the wire
coil offers but little resistance to the flo\\' of the vapor, and the vapor flow
TABLE VIr
Ih;c'r!J.i'ICATIO:-i CHAR,\CTERIS'rlCS 01' SINal.!!] WIRE-COIL PACKED COLUMNS
Pressure
droPr
Throllghllut Liquid Inm. Hg per Efficiency
liquid, H.E.T.P., holdup, theoretical factor,
ml./min. em. ml./plate plate Jllates/hl'.
0.3-em. LD. tube, wire coil 11 turns/in. No. 26 wirc 3 !

O.flO 3.G
0.70 ,'>.1
1.00 6.1
1.60 5.1
2.00 5.1
2.00 3.8
O.4-cm. LD. tube, wire eoil 6-7 turns/in. No. 20 wire 33
0.80 10.7 0.17 0.02 280
1.6 10.0 0.19 0.05 510
2.7 0.4 0.29 0.10 560
D.G-em. J.D. tube, wire coil 11 turns/in. No. 26 \Virc 3 !
1 4.7
5 7.6
is very similar to that in an empty-tube column. The reflux, however,

bridges the interstices between the wire coil and the tube n'ith capillary
films, which greatly increases the liquid surface exposed to the ascending
vapor. Compared with the corresponding empty-tube colum.n, the added
film surface caused by the presence of the wire eoil increases the holdup
(Table VI), and at high throughputs appears 1;0 increase the fractionation
n Rose, A., Ind. Eng. Chem., 28, 1210 (1936).
32 Podbielniak, Ind. Eng. Chem., Anal. Ed., 5, 119 (1933),
33 Whitmore el al., J. Am. Chem. Soc., 62, 795 (1940).
R4 Mair, SchicktlLnZ, and Rose, F. W., J. Research Nail, Bur. Standards, 15, 557
(1935).
36 ObCI'fell, Oil Gas J., 27, 142 (1928).
36 Podbielniak, Ind. Eng. Chem., Anal. Ed., 3, 177 (1931).
37 Podbielniak, Ind. Ena. Chern., .1nal. Ed., 5, 135, 172 (1933).
38 McMillan, J.lnst. Pelro18'lun Technol., 22, 616 (1936).
192 A. L. GLASEBROOK AND F. E. 'WILLIAMS
efficiency. As shown in Table VI, in the case of small-diameter tubes, the

wire coil retards the tendency of the reflux tll bridge the tube and flood;
the coil promotes draining and increases the maximum throughput. In
x
I
30
28
26
24
22
20
18
E
u
a;
. 16
x
I-'
LLi 14
:C
12
10
8
"
6 / A,\
" If r:,. A.
4 .j I \
><' r:,.
I
2
0 2 3 4 5 6 0 2 3 4 5
THROUGHPUT, ml./min
Fig. 9, Comparison of performallce of empty-tube and single wire.coil

packed columns: (0) O.G-cm. tube with single wire-coil packing; (IJ) O.6-Clll.
empty tube; (b.) O.3-cm. tube with Ringle wire-coil paekillg; (Xl O.3-cm. empty
tuhe.
the larger-?iameter tubes, the maximum throughput of the tube is greater

when the wire coils are absent.
The available data on the fractionating characteristics of wire-coil
packed columns arc given in Table VII, and in Figure 9 a porl,ion of the
data are compared with corresponding data for empty-tube columns
(1'able I).
Figure 9 shows t.hat at low throughputs t~e rectification efficiency of wire-
II. ORDINAHY FIL\C'I'IONAL DISTILL-\'rlON HJ3
coil packed columns is approximately the same or slightly less than empty-
tube columns. At high throughputs where the reetihcation cfIleieney of
open-tube columns is vel'Y low, the added film surface Ctlll~(erl by the wire
spiral results in improved vapor-liquid contacting compared with the
empty tube; consequently, the H.E.T.P. values for wire-coil packed
columns do not. vary with throughput to as great an extmlt as open-tube
colulllns, and throughput control is lef>:-l oritical. Due chieHy to the holdup,
t,he efficiency factor fol' wire-eoil packed column::; is not ::;0 high as for cmpty-
tube eolumns. Poclbielniak 32 has investigated double-wire coils. The
high holdup of tlii::; packing (Tuble VI) mnkes it unattractive. The gelwmI
construction and operation of a wire-coil packed column is similar to that-
of an llmpty-tnhe column. The wire coil is made by expanding a cl(]~e
wound coil made on a mandrel.
Distorted-Wall Columns. Many of the earlier investigators of recti-
fication columns sought t.o improve the separation efficiency by distorting
the wall of an empty-tube column. Such modifications increase the
liquid film surf:U'c in mueh the same manner as discussed under wire-coil
packed columns. Compared with the empty-tube column, the distorted-
wall columns may have advantages ut high throughputs or in cases in
which the column insulation is poor. Unfortunately, analytical infor-
mation on the rectification elu1l'acteristies of this class of columns is meager,
but in general it is believed distorted-wall columns are inferior to other
types of wetted-surfaee columns, and [my uclvant.ages they may have can
be secured more easily by other means,
The most popular column of the distorted-wall v[uiety is the Vigreux
eolumn. 3~ This column is made by heating a spot on a glass tube and then
indenting toward the tube axis with a sharp-pointed instrument such as t1
file or wire. The tube is then rotated 180 0 and a similar indentation Illude.
The indentations usually nre made so that they will have a slight dowllward
slope when the tube is vertical. The cleamnce bet\veen the tips is usually
equivalent to about one-fourth to one-third of the tube diameter. The
flame is then moved up the tube Ii few centimeters and two more indenta-
tions are made at right angles to the first two. The operation is repeated
until the entire length of the rectifying sectiOll has heen indented.
Seveml vucuum-jacketed Vigreux columns have been described which
give good separation for muterials with a 100e. difference in boiling point
and have holdups of 0.4 ml. or 1ess. 40 --12 Fenske and co-workers 43 report
an H.E.T.P. of 13.7 em. for [t column 1.1 em. in diameter.
,~ Vigrcux, Bull. soc. chirn. F1'aIH:C, 31, 1116 (100'10).
~B Shrader and Ritzel', Ind. Enq. Chern., Linal. Ed., 11, 54 (1939).
41 Weston, Ind. EII(J. Chern., Annl. Ed., 5, 179 (1933).
42 Oooper, O. M., and Fa,scc, Ind. Enq. Chent., 20,420 (1\)28).
The reader is referred to Y oung 44 for descriptions of other types of dis-

torted-wall columns. Hill and Ferris 46 have compared the rectification
performance of some of these columns under talce-off conditions.
Concentric-Tube Columns. In concentric-tube rectifying sections,
vapor-liquid contacting takes place in the annular space formed by two
concent.ric tubes. The formula developed for predicting the ideal per-
formance of parallel-plate columns 46 .'17 is applicable to the concentric-tube
column, and has been discussed in Chapter I (Sec. IV). The formula shows
that rectification efficiency of concentric-tube columns is improved by de-
creasing the vapor velocity, decreasing the thickness of the annular space,
or increasing the diffusion coefficient of the vapor. In these respects the
concentric-tube column resembles the empty-tube column.' However, in
an empty-tube column operating at a given vapor velocity, the diameter
cannot be changed without affecting the throughput. In a concentric-tube
column, on the other hand, the throughput can be varied independently of
the vapor velocity and the distance between the tubes by changing the
mean cir(~umfel'encc of the annular space. In other words, in a concentric-
tube column, the number of theoretical plates per unit length depends on
the difference in the diameters of the tubes forming the annular space and
is independent of their absolute magnitude.
Concentric-tube columns were used as early as 1909 by Rosanoff and
co-workers,48 but their development lagged until 1937 when Craig 49 showed
that columns of this construction had excellent rectification characteristics.
Since then concentric-tube columns have been the subject of several ex-
haustive investigations,r,o-54 which demonstrated that H.E.T.P. values
well below 1 cm. can be achieved with this type of construction.
The lowest H.E.T.P. values reported for concentric-tube columns were
obtained by Naragon and Lewis. 63 The fractionating section was insulated
with a vacuum mantle and pipe lagging. Construction is shown in Figure
10. The results of the efficiency tests are shown in Table VIII.
The pressure drop of the concentric-tube column described in Table VIn
43 Fenske, Tonberg and Quiggle, Ind. Eng. Chem., 26, 1169 (1934).
H Young, S., Distillation Principles and Processes. London, 1922, pp. 130-146.
45 Hill, J. B., and Ferris, Ind. Eng. Chem., 19,379 (1927).
46 Kuhn, W., Helv. Chim. Acta, 25, 252 (1942).
41 Westhaver, Ind. Eng. Chem., 34, 126 (1942).
48 Rosanoff, Lamb, A. B., and Breithut, J. Am. Chem. Soc., 31,448 (1909).
40 Craig, L. C., Ind, Eny. Chem., Anal. Ed., 9, 441 (1937).
50 Selker, Burk, and Lankelrna, Ind. Eny. Chem., Anal. Ed., 12,352 (1940).
61 Naragon, Burk, am! LankeJma, 100. Eng. Chem., 34, 355 (1942).
62 Hall, S. A., and Palkin, Ind. Eng. Chem., Anal. Ed., 14, 807 (1942).
53 Naragon and Lewis, Ind. Eng. Chem., Anal. Ed., 18,448 (1946).
64 Donnell and Kennedy, Proc. Am. Petroleum Inst., III, 26, 23 (1946).
II. OnDIN.\ltY F R.\C'l'IONAL DIS'l'II_,L,\'l'IOK 195
was not determinetl; however, on t,lw basi:> of other cittta it can be esti-
mated that the pressure drop was prohably less than n.F) mill. mercury.
TROUGHS 1 MM. 0.0.

TUBING CUT IN
HALF WITH I X3
MM.5LO;:..:T..:::S_ _--\\:""-"'I:T"1I
SECTION A-A
~-l-,I-G"'O""N.DENSER
GROUND GLASS
___-+----I--I-SEAT ON TAKE-
OFF VALVE
10 MM. O. D. TUBING
WITH 45' CUT
\\""~_ _ _ _--rrAN",D,,- DRIP. TIP
~
SCALE 30 MM.
Fig. 10. Concentric-tube fractionating column. 63
The observed and calculated H.E.T.P. values are in fairly good agree-
ment. The throughputs were measured at the base of the column;
196 fl. L. 'GLASEBHOOK AND F. E. WILLIAMS
TABLE VIII
REC1'IFICA'l'ION EI"FICmNT~Y A'I' T01'AI, H,wr,ux OF CONCICN'I'ItW-TUBIG COLIlMN"
Throughput H.l~.1'.P" CIll. EfliciUJICY

liqnid, factor,"
ml./lIlin. ObscI'vedb Theoretical C plltto./hr.
1. 28 O,af) O,·Hi 4800

1. 53 OAO 0,55 '1600
1. 78 0,49 0,63 4500
2,06 0,58 0.72 4300
" 30.5-em. long, O.S-em. LD. tube, O.04-cm. O.D. tube.
b Determined with n-heptaue- methy!cye!ohexlwe test mixt.ure.
C Calculated by the equation: l' = (17/35)(VaS 2 /Dv)(DvVa) using Dv = 0.04 CIIl.2/
sec. See Chapter I, section IV.
a Based on an estimated holdup of 1.5 m!.
at throughputs lower than 77 ml. per hour no reflux was observed at the
top of t.he column. Hence the l1verage throughput was probably less than
that observed. This fact may explain why the observed H.E.T.P. values
were somewhat less than thOHe predicted. Nevertheless the agreement be-
tween the calculated and observed H.E.T.P, values is sufficiently good to
justify the use of equation (5) in design work.
The results of other investigations of concentric-tube columns (two-tube
type) are snmmarized in Table IX.
TABLE IX
COMPARISON OF CONCENTRIC-TUBE COLUMNS
Outside
diam. Annular,
of space Column Through- Liquid
inner tbickness, length, put, holdnp, H.E.T.P.,
tube. cm. CIll. C1n. ml./min. ml./plate Cl11. Rei.
0.65 0.075 30,5 1.20-2,06 0,018-0,028 0.36-0,58 53

0.3 0.1 31.5 0,5 5l
1,255 0,07.'5 100 0, (j7·-l.(i7 0,022-0.081 0.7-0,g 54
0.4 0.15 10 0.015 1.3 49
l.0 0.15 238 1. 25-5.0 0.1-0.4 2.2-5.6 52
Figure 11 shows the results obtained in the fractionation of a known mix-

ture (boiling-point spread 2.3 ac.) in a eoncentric-tube column. The sepa-
ration is excellent, and the consistency of the data points shows that the
column gives smooth, st.able operation. However, the distillation curve
represents ()O hours of still operation during which time only 9.8 ml. of
overhead product. was obtained.
The construct.ion of a eoncent,ric-tube fractionating section of high ef-
ficiency is somewhat diffieult. and requires more precision work t.han many
other types of columns. To meet the requirement of a uniform annular
II. ORDINARY FRACTWNAL DIS'!'ILId~'I'ION 197
clearance, the tubr:s llsed in the constructioll mllst be straight and per-
feetly rounel, and tlwy must he assemhled concentrically and held in that
position. In order to have a thin even now of reflux over the tubes, the
surfaces mnst he smooth and uniform, and the reetifying section mounting
must not deviate from tlw verLicnL In addition, concentric-tube columns
require a reflux distributor to maintain even distribution of the reflux over
both tube surfaces.
Most designers of eoncent.rie-tube columns agree tha,t precision-Lore
tubing is more satisfactory for the outer tube. Ordinary tubing, care-
fully selected for straightn~~ss, roundness, and uniform diameter, is gen-
erally acceptable for the inner tube. Precision-bore tubing is not always
\oN
CHARG1e: 5.1 M~T.rI-HEPTANE 1
" ..
~
9 3~.O M~. METHYLGVGLOHEXANE_
eo ~ ~~~~EU6F~A~;i ..6.a g~~~~/~~~.
7 DISTI~LATE-/ ~
~
'"
fQ_
5
40
\
a
[\
20
'u
t:f:tt_.. 74.8%
lob OT ~tIXTU E
'"V
PO< Ion. l~~ fa: ~~;-
0
o 2 4 6 Il 10 12 14 16 18 20 22 24 26
LIQUID VOLUME PER CENT DISTILLED
Fig. 11. Concl.lnttic-tube fractionating column distillation of known mixture. 5'
straight, and this feature must be checked. Also, precision-bore tubing

is not ordinarily available in lengths of more thn,n 3 ft., which auto-
maticn,lly limits the length of the rectifying section. 1'0 compensate for
bowing of the outer tube, Hall n,ncl Palkin62 use a series of short interlocked
tubes for the center tube. In addition, these investigators wind the center
tube with a fine glass thread to obtain uniform reflux distribution. Both
these devices appear to increase the holdup (see Table V). With the ex-
ception of a column made by Kuhn,46 metal tubes have not been used in
the construct.ion of concentric-tube columns. However, it appears that
metal tubes should eliminate many of the difficulties encountered when glass
tubes are used.
Spacing lugs, located above and below the rectifying section, are used to
separate the inner and outer tubes. The length of the lugs must be care-
fully controlled, usually by grinding. Naragon and Lewis53 use a metal-

spiral spacer in the ann1l1:1i· space of the rectifying section for preliminary
spacing of the tubes. The tubes are then fused together through glass
spacers ll,bove and below the rec.tifying section, and the metal spiral is dis-
solved in aeid. (Detailed directions for the construction of eoncentric-
tube columns are given by Naragon and Lewis. 53 )
Kuhn 46 solves the difficult problem of reflux distribution by forming re-
flux OIl both tube surfaces. This is accomplished by extending both tubes
into a jackelied space through which the condenser liquid flows. The in-
ner and outer tubes are joined by a Dewar seal. Naragon and Lewis53
direct all the reflux liquid to the top of a sphere. Troughs attached to the
sphere at the circumference direct the reflux to the tube surfaces. Since
the distribution of reflux will be uniform around the circumfcrenee, the
width of the trough automatically determines the amount of reflux going
to each tube surface,
Sclleer, Burk, and Lankelmuno constructed a multiple concentric-tube
column. This type of construction has a much greatel; throughput than
the two-tube construetion, and the effects of heat losses are minimized.
Usually, the difficulty of proportioning reflux liquid more than offsets the
advantages of this type of construction. However, the spherieal reflux-
distributing device descdbed by Naragon and Lewis 63 may make it possible
to construct highly efficient multiple-tube columns.
Like all low-throughput colnmns, the concentrie-tube column requires
very efficient insulation. Most investigators have used a vacuum jacket
supplemented by lagging or by an auxiliary heater. The still pot should be
designed to secure even boiling; heat input must be stable. This can be
secured by llsing constant voltage in the heater and enclosing the still pot
in a vacuum jacket.
Owing to their high rectification efficiency and relative ease of operation,
concentric-tube columns are recommended for all applications requiring
a film-type column.
Packed Columns. (a) Dllmp Packings. In packed columns, the
rectifying section consists of a tube filled with an inert material or (' pack-
ing" in the form of small regularly shaped particles, or irregular-shaped
pieces turned or chipped from larger masses. PackingR consisting of small
individual particles are called" dump packings," from the mode of in-
troducing them into a column. The packing may be nn integral unit
packed at random in the column in the case of chain packing, or may be a
preformed close-fitting, rigid structure as in the case of Heli-Grid packing.
The downflowing reflux wets the packing surface to provide a large area
of liquid film for contaeting the ascending vapor stream. Both packed
and wetted-surface columns are film-type columns, and the vapor-liquid
n. ORDINARY FRACTIONAL DISTIT.JLATION 199
proeesses are similar. In the wmal weUed-surfnce eolnmn, h(JI\'ever, the

laminar flow of the vapor redneefl t.}18 rate of eXl'.hltllge between vapor and
liquid nec(_~ssitates 1/cry low vapor rates to seeure good reetitkation. In
packed columns the packing c:mlses frequent splitting of the vapor stream
and produces a more t.urbulent tim". The turbulence in('l"cases thc ntJ10r
diffusion coefficient and consequently permits higher yapor rates. or
throughputs.
The vapor and liquid streams in packed columns are not, eouflllcd to a
definite path hut follow tl haphazard (]OUl"se dictated by chance irregularities
in the paeking. VariatiOIls in the size or density of the packing are likely
to cause loss of metification efficiency due to channeling of the vapors or
loeaJized flooding. TheRo effeets are more pronounced in large-diameter
columns and are minimized when the packing is of uniform size, symmetri-
emIly shaped, and the ratio of column diameter to the diameter of the jn-
dividual packing units is greater than 8 to 1. In packed columns reflux
collects and is retained at the points of contact between the packing units
so that the holdup of packe(l columns is generally gl'enter than in wetted-
surface columns.
A good packing is one that can be packed uniformly and has the proper
balance between slLl'faee area [md free space to ensure efficient vapor-
liquid contacting and high throughputs without excessive holdup and pres-
sure drop. The rectificat.ion efficiency of most packings is increased if the
packing is thoroughly wet hy flooding at the start of the distillation. This
treatment. removes all ent.rapped ail' from the pll.eking interstices and. causes
the reflux to spread uniformly over the pncking surfnee.
'Packed c'.olumns combine good rectification efIiciency with high through-
put and usually the holdup and pressure dmp are not excessive. These
characteristics combined with CflHe of constTllCtion and. operation make
paelmd columns ideal for generallabomtol'Y use.
H dices. Of the numerous packings in many different shapes, and of
many materials of construction tlmt have been used in l:1boratory frac-
tionating columns, single-turn metaP5 and glass helices56 have had widest
appliclH,tion. Difficult problems of research in the pet.roleum industry
which required separation of complex mixtures of hydrocarbons were
largely responsible for development of metal helices for column packing.
Glass helices were developed as a substitute for metal hclices where the
latter had insufficient l'esistance to corrosion, as in the fractionation of
mllIly organie eompounds.
Several mothods and machines h[we been devised for const.ructing glass
55 Fellske, Tonherg, Itnd Quiggle, Ind. Eng. Chem., 26,1169 (1934).
'6 Laughlin, Nash, and Whitmore, J. Lim. Chem. Soc., 56, 1396 (1934).
200 A. L. GLASEllROOK AND F. E. WII,LIAMS
mld metal helices. 57 - G2 }'undamcntally, all methods consist in winding wire

or glass ftbers on a rotating; mandrel, adjacent turns being very nearly in
contact. A higlHipeed rotn,ting knife or stone then cuts, or, in the case of
glass, knicks each turn across the fiber to produce a single helix of paeking.
Commercial production of both types of helices has been so well developed
that t.here is no just.ification for building a machine to produce helices for a
few columns.
Metal helices are lLvailablc in soyeral sizes to allow both large- and small-
diameter columns to he packed with n. size having the proper ratio of
column diameter to helix dinmeter. If this ratio is 8 or higher, the packing
fits closely to the tube wall and provides maximum efficiency in a given
length. St::Uldard metal helices are manufadured, * usually of stainless
Ktccl, with inside diameter:> of 3/61 ,1/16,3/32 , and 5/32 in. of No. 40, No. 36,
No. 30, and No. 2G gage wire, respccth'ely.
The effect of helix diametcr on H.E.T.P. in a column 50 mm. I.D. is
shown ill Table X. The data include results on several sizes not commonly
used to cmphasize the effeet of helix diameter on efficiency.OJ
TABLE X 63
EnICIENCY 01> VAlUous-DLUUl'l'ER SINGLE-TUR!',' HELICES IN A COIJUMN
5.0-C,,!. LD. WITH PACKl!lD SECTION 259 eM."
Total press.
Boil-up drop No. of
ru.te, range, tlteor. H.E.T.P.,
P~cking mt./min. mm. tIg plates ent.
I/.-ill. I.D. rLiuminum h ••••••••• 325-355 17-17.5 15.2-14.7
'/u-in. 1.D. aluminum" ......... IS-180 1-6 27-16.5 9.7-15.7
5/'2-in. I.D. stainless steol' ..... 5S-1\)7 -to22 25.5-19 13.7-10.7
l/az-in. 1. D. nickel d • , •.••••••.• HiS-265 10-22 32.5 7.9
lis-in. I.D. niakel' ............. 168 45,0 5.8
'/32-in. LD. Eitnjnle~s steol d •• ' •• 33-12.5 4-23 68-44 3.8-5.8
• Tested with n-hept.lilIc-luethyl<lyclohcxl1.ne test mixture at total reflux.
b InCipiently flooded just prior to test.
e No flooding.
" Fully flooded prior to test.
It is apparent from the results of Table X that decreasing the diameter of

the helices leads to a marked reduction of H.E.T.P., particularly as be-
* Dr. ~1. R. Femke, Petroleum Refining Laboratory, State College, Pennsylvania.
Ace Glass rIle., Vineland, New Jersny. American Instrument Co,; Silver Spring, Mary-
land.
57 Young, W. G., and Jltomitis, J. Am. Chern. Soc., 58, 377 (1936).
58 Dostrovsky and Jacohs, Chemistry & Inrlustry, 23, 204 (J945).
6. Still, Chemistry l~ Industnl, 23, 130 (1\H5). ..
00 Stewart, W. W., Ind. Eng. Chem., Anal. Ed., 8,451 (1036). ,
61 PI'ice, R. W., and McDermott" Ind. Eng. Chem., Anal. Ed., li, 289.(1939).
02. Roper, Wright, H.uhoff, and Smith, W. R., J. Am. Chem, She., 57, 954 (1935).
43 Fenske, Lawroski, lind Touberg, Ind. Eng. Chem., 30, 297 (193~).
II. ORDINARY FRACTIONAL DtS'l'ILI,A'l'ION 201
tween helix diameter of 1/4 in. and 3ja2 in. Permissible hoil-up mte di-
minishes sharply with smaller diameter packing; the maximum effieiency
is attained at lower boil-up rates.
More complete data, including holdup, efficieney factor, anu pre:,;sure
drop, of columns packed with l/Io-in. and 6/32-1n. single-turn metal helices
are given in Table XI,64
TABLl'~ XI
Hr,;cTU'WA'rION CHA.RAC'fERlSTlCS 01' 1\1"''I'-'l.L HBLICI,s64
----~-~-
Column Ilrnssura
drop,
Length. Di~HnlJ "£hrougilput. 'l'heor. H.E.T.P., Holdu(l, Effie. Inm.
C111, em. lIlt/min. plates om. ml./plat. factor fIg/plato
1/,,;-111. I D. stainless steel of No. 34 wire

71 0.8 0.38 55 1.3 0.16 140
105 2.54 8.3 65 1.() 1.2 420 0.02
105 2.54 16.7 62 1.7 l.fi 630 0.07
105 2.54 .51.0 43 2.5 3.3 025 0.65
---~--.-
3/:,,-In. I.D. stainless steel of No. 30 ",ire

70 08 0.54 27 2.0 0.29 110
81 1.2 7.5 37 2.2 0 . .54 820
256 2.54 H.5 140 I.\) 1.2 425 0.13
2G6 2.54 Hi. 9 126 2. L l.!l 630 0.05
2(l1l 2.501 50.7 88 3.0 3.4 !JOO 0.57
225 5.08 22 58 3.9 13.5 HS O.OOU
225 5.08 44 54 4.2 16.3 162 0.048
225 5 08 132 44 5.2 31.0 255 0.72
Metal helices Ih6 and 8/ 64 in. in diameter are of chief interest for packing
high-efficiency columns of small physical size to minimize holdup and so
permit fractionation of small charges. Construction of the rectifying sec-
tion of fLn SO-plate analytical column packed with 31M-in. helices is shown
in Figure 12,66 H.E.T.P. of this column is 0.35 centimeter. Maximum
boil-up rate is about 2.5 ml. per minute; holdup is S mI., whieh gives an
efficiency factor of about 1000.
Metal helices have achieved eminence as packing for laboratory columns
chiefly because they provide high efficiency in reasonable column height at
relatively low cosL. Constmction is easy, and column characteristics are
fairly reproducible, particularly if the packing process is carefully and
properly performed .. Variations of H.E-T.P. in different columns of simi-
lar size and packing dimensions among different operators ,are to some ex-
Whitmore el aL, J, Am. Chem. Soc., 62,795 (1940).
64
Hall, H. J., and Jonach, SyiujJosium on High-Temperature Analytical Distillation.

6&
American Petroleum Ir-stitute, Meeting at Chicllgo, Il1., Nov. 11, 1946, p. 42.
202 A. J". GLASgllROOK AND F. E. WILLIAMS
tent u function of test. procedure, purity of components of It text mixt.ure,

and analytical [Lccuraey in making the tests, rather than vagaries in the
behn,viol' of the packing. Pressure drop acrosS helices-packed columns is
not excessive if reasonable throughputs are acceptable. Efficiency factor
tends to be somewhat low, but except in analyt.ical dist.illations of very com-
plex mixtures whose components are close boiling, t.his factor is not serious.
f ~ FEMALE
MERCURY SEALED
NO-LUB JOINT
:.,...---'"10 MM. TAPER
A - 25 MM. I. D. TAKE - OFF FUNNEL

B- 15 MM. I. D. NOT SHOWN
G- II MM. 1.0.
0- 15 MM. 1.0.
INSIDE DIAMETER
E- 8 MM. 1.0. OF OUTER
F- 8 MM. 1.0. JACKET- 35 MM.
:",...----10 MM. TAPER
~ ~ FEMALE
NO-lUB JOlrIT
Fig. 12. Design of vttellum-jl1cketed column fo!' analytical distillations. as
Columns packed with metal helices probably diminish in efficiency as

the packing undergoes a certain amount of corrosion inevitable under the
most careful operating and cleaning conditions, and as small amounts of
insoluble material deposit on the packing. These changes in efficiency will
vary widely, depending to a very large degree on the material distilled , and
.
II. OHDINAHY E'JL\C'l'IONAL DIS'l'II,LA'l'lON 20:3
on the extent of nil' leakage into the still pot as well as the effeetivclleRs of
the cleaning process after a distillation has been performed. No data are
available on the effect of corrosion and deposits on the efficiency of he1ices-
packed columns; however, a large Stedman column was tested after 3
years of use and found to have lost nearly 50% of its efficiency at high
throughputs, but only a few per cent at low throughputs. B6
Glass helices may be substituted for metal helices as packing for rectify-
ing sections when the metal type is not resistant to chemical action of a dis-
tillate. Glass helices are not available in as many diameters as metal
helices, they are less uniform in fibcr diameter, and their fragile character
prevents tamping to provide a uniform paclced section; consequently, a
column packed with glass heliccs may have a greater H.E.T.P. than one
packed with metnl, other factors being equal.
Table XII gives the reetification characteristics of lis-in. glass helices. 64
TABLE XII
Hr;CTIFICATlON CIL\.RAcnmISTlCS OF' lis-INCH GLASS HELICES AT TO'I'M" REFLUX
Column
length, LD., Throughput" Theor. H.E.T.P., Holdup, Efficienoy
em. eru. lIlI./min. plate~ CIll. IIlt/Jllate factor
')
43 1.0 ~.
~
I 11 3.9 0.70 230
43 l.0 7.8 11 3.9 0.77 (HO
80 l.2 3.7 21 3.8
80 1.2 11.3 18 4.3
-~~----.---
TABLE XIII
D
RECTIFYING EFFICIENCIN!3 OF lis-INCH GLASS HELICES
Column Packed lwight, Theor. n.E.T.p .•

dinIn" eUl. cm. [jIM." em.
1.1 26 9.0 2.8

1.3 38 5.0 7.6
0.8 '10 6.7 6.1
1.4 44 8.6 5.1
1.3 44 11.5 3.8
1.4 '15 11.0 4.1
0.9 53 14.0 3.8
1.6 60 9.7 6.1
1.3 63 9.0 6.9
1.7 82 12.5 6.6
1.5 87 18.5 4.8
1.4 135 21.3 6.3
a Tested with benzene-carbon tetrachloride at maximum throughput and total reflux.
66 Borns, Coffey, and Garrard, Symposium on High-Temperature Analytical Distilla-

tioll, American Petroleum Institute, Meeting at Chicago, Ill., Nov. 11, 1946, p. 26.
204 A. L. GIJASEBROOK AND F. E. WILLIA1\fS
~~g ~~8 ~~$

000000000 000 0101'- 0 OlClO
C'i"';c-ic-iC'i~C'i"';"";
""'~""'C'<I", .... t-<.01O

..... c<t1'-oot-OC'l>t:l
000000000
.,,-..,.-....- ,-......-...-.., ..-..--,......

.-IU~O ~lCIQ ~lOC:> X
c:-lC'l<¢ C'lc<te<:l cr,)cr,)cr,) ~
.._,..._.,.__ ~._..._.. '-"'..._...._., ~ ..j<C'l CI:)..j<
,...,cr,)<O CO~~ ""'0:0:: ... ~e.'><oIr:¢~IO~~
rll""'1~
...-Il"""irll""""'i"P""l
u:> ,.;l t- .0 <.0 '" 00 co '" OJ
~
.£
. ~
~ ..j< 0 C'l .0 Ol .... ":~~~t-:~c:~"'!
~~~~Cl':)ct:>}.r::.l~l~
~ !;i~;?i ~~g
~
t!
t-:.-;~,,:c:~~~c:
1:Oe.'>C'l'""'OC'l,"",0>00
C'lNC'lt-t-t-IO"<I'..j<
r-.---.""
OOOOi/:J.-....-..-.".......,-..-..
-':-1
IOlQlO'O"'lQ..j<~..j<
C'i C'i 00 oci cci oci cci 00
.......... ..__...._....._.."-'..._.,,~.._.
r:;r::r::ggg~~~
C'lC'lC'lC'lC'lC'l
";<~..j<";<";<~<O<.O",
XXXXXXXXX
";<";<..j<"<I''';<~<O<C<O
"
II. ORDINARY FHAC'l'WNAL DISTILLNl'ION 205
'""
In M I!'l "'i"! C"l
OC"l'<l'OOM
Ir: C"'"i"!
C'1:)~t-.
0000""" 00,....
0000",,1~
...... 00000 000
O>l'-ol;\l ..... Oc:-l

OD:lQ':l"""'Cf,)~
000000
I
~~l
Olt) I I la
C"l
!:'I 00 e.c
~ C'I? "!:'f"I It.,
"'" ~;'6~gS
00000 OOO'l""""lr-l
I I I ~~OSo2
O;;O'"'.-!C"I
c:. .,-t eq "1't lQ
.S O,.-..j....-lr-t,..._j 0 0 0 .............
,t....~~('_,CJlQlt"J
......
.;<.;<.;< It:lL~ Ii?
~~~~~~
"""O>Me<lM"->
c.o 1.Q ~ H'~ LQ lQ
o:i
...--...--._.---..
1~l.1':ll.QU~lQ1Q ...s
....,;.I
~~~~~~
_..._....._.,._,._...,-",
000000
~~O";Cj.lO';lCJ
C:-le<1l;\le<le<lG'l
.;<.;<.;<"'''''''
xxx XXX
206 A. L. GLAFiP;BHOOK AND F. E. WILIJIAMS
C 8<:"1 ~t.... ..,.

_.. o.,_, ""1'40;:,0 0......-1 t:.O~c..l
00 0 'I""""IC"l
~o~'""!~
oooooc 00000 oocoo
.""'1'1 rfJ C'1

..-I I"""'l
~Q c;:. ....-.
~ C1':l fd8~~~
00000 ~ C'1 M"'i'I ~
r.:;J I"-I~O~ f'.. 00 MC'l"'"LQt-. ~ l-O 1- ~ ~ 1'-00-;<"'0:>

c.ot-CfJ~O~ 1.0 t............ LQ 0 C'l ..-I c-.1 H":) c:: 0 ~o;....-4M~
000000 .-< 1:'1 1:'1 ~ 0":> rl,-!-,'I"""Icq OO.-l~__.t
~OlnOl~e-:J
...... N«:>lI:'OOt-.OOO":><oO«:>O
....-I ~ e.c a M 1t;)
...............
u') 10":>
.... c-i
X X
<t':> '"
.... c-i
X X
«:> 10":>
c-i ,...;
n. OHDINARY IIlUC'rIONAL DIS'l'ILLA'l'ION 207
0 0 r- ~ r-I
i.~ 1..":) "I'i'i"l::f'i ~
00000 0'-<0'-< ....

,b ..1 ,~ d, ..;< .\.
t-MOOOMO
000000 00
..c .co .0 .0 .e
(fJ.-(rfj10M 00 " ,.... U
,~
o Cl ... '1' t- t-
Cfl7711
C/) C\l t..... '"'tI- t-
'I '1 l'
l'- r-I,.....,
f'~ ~ ~ ~ l-l"e~C'1~O~C'l~lC
1""""t1"'o\1""""tP""1._'r-Il"""'I~
l:'lc<ie<lN
000 lr':' C'1 It'"J

0:.0 l.~ U':i ~ ~
Lt:I H,,:! 0 1 0
~~c:i~~
208 A. L. GI,ASEBROOK A.."N"D F. E. WII,LIAMS
A tabulation by Fenske et al. 67 of a large number of columns built and

tested by different workers is given in Table XIII.
The large variations of H.E.T.P. shown in Table XIII are not only the
result of differences in column diameter, length of packed section, and in-
evitable differences in testing techniques, but are indicative of variations
in packing density which results when helices may not be tamped. The use
of glass helices as pacldng is justified only when a noncorrosive packing is
required, as in the frnctionation of acids, organic halides, certain sulfur
compounds, and SOIlle phenolic products.
Jl1cMahon packing: Although single-turn stainless-steel helices are one
of the most effieient packings for laboratory columns 2 in. in diameter and
smaller, they show increasingly poorer results as column diameters are in-
creased. Values of H.E.T.P. increase, and under the most favorable
conditions, a column 15.2 cm. (6 in.) in diameter would have an H.E.T.P.
equal to the column diameter or larger; McMahon packing68 appears to
have more favorable characteristics in this respect. This packing is
made* by bending 1/4-in. squares of wire cloth into the form of saddles;
the cloth is 100 X 100 mesh with a wire diameter of 0.0045 inch. The
characteristies of the 1/4-in. packing are given in Table XIV.59
The lower H.E.T.P. of the brass packing appears to be the result of bet-
ter wettability of the brass as compared to Monel packing. Further investi-
gation of the effect of mesh, wire size, and particle size on the efficiency of
McMahon packing may lead to more favorable parameters for the packing.
McMahon packing made from 8/8- and l/rin. squares of 60- and 80-mesh
stainless-steel wire have great.er values of H.R.T.P. bllt lower pressure
drop than the 1/4-in. packing. 70
Protruded packing: A recently developed packing for use in columns 3.7
cm. in diameter ancllarger is known as prot.ruded packing. It "consists of
small units made from thin metal which contain more tluLll 1000 holes pel'
square ineh of surface and is shaped into a half cylinder ~wit.h corners or
edges bent in to prevent stacking. The holes are not clean-cut, but a burl'
or protrusion is fonned on one side. These burrs and holes impart an im-
portant property to the metal-because of them it. is readily wetted by
liquids. Preflooding is not necessary with this packing, a fact which greatly
lowers the time required for initial equilibrium to be reached. "t
* Wire Cloth Products, Inc., 332 S. Michigan Ave., Chicago, Illinois.

t Manufactured by Scientific Development Co., State College, Pennsylvania.
67 Fenske, Tonberg, and Quiggle, Ind. Eng. Chem., 26, 1169 (1934).
68 McMahon, Ind. Eng. Ghem., 39, 712 (1947).
6P Forsythe, Stock, Wolf, and Conn, Ind. Eng. Chem., 39, 714 (1947).
70 Fisher, A. W' f and Bowen, Chem. Eng. Progre88, 45, 359 (1949).
II. ORDINARY FRACTIONAL DISTILLATION 209
Table XV gives tile rectificat.ion characteristics of protruded packing

in 2- and 4-in. columns.71
This packing seems suitable for largcr cnhunns of high efficiency and
large throughput. When throughputs were in the ratio of the cross-sec-
tional area of the columns, the lO-cm. column had an H.E.T.P. only
slightly greater than the 5-cm. column with the coarser packing. With
the smaller pncking size, the 5-cm. column had the lower H.E.T.P.
"Heli-Pak" packina. The most recently developed dump packing ii'l
designed on the prineiples of the preformed Heli-Grid packing described
below. Each piece of paeking has a boxlike shape, and is constructed by
winding wire onto a rectangular mandrel, the spacing of the turns being
carefully controlled to ensure formation of a film by capillary action when
the packing is wet with liquid. Tbe packing has the desirable property of
pouring freely without tangling, and is easy to intl'ociuec into a column.
For best performance, the paeking should be prefloodecl.
Performance data of several sizes of Heli-Pak packing as determined in
the laboratory of the mallufactul'el'* are given in Table XV-A.
Ball pac/dna: Sphel'ieul pal'tides have the" least surface per volume of
particle of any possible eonfigumtion. When used as packing in a column
of diameter large compared to the diameter of an individual particle, balls
will pack uniformly, and have [L minimum surface in contact with any ad-
jacent ball or with the eolumn wall. Free space remaiIls rensonably con-
stant for two sizes of partic:le, one of about half the diameter of the other.
A summary of an investigation72 of copper, lead, and glass balls is given in
Table XVI. Column efficiency was independent of the material from
which the balls were made. Pressure drop per theoretical plate was high
which reflected the small free-space of the packing. The efficiency factor
is comparable to that of metal helices.
]l!Iiscellaneous packings: Table XVII gives efficiency data for several
miscellaneous types of packing.73 The majority of these are useful in
larger-diameter columns where high distillation rates are desired and low
efficiency is satisfactory. Carding teoth appear to be suitable for bigh-
efficiency packing; unfortunately, they are not available in stainless steel
or other corrosion-resistant material, and hence are 110t of interest as pack-
ing.
No. 19 jack chain, as shown in Table XVII, has no particular merit as
packing. Locket chain (link circumfel'enee 6.2 mm., rod diameter 0.33
mm.), while not outstanding, is a reasonably good packing for small col-
,. Podbielniak, Inc., 341 East Ohio St., Chicago, Illinois.
71 Cannon, Ind. Eng. Chem., 41, 1953 (1949).
72 Glasgow and Schicktanz, J. Research Nail. Bur. Standards, 19,593 (1937).
78 Fenske, Lawroski, and Touberg, Ind. Eng. Chem., 30, 297 (1938).
210 A. L. GLASEBHOOK AND F. E. WILT.. IAlIIS
umns; it showed an efficiency factor of 525 in comparison to 425 for ball

paeking when tested as illclieatecl for hall packing in Table XVl,12
Preformed Pad:inys. Stedman pad:ina: A number of wire-gttuze pack-
ings for laborat.ory and illclustrialrectification columns !lave been developed
by Stedman. 74 ,75 li'igure 13 shows the conical type which has been found
most suitable for laboratorv ('ohunns. I t consists of a series of wire-gauze
discs' stamped into flat, truncated cones and welded
together alternately, base to base and edge to edge,
to form a regular series of cells. Each cone has a
semic:iJ'cular opening which extends about two-thirds
of the distance from the edge of the cone to the flat
disc in the center. The cones are assembled so that
the perforations, whieh serve as vapor pass:1geways,
are located alternately on opposite sides of the cells.
SectioIls of paeking are inserted into a close-fitting
verticnl tube to form the rectification section.
In operation, 1ihe ascending vapor enters the vapor
hole in the lowest eell, flows practically horizontally
through the cell, :1nd lel1ves through the vapor per-
foration in the upper cone of the cell. The vapor
stream then divides and passes between the COIl-
taining tube and the gauze to the side of the column
where it entered the first cell. The vapor stream
then enters the second lowest cell. This flow pat-
tern is repeated until the Vl1por has passed through
all the cells.
The reflux liquid flows along the gauze as a thin
film which seals the openings of the mesh and makes
the vapor follow its prescribed course. The reflux
Fig. 13. Conical-type
also provides a liquid seal at the points of contact
Stedm:1ll pac:king. between the packing and the containing tube, and
this prevent.s by-passing the vapor as the reflux
descends the column it flows out toward the tube wall on the upper cone
of each cell and then back toward the axis of the column as it flows along
the lower cone of the cell.
To provide the proper meehanical strength, to fUl'1lish a surface of suit-
able roughness, and to give a. mesh opening of a size that ean be sealed by
the surface tension of the liquid require a careful choice of gauze; the most
satisfactory mesh sizes wel'e from 36 to GO Witl'p wires pel' inch and from 3fi
to 40 weft wires using lO-mil wire,14
74 Stedman, Can. J. Research, BIS, 383 (1037).
76 Stedman, Trans. Am. Inst. Chern. Engrs., 33, 153 (1937).
211
Stamping the r~ones for fabrication of Steelman packing requires pre-

cision dies; the welding together of the cones to form the packing requires
the use of special jigs. The eflieiency of the packing depends upon a close
fit with the containing tube. Precision-bore glass tubing with an inside-
diameter tolerance of 0.001 in. if> gencrally used for the r:ontaining tube.
Furthermore, the insertion of the packing in the tube is rather difficult;,
and if not done properly the column efficiency is greatly decreased. For
these reasons it is munh easier lLnd cheaper to purchase the assembled tube
:lud packing than to construct it. *
The redificatioll ehanwteristics of StedmlLll packings have been eu,re-
fully studied. 76.fT In order to obtain the full efficiency of the packing; it
was necessary to prefioOtl the packing before running the efficiency test.
The enhanced efficiency clLused by prefioocling packings was first noted by
Nickels. 73 In the ease of Stedman pttcking, the prefiooding operation ap-
pears to free t.horoughly the gauze rml'faces of entrapped air so thai a strong,
uniform liquid film forms which completely prevents channeling with re-
sultant loss of rectification efficiency. The data of Bl'llgg77 are given in
Table XVIII.
Inspection of Figure 14 (p. 213), a plot of the data from Table XVIII for
the 1,9-cm. diameter paekillg, shows that, in contrast to other types of
wetted-surface columns, the H.E.T.P. values of Stedman paeking are not
lowest at the lowest. t.hroughput. Instead, as the throughput of the Sted-
man packing is increased from t.he lowest rates, the H.E.T.P. values de-
crease sharply until a certain minimum value is reached; further increases
in throughput result in lL gradual inermtse in H.E.T .P. values until the
flood point is reached. Bragg explains this phenomenon on the basis that
Stedman packing requires a certain minimum throughput to keep the gauze
thoroughly wet and that t.he rectification efficieney is greatest at this mini-
mum throughput.
The H.E.T.P. values, presslll'e drop, and holdup of Stedman packing
compare favorably with those of other types of wetted-surface columns of
comparable throughput, such as the spiraL column:,;. However, the varia-
tion of H.E.T.P. with throughput is more pronouneed with Stedman
packing than with spiral colulllns (compare Figs. 7 and l'4). Consequently,
throughput control is more critical when'Stedman packing is used.
Hcli-Gl'irl packing: A prio1'i it nppears that properly designed packing
* Steuman pa.eked eolUllll1S :1l'!l sold by the Suirmt,ifif: GJass App:1mtus Co., Bloomfield,
New Jersey.
76 Bragg, L. B., Tmns. Am. Inst. Chem. Engrs., 3'1, 19 (194l); Nail. PetrolellnL New8,
28, R-34 (1941).
77 Bragg, L. B., Ind. Eng. Chem., Anal. Ed., 11, 283 (1939).
78 Nickels, J. E., Thesis, Pennsylvania State College, 1936.
TABLE XVIII. Rli:CTH'lCATlON CHAHAC'l'ElUSTICS OF S'rEDMAN PACKING
Tlwnll~hp"t Pressure drop, Liquid Efficienoy

liquid, H,E.T,p.,a 1II1ll. Hg/theor. holdup, factor,
ml./min.~ CIll. plate ml./plato plates/hr.
O.Bi}-I'Ill. di:tmr.ter panking

1.50 3.05 0.00+5
1.67 1.87 0.0028
1.83 1..17 0.0043 0.137 805
2.17 1.H 0.028 0.144 H05
2.34 1.08 0.030 0.126 1110
2.50 1.08 0.083 0.140 1070
2.83 1.1-1 0.084 0.150 1130
2.83 1.17 O.ODS
3.00 1.22 0.13 0.170 1060
3. Ii 1. :3fi O.li 0.19-1 980
3.2,1 1.38 0.18 0.197 1020
3.50 1.47 0.22 0.210 1000
3.75 1.5G 0.31 0.228 990
'1.37 1.M! 0.33
3.84 l<'lood pt.
1. 9-em. diameter packing
1.07 1. 41 0.0035 0.201 ,500
2.08 1. 27 0.0077 0.223 660
2.50 1.21 0.0093 0.213 705
2.92 1. 23 0.014 0.216 810
3.3'! 1. 26 0.025 0.356 560
5.00 1. '13 0.0054 0.421 710
6.67 1.66 0.229 0.564 710
8.34 1.81 0.342 0.628 800
10.0 1.!l3 0.450 0.808 740
10.0 2.19 0.605
10.1) lIlood pt.
2.5-CIll. diametor pncking
2.14 1.(;2 0.003D 0.217 590
2.50 1.25 0.0037 0.186 810
2.67 1.27 0.0038
3.33 1.39 0.0085 0.296 675
4.17 1.49 0.016 0.35'! 710
4.67 1.57 0.022 0.43H 6·10
5.00 1.58 0.027
6.07 1.75 O.O,;!\) 0.(343 620
g.OO 1.85 0.067
8.33 1.1)6 '0.097 0.772 650
10.00 1.06 0.151
11.70 2.06 0.214 0.915 770
13.3 2.16 0.320
15.0 2.32 0..160 1.25 720
16.7 2.48 0.647 L38 720
18.0 2.52 0.73 1.48 730
18.7 2.4!J 0.807 3.25 350
18.0 Flood pt.
a Det,ormined with benzene-ethylene dichloride test mixture.
U. ORDINAItY ~'HAC'rlONAL DIS·l'ILl,.\TlOl\ 213
made as !Hl integral unit aeeurding tu a detinite ptlttern Bhould have bet,tel'
rectification characteristics (md be more reprodueible constl'UetiollwiBe
than a packing depending on random arrangement of n large numher of
individual pieces of packing. Rigid onp-piece packings ntilizing wire turns
or loops spaced preeisely in uccordance with a (lpfinite [lesign have been
VAPOR VELOCITY - FEET PER SECOND

01 02 03 04 05 06 07
I2 55
STEDMAN PACKING NO 104
1 30
LEGEND
.', • - PR ESSURE ORO P
f - - - Io 50
,
0.5 !. \
a-HOLDUP
x-THEa . PLATES
If \
[7
v rl
9 ~ 25
8.3:
I-
«
45
(/)
w
:c
, \ /
u...
o
I
7 ~ 20 n..
0
U
U1
W
I-
<i
u
~
I
a.:
0.6 lZ / ::r:
u
6 ~
:::l
0
;5. 40
..J
a..
..J
<i
t-'
W II
/ \ 1/ n..
'5 ~
I
15
:c
.@
l)
5a:
:t
J IV 4gs
o
w
C'J
J
35
8:c
I-
0.7
/ao 1/\\ 38:
(/)
I{l
10
~,
II
0.8
) / \ ~
I-
z 2 30
II
/ / xO
0
t7 V v
5
09 9u...
o 25
o 100 200 300 400 500 600 700
REFLUX RATE - ce. PER HOUR
Fig. 14. Perforrnlillce of eonienl-type StPrlllUlIl paekillg. 77
investigated by Podbielniak. 79 He made an exhaustive study of wire pat-

terns, sizes, spacings, loop diameters, sectiol1H, etc. t.o determine the opti-
mum design rom the st.andpoint of H.E.T.P., holdup, and throughput.
Two forms of close-cleal'al1ce wire packings which Podbielniak79 found
79 Podbiellliak, Incorporated, Chicago, Illinuis, 1945 Catalog, Hyper-Cal High-
Tempera.ture Distillation Apparatus.

21-1 A, 1" GIAREBHOOK AND Ii', I" WILLIAMS
to }mve exceptionally good rectification characteristics are shown in Figure

15 and described as follows:
"TIle p~lcking in Figure 15£1 consists of coils of sectorlike section, wound around
eaeh and around a very small core wire, to yield a numherof uniform nonflooding
vapor passageways, 'liner]' perfectly with Cllpillary liquiu reflux films extending be-
tween vertically ad,iacent wire loops. Another form of packing (Fig. 15B) con-
sists of wire cage staircases, wound around a central solid core and around each
other in concentric telescoping layers, if necessary to ohtain larger diameters and
capacities."
(A) (B)
Pig. 15. HcH-Griel packings: (A) assembly of sector-section coils twiBted
around central core; (B) staircase assembly of roctangular section coils around
central core. 7'
See also Chapter IV.

These packings, which h!1ve been Gl1Iled Heli-Grid, have been made in
diameters ranging from 0.57 to 2.5 em. and in length up to 6 feet. Wire
sizes and spacings vary with the design but usually the wire is about 0.010-
in. diameter with spacings between wires about equal to wire diameter.
The packing is made from corrosion-resistant Nichrome or Inconel.
II. ORDINARY FHAC'l'IONAL DlS']'ILLA'l'IOJI: 215
The results of fractionating-efficiency tests on Heli-Grid rectifying sec-

tions of varions diameters arc given in 'Table XIX. The data show that
the Heli-Grid packings combine low H.B-T.P., Imv holdup, and low pres-
sure drop with a fairly high throughput resulting in an excellent efficiency
fuctor. Over most; of the useful operating range, the lLE.T.P. shows a
rather marked variation with throughput so that the packing requires an
extremely efficient adiahatic jacket and exact control of the heat input to
the still pot in order to s('(:ure o;table operation and full utilization of the
fnU'tionating effieiPI1cy of the Heli-Grid packing. A remarkably efficient
vaeuum jaelw1" and auxiliary H('ecH:-iories have been developed to attain this
elld.7 D
The time required for [t reetifying section to reach equilibrium is an im-
portant eritcrion for ('Ornparing column designs. Unfortu.nately, the litera-
ture docs not (:ontain much information on this subject. In the case of
the Heli-Grid paeking, however, such data are tLvailuble Ru and are plotted
in Figure Hi, for [1 column 2.5 em. in diameter nne I 3G in. long. Inspection
of (lata shows that [L :Hi-in. section of Heli-Grid packing requires about 8
hours to roach equilibrium; furthermore, the equilibration time appears
to be independent of throughput.
Like many types of eOlUll111 packing, the Heli-C}rid exhibits improvement
of fraetioIlt1tion pfficiency upon prefiooding. The effect of pl'eflood is illllS-
TABLE XIX
I{J~C1'IJ.'ICA'rlON CILIH·\Cn;m>n:ICS 01' HJ~J,I-GRl[) PACKINl,a
Efiiciency
Throughput, H.E. T.p .. r, preS"lll'e drop, Liqllid hold liP, factor,
Illl./min. ern. JIlIn. I-Ig/plato lIlI./pl.to plates/ltr.
O.ii7-mn. dillIlletpl'
0.83 0.75
1.25 1.20 0.02 3750
1.07 1.iG
2.50 2.29
1.2.5 0.5n 0.02

1.07 0.78 0.0·1
2,08 0.\)1 0.06 0.0·1 3120
2.50 1.m! o.on
3.33 1.36 0.14
.f. 17" U10 O.H)
Table continued
so Brandt, Perkins, and Halverson, Oil Gas J., Dee., 1946; Symposium OIl High-
Temperature Analytieltl Distillation, Americ!tn Petroleum Institute, Meeting at Chi-
cago,_IIlinois, Nov. 11, 1946, p. 51.
21fi A. L. GLAf:iEllnOOI( AND F. E. WILLIAMS
TABLE XIX (colltinued)

Efficiency
1'hro~ghput, H.E.1'.P .. ~ Prcp:;urc drop Liquid holdup, factor,
1n1.,unll. l',rll. mill. fIg/plate IlIl./pjlltc plateR/hr.
----"
1 . lO-elll. dhmetcr
208 0.02 0.05 3333

3.33 0.1l3 0.07 2860
4.17 l.17 0.20 2200
1).00 1.36 1950
(i.2[j' 1.(\0 1700
1.3-rm. diameter
'1.17 0.63 0.03
5.0 0.02 0.06 0.11 2720
5.83 1.05 0.08
6.67 1.19 0.11
7.50 1.34 0.13
8.:33 1.47 0.16
n.18' 1.00 0.20
1 . 3-em. ditLllll't[Jl'"
0.5
l.0
2.0 0.76 0.015 0.10 1200
3.0 O.H!) 0.03 0.15 1200
4.0 1.14 0.05 0.18 1330
5.0 1.3G 0.08 0.23 1300
5.5 lA7 0.10 0.26 1270
2 ,2-em. din meter
7.0 0.1l5 0,03
10.0 0.87 0.06
13, 1.12 0.09 0.38 2050
16. 1.39 0.14
20. 1.73 0,21
2fi. • 2.18 0.36
Z.5-I'm. diamet.er
10 0,71 0.03
]4 0.[15 0.07
2.li-pm, diullwtpT'
18 1.17 0.10 0.50 2160
22 1.47 0.16
25 1.09 o 21
30 2.13 0.33
35 2.41 0.45
Ta.ble continued
n. OHDINARY FRACTIONAL DISTILLATION 217
TABLE XIX (concluded)

Effieiene).·
Throughput. H.E.T.l'.,b Pressure d,'op, Liquid holdup, faetor,
>nl./min. em. mill. Hg/pillte ml.!plate pintos/hr.
2. ;3-mn. dhlllet(,r"
8.3 O.BIl o :36 13!10

16.7 1.24 0.5::\ 1890
25.0 1.58 o 76 1070
3a..! l.lll 1.07 1870
• 91.5-cm. length. Interpolated from curves pub~ished by Podbielnia.k. il Data at
H.E.T.P. values below 0.9 cm. obtained in 35-cm. sectlOn.
b Determined with n-heptane-methylcyc!ohexane test mixture, packing prefiooded,
at total reflux.
e Flood poi lit. d Unpublished dat.a of the authors. 'Data of Brallclt, Perkins, and
Hnlverson. "
tratecl in Figure 17, fIJI' a column 2.5 em. in diameter and 30 in. long. The
data show that the number of theoretical plates in the Heli-Grid packed
eolumn is increased ahout one-third as a result of preflooding. Of interest
is the faet that the prefloocled and unpreflooded columns both require the
same equilibrntion time.
To secure the best results from a fractional distillation, the u:';11al prac-
tice is to use a still-pot charge whieh is large with respect to column holdup.
Podbielniak,84 however, obtained some remarkable fractionation results
when he used a still-pot charge which was roughly equivalent to the total
column holdup, and used a column containing more plates than theoreti-
cally nccessary for the separation. The distillation curve shown in Figure
18 shows an analytical separation of n-heptane and methylcyclohexane
(boiling-pqint. difference 2.5°0.) in a 2.2-cm. diameter, (i-ft. long Heli-
Grid packed column. The total distillation time was only 10 hours and a
reflux ratio of only 12:1 was used. The sample size of 75 ml. was approxi-
mately equal to the column holdup. Toluene was used as a chaser to
"push" the sample up the column. Podbielniak's procedure should be
applicable to other types of rect.ifying sections.
The rectification characteristics of the Heli-Grid packing make it especi-
ally useful for analytical fraction:1tions, pal'ticularly when the available
sample size is limited. An example of an analytiC'111 fractionation obtained
in a Heli-Grid paeked column opel'l.lting in the r:oIlventional manner is
shown in Figure 07. A special type of fractional-distillat.ion apparatus
81 Podbielniak, Ind. Eng. Chem., Anal. Ed., 13,639 (1041).

82 Brandt, Perkins, and Halverson, Oil Gas .T., Dec., 1940; Symposium on High Tem-
perature Analytical Distillation, Americo.n Petroleum Inst,itute, Meeting at Chicago,
Illinois, Nov, 11, 1946, p. 51.
218 A. L. GLASEBnOOI( AND F. E. WILLIAMS
(Hyd Robot) incorporating [L lIeH-Grid packing is more 01' less standard

i~l the petroleuIll industry for analysis of light hydrocarbon gases.
Theoretical Plates Vs Times After Preflood-

Typical Curves
For Various Boilup Rates
II>
'"
[5
a:
0 60
(J
i!0
'"
.c:
I- 40 _~
20
2 4 8 10 12
rrrTIe Aner Pre flood
Fig. ]6. Heli.Gl'id {Junked rel'tifyiflg I!oilllllll, 2..5 cm. diameter, 36 ill JOlJg."o
Effect of Preflood on Efficiency-

1000 M~ Per Hour
BoilLJprate
'0
0::
'0
Q
~o
<lJ
.c
I-
2 4 6 a 10 12
Time at Toto I Reflux. tlours
:Fig. 17. Heli-Gl'id packed rectifying column.so
Full details for constructing the patented lieli-Grid packings have not
heen disclosed. Furthermore, the construction is difficult and requires
II. OUDINAHY FIlAC'l'IONATJ mSTILLATION 219
precise equipment and considerable experience. For these reasons the

purchase of the assemhled eolumns is l'Bcnmmcndcd. sa Constructional
Fig. 18. A nf~li-Grid p:u,ketl (,nJ-

Ulllll. liradiollatioll of Il-heplane-
I11uthyleycJuh('xaIH' tI·st. mixtul'f.',
12: 1 rnflllx ratio, total distillatioll
tinH' 10 hOlll'H."
960 10 20 30 40 50 GO 70 ao 90 100
DISTILLATE IN ee.
details of a glass helieal packing similar to Hcli-Grid packing in chnracteris-
tics have been publishccl. '5
B. PLATE COI,Ui\INS
A plate column consists of a tube containing a vertical series of equally

spaeed horizontal plates or trays. The reflux liquid collects on each plate
until it reaches a definite level and thereafter overflmvs to the next lower
plate through an overflow or down pipe. The reflux liquid on this next
lower plate and on each successive lower plnte seals the bottom end of the
overflow pipe and prevents vapor flow. The down pipe from the first 01'
lowest plate in the column is also provided with a liquid seal.
Vapor-liquid contacting takes place during the passage of the vapor
through the reflux trapped on the plates. From this it can be seen that
the separation takes place in a number of discrete steps. If the vapor-
liquid interaction on the plates were idea'!, each actual plate would be
equivalent to 1 theoretical plate; however, the vapor and liquid leaving
the plate are not usually in equilibrium so that each actual plate is equiva-
lent to only about 0.50 to 0.75 of a the01'eticI11 platc, or each actual plate
has an efficiency of .50 to 75%.
Two mot,hods of plate construction are commonly used for dispersing
the vapor in the liquid: bubble-cap platcs and perforated p1ates. Each
plate of a bubble-cap (lolumn contains one or more vapor-riser tubes which
extcnd above the liquid level on the plate. A circular cup or bubble cap
of larger diameter t.han the riser is inverted over each vapor riser and de-
83Heli-Grid Packed columns are manufactured by Podbielniak, Inc., Chicago, Illinois.
8iPodbielniak, Incorporated, Chicago, Illinois, 1945 Catalog, Low-Temperature
Fractional Distillation Apparatus.
8. l\1itchell and O'Gorrnan, Anal. Cham., 20, 315 (1948).
A. L. C,LASBBlWOK AND F. E. WILLIAMS
Heets I;hc vapur bCllPuth the surface of the liquid. The vapor passes
through slots in the periphery of the cap and enters the liquid as a stream of
small bubbles. In perforated-plate columns vapor rises through perfora-
tions in the plat.e. The preS~iUre and velocity of the vapors passing through
the perforations must be sufficient to hold up the liquid on the plate. If the
pressure and velocity nre too low, the liquid descends through the perfora-
tions, and if too high the liquid is pushed off the plate and the contacting
efficiency decreases.
Plate columns have much higher throughputs than film-type columns,
and the rectifying efficiency is nearly independent of boil-up rate so that
operation is simple and uniform. Owing to the necessity of maintaining
a liquid layer on the plates, the holdup and pressure drop are compara-
tively high. '1 he rectifying efficiency of plate columns as measured by the
H.E.T.P. is usually lower than in film-type eolumns. Generally speaking,
the construction of laboratory-size plate columns is more difficult than that
of nn avemge lnboratory film-type eolnmn.
The characteristies of plate columns make them extremely useful for the
purification or analysis of large samples.
Bubble-Cap Columns. Laboratory bubble-eap columns were first
DE.TAIL OF CAP
r;:::t:t:!===;""-,LAT HERE
1JL.~~~-NOrc CURVATURE
(a) (b)
llig. HI. Bruul! hubble-pap rectifying :;ection. 91
(a) Bubble-cu,p column de>lign:
~i()ts ill bubble ('ap. tLbout 0.5 to 1 mm. wide, 3 mm. deep, at 30° angle with radius fire
polished; Pyrox iabomtol''y glnss; all dimensions in millimeters, exact dimensions
rireled. (b) Open.tion of bubble cap: lA) flood point; (B) maximum operating rate;
(C) normal operating rate.
II. OliDINARY FRACTIONAL DISTILI,A'l'ION 221
emplo)'pd by LeBel and Hellningnr. RB Their column and ~evernl other

early t,ypef;R7.8~ are dmwriheti by YotIng'ili unfortunatuly, til<' te<'lt data pre-
sented are insufiieient to permit an accurate nppl'ai:-;al of tlu'ir reetifieatiull
eharacteristics Hnd efficieneieK.
Bnmn90 ,91 developed an i.mproved all-glaHs bubble-cap column. Figure
19 shows an assembly of a recti.fying seetion and describef\ the vapor and
liquid paths. A Bruun ('olumll of 100 bubble-cap sed ions of the size shown
in Figure 19 and equipped with a heating jacket and accesKories fol' semi-
automatic control has abo been deseribcd. 92 The test data obtained with
this column are given in Table XX. The Bruun column has many desir-
able fpatures. The all-glass construction makes possible the fractionation
of many cOI'l'oRive mutel'ials. 'The column reaches equilibrium in a com-
paratively short period of time; for example, the 100-section column re-
quires only 2 to t,
hours, and furt.hermore, owing to th(~ nonsiphoning
TABLE XX
HSCTIPICA'l'ION CHARACT~aUSTICS OF BRL'UN Bl_lBBLJo;-C.U' CnU)"AIN"
Through.r.ut '1'otol No. Liquid holdup, Efficiency

liqui . thear. H.E.T.P., mL/theor. factor,
ml./rnin. plate. b em. pln.te C plate./hr.
5 70. (1 2.8 320
8 8·.1.2 2.4 0.93 450
10 72.11 2.8 050
30 Floodpoillt.
-----------------------------------------------.------------
a2.5-cm. diameter, 100 bubble-eap section, 2-cm. spacing between plates.
b Determined with n-heptane-methyleyclohexane test mixture .
• Dynamic holdup determineci at an 'undisclosed but normnI throughput. Static
holdup is much larger.
construction of the bubble-cap seetion, an uncompleted dist.illation can be

shut down, and the partially fractionated portions of the liquid will be re-
tained on the plates. When the distillation is resumed, product with-
drawal can commence almost as soon as reflux reaches the condenser.
The throughputs are fairly high. Since the H.E.T.P. does not change ap-
preciably with throughput, the Bruun column is easy to operate.
The Bruun column is also useful for the distillation of materials contain-
ing dissolved water. Frequently during the early stag~s of a, distillation
86 LeBel and Henninger, Ber., 7, 1084 (1874).
87 Brown, F. D., .T. Chern. Soc., 37,59 (1880); 39,517 (1881).
88 Clark, H. T., and nahl's, E. J., Ind. Eng. Chern., 18, 1092 (1926).
89 Young, S., Distillation Principles and Processos. London, 1922.
90 Bruun, Ind. Eng. Chern., Anal. Ed., 1,212 (1\)29); Bruun and SC'hicktanz, J. Re-
search Natl. Bur. Standards, 7, 851 (1931).
91 Bruun, Ind. Eng. Chern., Anal. Ed., 8, 224 (1936).
92 Bruun and Faulconer, Ind. En(J. Chern., Anal. Ed., 9, 192 (1937).
A. L. GLASV,IlTIOOK AND l~. E. WILLIAMR
tlw prcHcnce of water will be detected in a supposedly "dry" still-pot

eh:uge. The water distils as an azeotrope, and usually the azeotrope is
heterogeneous and is the lowest, boiling constituent. Thus when the
eolumn is being brought to equilibrium, water condenses in the fractionat-
ing section. The presence of two liquid phases in the fractionating section
usually makes column operation very difficult. The Bruun column is an
exception in this respect. The main disadvantage of this column is the
exceptiomdly high pressure drop. Tests in the authors' laboratory of a
50-section column of the type described in Figure 19 and Table XX indi-
cate a pressure drop of 0.7 to 2.0 mm. mercury per theoretical plate at the
minimum and mr.tximum throughputs, respectively. The high pressure
drop mltkes the column unsuitable for many vacuum fractionations.
The Bruun column is difficult to construct but is available from several
glass fabricators, who manufacture this column as a stock item. * Despite
its appearance of fragility, it properly manufaetured, installed, and oper-
ated Bruun column is as rugged as most types of glass equipment and has a
long life. Vacuum-jacketed columns of the same type are also available. 93
Perforated-Plate Columns. Perforated-plate columns have been used
"01' laboratory distillations since 187I.D4 The early columns generally
employed metal gauze or screen for the plates and glass for the retaining
tuhe and down pipes. The published data on these older columns do not
permit. a reliable evaluation of their rectification characteristics, but the
de::;igns indicate H.E.T.P. values in the order of several centimeters and
large holdups. Readers interested in the earlier designs are referred to the
original articles.%-98 Several of the designs are discussed by Young. 89
A well-designed all-glass perforated-plate column ,vas developed by
OldCl'shaw. 99 The column consists of a series of perforated glass plates
sealed into a tube. Elwh plate is equipped with a baffle to direct the flow
of reflux, a weir to maintain a definite liquid level on the plate, and a down
pipe to supply reflux to the next lower plate. The details of the plate con-
struction and an assembly drawing of the column are shown in Figure 20.
Efficienc:y measurements made on an improved Oldershaw column (using
80 holes, 0.89 mm. in diameter per plate instead of the 42 used :n the original
* Ar'fJ nla~s COmpltllY, Villcbnd, New Jersey. Otto R. Greiner Company, Newark.
Now Jersey. Seipntitie lH!t';" App:mttus Company, Bloomfield, New Jersey.
93 Bruun ltlld W<'st, Ind. Eng. Chern., Anal. Ed., 9,247 (1937). .
94 Linnmilallll, Ann., 160, 105 (1871) .
• 5 Young, S., and Thomas, Chern. News, 71, 177 (1895).
98 Glinsky, .<lnn., 175,381 (1875).
'7 Du Pont, Chirnie S; indu8irie, 8, 549 (1922).
Il8 l'alkiu, Ind. Eng. Chern., Anal. Ed., 3, 377 (1931).
Il'd OlJershaw, Ind. Eng. Chern., Anal. Ed., 13,265 (1941).

II. OHDINA HY FHACTION.H, DIS'l'ILI,A'l'ION 223
column) IDD arc shown in Table XXI. It will be notic:ed that the column
has [L large throughput, and thf~ H.KT.P. is nemly iudependent of through-
put. Furthermore. the 30-plate l'olumn requires only about n.n to 1.5
houri! to reaeh equilibrium. These bet!'> indicate that tlw tolumn i:-; eagy
to operate ::md very useful in eases where the total distillation time is an
important fadoI', as in the purification of htrgc batches of material. The
all-glt\ss construction permits the fnu~tionation of corrosive materials.
HEIGHT OF
BAFFLE'IOMM.
3.5MM.I.0.
PLATE SPACING 1/2. MM.
'25 MM. GLASS BEAD
22 HOLES
129/42 Ground 20 HOLES
Joint
HOLES IN PLATE 2.5MM.
0.85 "'M.I.D.
Bailie
SLOT WIDTH 4M ....
SEAL PLATE
lI,pansion
e.llo,"s !~
~ fl4HOLES
,~ 6MM.
12MM.O.D."UBE "
ENLARGED TO • •
~ES~IUSl1>E(,R /
Ground Joint AND LIQUID
(a) (b)
Fig. 20. Oldmshaw plJrfomted-plate column: (n) columnj (Ii) phte COllstructiLln.9l.100
The Oldershaw column has [L plate efficiency of GO-G5%, which is slightly

lower than that of the Bruun column and uses a 2.5-cm. spacing between
plates compared with 3.0 em. for the Bruun column. As a consequence,
the H.E.T.P. of the Oldershaw column is somewhat higher t.han t.hat. of the
Bruun column. However, the Oldenilmw column was designed for ease
of operation and flexibility rather than low H.E.T.P. The plate efficiency
of the Oldershaw column is increat>ed by decreasing the size of the perfora-
tions99 ; when fritted-glass plates are used, extremely high plate efficiencies
100 Collins and Lantz, Inri. Eng. Chem., Anal. Ell., 18, 673 (1946).
224 A. L. GLASEllI!OOK AND F. E. WILLIAMS
TABLE XXI
a
RE("J'IFlCNl'l(l~ CHARM'TI':HIc;'fJCS OF OLDlmSHAW Pgl\FOR,-\.TED-PLA'l'1oJ COioUMN
Pressure
drop, I..Iiquid
'l'brollg1Iput 'rota.tllil. mm. Hg/ holdup. Efficiency
Jiqrtid, theor', H,E,T.P., theOl·. ml./theDr. :factor,
ml./rnin. [llntm~b mu. plate plate plo.te./hr.
25 l7,2 '1.'1 2.7 550

42 18,8 4,0 0,98 2.7 930
50 19.4 3.H 0,98 2,7 1100
58 18,2 4,1 1.11 3,0 1150
67 16,7 4.5 1.29 3.5 1140
80-85 Flood point,
.2.6 to 2.8 em. dirtmeter; 30 actul11 plates; 80 0.89-mm. diameter holes per plate;
2.5-cm. spacing bf'tween plntes. .
• [lntormined wlth n-hepktne-methylcyclohexane test mIxture.
TABLE XXII
O],D1,RSHAIV PEIlFORA'l'lm-Pr,A'l'E COLUMN PnESSUl\B DROP WlTH LIQUIDS
O~' VARIOUS SURFACE TBNSIONS
. _ - - - - - - - - - - - - - - - _.. _---------
l're,.ure drop per aetual plate. mm. Hg
Throughput liqni(l, Methyloyclo- Formic
ml./hL·. n·Hepto.ne hexane Benzene acid
200 0.88
400 0,95
!lOO 1.02
800 1.09
lOOO 0,7() 1.18
1125 1.23"
1500 O.liG 0.62 0.82
2000 0.53 0,63 0.84
2500 0,55 0,65 0.87
3000 0.57 0.67 0.91
3500 0,60 o.n 1.00
4000 0.64 0.77 1.11
4280 1.17"
4500 0.70 0.85
5000 0.77 0,93
52:30 0.80" 0,97"
PHYSICAL CONSTAN'r8
Boiling pt., DC." •.... ,., ... , !)8.4 100.8 80.1 100.8
Density, g./ml.
at 20°C...... " ............ . 0.6837 0.7601 0.8790 1.220
at b.p ....... , .... , .. , ... ,. 0,6180 0.6989 0.8153 1.170
Bllr-i:we t{)n~j()ll, (lynes/em.
at. 20°C........... , •....... 20.26 23.73 29.02 37.6
Itt b.p...... , ... , .... , , ... . 12.5 1,'5,7 21.3 2().0
.. Flood point with given liquid.
-------------------------------
II. ORDINARY FRACTIONAL DISTILI,ATION 225
can be obtailled 101 ; inereasing the weir height also increases the plate ef-
ficiency.lo2 It should be pO;lsible to reduce the plate spacing of the Older-
shaw column to give a lower H.E.T.P., but this change would cause il1-
ereased operating difficulties, 10\\'['1' throughput, and higher holdup and
pressure drop.
Owing to the eapillarity of the holes in the plate, the t.hroughput and
pressure drop of the Olderslmw column nre dependent upon the surface
t.ension of the material being distilled. This has been illustrated (Table
XXII) by pressure-drop measurements of the column at various through-
put.s using several !:ompounds of different surfaee tension as the diHtil-
l:lIlfPol) The data of Table XXII indic,1te the column is not suitable
for the fraetionatioll of materials of high surfa.ce tension; sm:Lll amounts of
undissolved water in the reflux are very troublesome and may causet4'the
column to become inoperable. In this respect the Bl11un column is de-
cidedly superior to the Oldershaw column. The high pressure drop of the
Oldersh:nv column makes it unsuitable for many types of reduced-pressure
operations. The 3D-sccHon column is not recommended for head pressures
much below 250 millimeters. 100 The Oldcrshaw column has a high dynamic
holdup (Table XXI). However, insofar as sharpness of separation is con-
cerned, the extremely high throughput tends to offset the holdup and en-
ables fairly sharp separations to be made. This same effect also results in
a good efficiency-factor rating for the eolumn. As a consequence, the
Oldershaw column is satisfactory for analytical fraetionations provided the
available quantity of still-pot charge is not limited. The column is self-
draining and tests 'with n-hcptftuc indicate a sts.tie holdup of about 0.02
ml. per actual plate. 10o
Instructions for the construction of the Oldershaw column are :1vailableo9 ;
however, the constmction is quite difficult and requires the services of a
skilled glass blower and special tools. The columns are available commer-
cially. * A simplified version of the Oldershaw' column has been described
in which the plates are supported by spacers mther than sealed into the
column tube. 102 However, the difficult comltruction of the perforated
plate remains. t
In many laboratory distillations t.he pr'(lserwe of metal in the Oldershaw
type of column is not harmful. In such cases an all-metal perforated plate
offers distinct advantages from the standpoint of construction; the per-
* Glass Engineering Laboratory, Belmont, C:tlifomia, manufactures a complete line

of Oldershaw columns and ltCCessory equipmeut.
t Plate sections can be purchased from G. Frederick Smith Chemical Qompl1ny, Co-
lumbus, Ohio.
101 Schicktanz, J. Research Natl. Bnr. Standards, 12,259 (W34).
)02 Langdon and Tobin, Ind. Eng. Chern., Anal. Ed., 17,801 (1945).
226 A. I,. GLASEBROOK AND F. E. WILLIAMS
forat.ed plate::;, for example, ean be stamped from metal screcn. Further-
more, metal perforated-plate columns can be designed with rectification
<:haracteristics elosely approaching those of the all-glass Oldershaw column.
Such a column along with complete constrllctional details has been de-
Hcribed. 1D3
C. REC'rIFYING SECTIONS EMPLOYING MOVING PAR'l'S
Recently several investigators havc advucated the use of mechanical

agitation in the rectifying section to improve the rectification characteris-
tics. All the columns possessing moving parts which have been described
Fig. 21. Podhielniak centrifugal Su[mrfnwtiow.tol'.los
are adaptations of wetted-surface columns and with minor modifications

most of them will function as such when the moving parts are at rest; con-
sequently, they are described in this section.
The first fractionat,ing unit of this type was described in 1935. 104 Al-
though this apparatus has found its chief use in larger-scale operations
than are customarily carried out in the laboratory, it seems worthwhile to
describe briefly the apparatus and its operation. A spiral (Fig. 21) fabri-
eated from a thin strip of metal retained Letween two spirally grooved end
103 Griswold, Monis, and Van Berg, Ind. Eng. Chern., 36, ll1ll (11.144).
104 Podbielniak, Am. Chem. Soc. Meeting, New York, 1935.
II. OHDINARY FRAC'rIONAL DISTXLL_\'I'lOK 227
platen eonstitutes the rotating part of t.he apPILratun. The reflux is intro-
dueed in the center of t.he Rpiral, and beeause of the centrifugal force gener-
ated by the rotation, the reflux liquid fl()w~ in :L thin film along the outer
wall of the spiml passageway and outward from the ll.xis of rotation. After
leaving the spiral the reflux is eolleeted nnd returned to the still pot. Owing
to the pressure differemf' betwnen the fitill pot and the cOIl(lenser, the vapor
traverHes the spiral in au inward direction COllntercurrent to the reflux
stream. Vapor-liquid interaction occurs nt the interface. '''lith this ap-
paratus, exceptionally high throughputs can he obtrLined along with com-
plete absence of entrainment. A lis-in. spiral passageway 4 in. wide has a
throughput as high as 2no ml. pC'l' minute when rotating at 1200 r.p.m. A
IGO-ft. long spiral passageway of these dimensions is reported to have an
efficiency of 80 plates, it holdup of 500 ml., and a preSAme drop of 10 to 20
nun. mercury. IOn
Fractionating sections consisting of a succession of alternate rotating
and sta tionary inverted metal eOllCS (Fig. 22) Imve been used. 106,107 Reflux
liquid from the eondenser flows elown the uppermost stationary cone by
gravity and drops to the top rotating cone where centrifugal action runs it
up over the cone surface and throws it off to the next stationary cone.
The vapor flows through the spae<'s between the fixed and stationary cones.
Inspection of Figure 22 shows that the amount of wetted surface available
in a given column height ean be greatly increased by using this type of con-
struction.
The Mair-Willingham all-metal rotating-coIle column 107 employed a 56-
em. long fractionating sec:tioll. The stationary cones had an outside di-
ameter of .5.08 em. and a 2.54-cm. diameter center hole. The rotating
eones had an outside diameter of 3.33 em. and were mounted on a 0.g5-em.
eliameter shaft. The cones haJ base angles of 30° ::tlld WCl'C spaced 0.64
em. apart to give it elcal'flnce of 0.32 centimeter. Experimental data on the
Mair-Willingham column are given in Tnble XXIII; the speed of rotation
had no effed on the reetificatioll efficiency.
The throughput of the Mair-Willinghnlu columIl is fairly high for a wet-
ted-surfnee column, and the rectification efficicney is almost independent
of throughput. On the other htLllcl, the holdup is mueh higher than in many
other types of wetted-surface eolumns and results in a low efficiency factor.
The constructioIl of the parts and the assembly of the rotating-tube
column require a considerable amount of precision work. The cones are
made either with a die or by spinning. The rotating shaft mnst be accu-
105 Circular No. 14, Podbielniak, Illcorporatecl, Chicago, Illinois, 1937,
105 Huffman and Urey, Ind. Eng. Chem., 29, 531 (1937); Pegram, Urey, and Huffman,
Phys. Rev., 49, 883 (1936); J. Chem. Pllys., 4,623 (1936).
ID7 Muir and Willingham, J. Research Natl. Bur. Standards, 22, 519 (1939).
228 A. I,. GLASEfiROOK AND F. E. WILLIAMS
__------A
1..-
9.5~.6
o--~l
iii;; /
~i::lbt:).:~:.
~~3.~
P _ _ ~1.6
M
'ijIe,4
102
Fig. 22. Rotating-cone distilling column (1l,1l dinlenAions are in millimetBfs).107
A, shaft extending through condenser B G, cylinder of fractionating column

C, inner seal so that liquid returning from I, short section of experimental column
condenser to column first flows with basket.s
through glass sampling tune D ~T, rotating members (bat>ketR, p]ateH, or
E, lliquid level :cones)
]i', stopcock L, manometer
IM,O, cones
II. ORDINARY FHACTIONAL DISTIL14TION 229
TARLE xxrn
RE(~'I'IFWA'l'ION CUAR.\C''l'lmlSTWS OF HO'l'A'I'I:-IG-Cm.;g COl.U~!N
PrUB~ure
Throughput drop, Liqllid Efficiency
Hatlltion, Ii(luid, mm. JIg/tImor. H,E.'f,P., holdup, factor.
r.p.lll. 1Il1./min. plate cm.a lul./pll1te !.l!ltcs/hr.
l;jaO 1.8 0.018 0.87 1.0;- 100

1500 4.7 0.022 0.1l5 1.35 210
1500 Ii. 3 0.02!i 1.12 1.63 230
1500 7.0 0.030 1.0'l 1. 57 270
1500 \lA 0.034 1.13 1. 73 330
500 2.8 0.021 1.0n 1.48 110
500 4.8 0.010 0.\13 1.33 220
500 - ?
a._ 0.021 0.11-1 1.36 230
500 6.7 0.027 1.02 1.50 270
500 8 5 0.032 1.13 1. 70 300
250 1.0 0.012 1.0G
250 3.7 0.017 l.lf)
250 5.6 o 020 1.23
a Determined with n-hcptane-methylcyclolwxane test mixture.
TABLE XXIV
REf''l'rr'WA'I'IOI'< CIHRAC'')'ERISTW;:; OF SPINNING-BAND Cor,VMNS
Pressure
Throughput drop. turn. Liquid Efficieney
liquid, Speeu, R.E.T.P., Hg/theor. holdu!), factor,
tnl./m.in. r,p.ln. etn. plate llll./I'late plates/hr.
------------------
37.5-mn. long, O.B-em. diam!,tel' tube, O.4-em. wide spinning band llO
o - -b
D.I
-b
1000 2 ,[)
1000 3.S b
37.5-em. long, 0.6-!:m. diameter tubo, OA-cm. wide spinning bamllo9
1000 2.5 0.013

75-cm. jong, o.n-I·m. diametel' tube, O.32-clII. wide spinning bUild 108
1500 3. I' 0,021
550-em. long, O.fii-('.m. ditLm('ter tube, O.B-em. Wirill Rpiuning bn,nli lll
2.7 080 0.01' 0.10 1600
5.0 980 0.1.5 2000
01-CIll. long, 3.B-em. diamet.r!J' tube, 3,2-cm. wide spinning band112
4.3 o
4.3 200 0.002'
• Estimated from observed throughputs of 1 and 2 drops per second.
~ Determined with carbon tetrachloride-benzene test mixture .
• Determined with 1,2-dichloroethane-bemene test mixture.
d Determined with n-heptane-methylcyclohexane test mixture.
, Measured when operating at pressures below 1 mm. Hg.
230 A. L. GLASEUHOOK AND F. E. ,VILLIAMS
rat ely SPLlccd and ccntered, and for best operation should be mounted on
bearings. Huffman, Urey, and Pegram lOu give complete directions for the
constrLIetion and assembly of a rotating-cone column.
Mail' and Willingham IIl7 also investigated columns with stationary cones
and with rotating baskets and rotating flat
plates. Both of these modifications gave higher
H.E.T.P. values thah the rotating cones.
One of the most widely used types of labora-
tory columns employing rotating parts is the
spinning-band column.10B-112 The fractionat-
-Water
Condenser ing section of a spinning-band column (Fig. 23)
consists of a straight, vertical tube in which a
thin metal strip, whose width is only slightly
less than the diameter of the tube, rotates at
fairly high speeds, usually about 1000 r.p.m.
The available dat~ on the rectification char-
acteristics of spinning-band columns (Table
XXIV) show that this construction retains the
low holdup and low pressure-drop charac-
-RuMer teristics of empty-tr\.be eolumns, but at com-
Tuhe parable rectification efficiences the throughput
of the spinning-band column is about fivefold
greater than for an' empty-tube column of
equivalent diameter. -- The greater efficiency of
the spinning band column compared to that
of an open-tube column is due to the larger
diffusion coefficient of the vapor which results
from the turbulence created by the rotation of
the band. See Chapter I, Section IV-5.
Most investigatol'susing tubes of about 0.6-
Fig. 23. Spinning-band recti- cm. diameter prefer a rotor speed of about 1000
fying column. r.p.m.109-111; Bjorkman and Olavi 108 claim
that a speed of 1500 r.p.m. provides the beRt
operation and that a reduction of efficiency is observed at speeds below
1000 l'.p.m. Bakel', Barkenbus, and Roswell II 1 found no change in ef-
ficiency when the speed was increased from 980 to 1900 r.p.m. Birch,
Gripp, and Nathan 1l2 used a much larger dianl,eter (3.6 cm.) tube and
108 Bjorkman and Olavi, Svensk Kem. Tid., 6, 145 (1946).

lOP Kock, Hilberath, and Weinrotter, Chern. Fabrik, 14,"387 (1941).
JlD Lesesne and Lochte, Ind. Eng. ahem., Anal. Ed., 10, 450 (1938).
III Baker, R. H., BarkellbuB, and Roswell, Ind. Eng. ahem., Anal. Ed., 12,468 (Ul4.0).
112 Birch, Gripp, and Nathan, J. Soc. Chern. Ind., London, 66,33 (1947).
II. ORDINARY FHA.CTIONAL DlSTILLA'rrON 231
observed [1 peak efficiency at about 200 r.p.m. However, the peripheral

speed of the bands for optimum eft1eieney in the 0.5-cm. and 3.6-cm.
diameter columns was about the same.
Spinning-band columu:l appear to be particularly well adapted to vacuum
operation. The O.G-em. diameter, 18-ft.long column of Baker, Barkenbus,
and Hoswelpll had a pressure drop of only 0.7 mm. mercury when operating
at a head pressure of 1 mm. mercury. The Birch, Gripp, ~Uld Nathail
column 112 designed for vacuum operation had a total pressure drop of only
0.04 mID. mercury and a throughput of more than 300 mL per hour when
operating at pressures below 2.fl m~l. mercury. Bjorkman and OhlVP08
found that t.he rectification efficieney of their spinning-band column was
constant over the prOfisnrc runge 9 to 700 rom. mercury.
The cOll::;tl'\letion of the spinning-b!Lud eolunm ii:l simpler than that of
other types of rotating colUIllns. At. the top of the fractionating section
the band is fastened to a henvy wire which extends through the condenser
section and connects t.o the stilTing device (Fig. 23). The band is so light
that only an upper bearing is required, and the lower end of the band is free.
In the 5.50-crn. (18 ft.) long fractionating section of Baker, Barkenbus,
and Roswell,111 the Rpinning band was composed of a number of short
lengths connected with wire links. Birch, Gripp, and Nathan ll2 twisted
the band through 350 0 on its long center axis to provide smoother rotation.
They also demonstrated that, when the twisted band was rotated in a direc-
tion to assist the flow of vapor, the pressure drop was reduced. These inves-
tigators introduced the further modification of connecting a still-pot stirrer
to the spinning band. The stirrer aided still-pot vaporization and allowed
smooth, bump-free operation during reduced-pressure operation.
The data on spinning-hand columns are rather sparse, and the knowledge
of the effect of column diameter, clearance between the band and the tube,
and speed of rotation !1re incomplete. When such information is de-
veloped, it is believed th~t it will be possible to effect marked improvement
in the efficiency. However, even in their present state of development,
spinning-band columns are one of the most useful types of laboratory col-
umn.
The recently described'rotating concentric-tube column ll8 combines the
low holdup, low pressure drop, and high rectification efficiency characteris-
tics of film-type columns with the high throughput characteristics of plate
or packed columns. In this type of rectifying section, vapor-liquid con-
tacting takes place in the annular space formed by a stationary outer tube
and a rotating concentric inner tube. At sufficient speeds of rotation the
vapor phase is thrown into turbulence, which increases the diffusion coef-
)13 Willingham, Sedlak, Westhaver, and Rossini, Ind. Eng. Chem., 39, 706 (1947).
ficient, and in aecordance with theory, allows the vapor velocity or through-
put to be increased without decreasing the rectification efficiency.
The column studied at the National Bureau of Standards u3 used a 7.658-
cm. J.D. stationary tube and a 7,440-cm. O.D. rotating tube. The annular
space was 0.109 em. wide and the rectifying section was 58,4 em. long.
The column assembly is shown in Figure 24.
(A) Motor assembly. (AI) Electric motor, 110 v. a.c.,

3450 r.p.m. (A2,A3) Pulleys, A-section V-belts (3, 4,
and S-in. diam. pulleys used in different combinations to
vary speed of rotor C2). (A4) Shaft for driving rotor.
(A5) Ball-bearing assembly of shaft A4. (Bl Head as-
sembly, Pyrex. (BI) Opening for shaft A4 (also outlet
to atmosphere pressure for column). (B2) Head con-
denser, water-jacketed. (BS) Tungsten rod (No. 12
AWG), control for valve B7. (B4) Neoprene stopper.
(B5) Side condenser, water-jacketed. CB6) Glass rod to
direct flow of liquid reflux onto valve. (B7) Glass valve
for sampling liquid reflux in head. (B8) Outlet for head
sample. (B9) Thermal insulation, asbestos wool. (BIG)
Aluminum foil. (Bll) Thermometer well, AI (soldered
to AI plate fitted to glass well of head). (B12) Nichrome
heater uIlit for head. (Bl3) Metal flange (standard type
connector for 3-in. Pyrex pipe). (C) Rectifying sectioIl,
steel. (01) Retainer for ball bearing at top of cylindrical
rotor (openings for passage of liquid refiux and vapor to
and from column are shown). (02) Cylindrical rotor,
closed end, smooth surface, 2.928-in. O.D. (CS) Annular
space, 2.928-in. I.D., 3.015-in. O.D .. and 23-in. long. (C4)
Wall of rectifying section, 3.0I5-in. LD. (C5) Asbestos
paper (0.25-in, thick). (C6) Hel1ting jacket, metal with
asbestos covering. (07) Nichrome heating elements, 3
separate units, one each for top, middle, and bottom por-
tions of jacket. (C8) Thermal insulation (85% mag-
02 nesia). (e9) AI foil. (010) Retainer for ball bearing at
base of rotor (similar to Cl). (D) Pot, Pyrex. (D!)
Ma.nometer, butyl Carbitol. (D2) Tube for withdrawing
sample of material fl'om pot. (D3) Metal flange (standard
connector for 3-in. Pyrex pipe). (D4) Wall of pot. (D5)
Pot heater, Glas-Col special sleeve type. (D6) Thermal
insulatioll (85% magnesia). (D7) Al foil. (T) Thermo-
elements, copper-constantan. (TI) Thermoelement used
to determine difference between temperature of top por-
tion of rectifying section and liquid-vapor equilibrium in
head. (T2) Thermoelement. to determine difference be-
tween temperature of middle portion of rectifying section
and mean temperature of top and bottom sections of recti-
Fig. 24. Rotating con- fying section. (T3) Therlhoelement, to determine differ-
centric-tube distilling col- ence between temp~rature of bottom of rectifying section
umn."l and temperature of liquid in pot.
Operating at 4000 f.p.m. and the very high throughput of 50 ml. of

liquid per minute, this column had an H.E.T.P. of 1.61 em., 0040 roL
II. ORDINARY I~HACTIONAL DIS'l'ILLA'l'ION 233
holdup per theoretical plate, :1nd a pressure drop of 0.032 mm. mercury
per theoretical plate. Under the,;c operating conditions the efficiency
factor was 10,600 plates pel' hour, which is considerably higher than the
efficiency factor of any other type of laboratory (~olumn. The effects of
throughput and speed of rotation on the pressure ell'OP and rectification
efficiency of the rot.ating t:olll'(,lltl'ie-tuhe eolumn were thoroughly investi-
gated at the National Bureau of Standards. The data given in Figures
25-27 show that the prCHSlll'e drop inereases with increasing throughput,
z
0
;:: 3.5
I,)
w
rI)
~~
.>=
::E lJ..
::E13
~~
d. lJ..
~o
:r
I-
w!,g
~~
I.JJ tr:
!E~
::E
°Owe~~---IOO~O---L---2~O~OO--~--~~--~---J~--~---'
THROUGH-PUT IN
l!,ig. 25. Pressure drop vs. throughput for rot[lting concentrie-tube column ut
~evernJ spl'l'ds of fotl1t.iou.113
and at a given throughput increases with the speed of rotation. The

number of theoret.ical plateR decreases ItS the throughput is increased. At
a given throughput the number of plates increases as the speed of rotation
is increased. The curves of Figure 27 show that the rectification efficiency
increases markedly when the rotor reaehes a speed of about 2300 r.p.m.
This is the speed at which the vapor flow becomes turbulent.
The holdup of the rotating concentric-tube column was not determined
under actual distillation conditions. However, by means of equation (2),
the holdups were calculated for a number of different throughputs. These
values are given in Table XXV. The efficiency factor of the column under
various operating conditions is given in Figure 28.
THROUGH-PUT IN ML. (LIQUID) PER HOUR
Fig. 26. Theoretical pll\tes vs. throughput for rotating concentric-tube column at
severs.! speeds of rotation.n a
<II
IJJ
E
u
W
) I ) I I
~
_, (/)
0.. (!)
_, >=
Z
«
l)
!!:
i=
w
(t: ~
@ U.
0
:I:
I-
"-
0
~z
a: ~
w
to
::>
~
:2
~
l1J
:::;:
(t:
~
SPEED OF ROTATION IN REVOLUTIONS PER MINUTE
:Fig. 27. Theoretical phLttlS vs. speed of rotation fat· rotating concentric-t.ube
eolumn at several values of throughput.1l3
II. OHDIX.\.HY l!'IL\C'l'IONAI, DISTILLA'l'IOX
A-4000
A-3000
~-B' ,,~
2000 ~ ~'~A-~,I~OO~O~--J-'-----------
0/ F A-ZERO-=-
~ H.., §=~" I+' LA-ZERO
o 1000 2000 3000 4000

THROUGH-PUT IN ML (LIOUID) PER HOUR
Fig. 28. Eflieiency faetor n. speed of rotatIOIl for rotating concentric-tube column
at s(lvnraJ values of throughput. m
TABLE XXV
CALCULATED VALUES OF HOJ,DUP 0];' ROTA1'ING CONCEN'rRlC-'l'UBf~ COI,UMN
Holdup,
Throughput, liquid/meter length, Li;Uid holdup,
m!. liquid/hr. ml. mi. theor. plate"
1000 12,6 0.10

1500 14.5 0,14
2000 15.9 0.19
2500 17.2 0,24
3000 18.2 0.29
3500 19.2 0.36
4000 20,1 0040
a Holdup Ilssumed independent of the speed of rotation. Rotor speed 4000 t.p.m.
2. Construction of Columns
A. COLUMN TUBES
Packed columns are usually constructed of a straight tube of uniform

bore. The length and diameter are chosen to give a prescribed efficiency
and boil-up rate with a given type of packing. A packed section of uni-
form diameter and packing is most suitable for narrow-range distillands;
if the latter are of wide boiling range, 2-6 in. of the lower end of the column
may be packed with somewhat coarser packing than that to be used in the
23t1 A. L. GLASEBROOK AND F. E. WILLIAMS
major length of the packed section. For example, 3/I6-in. helices may be
used in the lower section and 3/a2 in. in the main section. To avoid use of
two sizes of packing, the lower end of the tube for a distance of 2-6 in.
may be enlarged to have a cross-sectional area approximately twice that
of the main section. (A 2.1-em. LD. tube has nearly twice the cross-sec-
tional area of a 1.5-clTI. tube, for example.) Vapor of wide boiling range
entering the bottom of the column is rapidly fractionated into a higher-
boiling reflux and a narrow-boiling vapor, which proceeds to the upper
rectifying zone. The higher-boiling reflux would tend to overload and flood
the lower end of the column if the packing and cross-sectional area were
uniform, whereas with the coarser packing on the enlarged section at the
bottom, the vapor and liquid capacity is greater and flooding is prevented.
Figure 12 shows a design incorporating an enlarged section at the bottom
and in addition two vapor disengaging zones at the top of the column to
prevent flooding at the top.
B. PACKING SUPPORTS
Proper support for the packing is of utmost importance to ensure that

the supporting mechanism will not be a limiting factor in the column capac-
ity, and will not allow the packing to drop into the still pot. For metal
packing, a corrosion-resistant cone of screen provides a satisfactory sup-
port. The mesh of the cone should be such that the greatest dimension of
any mesh is slightly less than the diameter of a particle of packing. An-
other type of support consists of a cylinder of wire screen, closed at one end
by a cone or spherical section of the salTIe wire screen. Either the cone or
cylindrical type provides an area for liquid and vapor flow that is greater
in area than the cross-sectional area of the column, providing the column
is enlarged in the region where the cone or cylinder is to be located. A
Rlight but sharp constriction in the column wall will provide adequate sup-
port without offering significant impedance to vapor flow. If a section of
coarse packing is used at the bottom of the column, a second cone may be
set on top of .the coarse packing to act as a secondary support for the finer
packing of the main part of the rectifying section.
Noncorrosive packing (glass helices, etc.) may be supported all a COlle of
platinum screen (poisoned with hydrogen sulfide or otherwise to eliminate
catalytic activity) in the same way as metal packing. The relatively ex-
pensive platinum cone may be eliminated by using an enlarged section at
the bottom of the column in which a glass cross is sealed. A shallow layer
of eoarse glass spirals rests on the cross. .Progressively smaller diameter
spirals are then dropped into the column until a layer deep enough to hold
the finer main packing has been formed. The coarser spirals tend to break
II. OHDlNAHY l~llAC'l'IONAL Dlf>'l'ILLA'i'ION 237
up on heating and cooling which may eventually necessitflte l'ppaeking of

the column.
In the usual preflooding operation, column packing is subjected to force8
which tend to loosen the packing, particularly small part.icles neal' the top
of the packed section. To minimize this action of preflooding and prevent
packing particles from being carried into the condenser section, a tight. roll
of wire screen may be constructed from a strip 1 in. wide and of sufficient
length to make the resulting roll fit fairly tightly in the top of the column
against the packing. Another method consists in placing a circular di.;;e
of wire screen, having It diameter equal to the inside diameter of the eolumn,
on top of the packing, and forcing several broken eircles of wire down
against the screen. The diameter and circumference of the broken circles
should be such as to allow them to make a spring fit against the inner walls
of the column. Glass spirals and other inert packing may be held in place
by a layer of progressively coarser spirals placed in an enlarged section of
the eolumn above the packed section.
C. PACKING PHOCEDURE
Packing:;;, such as balls, Raschig ring::;, Berl saddles, ete., whieh have no
tendency to form eiuste1'8 or groups by intertwining will pack properly if
the individual pal'tieies are dropped into the column singly or a few at a
time. Single and multiple-turn helices, on the other hand, form large clus-
ters by intertwining, and must accordingly be separated and dropped into a
column singly, or in clusters of only a few rings, if the packing is to he ef-
ficient and the column easy to operate.
Metal helices of the sizes used in laboratory columns attain their maxi-
mum efficiency only when they are properly introduced into the column
and tamped to eliminate loose clusters and voids in the finished packed
section. A method which works well hail been described. lH A 6-in. glass
funnel is placed in the top of the column 1;0 be packed; then a piece of \vire
screen having a mesh which will pass the helices is bent into a rough cup
shape and set into the top of the funnel. Small clusters of helices are
dropped into the mesh cup and, by properly vibrating the screen, the cluster
gradually breaks up into single helices or clumps of only a few particles,
and drops into the column. When enough helices have been put in the
column to cover the top of the supporting cone, tamping should begin.
For short columns up to 3 ft. in length, a lis-in. welding rod may have a
metal cross attached at the lower end; the arms of the cross should be
slightly shorter than the column diameter to prevent jamming of the tamper
114 Hall, H. J., and Jonach, Symposium on High-Temperature Analytical Distillation.

American Petroleum Institute, Meeting at Chica.go, Illinois, Nov. 11, 1946, p. 42.
238 A. L. GLASEBIWOK AND F. E. WILLIAMS
by the helices. The tamper rod passes up through the stem of the funnel
and the screen cup.
For longer columns, the tamper may consist of a brass block cut in the
shape of a 4- or 6-sided pyramid. The major dimension of the base should
be somewhat less than the column diameter. A small hole drilled through
the apex of the tamper provides an attachment point for a piece of No. 26
Nichrome wire to operate the tamper. The upper end of the wire may be
attached to a spool on which to wind the wire as the required wire length
diminishes during the packing operation. Occasional examination of the
tamper should be made during the packing operation so that any helices
which may have collected on the wire may be removed.
The fi'agile nature of glass helices prohibits tamping. Dust and broken
helices should be removed by screening. This is most easily accomplished
by construct,ion of an all-metal sieve, the bottom of which is made of a
metal plate slightly thicker than the diameter of the helices to be screened.
A large number of holes slightly smaller in diameter than the outside diam-
eter of the helices should be drilled in the bottom. This type of sieve 1l6
will remove dllst and broken helices, but will not allow a complete helix
to hook through a hole as might occur with ordinary wire sieves of suitable
mesh. Small clusters of glass helices break up into smaller clusters 01'
single units during the sifting, and may then be dropped. into the packed
zone.
D. COLUMN INSULATION
Because of their large ratio of surface area to volume, laboratory frac~

tionating columns require adequate insulation to minimize heat loss, re-
duce internal reflux, and increase column efficiency and vapor capacity.
Internal reflux resulting from excessive heat loss may under some conditions
bring about a slight increase in fractionating efficiency, but such results are
probably fortuitous to a large extent. Poorly insulated columns are gener~
ally less reliable in operation and lower in fractionating efficiency over a
wide temperature range than are columns properly designed to minimize
heat losses.
Two general methods are available for preventing heat loss from frac-
tionating columns. The first and simplest of these consists in covering the
column with a layer of material of low heat conductivity, such as 85%
magnesia, glass wool, asbestos, or mineral wool. For very large commer-
cial columns, this method is commonly used and is entirely satisfactory;
for small columns of laboratory size, simple insulation is not effective if the
temperature difference between the inside of the column and its surround-
116 Cook, N., private communication.
II. ORDINARY FllACTION,\L DIS'rILLA'l'IO~ 239
iugs is large. Increasing the thickness of the insulation does not reduce the
heat losses indefinitely ldter a eertain depth of insulating mat.erial has been
used. On very small diameter columns of the order of It quarter of an
inch, increasing the thiekness of the insulation beyond a certain value will
lead to an increase of heat loss because of the large surface area. of the insu-
lation a.vailable for radiation.
In Table XXVI is a summary of heat losses (neglecting end effects) to be
expected for a column 1 in. in diameter and 1 ft. long when insulated with
[},5% magnesia of the thickness shown.
TABLE XXVI
HEA1' LOSSE,~ THROU[m 85% MAGNESIA, AND EQUIVAI,ENT CONDENSATION
Ins(t1.tion thickness
One inch Twoinohes
Heat 108H, Ml. Heat loss, lvU.
AT' = cal./hr. condensed/hr. cul./br. condensed/hr.
55·, benzene ......... 4,800 54 3450 42

85", toluene, ........ 7,600 lOt 5400 73
151 0 , p-cymene .. , ... 13,800 238 USDa 170
AT* = normal boiling point minus room temperature (25 ·C.). All caloulations based
Oil heat-loss dl1tu from J ohns-l\lanville bulletins.
In striking contrast to the insulating efficiency of low-conductivity lag-

ging materials mentioned above is the extraordinarily high insulating ef-
ficiency of vacuum iackets. These consist of two concentric glass tubes of
suitable relative diameters sealed one to the other with a glass-to-glass seal
at each end. Through a small tube sealed near one end of the outer tube,
ail' is removed from the annulus down to a pressure of 10-5 to 10-6 mm.
of mercury. During the evacuation, the whole jacket is baked at high tem-
perature (300-400 0 0.) to remove occluded gases and moisture to ensure a
permanent vacuum. Although a jacket of the above construction has good
insulating qualities, its efficiency in this respect can be greatly improved by
silvering the inner walls of the jacket or by placing a highly reflecting radia-
tion shield in the annular space during the fabrication of the jacket. In
either case, the annular space is highly evacuated. In order to observe
the packing through the jacket, a strip may be left unsilvered, or in the case
of jackets with an inner radiation shield, small holes about 1/4 in. in diam-
eter are punched near the top and bottom of the shield. Vacuum jackets of
this construetion have been used as separable insulation for a fractionating
column by slipping the column into the jacket and closing the small an-
nular spaces between the column and each end of the jacket with solid insu-
lation, but the more common practice consists in making the inner tube of
240 A. f,. GLASEBIWOK AND F. E. WILLIAMS
the jacket serve as the column proper. Extensions of the inner member
are used to attach the still head and still pot by suitable joints.
A column of the above construction would be usable, but the permissible
temperature differential between the inner and outer tube would be small,
at least for a column of signific:ant length, if intolerable strain and breakage
were to be avoided. To allow for the differential expansion between the
inner and outer members of a vacuum-jacketed column, an expansion
member is made part of either the inner or outer tube. When the expan-
(Not satisfactory
on EwellpClcklld
~J
Fig. 29. External oxpansion section Fig. 30. Internal expansion section for
for vacuum-jacketed column. vacuum-jacketed column. 11.
sian element is part of the outer tube, one or more short sections of this
member are made into the form of a glass bellows; the number of such
bellows depends on the length of the jacket, a larger number being required
with a long jacket. When the expansion member is part of the inner tube,
it consists usually of two or three small-diameter glass tubes in the form of
coils attached to the inner tube at the lower end to fOIm parallel vapor
lines feeding the larger tube. The flexibility of the coils allows the neces-
sary expansion of the inner tube to take place without placing undue strain
on any part of the ColUlhn. Difficulties in producing satisfactory expansion
elements have limited vacuum-jacketed columns to diameters of about 1
II. ORDINARY FRACTIONAL DISTILlu\'l'ION 241
in. and to lengths of about 12 feet. The two types of construction are
shown in Figures 29 and 30.
In an earlier design,116 expansion elements were avoided by constructing
the inner member of the jaeket of fused quartz ancI the outer jacket of
Pyrex, the union hetween the two ends being a graded glass s(ml. The ex-
tremely low eoeffir:ient of expansion of quartz permitted the inner tube to
be heated to a high tempemture without putting excessive strain on the seal
between the inner and outm' tubes. The high cost of the quartz inner
tubes and the limited lengths and dial"neter::; available have caused this
design to be abandoned.
The comparative insulating qualities of several designs of vacuum jack-
ets have been thoroughly investigated by Podbielniak.H s Figure 31 shows
the type of jacketfl he studied, and Figures 32-34 summarize his results.
It is apparent that the insulating efficiency of the several designs varies
widely, but these results when compared with those of Table JL"'{Vlshow
the deeided superiority of vaeuum insulation. For fractionation of liquids
boiling below room tempemture, vaeuum jackets are the only satisfactory
type of insulation; in this application they function to prevent superheat-
ing of the column. In columns opemting above room tempemtufe, vac-
uum jackets minimize heat losses. The most efficient vacuum jackets lose
some heat (largely by radiation) when liquids of boiling points far removed
from ambient temperatures are distilled; to minimize these losses still
further, a small heater is inserted into the evacuated spRee between the
column proper and the radiation shield.ll7 The energy dissipated in the
heater (if properly adjusted) just compensates for the residual radiation
losses of the column.
Willingham and Rossini llB enclose long vacuum-jacket columns with a
metal tube on which are wound electric heaters. A layer of 85% mag-
nesia and finally a layer of aluminum foil cover the heaters. Nat on1y does
this method reduce heat losses to an extremely small value, but it serves to
bring the outer surfaee of the vacuum jacket to a temperature close to that
of the inner tube, and consequently greatly reduces the strain on the ex-
pansion members of the column.
For larger laboratory columns made of either metal or glass the second of
the general methods is used for preventing heat loss. This method con~
sist.s in maintaining the region immediately adjacent to the column at the
vapor-liquid temperature in the column by circulation of suitably pre-
heated air, vapor, 01' liquid in a jacket surrounding the column, 01' oy dis-
sipating energy in an insulating medium around the column. To be com-
116 Podbielniak, Ind. Eng. Ghem., Anal. Ed., 5, 121 (1933).
117 Podbielniak Company, Chicago, Illinois.
118 Willingham and Rossini, J. Research Natl~ Bur. Standards, 37, 15 (1946).
on
V
--$-/
II
1--
""f - --
i
~~I
_~I i
I T
I
"!
Ij 1
i:
! ::>
v
~
'I z
9
I-
/il "'
0
11.
S
\, "<
It
~
I
,I I
i
,. ! (l
--~
_1_ _- -- .
cj\-~I
'--
r--~I
~I
11'\ ""
~
r-----..
P')
~J.'OI\I1'11"(
l'--..
N
HI 3IfnJ.VIIldI't3J.
0
Q
II. OIWINAT{Y FRACTIONAL DISTILLATION 243
2±4 A. L. GLASEBROOK AND F. E. WILLIAMS
pletely effective, either of these methods should establish theoretically a

temperature gradient throughout the length of the column on the outside
40 BO '20 160 200 240 280

40 BO' 12~ 160 200 240 280 AVERAGE DIFFERENCE iN TEMPERATURE BETWEEN
AVERAGE DIFFERENCE IN TEMPERATURE BETWEEN INSIDE OF YACUUM JACKET e. SURROUNDING. ·C.
INSIDE OF VACUUM JACKET a SURROUNDING. ·C.
Fig. 32. Efficiency of column insulation. h. Fig. 33. Efficiency of columI! insula-
Jackets 1, 2, 5, 13, and 13A of Figure 31. tiollS,llo Jackcts 6, 11, 12, and 15 of Fig-
ure 31.
o 40 ao 12Q 160 200 240 <90

AVERAGE DIFFERENCE IN TEMPERATURE BETWEEN
INSIDE OF VACUUfo,! JACKET a SURROUNDING. 'C.
Fig. 34. Efficiency of column insulation.1l6 Jackets 3, 4, 7-

10, and 14 of Figure 31.
exactly equal to the temperature gradient prevailing inside the column

caused by fractionation and to some extent pressure drop; practicallimi-
tations will permit this condition only to be approximated.
II. OHDINTItY FHACTION.\L DIS'fII.I,A'l'roX 245
Seyeral urrangemtmts have been used to provide a region of suit,able

tempemt,llI'c adjacent to the column. Kistiakowsk.:i 1lG enclosed the packed
section with a jacket of somewhat greater diameter t,hrough whieh a liquid,
heated to the propel' temperut.ure, was eirculated. For short columns, this
design is probably satisfactory at temperatures not too far remuved from
the ambient temperatlll'e; at high tempeI'atures heat los~es from the jacket
would be exeessivc, and insub.tion would be necessary. Some tempem-
ture gradient would exist as n result of natural eooling of the liquid as it,
moves up the jaeket; on tall columns, this typt\ of construction would
probably not supply the proper temperature gradient. The chief objec-
tion to this design iel the necessity for circulati.ng a hot liquid. This not
only poses a medu1nieal problem, but introduces a certain amount of haz-
ard in the event of fracture of the glal'ls jacket.
Wirlmerl~() aehiev(]!l the same result automatically by surrounding the
eoIumll with a jacket 80 arranged that the vapors of the materinl being dis-
tilled were fOl'eeu to pass to the top of the jacket, reverse their direction
down t.he inside of the jacket, and thence up through the fractionating
column proper (see Fig. 3). This design requires t:onsiderably greater heat
input to the still pot to attain ~L given throughput in the colllIlln proper
than would bE) required for a vacuum-jaeketcd column, or for other de-
signs in which heat losses are eompensated independently. The Widmer
column has greater dynamic holdup for the sa,me reason, and hence would
not separate components of a charge as sharply as other designs. The
jacket section of the column has little fraetionatillg effieiency compared to
the packed rectifying section; consequently, the temperature g1'l1dient
in the column would be upset to some extent. For short columns of moder-
ate efficiency, the Widmer column should be satisfactory.
The most widely used system, and one of widest applicability, for Illini-
mizing heat l()sses from fractionating columns eOllsists of electric heaters
appropriately placed around the column. Fenske et al. 121 used this method
in a column 52 ft. high and 1 in. in diameter in which were separated the
two isomers of diisobutylene; Boms, Coffey, and Garrard ln used it on
a Steelman column 3 in. in diameter and 12 ft. long, and it is being used on
Stedman columns (j in. in diameter Hnd 18 ft. long. 123 Electric heaters are
particularly suitable for those who build their own coluIlllls and lack the
119 Kistiakowski, Ruhoff, Smith, H. A., atlLl Vn,ugh::ltI, J. Am. Chern. SOl;. , 57, 878
(1938).
l~O Widmllf, Hclv. Chim. Acia, 7, 59 (1924).
121 Fenske, Quiggle, and Tonberg, Ind. Eng. Chern" 24, 408 (1932).
122 Borns, Coffey, and Garral'd, Symposiuill on High-Temperature Ana.lytical Distilla-
tion. American Petroleuill Institute, Meeting at Chicago, Illinois, Nov. 11, 1946, p.
26.
m Foster Wheeler Corporation, New York City.
246 A. L. GLASEBIWOK AND F. E. WILIJIAMS
facilities for building vacuum-jacketed columns of glass; their use is the

only feasible method of heat-loss compensation for large or small columns
of metal.
In order to attain the best extemal temperature gradient throughout the
length of the column, the electric heaters are put on in sections, each inde-
pendently controlled by a variable transformer. Short columns up to
about 3 ft. long operate satisfactorily with a single heater; longer columns
should have heater sections not exceeding 3 ft. in length.
When all-glass construction is to be used, the column is placed concen-
trically in a second glass tube of an inside diameter a/ 4 in. larger than the
outside diameter of the column, and slightly longer than the proposed
Fig. 35. Cross section of column with compensating jacket.
packed section. Four to six lengths of asbestos cord are then fastened
longitudinally on the jacket at equal angular spacing to prevent contact of
the heater wires and jacket, and to minimize slippage of the heater element.
Sufficient resistance wire or tape of the proper current-carrying capacity
and resistivity should then be wound on the jacket to carry the full load
of the variable transformer to be used with the jacket (see Table XXVII).
Adjacent turns should be 1/2 to 3/4 in. apart. The variable transformer
used for this purpose usually has a maximum rating of 5 amp. at 110 v.;
the resistance should be accordingly about 22 ohms. The full heater dissi-
pation may not be required, but by using 22 ohms there is no danger of
exceeding the rating of thc transformer in the event the full voltage is
thrown on the winding. A second glass tube of suitable diameter for a
snug fit is then slipped over the jacket for protection if full visibility of the
packing is desired; otherwise, a layer of 85% magnesia may be put over
the winding to reduce power requirements, and to minimize the effect of
drafts on the regulation of the jacket temperature (see Figs. 35 and 36).
247
When several jaeket sec:tions of the ahove GOllstrnetioll arc used OIl II col-
umn, a Transite plate drilled to slip onto t,he column with sllug fit should
be placed between adjacent iiectiolls to prevent circulation of hruted ail'
between the sections.
HEATER WIRE
LEADS
~.$OIf-~-METAL RING TO HOLD

I j JACKET THERMOCOUPLE
"-.._ ASBESTOS TAPE HOLDING
COLUMN THERMOCOUPLE
JACKET
HEATING JACKET
- 25 MM. 0.0. COLUMN
Fig. 3B. Vert.ical view of hO!1t-compens(l.tillg jacket.
To indicate the proper power input to a given heater section, thermo-

couples hooked up in opposition offer the most convenient method. If
glass jackets are being used, one thermocouple is silver-soldered to a circu-
lar band of Nichrome ribbon about 3/Sto 1/2 in. wide, and slipped into the
jacket to a point equidistant from the ends. Spring action of the ribbon
will usually hold the band in place. A second thermocouple is silver-
soldered to another strip of Nichrome and fastened to the column, directly
opposite the jacket thermocouple, by means of asbestos tape and fine-

gage Nichrome wire. The foul' leads are brought out the ends of the jacket.
If a copper-constantan thermocouple is used, the two constantan leads are
silver-soldered together, and the copper leads run to a galvanometer of
fair sensitivity. When the heat. input to the jacket is cOl'l'ect, the galvanom-
eter will give zero reading. One galvanometer can be used to indicate
t.he proper heat balance in sevel'lL1 jacket.s on one or more columns by t.he
use of simple rotary switches of t.he type found in radio circuits.
Prevost 124 has designed an automatic control to regulate the heat input
to jackets of t.he above construction. Two additional windings of fine
wire having a high coefficient of resistance are wound, one directly on the
column and the other located on the inside surfaee of the jacket; they
function as two arms of a Wheatstone bridge. Any unbalance of the
bridge caused by a difference of temperature of either winding (change of
resistance of either winding) actuates a relay through a small thyratron
tube, and increases or decreases the power to the heater winding to restore
t.he heat balance. Control to 0.5°C. is claimed for the circuit.
Variations of the jacket design described previously haNe been used satis-
factorily for both glass and metal columns. Fenske et al,121 used a 2-in.
thiclcness of 85% magnesia directly on the column; resistance tape was
then wound onto the magnesia, and a second layer of magnesia covered the
winding. Thermocouples were placed on the column at the middle of each
section, and in the first layer of magnesia at points directly opposite the
column thermocouples.
Borns, Coffey, and Ga1'rar<1122 modified the design for a Stedman column
3 in. in diameter. A layer of mica was wrapped on the column to serve as
electrical inKula~ion for a Nichrome winding wound directly over the mica.
This winding was used only during the start of a distillation to heat the
column and to reduce the time required for refiuxing to start. A fils-in.
layer of magnesia covered the first heater windings, followed in turn by
copper sheet, mica, a second Nichrome winding, 1 in. of 85% magnesia, and
finally 2.5 in. of glass wool. Thermocouples were located on the column
and ill the first layer of magnesia as in Fenske's design. Marschner and
Croppe1'126 insulated their high-efficiency columns with 1.5 in. of shredded
magnesia inside a steam-heated jacket. Steam pressure was automatically
controlled to minimize heat losses and to prevent excessive column temper-
ature. Peters and Balcer l26 iacketed their column with an electric heater,
and provided, in addition, an electrically heated stream of air to the bottom
Prevost, C., Thesis; Pennsylvania State College, 1948.
124
m Mar!!chncl' and Cropper, Symposium on High-Temperature Analytical Distillation.
American Petroleum Institute, Meeting at Chicago, Illinois, Nov. 11, 1946, p. 35.
1M Peters and Baker, T., Ind. Eng. Chern., 18, 69 (1926).
I!. ORDINARY li'RAC'l'IONA1 DU:i'l'ILh,\'rION 249
of the jacket as a meum of estahlbhing the proper tcmp(wuturc gl'lldient

along the packed section.
Either vacuum jacketing or heat-loss compensation provides satisfactory
column insulation. For small-diameter columns not, exceeding a length of
about 10 ft., vacuum jacketing may be prefemble. Actual experience has
shown that somewhat lower boil-up rates can be maintained in tt \'twuum-
jacketed column. On larger-diameter columns (> 1 in. in diameter) and
for all-metal columns, heat-loss compensation is the only satisfactory de-
sign. All-metal vacuum-jacketed columns hnve been built 127 and the
jacket given a long period of evacuation to degas the metal. When further
degassing of the metal under opemtion rai~es the jacket pressure to 10-,1
mm., the evacuation is repeated. This procedure introduces complexities
scarcely justified for the little nclditiolll1l efficiency and operating smooth-
ness achieved.
IIi. CONDENSER SECTION

1. General
Vapors from the i:!till pot pass up through the column proper, or recti-
fying section, where tl,ley are enriched in the more volatile components,
and emerge into the condenser section at the top of the column to be totally
or partially condensed to a liquid. If the vapors are partially condensed,
condensate automatically returns to the top of the rectifying section as
reflux, and the remaining vapor passes into an auxiliary condenser in whieh
the vapor is liquefied and withdrawn as product. If all the vapors emerg-
ing from the rectifying sections are liquefied in the condenser section, the
system operates as a total condenser, and a portion of the condensate is
withdrawn as product. Both types of condensers are operated as total
condensers while the column is being prefloocled and brought to equilibrium
hefore product collection begins.
Historically, there has been some controversy on the relative merits of
partial versus total condensers. Peters and Baker126 and later Podbiel-
niak 128 favored the partial condenser because the unavoidable holdup in the
total condenser works against the sharpest separation. Podbielniak's
Hyper-Cal column uses a total condenser.128 Young129 contends that no
plall.t column can be thoroughly efficient without some form of partial
condenser (clephlegmator). Robinson 130 argues that the usual type of COll-
127 Smith, V. C., Glasebrook, Begeman, find Love]l, Ind. Eng. Chem., Anal. Ed., 1'7,47
(1945).
12.9 Podbielniak, Ind. Eng. Chem., Anal. Ed., 5, 127 (l!l33).
129 Young, S., Distillation Principles and Processes. London, 1922.
130 Robinson, C. S., Elements of Fractional Distillation. New York, 1930, p. 88.
denser is peculiarly unadapted for good fractionation. H.ecent.ly, Birch

ct. al.'81 published the design of a low-pressure column using a partial con-
denser to which was ascribed a part of the fractionating efficiency. Leslie 132
has shown mathematically that fractional distillation is superior to frac-
tional condensation.
Regardless of the theoretical merits of partial condensers (and these are
not imposing), their use on modern laboratory columns for high-tempera-
ture distillations is very limited. A partial condenser must be supplieel
with a carefully metered quantity of coolant, or with a suitable quantity of
coolant at a closely controlled temperature, in order to maintain the proper
reflux ratio while a product of constant boiling point is being collected.
During breaks, the rate of circulation of the coolant or its temperature must
be closely watched and changed as necessary to maintain flow of product
to the receiver.
Total condensers, on the other hand, need only be supplied with enough
coolant to condense all vapors entp.ring the condenser section; excess cool-
ant does no harm, and is of no significance for laboratory columns. If the
heat picked up by the coolant is to be used for calculating the boil-up
rate, the rate of flow of the coolant must be closely controlled during the
time the measurement is being made. This applies to the partial condenser
system also; here, two rates will have to be determined, one of which is by
temperature difference; and product rate directly. A portion of the COll-
densate can be withdrawn continuously, or all the condensate for a frac-
tion of unit time can be withdrawn to establish the proper reflux ratio;
all the vapor can be withdrawn for a fraction of unit time as product and
thus establish the reflux ratio. The simplicity of control of condensation
and produet collection with a total condenser has made it the preferred
system. When several columns are attended by a single operator, as
frequently occurs in large laboratories, the impracticability of partial con-
densers is immediately obvious. '
The condenser systems of a fractionating column should be construct.ed
with sufficient cooling surface to condense all vapors entering the con-
denser. It is assumed that the coolant cireulating through the eondenser
is significantly colder than the lowest-boiling material to be condensed.
Inasmuch as many types of laboratory columns have to be preflooded to
achieve their maximum efficiency, the condenser system must be capable
of condensing the abnormal supply of vapor during the preflooding opera-
tions; if it does this, the condenser will be more than adequate for the
normal vapor load. In addition to condensing the vapors entering it, the
condenser system of laboratory columns is provided with a partitioning
131 Birch, Gripp, and Nathan, J. Soc. Chem.Ind. Lrmdon, 66, 33 (1947).
)3~ L(lSlie, Motor Puela. New York, 1923.
n. ORDINAnY I1HAC'l'IONAL UlR'l'II,LN[,ION 251
device by which a regulated fraction of the conciens[tte or vapor is with-

dra\vn as produet, and the very important "reflux ratio'" iF; established.
2. Total Condensers
For ease of construction, ruggedness, low holdup, large vapor capacity
with no tendency to flood, and low pressure drop, a form of the modified
Liebig condenser is frequclltly used. Somewhat gren,ter condensing sur-
face for a given length of condenser can be obtained by introducing a "cold-
finger" into the center t.ube. This design has a smaller liquid ca,pacity
than the simpler form, and may more readily discharge reflux overhead into
the fmction reeeiver during the preflooding operation. Another design
eonsists of a plain glass jacket with a "cold-finger" in the center. This
may consist of a plain glass tube through which the coolant is circulated,
or may be a glass or metal eoil of' small-diameter tubing. The turns of the
coil must be sepamted far enough to prevent the formation of a bridge of
condensate between adjacent turns of the coil if condenser holdup is to be
minimized. The simple Liebig type of condenser in a vertical position
probably has the lowest holdup of all the designs; this supposition has not
been verified, however.
Physical dimensions of condensers are not critical; 150 sq. cm. of con-
densing surfaces (30 cm. long by 1.6 cm. LD. standard Pyrex tubing) for
high-efficiency columns having a maximum boil-up rate of 1560 ml. per
hour of hydrocarbon is satisfactory.133 If there is a small differential be-
tween vapor and coolant temperature, or if the vapors have high latent heat
somewhat greater condensing area may be required.
As mentioned above, a portion of the vapors entering the condenser sec-
tion is withdrawn as product, either as liquid or as vapor which is subse-
quently liquefied. Some form of' partitioning device regulates this process.
How well this is done determines to a marked c;\."tent the success of a frac-
ti.onation.
Two general methods are used to control the product rate. One con-
sists of a flow-regulating device, which allows a steady stream of condensate
to flow from the reflux condenser to the product receiver, excess condensate
flowing back to the rectifying section as reflux; the other consists of a
mechanism by which all the condensate can be diverted either to the prod-
uct receiver or to the rectifying section of the column. This mechanism
is operated by a timing device. If the timing deviee, for example, holds the
column on total reflux for 59 seconds, and then on total takeoff for 1 second,
out of every 60 seconds, the time ratio is 59: 1 and theoretically the refllLx
ratio is 59: 1. (The boil-up rate is assumed to be constant during the full 60
133 Willingham and Rossini, J. Resea1'l:h Nail; Bur. Standards, 37, 15 (1946).
252 A. J,. GLASFlBltoOK AND l~. E. WILLIAMS
seconds.) This system should give the same reflux ratio regardless of
boil-up rate; however, holdup in the diverting mechanism, time lag in the
operation, and condensation in parts of the column other than the condenser
tend to give a higher reflux ratio tha.n the time ratio indicates. A modifica-
tion of this system uses a magnetically controlled valve to close the en-
trance to the condenser and open an exit valve and thus direct the vapor
into an a.uxiliary condenser where it is liquefied.
I"ig. 37. Details of condenser section and product

SECTION A-A valve. 107
g
,... CA) Section showing opening in jacket of COIl-
o denser.
:g (B) Portion cOIlnecting condenser with grouIlll-
glass valve.
9 (C) Tungsten rod, No. 12 A WG.
O'D~
(D) Male ground joint, standard taper 12/20.
B RIGHT SIDE (E) Portion having extra-heavy wall.
VIEW OF (F) Apron to direct liquid from the condenser
B
to the thermometer well.
(0) Bottom edge for sealing to rectifying tube
E (should be smooth and have plane perpendicular
to axis of thermometer well).
(H) Clearance to be not less than 15 mm.
(.1) Ground-glass valve, taper 1 in 5, small end
4 mlll., large end I) mm., length 10 mm., ground
for high vacuum. Material is Pyrex laboratory
glass.
Historically, the ubiquitous stopcock was the first device used to control
product rate from a column. It served this purpose admirably, particularly
in that its vagaries led to decidedly better methods of control. The neces-
sity for using a lubricant which sooner or later dissolved in the product, air
leakage during reduced-pressure distillations, difficulties of precise adjust-
ment, and intermittent. and irregular flow when two-phase condensates
appear make the stopcock a generally unsatisfactory device for control of
product rate. Some improvement, so far as adjustment of the rate was
concerned, was made by the development of the variable-flow stopcock,
but this did not overcome the disadvantages mentioned above.
n. OlWINARY FRACTIONAL DIP,'l'ILTA'l'ION 253
A design which eliminated in one step all the ditlieulties eneountered

with stopcocks except that of irn·gulal' flow of two-phase eondensates is
shown in Figlll'e 37.133 FigUl'P 38 gives the detnils of n micrometer adjust-
ing screw for regulating the flow of produet. 134 This design is sueh that
the ground-glass valve, ,J, cannot be forced froIIl the female section if it
Brass
=--rr I'> Ilia"
F=======::b~_j
~------.-_--
I
3 ',,' ---.--- ---~
~;+~~
_.-.._._.~ ./ Ream hole
I L._ ___ .:·;--r
3/6 ; 2.~/6~'
11
... 5/0
,
I-r--- -- , __....l
.,
_._- \ Drill 33 and top 6-'10 thread
Bross for '/a" set screw
28 threads/inch
'/8" Stainless steel rod 5" long

/,a::====::::=========:::J'!El"tlll"
'Is"
<Drill '/,s" hole deep ,,""
5- 40 thread
Fig. 38. Modified vl1lve control to use with !:ondenRer ~()ction of Figul'fl 37. [34
sticks. The valve may be readily released if thiR ommrs by earefully heat-
ing the female section with a small flame and t,hus avoid breaking the glass-
to-tungsten seal. As will be noted in Fignre 37, the condenser delivers
reflux to the plug and keeps the orifice covered with reflux at all times to
maintain a very small but adequate supply of product. No lubricant is
needed since the product acts as its own lubricant. When the valve parts
are properly constructed and ground as recommended, the valve does not
leak with the eolumn on total reflux .. Product rate is very constant so long
as the surface tension and viscosity of the reflux do not change. In the
distillation of pure hydroearbons, tests have shown the rate to remain con-
stant within 2% 01' less for periods of 16 hours or more. The design is
1M Hercules Powder Co. design ..
254: A. L. GLASEBHOOK AND F. }<;. WILLIAMS
satisfactory for both atmospheric andlower-pl'essul'e distillations because

the valve has substantially the ,same pressure on both sides. Holdup is
minimum and is almost entirely the normal holdup of the condenser and
associated surfaces.
Vacuum Jacketed Reflux Condenser
I 29142 Ground Joint
Producl Condenser
Joint
Fig. 39. Automatic liquid-dividing head. 136
The generally unsatisfactory regulation of product rate by means of

stopcocks, the difficulties of determining reflux ratios with the best systems
of continuous product collection, and demonstration of equality or slight
superiority of intermittent take-off compared to continuous take-off135
have led to wide use of intermittent take-off systems. Figures 39 and
136 Oldroyd and Goldblatt, Ind. Eng. Chern., Anal. Ed., 18,761 (1946).
II. OUDINARY FRAC'rrONAi. DISTILLATION 255
40 show two designs for liquid and vnpor take-off, respectively.l3G In the
('ase of liquid take-off (Fig. :39), a timer energizes the solenoid ,,,hieh moves
the tipping funnel over the opening of the product line to discharge the
reflux to the product cooler and receiver; interruption of the current to
the solenoid by the timer then diverts the reflux back to the column. For
vapor take-off (Fig. 40), the solenoid when energized closes the opening of
the condenser and allows the Vltpor to flow into the product condenser fot·
a given interval of time. The ratio of time OIl totall'eflllX to time on total
Vacuum Joe~.'.d Rollu.
i 29/42. Ground Jalnt - - - - ;
li'ig. 40. Automatic vapor-dividing headY&
takeoff is the reflux ratio, theoretically. Studies by Collins and Lantz l3S
of these two designs show that, the time ratio is smaller than the actual re~
fiux ratio, that the actual refiux ratio varies with boil-up rate, and that the
vapor take-off system gives a smaller reflux mtio than the liquid take-off
us Collins lind Lantz, Ind. Eng. Ghent., Anal. Ed., 18, G73 (1946).
256 A. L. GLASlmlWOK AND F. E. WILLIAMS
system. Imperfect insulation and cold reflux, requiring the condensation

of a certain amount of vapor to heat the reflux to column temperature, are
hugely responsible for t.he difference noted.
1 i " -2
'MALE er.LL' I
r--!
JOINT
Fig. 41. LiqUid-dividing head lInd cOlldenaer scetion for analytical dis-
tilling column of Figure 12.137
A careful examination of Figure 39 wiII show that the reflux entry port is
closely adjacent to the product discharge line. This region of relatively
high temperature is fi11ec1 with vapor. Some of this vapor diffuses into the
product line and increases the product rate. For the ~ame reason, a certain
amount of product will collect during the initial equilibration of the column
under total reflux. A suitable valve in the product line as close as possible
to the product port will hold this flow to a minimum during the period of
total reflux (see Fig. 41). Figure 41 137 is another design of liquid-dividing
m Hall, H. J., a.nt! Jont1ch, Symposium on High-Temperature Analytical Distillation,
AInerican Petroleum Institute, Meeting at Chicago, Illinois, Nov. 11, 1946, p. 42.
II. OHDI~ARY FRA(;'l'IONAJ. DISTILLATIoN 25i
head, similar in principle to Figure 3[1. A somewhat different design of

liquid-dividing head is shown in Figme 42. 13S Drainage of the condensate
12
5B 1817
1 24/40 -_
o
o
E)(PANSION
'"
BELLOWS. NO TAKE-OFF VALVE.
LIQUID TRAP. LAPPE:D IN TOil.
FLAT-ANGLED SEAT,
~~D~nlt~'Wdr MORE
THAN I NIL MA Y BE.
LOWEST POINT OF TRAPPED HERE.
DRIP TIP MUST BE
OPPOSITE SIDE
TUBE.
!l'12,2MM.BORE
2~teSWv~'r.\af>S
LAPPED IN TO
FLAT -ANGLED
SEAT. CLEARANCE
FROM SEAT WHEN
OPEN-' TO 2 MM.
PLUG MUST FIT
CLOSELY BETWEEN
CENTERING TIPS.
Fig. 42. Condenser section and liquiu-dividing head (1111 dimensions in millimet(ll's).l3B
138 Kieselbach, Anal. Chem., 19,815 (1947).
258 A. L. GIASfiJBROOK AND F. E. WILLIAMS
from the condenser to the valve is positively directed by the drip tip; the
short extension in the bottom of the valve plug produces a "pouring rod"
effect by virtue of the surface tension of the condensate to aid in carrying all
condensate into the receiver during the time the valve is open. Excessive
Fig. 43. Gomd condenser section and reflux divider.140
boil-up rates will probably overload the product valve unless the valve size
is increased considerably. Leakage of condensate can be kept to a very
low value jf the valve is properly ground, and given a final lapping with
jeweler's rouge.
To avoid inadvertent heat losses the product valve and condenser are
located in an enlarged section of the vacuum jacketed Hyper-Cal fraction-
II. ORDINAHY FItAC'fIONAL DISTILLA'fION 25D
ating eoluu1I1 just above the top of the pad.::ing. 139 Some holdup prevails
around the valve plug. At high boil-up mtes, the amount of holdup is not
significant, but at very low boil-up rates as used at reduced pressures (if
the pressure drop is to be kept low), the holdup l'equires the ratio of time
on total reflux to time on total take-off to be mueh larger than the clc8il'ed
reflux ratio.
For controlling reflux ratiol'! of 30:1 and below, the design shown in Fig-
ure 43 was developed and designated the Corl1d head, COnstant RAtio
heaD.l4fJ Several condensation strips of different [treas are provided on the
eondenser surface to vary the fraction of the total condensate dropping off
anyone tip. The condenser surface can be rotated to bring any drainage
tip over the produet. discharge port. i drainage from the other tips returns
to the column as reflux. During the construction of the condenser, the
area and numher of s1;1'ips are set, [Lnd hCIlce the seveml possible reflux ratios
are pre-established. A reasonably good estimate of the amount of vapor
reaching the condenser can be made by maintaining the boil-up rate con-
stant by appropriate means, collecting the condensate from each drainage
tip in turn for a llleasured time interval, and dividing the total weight or .
volume of material eollceted by the total time required. This will give the
rate of boil-up at the condenser, 'rests have shown that the l'efhnc ratio
established by anyone drainage tip will vary somewhat with the amount of
vapor reaehing the condenser. Product rate is also increased by vapor dif-
fusion into the pro duet line; this is probably more pronounced at the
higher reflux ratios. The Comd head has the advantage of simplicity in
that an auxiliary timer and solenoid are not required for its operation.
At reduced pressures, the ground-glass joint h!lS a tendency to bind when
rotated; this eondition may lead to breakage if the force applied is too
great.
3. Partial Condensers
Partial condensers have been almost entirely superseded by total con-
densers on laboratory columns. Doran,w Kistiakowsky,142 amI Laughlin,
Nash, and Whitmore H3 have described columns using partial condensers.
Birch et aU44 used a partial condenser in a low-pressure spinning~band
column. An extension of the column propel' is surrounded with a short
la9 Podbielniak, Incorporated, Chicago, Illinois.
1'0 Lloyd lind Hombacher, Anal. Chelll., 19, 120 (1947).
141 Doran, Anal. Chem., 5, 101 (1933).
u~ Kistiakowski, Ruhoff, Smith, H. A., !md Vaughan, J. Am. Chem. Soc., 57, 878
(1935).
143 Laughlin, Nash, and Whitmore, J. Am. Chem. S()c., 56,1396 (1934).
144 Birch, Gripp, and Nathan, J. Soc. Chem.Ind., London, 66, 33 (1947).
2(10 A. L. GL.\SEBROOK AND F. E. W[LLL\MR
jnckf~t. through which nil' is blown to condense the vapors partially, and
to regulate the product rate. Inasmuch as the reflux must flow down the
column walls in a thin uniform Iilm, the partial condenser offers an ad-
vantage in this type of column. For low-temperature columns, partial
condensers offer simpler control of product take-off rate than total con-
densers, chiefly because the distillate is collected as a gas, not as a liquid.
The best systems so far devised for establishing reflux ratios give only
approximate values. Unavoidable condensation of V[1POl' in inaccessible
parts of the condenser system-product eolleeting systcm and internal
reflux caused by imperfect insulation tend to make the aetual reflux ratio
greater than the measured value. For example, one group of six vacuum-
jacketed perforated-plate columns had internal condensation rates varying
from 170-2GO ml. of n-heptane per hOllr; the internal reflux was independent
of the boil-Up rate, as would be expected. 145 Less efficient insulation would
allow greater internall'eflux. As a practical mat,ter, uncertainty regarding
the reflux ratio is not too important. If a given column will produce a
product of satisfactory purity at some reasonable reflux ratio, its absolute
value is not generally of major importance.
IV. STILL POTS

The still pot of a batch distillation column is the part of the column into
whieh the distilland is charged. It iB attached to the lower end of the frac-
tionating section. Heat is applied to the still pot from a suitable source;
the still pot transmits the heat to the distill and and brings about evapora-
tion at a more or less uniform rate. The vapors so generated pass up
through the column where they are rectified or fractionated, a portion is
withdrawn as product, and the remainder flows down the column as liquid
reflux, finally returning to the still pot. At the end of a distillation the
still pot will contain the undistilled part of the distilland and the draiF,age
from the rcct.ifying section. :w:.
The most commonly used still pot for laboratory fractionating col~rons
is a round-bottom or balloon flask of suitable size. These are available in
several standard sizes, equipped with either standard-taper joints or spheri-
cal joints, for easy attachment to a companion joint, on the lower end of
the column. The same types arB also available with sealed-in thermometer
wells for distillationsreql.lil'ing a knowledge of still pot temperature. When
the pressure drop across the column is to be determined, or when it is used
to control boil-up rate, a tubulation in the neck of the flaskis provided for
(~onnecting the back-pressure manometer. .
Permanently attached still pots of the same generalt~e;~ uently
·t··l: \i' .:;:,;\,.:1:/
146 Collins, and Lantz, Ind. Eng. Chem., Anal. Ed., 18, 673 (1946);
II. ORDINAHY l<'HAC1'IONAL DIS'I'ILLATION 261
used. A satisfaetol'Y design 146 is shown in Hgure 4,1. The sealed-in

thermometer well haR been eliminated to avoid the use of a fragile ring seal
necessary in this COllstruetion. If the still-pot tempemture is desired. It
suitable thermometer well ean be slipped into the charging line ill place of
the intcmal heater shown. The connection to the baek-pl'essurc manom-
eter is made in the cold region of the charge line to avoid diffusion of
"tipO!' into that part of the equipment.
TO TRANSFORMER
TO B.R
MANOMETER
25 MM. 0.0.
TOV;-~
7! 5 LITER
TO VARIAC
Fig. 44. Still pot for pllrmanent attachment to column.
Spherical still pots offer great convenience of construction and a maxi-

mllpl of safety from implosion when distillations are made at greatly re-
duled pressure. Their chief disadvantage is t.he low ratio of area to vol-
ume; this requires high heat tmnsfer per unit area, and hence necessitates
increased int.ensity of heat per unit area for a given boil-up or evaporation
rate. Except in case of ext.reme thelma! sensitivity of the charge, this is
not usually serious. Still pots of other shapes have been used to some ex-
tent. Willingham and Rossini 147 used a cylindrical type, Hall and Jonach 148
the design of Figure 45.
Laboratory columns of all-metal construction usually have cylindrical
still pots, positioned either horizontally or vertically to the cylindrical
146Herdpl¢S' ;r~~der Co. design.
147Willi 'I ;;~~l Rossini, J. Re8earch Nail. Bur. Standards, 37, 15 (1946).
1481(: •.... , .. :~d Jonach, Symposium on High-Temperature Analytical Distillation,
Ameri~3ll~l~uin Institute, Meeting at Chicago, Illinois, Nov. 11,1946, p. 42.
'-:'if.:
:262 A. L. GLASEIllWOK A:ND F. E. WILf,IAMS
[lXiS.149-151 Others,152 on the other hand, have a spherical still pot con-
structed from a large-volume stainless-steel float.
14
3 GLASS HOOKS • $ 1'35 NO-LUB MALE
12.0 0
APART ~ JOINT WITH MERCURY
SEAL
165 MM.
OUTER TUBE 18 MM. 0.0.
60 MM. 7 MM. 0.0.1 NNER TUBE

ATTACHED TO BOTTOM.
TWO V NOTCHES 5MM.
HIGH 1800 APART AT BASF
____-,,--GLASS RING TO SUPPORT

HEATING WIRE
Fig. 45. Still pot for ltntt\yt,ica\ distilling eo\umn of Figure 12.148
V. HEATING~ CONTROL OF HEAT INPUT, AND

TEMPERATURE MEASUREMENTS
Heating of laboratory apparatus has been discussed rather completely
by Egly in Volume III of this series (Technique of Organic Chemistry).
149 Borns, Coffey, and Garrard, Symposium on High-Temperature Analytical Distilla-
tion, p. 2U.
160 Marschner and Cropper, ibid., p. 35.
151 Fenske, Quiggle, and Touberg, Ind. Eng. Chem., 24,408 (1932).
In Smith, V. C., Glasebrook, Begeman, and Lovell, Ind. Eng. Chem., Anal. Ed., 17,
47 (1945).
II. ORDINARY FRACTIONAL DIi';TII,T~ATION 263
For heating; laboratory still pots and columns, eleetrie power ill almost uni-
versally mied. Ease of control, safety, and adaptn.bility to automatic
operation have made it supplant all other methods. Heat is generated
by dissipat.ing pm,'er in an element of high resistivity a('cording to the
formula P = RI2, where P is the power dissipated (in watts) and I is the
current (in amperes). Consider, for example, a resistance of 22 ohms and
a supply voltage of 110 volts. The usual type of variable transformer used
in laboratory work at its maximum setting will supply 5 amp. to this ele-
ment, the safe output for the transformer. Table XXVII gives the power
(wntts) diHsipated in the resistor at the various voltages, and the equiva-
lent elllol'ies per hOlll' dissipated (1 watt haul' = 860 caL). If we assume
10 to he the lowest and 110 the highest voltHge used, there is It voltage
variation of 11: 1 ttvnilnble OIl the adjustable tmnsformer; the power ratio,
however, will be 112:101' 121.
T ABLE XXVII
POWER D18SII'A1'lm IN 22-0lfM RgS1S'rOR AT VARIOUCI VOLTAm;s
y I P = RI'. watt. clll./hr.
0 0 0 0
10 {JAM 4.5 3,870
20 0.\)10 18.2 15,300
30 1.365 30.0 21),800
40 1.820 74.3 63,800
50 2.275 117.7 100,000
60 2.730 164.0 141,000
70 3.185 235.4 202,000
80 3.040 2\)0.4 250,000
\)0 ·1.095 367.4 316,000
100 4.550 4ii3.2 390,000
110 5.000 550.n 474,000
Energy supplied to a still pot is used in part to oven:ome the heat losses
arising from imperfect insulation and thereby hold the charge at its boiling
point; the rcmainder provides the heat of vaporization. Heat losses are
greater, the higher the operating temperature of the still pot. When dis-
tilling benzene, for example, at a rate of 1000 ml. PCI' hoUl', the energy sup-
plied for evaporation would be 95 watts per hour; the same volume of p-
cymene (b.p. 175°C.) would require 57 watts per hOUl'i yet the much
higher heat losses from the still pot at 175°C. would make the total energy
requirements considerably higher than when distilling benzene at SO°C.
Little information is available in the literature on the power require-
264 A. L. GLASEUnOOJ{ AND F. E. 'WILLIAMS
ments for laboratory still pots. Goldsbarry and Askevold 153 using Glas-Col
heating mantles as their heat source showed that a mixture of n-heptane
and methylcyclohexanc could be distilled at a rate of 1000 and 2000 m!.
per hour with heat inputs of 125 and 215 watts, respectively. In this case,
heat losses were apparently 35 watts per hour.
Two general methods are used to heat still pots. The more common
method consists in surrounding more or less of the still pot surface with the
electric heating element, a thin layer of insulating material being used to
prevent direct contact of the element with the still-pot surface. A less fre-
quently used method of heating consists in submerging a heating element
directly in the material being distilled. Sometimes in the interest of close
control of the evaporation rate, the two methods are combined.
1. External Heating
Heated-oil baths, sand baths, molten-salt baths, and lead-shot baths,
heated either by gas 01' electric power, have been used to some extent for
supplying heat to still pots, but the general inconvenience of several of
these and their slow response to control have nearly eliminated them as a
source of heat. Cylindrical-~haped still pots can be readily equipped with
an electrical heating element of suitable resistance wire or tape. The in-
sulated wire may be wound directly on a glass still pot and covered with a
thick layer of 85% magnesia to minimize heat losses. If bare wire or tape
is used, a thin layer of asbestos paper should be applied to the surface of the
glass to support and hold the wire or tape in position. Additional heavy
insulation is then put over the resistanee element. If the heater element
is to be controlled by a variable transformer, the current-carrying capacity
of the wire or tape should be of a size to carry safely the total current avail-
able from the transformer, and the total resistance of the element should be
great enough to ensure that the current will not exceed the transformer
rating at the highest voltage setting. For example, a variable transformer
supplying 5 amp. at 110 v. should be used with a resistance of at least 22
ohms.
Table XXVII-A gives data to guide the designer in choice of the proper
resistance wire or tape to use as heating elements for still pots and column
jackets. Temperatures given are for straight wires of the indicated sizes
in open air. Enclosed wires will reac.h considerably higher temperatures.
Maximum safe operating temperature for Nichrome V wire is 1175°C.
Electrical heaters for cylindrical metal still pots can be similar to those
101 Goldsbarry and Askevold, Symposium on High-Tomperature Analytical Distilla-

tion. American Petrolaum Institute, Meeting at Chicago, Illinois, Nov. 11, 1946, p.
12.
II. ORDIN.\RY FRACTIOKAL DISTIr,L.\'l'[OX 2(i;j
TABLE XXVII-A.
R('sistlLn(Je, Current, Temp ••

'V'ire size ohms/ft. ump. 'C.
l/lIj"·in. X O.OORD-in. t,ajJe ......... 0.97 5.7 ijfl.5

20 gage ........ -, .... ,- .... . . O.H3 5.0i 315
22 gag!L .••.... ............ .. 1.02 5.0 .~~
i}h)
24 gagl•........ .. . .... - ......... LIn 5.0n (i50

2ti gagl~ ........ .. ....... ..... . . 2.57 5.25 8i1
a D:1kl from Catalog R-4(j, Driver-I-brris Comp:lllY, Harrison, New Jerser.
described above. A sufficient thickness Qf a::;bestos paper should be ap-

plied to the surface for adequate electrical insulation before the resistance
wire is put in place. Fenske et al. 154 used this method, and provided addi-
tional heat on the bottom by means of a ring heater. To aid heat transfer
to the liquid, copper fins were brazed onto the inside surface of the still
pot. Strip heaters offer a convenient means of heating metal still pots
having horizontnl axes. These heaters are available shaped to fit the con-
tour of surface, the heater being placed parallel to the axis of the cylinder.
Adequate insulation over the whole still pot reduces heat losses.
Electric heaters for spherical still pots are difficult to (j()llstruct. For-
tunately the standardization of round-bottom flasks has led to the develop-
ment of woven-glass heating mantles well suited for heating spherical still
pots.* These mu,ntles are available in several sizes ranging from 500 ml.
t.o 22 liteI'll. The smaller sizes up to and including 3 liters have OIle pair
of input terminals; larger sizes have two or three sets of elements to be
controlled by a corresponding number of variable transformers. On the
smaller mantles a part of the heating element is in the upper half of the
mantle to minimize condensation at the top of the still pot. Heat losses
are kept at a reasonably low value by the use of a thick layer of glass-wool
insulation. The mantles may be easily removed from detaehable still
pots; when used on permanent still pots, years of service may be expected
from the mantle, provided the manufactul'ers' reeommendations are not
exceeded.
A somewhat unusual de::;ign of external heater ::;uitable for very small
still pots hUH been used. m This consists of it 30-watt heater,3/s in. in di-
ameter and 1.5 in. long. It, slips into the heater well of the still pot shown
* Manufactured by GIas-Col Apparatus Company, Terre Haute, Indiana. Scientific

Glass Apparatus Company, Bloomfield, New Jersey.
1&< Fenske, Quiggle, and Tonberg, Ind. Eng. Chern., 24, 408 (1932).
166 Donnell and Kennedy, Symposium on High-Temperature Analytical Distillation.

2GG A. L. GLASEBROOK AND F. E. WILLIAMS
in Figure 46. Birch, Gripp, and Nathan 166 surrounded the spherical still
pot of their spinning-band column with an insulated steel box to provide
an ail' bath; It lOOO-watt electric resistance element placed in the bottom
of the box supplied heat.
Vacuum Jacket
Tapered Joint
50 ml. Stillput
Heater Well
Chimney
Fig. 46. Vacuum-jacketed still pot for small chal'ges. 156
2. Internal Heaters
Internal heaters are of two types: those in which the resistance element
is submerged in the distilland, heat transfer being by direct contact of the
liquid on the heating element; in the other type the resistance element
transfers its heat to a surrounding sheath which in turn heats the liquid in
. which the sheath is submerged. Direct-heating internal heaters when
used in conventional types of glass still pots are of necessity small ill order
166 Birch, Gripp, and Nathan, J. Soc. Chem. Ind. London, 66,33 (1947).
II. OUDINAHY FH.\C'TIO!\ AI, DI5TILLA'l'IO.N 2ti7
that they may he inserted anJ properly located. Consequently the

amollnt of heat dissipated is small if excessive temperature of the resistance
element leading to possible thermal decompmlitioll of the di:')tilland is to be
avoided. This limits the usefulness of this type of heater to low vaporiza-
tion rates. Special still pot design will allow the use of larger heaters and
greater vaporization rates. As a matter of safety, a chaser should be added
to a charge whell un internal heater is used to permit complete distillation
of the charge, and still have the heater submerged at the end of the distilla-
tion. Some elect,rolytic eorrosion occurs with distillands containing water
or other eonducting components and limits the life of an internal heater,
Still-pot charges eontaining substances corrosive to the heater element
should be avoided. Direct-heating internal heaters arc most suitable for
hyclroearbon fractionations.
Sheathed types of internal heaters offer much larger dissipating surfaces
thau the direct-heating type; the surface must be kept submerged at all
times in the distilland. This type of heater is most suitable for use in
eylindrical still pot.s made of metal where the design will allow a liquid-
tight SCL'ew connection between the heater shank and the still pot to be
made.
Internal heaters function to best advantage when they operate in con-
junction with an external heater, the latter providing the major portion of
the heat input to a still pot, and the former providing a varying amount of
the small remaining heat requirements to hold the boil-up rate constant as
controlled by a back-pressure manometer. }i'igure 44 shows this principle
applied to a glass still pot, the external heat being supplied by a mantle-
type heater.
The most f:1Vorable characteristic of the combination of internal and ex-
ternal heaters is the rapid response of the system to changes in back pres-
sure when a back-pressure manometer is used to control vaporization rate
from the still pot. Of the energy supplied to a still pot, a portion is used
to hold the distilland at its boiling point, and the remainder supplies the
heat of vaporization. If the external heater supplies all the former and
90% of the latter, the internal heater will supply 10% of the heat of vapori-
zation to maintain a const.ant evaporization or boil-up rate. The mass of
the internal heating element can be kept small to reduce heat lag because
the total heat dissipated is smaller; response of the internal heater to
changes of power input as controlled by the baek-pressure manometer is
accordingly quite rapid. As an example, consider a heating system de-
signed for evaporating 1000 ml. benzene per hour, or 880 grams, Heat of
vaporization is 93 cal. pel' gmm; total heat of vaporization will be 880 X
93 = 81,840 cal. per hoUl', or 81,840/860 = 94 watts. The external heating
mantle therefore will be required to supply the heat losses plus 90% of 94
268 A. L. GLAS]JRROOK AND F. E. WILLIAMS
watts, or 85 watts. The internal heater will, accordingly, supply 9 watts.

If its resistance is 2 ohms, the current will be, from P = RI2, I = VP / R =
V9/2 = 2.06 amperes. This will be the average current; to allow the
necessary control by the back-pressure manometer, a resistance of 1.4
ohms will be hooked in series with the heater, and 5.16 v. will be impressed
across the pair in series; during this period the internal heater will dissipate
4.5 watts. As the boil-up rate drops, the back-pressure manometer will
short out the series resistor, and increase the dissipation of the internal
heater to 13.5 watts to restore the boil-up rate of the benzene. A small
transformer with a no-v. primary, and 10-v., 5-amp. secondary, powered
from a variable transformer will allow a wide range of adjustment of the
power dissipation in the internal heater; direct operation of the low-voltage
,internal heater from the variable transformer is neither safe nor convenient.
. The internal heater of Figure 44 is made by winding 38 in. of No. 20
Nichrome wire on a steel mandrel I in. in diameter; spacing between turns
is one wire diameter. After the winding is completed, the wire is heated to
redness in an air-gas flame to "set" the turns so that the coil will hold its
shape and size when it is removed from the mandrel. The ends of the
wire are then bent to meet the two tungsten leads of the glass closure tube
of the still~pot charge line; connection to the tungsten leads is by SilV8l'
soldering.
3. Control of Heat Input
Heat input to the still pot of a fractionating column is one of the impor-
tant operating variables in a distillation. Too little or too much heat lead,
respectively, to interruption of reflux or to flooding; between these ex-
tremes the proper choice of heat input will determine the column efficiency
to sotne extent, the reflux ratio in columns having constant take-off, and
the optimum productivity of the column. It has been shown 167 that the
efficiency of a Hyper-cal fractionating column varies from 92 theoretical
plates at a boil-up ra.te of 500 ml. per hour to 48 plates at 2000 ml. per hour.
These results were confirmed, in a general way, in another column of the
same type. 108 The same investigators showed, however, that a perfo-
rated-plate column and a helix-packed column of the same packed length
and diameter lost efficiency quite slowly with increasing boil-up rate.
Collins and Lantz 159 confirmed the work on perforated-plate columns.
Reed'sleo work on a Ewell column indicated some decrease in efficiency with
l57 Brandt, Perkins, and Halverson, Symposium on High-Temperature Distillation.
American Petroleum Institute, Meeting at Chicago, Illinois, Nov. 11; 1946, p. 51.
161 Goldsbarry and Allkevold, Symposium on High-Temperature Distillation, p. 12.
m Collins and Lantz, Ind. Eng. Chem., Anal. Ed., 18,673 (1946). .
IGI Reed, C. R., Symposium on High-Temperature Analytical Distillation. American
Petroleum Institute, Meeting at Chicago, Illinois, Nov. 11, 1946, p. 4.
II. OHDINXHY FHAC'l'ION AL mS'I'ILL.\TlOX 20fl
increasing boil-up rate. High-efficiency helix-packed columns (of 100

t.heoretical plates) have been found to lose efficiency aH boil-up rate in-
creases. If it is assumed that these results as measured under total reflux
are indicative of column behavior with finite reflux ratios, the importance
of control of boil-up rate with many types of columns is apparent, particu-
larly in those instances requiring the highest efficiency from a given column
for a critical separation.
Columns using constant take-off devices n,re more sensitive to clHl,nges
in boil-up rate than are those using the intermittent type of take-uff. With
the former, an inerease of boil-up rate will change the reflux ratio in a favor-
able direction, and a decrease in boil-up rate will lower the reflux ratio for
an adverse effect. A certain amount of compensation may take place if
column efficiency increases or decreases, respectively, with decreases or
increases in boil-up rate. With intermittent take-off, where t.he partition-
ing mechanism gives reasonably good approximation~ of the true reflux
ratio, changes of boil-up rate are less serious except. insofar as such changes
adversely affect column efficiency. AA a priwtical procedure, operation of a
column should be confined to conditions in which effieient'y it> not unduly
sensitive to boil-up rate.
Control of heat input 161 to the still pot can be eit~ler manual 01' auto-
matic i both require some means of measuring boil-up rate, either in arbi-
trary terms or in absolute terms by precalibration, and of detecting changes
in boH-up rate in order that appropriate nhanges of controlling elements
may be made. The most commonly used device for measuring boil-up
rate is a back-pressure manometer, one leg of which is connected to the
still pot and the other leg to the top of the condenser seetion. Either
mercury or other liquid of low volatility it! used as the manometer fluid.
Columns operating with small pres~mre drop require low-density liquids to
show signifieant changes for small variations in boil-up rate.
A back-pressure manometer !:limply measures the clifferenee in preSl:\ure
arising from the flow of vapor through the const,rietions of the column pack-
ing bet.ween the still pot and the top of the reflux condenHer, as so many
units of length of the manometer fluid. 'To be a measure of boil-up rate,
the relationship between back pressure and the grams or milliliters of liquid
condensed in the head, per unit of time, or evaporated from the still pot
must be established by independent means. A calibration of this kind ap-
plies strictly only to the calihrating liquid in a particular column at a single
temperature level. Two or more liquids silUilar in structure and boiling
point should give closely related calibration curves in a given column.
During fractionation of such mixtures, the boil-up rate should remain con-
IDI See Sturtevant, "Temperature Control," in Volume I, thi':! series: Phllsicalltf ethods
of Organic Chemistry, 2nd ed., 1949, Part I, Chapter II.
270 A, L, GLASEBROOK AND F, E, WILLIAMS
stant for a given back pressure. On the other hand, a mixture of structur-
ally different components whose molal heats of vaporization differed widely,
and whose individual calibration curves of boil-up rate versus back pressure
showed wide differenees, would probably exhibit a variety of boil-up rates
for a given back pressure during a fractionation of the mixture. The initial
boil-up rate of the mixtme would be some average of the curves of the indi-
vidual components; as the fractionation proceeded, the composition of the
distilland would ehange, and hence the boil-up rate would change. It
would be necessary under these circumstances to determine the boil-up
rate during the distillation by an absolute method in order to maintain a
constant flow of vapor at the head of the column. With a properly con-
structed condenser system operating on the principle of intermittent take-
off, changes in the boil-up rate are indicated by an increase or decrease in
the time required to collect a given amount of product, provided the ap-
parent reflux ratio is kept constant. Necessary adjustment of the back
pressure may then be made to restore the desired boil-up rate,
Available information on pressure drop versus boil-up rate is meager.
Hall and Jonach l62 report pressure drop in one column in mm, of mercury
at n. boil-up rate of 150 m!. per hour as follows: Cs alkylate 2.5, C 12 alkylate
3.5, benzene 3,6, toluene 3.9, and xylene .5.0. Goldsbarl'Y and Askevold 158
made measurements 011 methylcyclohexane-heptane mixtures in three
columns. Their results are shown in Figure 47. ("Feed rate" is the total
vapor leaving the still pot and represents the vapor condensed at the head
plus internal condensation in the column arising from imperfect insulation
or heat-loss compensation,) It is apparent that back-pressure readings
give substantially no measure of true boil-up rate over the range 1000-3000
ml. :per hour in the perforated-plate column. In the Heli-Grid packed
column, the back pressure is very nearly a linear function of feed rate be-
tween 800 and 2600 mI., while in the helix-packed column the back pressure
changes slowly and nonlinearly with feed rate up to about 1800 ml. per
hour, and linearly to about 3000 ml. per hour, The helix-packed column
was maintained adiabatic by heated jackets which allowed internal reflux
of 620 ru!. pel' hour, whereas the Heli-Grid column was vacuum-jacketed
and had internal reflux of 305 ml. per hour, These differences in internal
reflux: may have contributed to the differences in the character of the back-
pressure curve.
It is apparent that the back pressure or pressure drop through a column
may be kept constant, either by manual or automatic control, but the abso-
lute value of back pressure will indicate the true amount of vapor reaching
the condenser only when sufficient calibration data have been established.
16~ Hall, H. j" and Jonach, Symposium OR High-Temperature Analytical Distillation,
II. ORDINAHY FHACTIONAL DIS'l'ILLATIO:-i 271
An effective method IfjJ for determining boil-up rate con::;if-its of introducing

thermocouple into the outlet and inlet lines of the head condenser, the
:.t
couple in the inlet line serving as the cold junction. In this way changes
in temperature of the entering water will not affect the measurements.
During the time u. meaf:;urement is made, t.he How of water has to be held
at a known constant value. Suitable baffles should be placed in the con-
denser to ensure thorough mixing of the W!Lter, [lnel proper insulation should
6oor---------------------------------------------~
~.
.~.
~
_.--------
2!.
o PERFORATED-PLATE COLUMN
30
E 400 _ _ _ _I _ e -
OI
E
0.."
/".'-
o
a:: ./.
o
w
0::
;:) 200
~
0::
a. HEll-GRID COLUMN / _ . /
/" .../ GLASS HELIX-PACKED COLUMN
,..,.G ./
.----. _...,~.---.,..,. /
OL-~~~,~O~OO~~~~2~O~0~O~~~3~O~O~O~~~740~O~O~~~~50~O~O
FEED RATE, ml.lhr
Fig. 47. Pressure drop VOl. feed mw for three types of columns.
be placed u.round the condenser to prevent heat losses. Calculation of the

boil-up rate is simple:
17 B = twlA
where VB is boil-up rate grams per hour, w is grnms of water flowing through
the condenser per hour, t.. is latent heat of vapor in calories per gram, and
t is temperature of the water (t is not the t,l'ue temperature of the water,
but the increase in its temperature resulting from heat picked up; the ar-
rangement of the thermocouples, however, gives the apparent temperature
of the water). Boil-up rate determined by this method is accurate to
about 5%, if the water rate is held sufficiently constant during the time a
temperature measurement is being made, and if the temperature is great
enough to permit an accurate measurement to be made. This method is
excellent for determining boil-up rate versus back pressure for a series of
pure compounds whose mixtures are to be fl'actionated~ Its usefulness is
limited to those compounds whose heat of vaporization is known or may he
163 Fenske, Quiggle, and Tonberg, Ind. Eng. Chem., 24, 408 (1932).
272 A.. L. GL,\SEBlWOK A.ND F. E. WILLIAMS
calculated from vapor-pressure data. By suitable construction of t.he

condenser system, this met.hod can be used to check the boil-up rate during
a distillation and indicate appropriate changes in the setting of a controlling
back-pressure manometer to maintain a pre-
scribed boil-up rate.
The usual back-pressure manometer con-
sists of a simple U-tube partly filled with a
fluid of low volatility. One leg of the tube
is connected to the top of the reflux con-
denser, and the other leg to the still pot at a
point not exposed to the hot vapor from the
distill and. Safety traps are usually attached
to each leg to catch the liquid contents of
the U-tube in the event the back pressure
exceeds the range of the manometer, either
as a result of excessive heat input to the still
pot or by violent bumping of the charge as
sometimes occurs, particularly when water is
present in the charge. Figure 48 is an ex-
tLmple of a satisfactory back-pressure ma-
nometer. Mercury, dihutyl phthalate,
propylene glycol, diethylene glycol, or lubri-
cating oil of light grade maY,. be used as the
manostatie fluid; mercury offers the great-
est possible range of pressure i11 a given in-
strument, hut low sensitivity; lower-density
liquids reduce the pressure range but offer
the advantage of increased sensitivity,
greater by a factor of 10 to 15. Mercury
or the glycols containing a small amount of
dissolved sodium nitrite may be used when
Fig. 48. Back-pressure ma- conductivity of the manostatic fluid actuates
nometer with terminals for con- an electronic switch for control purposes.
trolling heat input. Nonconducting liquids may be used as
manostatic fluids in controlling manometers if a photoelectric control is
used. 164
Certain precautions should be observed in working with back-pressure
manometers. During the prefiooding operation, the high-pressure side
of a low-range manometer should be closed by a valve just before the opera-
tion. is started, and the valve opened when normal boil-up is resumed.
114 Hall, S. A' I and Palkin, Ind. Eng. Chern., A.nal. Ed., 14, 6§2 (HI42).
II. ORDINARY FRAU'rrONAL DISTILLATION 273
Introduction of a very slow stream of carbon dioxide or nitrogen gas into

the still pot through the back-pressure line will stop any diffusion of vapOl'
into the line during the prefiooding period. 1 or 2 ft. of 2-mm.. capillary
tubing used as a section of the high-pressure line next to the still pot will
help to m.inimize diffusion of vapor toward the manometer. Variations in
the temperature of the gas in the high-pressure line to, and in the hody of,
the back-pressure manometer undoubtedly lead to variatioIls of boil-up
rate if the back pressure is maintained constant. Diffusion of the gas into
the still-pot vapor is an unknown factor in the behavior of the manometer.
Its simplicity and general adaptability to automatic control of boil-up rate
has made the back-pressure manometer a popular auxiliary of many distil-
lation columns, but unless boil-up rates are checked frequently by n more
reliable and independent method, considerable uncertainty attends its use.
4. Reflux-Temperature Measurement
Fractionating columns tue usually equippecl with a thermometer or other
temperature-measuring deviee l66 located at a position immediately outside
the fraetionating section where the issuing vapor impinges on the measuring
element on its way to the condenser section. In a properly constructed
column, the temperature indicated is the vapor-liquid equilibrium tempera-
ture, or very netlrly so. If the column is producing a substantially pure
compound, the indicated temperature will be the boiling point of the com-
pound at the prevailing pressure; if the column is producing a mixture of
two or more compounds, the indicated temperature will be the boiling point
of the mixture of a specific composition. (Azeotropic mixtures, either bi-
nary or ternary in composition, act as a pure compound at a given pressure.)
The above statements are true if the quantity of vapor is sufficient to supply
all heat losses, either of radiation or conduction, of the thermometric ele-
ment, and if the vapor is not superheated. Unavoidable small heat losses
from the element provide a thin layer of condensate to establish the desired
liquid-vapor equilibrium condition for correct temperature reading.
Proper placement of the thermometric element in the vapor atream is of
primary importance to ensure the closest approach to true vapor-liquid
equilibrium-temperature readings. In low-temperature columns, this con-
dition is most easily established by the use of an efficient vacuum jacket to
conduct the vapors from the rectifying section to the thermometric ele-
ments i in addition, means should be provided to keep the thermometer
wetted, either by refrigeration, or by directing a part of the reflux onto the
thermometer. In high-temperature columns, the possibility. of super~
185 See Sturtevant, "Temperature Measurement," in Volume 1, this series: PhySical
Methods oj Orqanic Chemistry, 2nd ed., 1949, Part I, Chapter 1.
274 A. L. GLASEBIWOK AND F. E. WILLIAMS
heating is unlikely under proper operating conditions, and hence insulation

is used to prevent excessive heat losses. At very low pressures (1 to 10
mm.), polished-metal radiation shields and external heaters may be neces-
sary to minimize heat losses to enHure correct temperature readings.
In many column designs, the thermometric element is part of the con-
denser system, as is indicated in Figures 37 to 413. Fenske et al.,163 on the
other hand, using thermocouples for
temperature measurements, buried them
several inches in the packing of the rec-
tifying section to ensure true vapor-liquid
TO CONDENSER equilibrium. This system requires that
SYSTEM
the measuring element be far enough
/ below the top of the packing to allow the
/ cold reflux to reach equilibrium tempera-
ture before contact is made with the
thermometric element. The vapor-liquid
equilibrium temperature so indicated may
be higher than the temperature measured
just above the top of the packing because
of the fractionation occurring in the
packed section above the thermocouple.
TOP OF PACKING Willingham and Rossini1 86 used a design
in which a resistance thermometer was
located in an unpacked section of a
column immediately above the top of the
packing. The vacuum jacket extended
very nearly to the terminal block of the
thermometer and hence provided excellent
b'ig. 49. Sealed-ill thermocouple insulation against heat losses and con-
well. sequent errors in temperature readings.
The vapor conduit to the total condenser
was sealed to the column just above the vacuum jacket. This design is
probably ideal for proper placement of the thermometric element, but is
usable only with resistance thermometers of suitable length, or with ther-
mocouples, as shown in Figure 49. Partial-immersion thermometers do
not have enough immersible length to permit the thermometer to be in-
serted to the correct depth and still provide for connection of the vapor
conduit to the condenser.
Measuring Devices. Three devices based on three different principles
are commonly used to measure vapor-liquid equilibrium temperatures:
1&, Willingham and Rossini, J. Research Nail. Bur. Standards, 37. 15 (I94G).
II. ORDINARY FRACTIONAL DISTILLATION 275
(a) the liquid (usually mereury)-in-glass thel'll1ometer, (Ii) the thermo-
couple, anci (c) the resistance thermometer.
Thermometer8. The liquid-in-glass thermometer is the oldest, simplest,
and least expensive of temperature-measuring imltrumentR; for these
reasons it is still a fa,vorite means of determining vapor-liquid equilibrium
temperatureR in fractionating columns. The working range of liquid-in-
glass thermometers covers the interval -HIO° to 500°C., and henee, is
entirely adequate for use in distillation columns. 167
Two designs of thermometers are a.vailahle for general use. The total-
immersion type requires the thermometer to be immersed in the medium
whose temperature is being measllred to a depth sufficient to bring all
liquid in the thermometer to t.he temperature of the medium; this type is
generally small and of limited range for each thermometer of a series. Suit-
able design of (L still head will allow total immersion thermometers to be
used in distillation work where the range of vl1por-liquid temperature is
small, but the general practice is to allow greater latitude by using partial-
immersion types of thermometers.
Partial-immersion thermomet.ers are constructed to give the correct or
nearly correct reading of a temperature when theyareirnmersed to a specified
depth in the vapor whose temperature is being me1Lsured. This commend-
able effort on the part of the manufactmers cannot overcome the variations
of indicated temperature arising from the variable temperature of the emer-
gent stem of the thermometer; nor is it generally possible to create a still-
head design which allows adequate circulation of vapor around the section
of the thermometer specified for immersion. Generally, a pocket exists
where the thermometer enters the still head and poor vapor circulation pre-
vails. These two factors introduce errors in the indicated readings that are
not easily corrected.
Thermometers must of necessity be located in the vapor stream leaving
the fractionating section; this position offers no problem in reading with
short fractionating columns operated from floor level; but taller columns
require the use of a ladder or superstructure to make the thermometer ac~
cessible for reading. When a battery of columns is under the care of one
operator, this expedient is entirely too burdensome and time consuming.
The objectionable features of liquid-in-glass thermometers can be largely
overcome by one of the electrieal methods of temperature measurement.
Thermocouples. In t.he practical application of thermoelectric thermom-
etry only a few pairs of metals have received wide application. These
are shown in Table XXVIII. Any of the four couples, commonly called
base-metal types, of the above list are sat.isfaetory for measuring vapor-
liquid equilibrium temperatures in distillation columns where the tempera-
167 Busse, in 'l'emperaturB, Its lJ1 easurement and Conlroll:n Science and Indu.stry. Rein-
hold, New York, 1941, p. 228.
27(1 A. L. GLASEIlUOOK AND F. E. WILLIAMS
TABLE XXVIII
E.M.F.
Couple lOUOC. 20QOC.
Coppcr-ConsbntlLll .......... . 4.28 mv. n.29 mv.

Iron-C(l[L~tu.nt!L!l ............ . 5.28 10.78
Chromel-Collstnntall ........ . 6.30 13.30
Chl'onwl-Alumel ............. . 4.1 8.13
ture will not exceed about 300 0 e. Iron-constantan and Chromel-constan-

tan have somewhat greater thermal E.M.F. for a given temperature com-
pared to the other two, but the difference is not marked. Copper-constan-
tan thermocouples offer certain distinct advantages. This pair is usable
over the range - 200 to 3.50°C.; since one element of the couple is copper,
a relatively short length of constantan cn,n be used to form one couple for
the high-temperature position and the other couple for the reference junc-
tion located nearby; long leads of low-resistance copper wire can then be
used to connect to the measuring instrument at a convenient position for
operation. Number 28 or 30 gage copper B.nd constantan wire, insulated
with enamel and silk, are satisfactory for most purposes, the smaller wire
being preferred from the point of view of low conductivity of heat from the
junction.
Construction of a junction is simple; insulation is removed from about
1/2 in. of the ends of the two wires forming the couple; enamel insulation
can be removed by use of fine emery paper. The two bared ends are then
twisted together, and given a very light coat of a flux suitable for silver
soldering. A loop about l/S to 3h6 in. in diameter is formed at the end of a
6-in. length of No. 18 Nichrome wire, and filled with silver solder* in a
small air-gas flame. After the solder in the loop is molten, it can be kept
in this condition by holding the flame slightly above the loop. The pre-
pared cpuple is then slipped through the molten solder and immediately
withdrawn. The couple will be found to have a thin uniform coat of solder,
and will not have been heated long enough to cause undesirable changes in
the character of the metals. About 3/Sin. of the tip of the couple is then
cut off, any adhering flux removed, and the construction is complete. The
other junction is similarly prepared. The finished couples are shown dia-
grammatically in Figure 50.
A thermocouple should be protected by a suitable well to prevent COITO-
sian of the couple by chemical reaction with components of the vapor
stream. The small physical dimensions of a couple constructed from the
'" Manufacturers of silver solder and £luxes: Handy I1nd Harman, New York, N. Y.;
Whitehead Metal Products Company, New York, N. Y.; American Platinum Works,
Newark, N. J.
II. ORDIi'\AHY FRAC:'l'I00;AL DlS'l'rLL.\,{,£OK
recommended size of win) permits the usc of sll1alhliaDwter wells. For

greater mechanical strength, only the lower end of the well, for rt h.'}lgth of
METAL A
TO POTENTIOMETER
METAL B
COLD HOT
JUNCTION JUNCTION
Fig. 50. Therllloelectric circuit for teJUIJemtul'c mCILSUI'CIllP.ut.
10 to 15 dimnetel'H, need be made Slllall diameter. A suitable design for

sealed-in construction is shown in li'igure 41). About 1}2 in. of high-boiling
liquid such as dibutyl phthalate should
be introduced into the test-tube cud with
a capillary pipet; before the e(Juple is
placed in position. This will ensure ex-
eellent heat transfer to the junction. As TO CONDENSER
soon as the column is in operation with it SYSTEM
liquid whose boiling point is neal' the

highest temperature at which the column
is to be operated, a cork should be in-
serted firmly in the top of the thermom-
eter well, and eompletely covered with
Apiezon W wax to provide a vacuum-tight
seal. This 'will prevent "breathing" of
the well and prevent hydrolysis of the
dibutyl phthalate and sllbsequellt COl'l'U-
sion of the couple.
Mllitiple-junction thermocouples) in
series, may be used to give greater thermal
E.M.lf. for a given temperature level to
provide greater sensitivity. Space limi-
tations in laboratory columns limit the
number of junctions to two or possibly
three. Installation is more difficult than
with single junctions since each junction
I"ig. 51. Thermor:ollple well for
must be well insulated from its com- double-junction thel'lllor;Ollple.
panion to avoid short circuits; this is
most easily accomplished by installing mUltiple-tip wells to isolate eaeh
element, or by use of the eonstruction shown in l~iglll'e 51. The U-tube
should have enough high-boiling liquid in the bottom to provide heat
278 A. L. GLASEBROOI( AND F. E. WILLIAMS
transfer to the two couples. (It; may be mentioned that, by removal of

the theTnlOCOuples and the high-boiling liquid, the U-tube may be used as
a condenser for measuring hoil-up rates by the method outlined in Section
V3.)
Thermocouples for determining vapor-liquid equilibrium temperatures
can be most readily calibrated in situ. A suitable amount of a pure com-
pound of known boiling point may be charged to the still pot, the column
put into operation, and kept under total reflux. Traces of water may be
removed from the system by withdrawing a few grams of the distillate.
The thermal E.M.F. developed at several boil-up rates should be identical
for a given charge of pure material. If significant variations of E.M.F.
develop under different boil-up rates, improper placement of the thermo-
couple is indicated. Three or four check points in the proposed working
range of the thermocouple will suffice to establish the necessary corrections
to be applied to measured E.M.F. to give true temperature readings on
conversion from standard tables.
The reference junction for thermoelectric systems on laboratQry columns
is most easily maintained constant by the use of a bath of ice and distilled
water; one-quart wide-mouth Thermos bottles are suitable containers
for this purpose. A well similar to that shown in Figure 49 should be pro-
vided for shielding the junction. The reproducibility of the ice point is
excellent, variations being 0.0002°C. or 1es8. 168 This is smaller by a factor
of fifty than the variation expected in the most precise column operation.
Automatic cold-junction compensators used in some industrial instruments
are not satisfactory substitutes for an ice~bath reference junction if tempera-
tures are to be measured to 0.1 ° C., or less.
Precise measnrement of the E.M.F. of a thermocouple is possible only
with a potentiometer. The principle of this instrument, the history of its
development, and the precision of its measurements are fully discussed by
White. 169 Potentiometers of moderate sensitivity are usnally suitable for
measuring the E.M.F. of thermocouples from which vapor-liquid equilib-
rium temperatures are derived. Portable, double-range instruments hav-
ing scales of 0-10.1 and 0-101 mv., and 0-16 and 0-80 mv. are available
commercially. * On the lower ranges of these instruments, the smallest
dial division of the slide wire represents 0.005 mv.: the temperature
equivalent is 0.1 °C. when a copper-constantan thermocouple is being used.
Interpolation will permit lcss than 0.005 mv. to be read on these instru~
* Leeds & Northrup Company, and RubiooIl Company, Philadelpliia, I)ennsylvania.

188 Thomas, in Temperature, Its Measurement and Control in Science and Industry, p.
159.
169 White, ill Temperature, Its Measurement and Control i11 Science and Industry, p. 265.
II. ORDINARY FRACTIONAL DIS'rILLA'£lON 279
ments, but unless more sensitive galvanometers than those built into the
instruments are used, interpolation is not justified.
Leads to the potentiometer should have low resistance to permit accurate
readings of E.M.F. If the thermocouples are made of fine wire, long leads
to the instrument may be of coarser low-resistance wire (No. 16 or 18 gage)
by use of the construction shown in Figure 52. The temperature of the
two cold junctions must be identical and known if true temperature read-
ings are to be obtained.
The advantages of the thermoelectric system for determining vapor-
liquid equilibrium temperature are apparent. Compared to Iiqllid-in-
glass thermometers, there is no error corresponding to emergent stem error;
the temporary or permanent changes in volume of the bulb of thermometers
have no counterpart in thermocouples; the mass of the sensitive part of a
METAL A
METAL a
HOT COLD
JUNCTION JUNCTIONS
Fig. 52. Circuit to provide low-resistance leads to potentiometer.
thermocouple can be made quite small and hence it responds to changes in

temperature more rapidly than it thermometer. The indicating instru-
ment can be located at a convenient position reasonably remote from the
column; when a number of columns are under the care of one operator, a
central operating position can be penntwently set up at which complete
thermal data on all the columns can be obtained.
Resistance Thermometers. The operating principles and methods of using
resistance thermometers have been discussed in detail elsewhere. 170.17!
As a means of measuring vapor-liquid equilibrium temperatures in distilla-
tion columns, they have not had wide use. At the National Bureau of
Standard:" however, they are used in preference to thermoconples. 172
For measuring vapor-liquid equilibrium temperatures in distillation
columns, the resistance thermometer offers the highest accuracy and maxi-
mum reliability and reproducibility. It is less convenient to use than ther-
170 Mueller, in 7'emperalure, lis Measurement and Cantrol in Science and industry, p.
162.
171 Weissberger, ed., Physical Methods of Organic Chemistry. 2nd ed., Interscienr.e,
New York, Part 1-1949, Part II-1950.
172 WiJlingh!lffi and Rossini, J. Research Nail. Bur. Standards, 37, 15 (1946).
280 A. L. GLASEBHOOK AND F. E. 'VILLIAMS
moelectric systems because each temppmture reading is derived from two

instrument readings, compared to one instrument reading for thermoelec-
tric systems; the latter, on the other hand, require proper maintenance of
the c~ld junction. The large physical size of resistance thermometers
compared to thermocouples makes their use less convenient on most col-
umns and impossible on very small columns. Their large heat capacity
compared to that of thermocouples reduces their speed of response to
temperature changes to a value far below that of thermocouples. Initial
cost of instruments for resistance thermometry is much greater than a suit-
able potentiometer and associated equipment for measuring temperatures
thermoelectrically. The latter system is an excellent compromise between
the high cost and reliability of resistance thermometry, and the low cost and
poor reliability of liquid-in-glass thermometers.
VI. PRESSURE CONTROL

1. General
Precise measurement of vapor-liquid equilibrium temperature has a
definite meaning only when the pressure at which the temperature is
760
w
a:
:::;,
(/)
~
0::
a..
a:
oa..
~
TI;M PERATURE
Fig. 53. Temperature-vapor preSRU1'8 curve.
measmed is closely controlled. If a pure compound is being distilled, for

example, the vapor-liquid equilibrium temperature is the temperature at
which the compound has a vapor pressure equal to the operating pressure
at the heacl of the column; if the operating pressure is lowered or raised,
vapor-equilibrium temperature wiII decrease and increase, respectively.
A. generalized picture of the manner in which vapor pressure changes with
temperature is shown in Figure 53.
II. ORDINARY FHACTIONAL DISTII,LA'l'ION 281
Consider T the ten.peril.ture at which the vapor prCS5Ul'C of a compound iR 1

atmosphere, or 7(}0 mm. of mercury. Smull change in temperature makes large
changes in vapor pressure in this region. At T' all equnl change in tem,pemture
will make a much smaller change in the vapor pJ'cssUl'e. In distillation columns
the pressure should be kept constant; changes in vapor-liquid equilibrium tempera-
ture then indicate a change in the composition of a distillate. If the pn',;;,qUl'e is not
constant, vapor-liquid temperature will change withont ncr:essluily indi.cating a
change ill the composition of the distillate.
I
Inspection of Figure 53 will show the importance of precise pressure con-
trol in column operation, and why the pressure must be more dosely con-
trolled at greatly reduced pressure than !tt atmospheric pressure if t,he
vapor-liquid equilibrium temperature is to have equal validity at both
operating pressures.
Many devices have been designed to control the operating pressure of
distillation columns. One investigatorl73 found about 30 pUblications on
pressure control covering the period 1840-1933; in the last Hi years the
number has nearly doubled. Many of these designs were related t,o specific
problems and were ent.irely satisfactory for the purpose. On the other
hand, the developments in high-precision thermometry require greater
precision in pressure control to utilize profitably the high preeision pos-
sible in temperature measurements.
2. Manostats
The majority of pressure-controlling devices operate on the principle of
t.he manometer and are called manostats. A simple design 174 is shown in
Figure 54. As the pressure is lowered, the mercury in the right leg of the
manostat approaches the adjustable contact rod, and finally, when contact
is made, an electric circuit through a relay is completed to stop the pump
motor or open a gas leak which slightly exceeds the capacity of the pump;
this permits a slight increase in pressure, the electric contact is broken, and
the pump starts, or the leak closes, and the cycle is repeated. The pressure
so established is an average value varying slightly above and below an ab-
solute value determined by the adjustment of the manostat; the extent of
the variation, assuming a constant temperature, is dependent upon the
inertia of the system as a whole.
The continuously adjustable manostat of Figure 54 has been modified to
operate at a number of fixed pressures by sealing in tungsten contacts at
different positions in the open leg of the manostat.176,176 To change the
173 Huntress and Hershberg, Ind. Eng. Chem., Anal. Ed., 5, 144 (1933),
174 Cox, Ind. Eng. ahem." Anal. Ed" 1, 7 (1929),
116 Sunier and White, C. M., Ind. Eng. Chem., Anal. Ed., 3, 259 (1931).
176 Willingham, Taylor, Pignocco, and Rossini, J. Research Nall. Bur. Standards, 35,
219 (1945).
282 A. L. GLASEBROOK AND Ie. g. WILLIAMS
operating pressures, one electrical eonncctioll is shifted to the appropriate

contlLct.
Greater precision of control is possible with a cascade system in which
a closely controlled operating system exl1austs into a roughly controlled
system at somewhat lower pressure. 17? The pressure differential of the
two systems ensures an exhaust mte from the closely controlled system
low enough to permit the man os tat to follow the pres-
sure without significant overshoot. The system, as
outlined, requires intermittent opemtioll of the evacuat-
ingpump.
To obviate the usc of an electric circuit and associated
relays, the manostat shown in Figure 55 was de-
veloped.17S-l80 Flow Df mercury from one leg to the
other in the manostat shifts the balance point of the
lever to open or close the bleed valve and to restore the
pressure to its equilibrium value. With a given capil-
lary, the regulator tends to "bounce" at certain pres-
sures. To eliminate this, suitable capillaries, covering
a restricted range of pressure, are used in conjunction
with a rubber pad whose angular position is adjustable
relative to the plane of the capillary tip. lSI This regu-
lator may be const.ructed with a closed-end manometer
instead of a stopcock and gas bubble if a restricted
range of control is permissible. ISO
The high density of mercury as a manostatic fluid
makes it undesirable from the standpoint of sensitivity.
To achieve greater sensitivity the manostat of Figure
56 was designed. 182 Sulfuric acid of density 1.71 is
used to give a sensitivity of 7.9 compared to mercury.
The maximum sensitivity with the acid is probably
not attained because surface tension maintains the con- Fig. 54. Simple
mercury manostat.
tact beyond the time when the general level of the liquid
in the contact leg is below the tip. The U-tube section of the manostat
must have a diameter about four times as great as a comparable mercury
manostat to counteract the greater viscosity of the sulfuric acid. Degas-
sing of the sulfuric acid b:y gentle boiling at 1 to 2 mm. pressure is necessary
In Palkin, Ind. En!!. Chem., Anal. Ed., '1, 436 (1935).

118 Schierholtz, Ind. Eng. Chem., Anal. Ed., '1, 286 (1935).
m Schierholtz and Thelin, Ind. Eng. Chem., Anal. Ed., 13, 908 (1941).
IBO Bailey, A. J., Ind. Eng. Chem., Anal. Ed., 15,283 (1943).
181 Dajin, G. A., Ind. Eng. Chern., Ana'. Ed., 15,731 (1943).
~2 Hershberg, E. B., and Huntress. E. H., Ind. Eng. Chern .. Anal. Ed., 5, 344 (1933)
II. ORDINARY FRACTIONAL DlSTILLA'l'ION
before the manostat will operate satisfactorily. Operation of the mano-

stat is simple; with the stopcock open, the pressure of the system to be
controlled is reduced to very nearly the desired value, the stopcock is
closed, and as the pressure continues to diminish in the system, contact it;
est.ablished and the pump is stopped or a leak is opened. A suitable elec-
o 5
CENTI METERS
Fig. 55. Mercury Imlunce mlLIlost:Lt. 1BO (A) I-mm. Fig. 56. Sulfuric acid man-
bol'o capilhry tuhing; (B) 8-nun. O.D. tubing; (C) ostat.lB2 (A) 2-Cill. LD.; (B)
19-mm. O.D. tubing; (D) spring (90 turns, No. 26 6-cm. LD.; (C) 2.5-cm. I.D.;
gage piano wire on 4-mm. mandrel); (E) pivot and (D) 17.5 cm.; eE) 5-CID. on
vacuum connection; (F) swivel joint; (a) adjusting centers; (F) O.025-mm. di-
thumbscrew; (H) supporting rod; (I) rubber cushion. ameter platinum contact
wire; (G) O.125-mm. diame-
ter platinum contact wire;
tron relay and solenoid valve for use with (H) tubes filled with mercury
the sulfuric acid manostat are easily con- to estlLblish contact betwecn
stl'ucted. 18S,184 platinum wire and external
circuit; (1) 12/20 standard
Another design 185 of manostat using mercury taper ground-glass joint; (J)
but having greater sensitivity than 'the con- 3-mID. bore stopcock i (K)
ventional type is shown in Figure 57. Opera- ground-steel joint.
tion is similar to that of the sulfuric acid type.
A very simple pressure regulator is shown in Figure 58. The controlled
pressure is equal to the depth of oil along the· bubble tube (measured in
mm. of mercury) J plus the pressure in the pump side of the manostat. 1&6,187
The Cartesian diver pressure control, based on the principles of a baro-
183 Jacobs, G. W., Ind. Enq. Chern., Anal, Ed., 7,70 (1935).
III Huntress, E. H., and Hershberg, E. B., Ind. Eny. Chem., Anal. Ed" 5, 144 (1933).
185 Ferry, Ind. Eng. Chem., Anal. Ed., 10,647 (1938).
186 Birch, Gripp, and Nathan, J. Soc. ChBm. Ind. Londan, 66, 33 (1947).
187 Todd, Ind. Eng. Chem., Anal. Ed., 20, 1248 (1948).
284 _\. L. GLASlmHOOK AND F. IG. Wn,LIAMS
scope,188 is diltgmmmed in Figure 59. The operating principle of this in-

strument ::mel the mathematics of its behavior have also been dcscribed. 189
Opcrt1tion is as follows: With the stopcocks open, the system is evacuated
to very nearly the desired pressure, and the stopcock is closed; as evacua-
tion continues, the pressure in the inner chamber increases relative to the
pressure in the outer cht1mbel' and raises the floating bell until the disc
closes the exhaust port to stop further reduction of the pressure in tll(l
system.
TUN
CHROMEL
Fig. 57. Meroury nlllllostat with enhanced sensitivity.185
Precise preStiure control at 10 mm. and below is somewhat difficult.

Mercury manostats become sluggish; manostatic liquids which wet the
glass walls overcome this if allowance is made for the viscosity of the usual
high-boiling liquids suitable for this purpose. A simple design of manostat
for 10-mm. operation and giving fair control at 1 mm. is shown in Figure
60. Diethylene glycol containing about 0.5% sodium nitrite in solution
provides sufficient conductivity to operate an electron relay for controlling
the pump. A thin layer of diamyl phthalate on the surface of the glycol
in each leg minimizes evaporation. A reference vacuum is provided on one
side of the manostat by a small mercury-vapor pump. Dry-ice traps are
provided between the manostat and the mercury-vapor pump and the
system under contra] to condense moisture and vapors which would dis-
solve in the glycol and change its density.
188 Caswell, Phil. Trans., 24, 1597 (170-0.
li9 Gilmout, Ind. Eng. Chern., Ana.l. Ed., 18,633 (1946).
J l. ORDINARY FRACTIONAL DIS'l'ILI,ATIOX 28.5
Pressure control at 1 mm. is possible with fair preeision (=+=0.01 mm.)

with the two-liquid manostat WO of Figure en. The upper layers are di-
butyl phthalate saturated with diethylene glycol, and the lmV(~r layer, di·
ethylene glycol saturated with dibntyl phthalate. A trace of sodium ni-
trite in the glycol layer provides sufficient conductivity to operate :1 COIl-
JOINT TIGHTNESS MAINTAINED

BY TWO SPRINGS
(ONE SHOWN)
JOINT PERPENDICULAR TO PLANE

OF DIAGRAM. THE APPARATUS IS
SUPPORTED BY AND IS PIVOTED
ON THE ..JOINT.
I
--j3 ern.f- 9mm.O.D.
3mm. I.D.
Fig. 58. BUbbling-type pressure reguhttor. 186 All tubing I-em. I.D. Tap
hits 5-mm. bore.
trolling electron relay. The sensitivity of the manostat is related to the

ratio of cross-sectional areas of the large chambers and the U.:.tube, and to
the relative densities of the two liquids. A reference vacuum and cold
traps are provided as described in the preceding design. Close control of
1BQ Williams, Ind. Eng. ahem., 39, 779 (1947).
28G A. L. GLASEBROOK AND F. E. WILLIAMS
the temperature of the manostat is necessary to avoid changes in the mutual

solubility of the liquids and to obviate density changes.
An entirely different type of pressure control for I-mm. operation is
based on the principle that at constant pressure the vapor-liquid equilib.
rium temperature of a boiling pure liquid is constant. 191 A suitable boiler
containing a small amount of diphenylmethane (b.p. 80°C. at 1 mm.) pro-
vides a vapor bath for a mUltiple-junction thermocouple system. The
TO VACUUM
TO SYSTEM
8 STOPCOCK
Fig. 59. Cartesian-diver manostat. Fig. 60. Manostat for 1 to lO-mm.

pressure using diet.hylene glyeol as
m!tllostat fluid.
thermoelectric voltage developed is balanced out ill a potentiometer circuit

with the aid of a mirror-type galvanometer. At the null point, the light
beam from the galvanometer circuit actuates a photoelectric relay circuit
which in turn closes a solenoid valve connecting the boiler to the vacuum
pump. If the pressure in the boiler and controlled system increases, vapor-
liquid equilibrium temperature increases slightly, the galvanometer de-
flects, and the solenoid valve opens to restore the balance. Inevitable
changes in the voltage of the potentiometer battery and shifts in the zero
setting of the galvanome~er cause a drift in the controlled pressure. The
lPl Fenske, U. S. Pat. 2,116,442.
II. ORDINARY FRAC'l'IONAL DISTILL!I.'l'ION 287
potentiometer circuit could probably be eliminated by having another pres-

sure system accurately controlled hy a conventional manostat.. A second
boiler attached to this system, and using a liquid having the same vapol'-
liquid equilibrium temperature at the pressure of the second boiler as the
diphenylmethane has at 1 mm. would provide a thenunl E.M.P. of equal
magnitude. The output of the two thermocouple systems when delivered
TO-=-_.r---.~~
CON-
TROLLED
SYSTEM
DIETHYLENE
GLYCOL
Fig. 61. Two-liquid m:wostat for l-mm. operation.lso
to the galvanometer in opposition would give no deflection until the pres-

sure in the I-mm. system increased when the photoelectric relay would
function as described previously.
A.. EFFECT OF TEMPERATURE CHANGES ON MANOSTATS
Temperature changes of a manostat affect the absolute value of a pressure

being controlled. A manostat of the type shown in Figure 54 is subject to
288 A. L. GLASEBROOK AND }'. E. 'VILLIAl\IS
three effects by temperature: (a) t.he total volume of mercury increases;

(b) the glass expands t.o increase its volume and to a slight extent compen-
sates for a; and (c) the adjustable contaet arm increases in length with in-
creases in temperature.
If in Figure 54 a horizontal line is drawn from the contact point to in-
tersect the left" or measuring, leg of the manostat, any expansion of the
mercury above this line will not affect the pressure because the product of
density, d, and height, h, ,vill remain constant. The mercury in the U-tube
sedion below the horizontal line will also ex-
pand to force mercury above the line and
hence increase the numerical value hd. The
adjustable contact arm will likewise increase
in length and have the effect of forcing still
more mercury above the line and increase hd
still more. The net effect of the increase in
temperature is an increase in the controlled
pressure. Expansion of the contact arm can
be completely elimina ted by the construction
shown in Figure 60 j the outer tube will in-
crease in length with temperature and tend
to raise the contact point, but the inner tube
will similarly increase in length to hold the
tungsten contact point in a fixed position.
Substitution of this contact point for the
metal contact arm of Figure 54 will eliminate
a part of the temperature effect. A mano-
stat, modified as indicated, to control pres-
Fig. 62. Fixed·pre8~ure, tem-
sures over the range 50 to 760 mm, of mer-
perature-conlpensated mercury
manostat. cury will give fair control at the higher pres-
sure and poor control at the lower pressure.
For example, if the manostat, when set at 760 mm., has 21 % of the total
volume of mercury in the U-tube section, an increase of lO°C. in tempera-
ture (from 25-35 0 ) wlll increase the controlled pressure to 760.4 mm. i
when adjusted to control at 50 mm., the U-tube would contain 95% of the
total mercury, and an increase of lO°C. in the temperature would increase
the controlled pressure to 51.65 mm. of mercury. It is apparent from
these results that the mercury in the U-tube section below the measuring
arm is chiefly responsible for the changes in controlled pressure resulting
from increased temperature.
A manostat constructed as shown in Figure 62 will be compensated to
changes ill temperature (excluding the slight effect of expansion of the
glass), as the following analysis will demonstrate: .
II. ORDINARY l~nAC'rION.\.IJ DIS'rILLA'rIOi\ 289
Let Y equal total volume of mercury ut sOllle stambrrl temperatllre T,; a is

volume of mercury in height h; b is volume of mereul'Y below height h; then V is
+
a b; A is area of upper bulb, 7l"D2j4; a: is coefficient of exp:msiOll of mercllry;
ds is density of mercury at Ts; p, = h. is pressure at T.; Pt is pre3S1Il'e at another
temperature (T. + At) = he; deis dCIlsity of mereury at (T, Ill): +
and PI = dlhl = (h., + il.h)(el" + il.rI)
When the manostat is completely compensated:
Ps - Pt = 0
and:
dA - (lb. + il.h)(d, + del) = 0
Simplifying:
h, Ad = - Ilhd, (1)
Now ds = M/Y s and dl = M/Ft, where M is the llln~s of mercury in the mtUlostat,
Then:
But:
dlY I = (d, + Ad)(Y. + AT')
il.d = -d.(il.V)/Y (lL)
By defillition:
a = (Ily)/ il.tV
Therefore:
(b)
(a) and (b) may be combined and simplified to:
Ad = -clsa:6.tl'/V = ...,cl,a:At (2)

Since point P is fixed:
!!;/i. = il.Y/A = aAtV/A (3)
Substituting (2) and (3) in (1), Y = h,A, which is the equation for expallsion of
mercury only.
Equations (1) and (2) are independent of expansion of glass. To allow for the
latter, equation (3) must be modified as follows: Let f3 = coefficient of cubical ex-
pansion of gbss. Subscripts Hg and gl. refer, l'espectively, to merclIty and glass.
(4)
2\)0 A. IJ. GLASEBROOK AND F. E. WIIJLIAMS
AVH. = dAW 1
A V.I. = f3 AtF by definition
AA gl • = 2/sM(3A
Uhgl. = 1/3tJ.t(3h.,J
Substitl1ting; these definitions in equation (4):
_ tJ.tV(O! - (3) + l/a At (3h (5)
Ahug - A(l + 2/3D.t(3)
Substituting (2) and (5) in (1):
[_AtV(~
h(-d.",At) = -d.
s , (1 + 2/a-D.t(3)(3) + l/s At f3 h]
lisci = nO! - (3) + l/a{3h.,
A(l + %t:.t(3)
h,(d - 1/3(3) = (~) (1 ~ ~3:t(3)

T
l' =
hoA(", - 1/3f3)(1
-
2/atJ.t(3) +
a - Ii
The terlll (1 + 2/3tJ.t(3) is very nearly unity:

y = h.,A(d - 1/3(3)
a-{3
For Pyrex:
(3 = (3.6 X 10- 6)3 = 1.08 X 10-·
For mercury:
'" = 18.186 X 10- 5
1" = ~8A(17.826 X 10- )

5
= 1.042h A
17.106 X 10- 6 '
The compensated manostat of Figure 62 maintains its compensation if

tilted in or out of the plane of the paper. This permits a fine adjustment
of the absolute pressure at which the instrument controls. To avoid capil-
larity effects, the open leg of the manostat should be constructed of tubing
15- to 20-mm. I.D., and the closed side of 40-mm. I.D. tubing for mano-
stats controlling at 50 to 200 mm.; for controlling at 725 mm., for example,
the manostat may be constructed entirely of 15-mm. tubing. If the COll-
trolling contact is placed in the closed arm of the manostat instead of the
open arm, the instrument fails to compensate for temperature change.
Manostats of the above design have been constructed to operate at 50
II. ORDINARY FRACTIONAL DlS'rILLATION 291
nun. and above. At lowcr presslU'CS, the volume VI becomes too small to
permit satisfactory glass-working operatiullfi.
Manostats of the type shown in Figures 55-57 and 59 are more seriouslv
affected by changes in temperature because the trapped gas in the close~l
arm expands or contracts in accordance with the gas laws. Consequently,
the controlled pressure will change about 0.3% per degree of temperature
change. Unless these types of manostats are thermostated, they are not
reliable for long periods of operation.
B. MISCELLANEOUS FACTORS IN PRESSURE CONTROL
Elimination of temperature effects on manostats, either by temperature

control or compensation, ensures precise control of pressure only when the
inertia of the whole pressure system permits a manostat to follow pl'essure
changes very closely. If lcakage into a pressure-control system is rapid, the
exhaust pump will of necessity function frequently to maintain the desired
pressure as controlled by the manostat. If the frequency of pump operation
is of the same order of magnitude as the natural frequency of oscillation of
the controlling manostat, the latter will be kept in a state of oscillation.
As a result, the absolute value of the controlled pressure will vary above
and below the nominal value at which the roanostat should control. In-
asmuch as the frequency of oscillation of a column of liquid is an inverse
function of the square root of its length, a manostat designed to control at
725 mm., for example, will be thrown into a condition of oscillation more
readily than will a manostat designed for 50-mm. control. (This assumes
that the length of liquid column in each instance is roughly proportional to
the pressure being controlled.) Precise control of pressure requires, there-
fore, that the leakage be small, so as to necessitate infrequent operation of
the pump. By increasing the tot111 volume of the controlled system, a given
amount of leakage will change the pressure more slowly and require less
frequent operation of the exhaust pump; by the same token, the pump will
reduce the pressure more slowly to nominal value.
Large surge tanks 192 of 30- to 50-gal. capacity, insulated and properly
located where ambient temperature changes are slow, are necessary if con-
trol is to be precise. Smaller surge tanks may be used jf the exhaust rate is
sufficiently reduced by insertion and proper adjustment of a throttling
valve between the pump and surge tank. A slight amount of leakage, pref-
erably of carbon dioxide or nitrogen deliberately introduced into otherwise
tight systems in order to activate the exhaust pump for 1 or 2 seconds of
each minute, will obviate the possibility of subnol'mal pressures, should the
192 Willingham and Rossini, J. Research Nail. Bur. Standards, 37, 15 (1946).
292 A. L. GLASEBROOK AND P. E. WILLIAMS
surge tanks drop in temperature; increases in the temperature of the tanks

will be taken care of autom[Lticnlly.
Motor-driven pumps coast for a fcw revolutions after the power is cut
off by action of the manostnt. The effect of this after-pumping may be
eliminated by placing a solenoid valve between the pump and surge tank
and connecting the solenoid in pamllel with the motor circuit so that the
valve opens when the motor is energized and closes immediately when
power is cut off. At pressures near atmm;pheric, the exhaust rate of con-
ventional oil pumps is too great for preciRe control; reduction of the pump
speed to I/O to l/tO of itt! normal value by means of a speed-reducing system,
and use of It solenoid valve mentioned previously permits satisfactory con-
trol.
Convenient location of surge tanks usually requires long lines to bring
the controlled pressure to a fractionating column. For permanent installa~
tion 1.5-in. copper tubing should be used. This should be soft-soldered to
suitable copper or bronze T's; dia,phragm valves 193 may be used at each
T to provide a vacuum outlet near each column.
VII. MISCELLANEOUS ACCESSORIES

1. Joints and Stopcocks
The most unsatisfactory parts of a laboratory distillation column are
the joints and tltopcocks. Rtandard interchangeable joints and stopcock
plugs, althQugh they arc within the allowable manufacturing tolerances,
permit unavoidable lORses of distilland or oxidation by air leakage,
======)J
============~~~=====~
Fig. 63. Longitudinal section of mltle and Fig. 64. Longitudinal seotion of male and
f('ma.le stanrlard·tltper joints. female spherical joints.
118 Keroteat Manufacturing Comp!my, Pittsburgh, Pennsylvania.

II. ORDINARY I'RACTIONAL DI8TILL,~TIO:N 293
especially at redllcrd pressures. Sometimes t1 lubricant call be uHed to

prevent the losses Of leakage when ('(~rtain typps of eompounds are being
distilled, but tihese illRfi:mees am excep1 ional. especially ut higher temperu-
tUl'es and low pressures. The simplm.;t solutlon to the prohlem of joints
is to connect the still pot and condenser seetion to the column by fused glass
seals. The design shown in Figure 50 is satisfactory because the one lubri-
cated joint can he kept. sufIicient,ly cool to be out of cont.act wit.h significant
concentrations of vapor or liquid. In other parts of t.he column and auxil-
iary equipment where the concentration of vapor is low, or where cont.act
of the distillate with the lubriented Rlufaces is minimized, joints nnd stop-
cocks are usunlly satisfactory, provided a lubricant insoluble in the distil-
lttte is used for lubric:.1tion.
J Dints are either of tapered construction (Fig. 63) or spherical (Fig. G4:).
They consist of male and femlLle sections and are availahle in a variety of
standard sizes. Spherieal joints offer eCl'tain advantages in lIse j the two
sections nced not be so preeisely lined up when equipment is being as-
semhled and clamped, strains are less likely to develop, and the sections
are more easily sepltrat.ed than arc tapered juints.
2. Lubricants
Glycerol-starch has been found suitable as a stopeoc:k lubricant when

nonpolar compounds are being hnndled. 104 Sucrose or mannitol in glycerol
containing 1-3% of medium-viseosity polyvinyl alcohol has been recom-
mended for the same purpose. IUD A partinlly esterified and polymerized
mixture of tetraethylene glycoll1nd citric acid 196 or the SlLme mixture modi-
fied by the addition of cellulose acetate 197 have been used as lubricants when
working with aliphatic and aromatic hydrocarbons. Alcohols, ketones,
and water' attack this mixture. The resinous mixture formed by reaction c
of sebacic acid and ethylene glycol hI insoluhle in aliphatic hydrocarbons,

alcohols, and diethyl ether, but is soluble in benzene, pyridine, and alkyl
halides.l 98 Polymerized phthalate esters of di- and triethylenc glycol are
effective lubricants in contact with aliphatic hydrmm.rbons. 198 Proprietary
mixtures availuble from many suppliers of laboratory equipment are geneI'·
ally of limited usefulness as stopcock lubricant,s.
194 Herrington and Starr, Ind. Eng. Chern., Analc Ed., 14, 62 (1942).
116 Puddington, Ind. Eng. Chern., Anal. Ed., 16,415 (1044).
)98 Sager, Ind. Eng. Chern., Anal. Ed., 4, 388 (1932).
197 PearIson, Ind. Eng. Chern., Anal. Ed., 16, 415 (1914).
198 Bruun and Schicktanz, J. Research Natl. BUT. Standards, 7, 871 (1931).
2\J-1 A. L. GLASEBllOOK ,\ND F. E. WILLIAMS
3. General Suggestions for Mounting Equipment

When one or more columns are to be se!; up permanently, a suitahle
framework should be instl111ecl to hold and pl'oter:t the eolumn. One satis-
factory design is shown in Figure GS.
Variltble transformel's* for c:ontl'olling the power input to the column
Diameter 1/4"
Seamless less tho n I.D.
steel of steel tube
tube
I" longer than

length of vacuum jocket
r- 18 " - - \
3/IS" steel
top plate
ts"
C<;lver 3 sides and
bottom with Yill" \
sheet metal. Sliding
door in front ~ USe I" x I/S"
angle iron
4"
t
Fig. 65. Self-supporting framework for mounting distillation column.
may be attached to the column panels or set on top of the base. If several
columns are located together, the variable transformers may be more con-
veniently mounted in standard steel cabinets.
The receiver may be attached to steel bars bolted to the column frame.
A single-turn glass spiral about 3 in. in diameter made of 6-mm. tubing
should be used to connect the product line to the receiver. This provides
enough flexibility to prevent breakage when the column expands at ele-
vated temperatures. For the same reason, the reflux: condenser should be
lightly clamped.
* Available from: Genel'!ll Radio Company, CambJ'idge, Massachusctts, and Super-
ior Electric Company, Bristol, Connecticut.
II. ORDINARY FRACTIONAL DIS'rILI.ATIO~ 295
Pul't: 2. P ROCED URE
1. INTRODUCTION
A batch distillation of the ,-,i1l1plest possible mixture, that of hm w,jatile

components, is not strictly amenable to theoretieul treatment, as is shown
in the chapter on theory. The complex interrelations of column cfIicicncy,
reflux ratio, column holdup, and ideality or nOlli<lcality of the mixture be-
come still more complieated I1S the number of components in the dii>tilland
(or charge) increases. :B'ol'tunately, so far as the practical application of
batch fractional distillation is coneerned, the shortcomings of theOl'y do not
greatly reduce the value of distillation as [L laboratory technique of great
power and u"efulnes~. Some theory, an adequate column, and the proper
operating procedure nrc sufficient to resolve all but the most complicated
mixtures of volatile cmnpoUllc]s and tLzeotropes.
This seet-ion discusses the proeedllre to be followed in making fL fractional
distillation. Tho choice of eolumn, method of sampling und charging, the
preliminary operations of pre flooding and estl1blishment of initial equilib-
rium, choice of reflux l'a.tio, collection of product, and interpretation of data
will be the chief points of discussion. Certain miscellaneous items found
helpful in many di"tillations will also be included.
II. PURPOSE OF FRACTIONAL DISTILLATION
Batch fractional distillations on a lahoratory seale usually have OIle of

two purposes: to determine quantitatively the composition of the distil-
land, or to isolate one or more of the components from a mixture for further
examination.
An analytical distillation is not of itself sufficient to establish the quan-
titative analysis of It mh.iure in many instances, particularly those involving
complex mixtures such as hydrocarbons or any mixtures which fonn azeo-
tropes. In such eases, an analytieal distillation serves to isolate and re-
move interfering components, 01' to concentrate eomponents present in
small amounts, and to permit other analytical means sueh as infrared, ul-
traviolet, or mass-spectral analysis to establish the final analysis with cer-
tainty ::tnd precision. Sometimes, this procedure is followecl of necessity
because columns having fractionating efficiency adequate to separate
every component are not available; in other instances, a combination of
distillation and other analytical means requires much less time than would
a complete analytical fraetionl1tioll when a column ha,ving adequate ef-
ficiency is available. .
It is obvious that all analytical distiUution eun accomplish both purposes
if a charge of sufficient size is used for the fraetionation. Normally, analyt-
29G F. E. WILLIAMS
ical fractionations are made on quantities of a few milliliters to several

hundred milliliters; considerably larger volumes are employed when
significant quantities of a component are to be isolated.
Depending on the purpose of the distillation, the column efficiency re-
quired may differ when a given mh. ture is being fractionated. If the frac-
tionation is for analytical purposes, the collimn efficiency required will be
governed by the relative volatility of the most closely boiling components
whose separation is desired; if from the same mixture only one component,
for example, is to be separated in high purity, Itnd the volatility of this
component is large compared to all other components in the mixture, a
column of lower efficiency and greater productivity will be satisfactory.
III. CHOICE OF COLUMNS
When small still charges are to be fmctionated, columns with low holdup
are desirable, lwd for analytical distillations, they are a necessity in the
intercst of sharp separations. In the majority of columns, particularly
packed columns, low holdup implies small size, which limits throughput
and, consequently, product rate for a given reflux ratio. If small quantities
(20-200 m!.) of material are to be fractionated, smf111 coneentric-tube col-
umns provide the best solution for the problem. (See Figure 10, Part 1,
this chapter.) Packed columns of low holdup sueh as the Hyper-Calor
the column of Figure 12, Part 1, this chapter, are suitable for eharges of
200 ml. or somewhat larger. Larger still charges can be satisfactorily
fraetion[Lted in columns of considerably greater holdup; boil-up rate can
be increased greatly and product rate likewise increased for a given reflux
ratio. Total distillation time will not be significantly different in the two
cases if the ratio of charge to holdup is about the same. Certain advan-
tages accrue in the use of larger columns and proportionately larger charges
when sufficient material is available. Components present in the charge in
small proportion relative to the holdup will not usually be isolable with
either the large or small column. Evidence of their presence may be in-
dicated by the character of the distillation or the refmctive-index curve.
Irractions from a large charge suspected of having such components will
frequently be of sufficient amount, when combined, to be refractionated
through an adequate column of low holdup to determine their composition.
This procedure was used in studying 5% fractions derived from gallon
charges of heptane alkylate containing 11 eomponents. 199 In working
with large clmrges, handling losses will be a smaller fraction of the charge,
and the material balance will be better than in the case of sm!111 charges.
199 Marschner and Cropper, Symposium on High-Tempomture Analytical Distillation,
America.n Petroleum Institute, Meeting at Chicago, Illinois, Nov. 11, 1946.
II. OIWINARY FRAC'l'IONAL DIS'l'ILLA.'l'ION 207
Although small holdup i~ neeeSSlLl'Y for sharp SCllllrutiolls, adequate

fractionating efii.eieney is required to ellHure a prescribed degree of separa-
tion. Unknown mixtures can be most profitably examined Hllalytiea.lly in
a column having an effieieney of the order of 100 theol'etit:al plate:,;; results
of this investigat.ion lllay then indieate Ui:le of a Im.;s efficient, lIlore produc-
tive eolnmll for flll'ther work. More COlllmonly, Bome knowledge exists
uf the qualitative eompositioll (If a mix1;ur'e to be .':icparat(~d. This muy be
deduced fmm the source of the material, the ehemieal reaetioll by which the
mixt.ure was produce!l, or by a qualitative ehemieal.examination to m;tab-
lish the key eompOlwuts. The [Lctual efficiency chosen for it given frac-
tionation i:.; determined by the relative volatility of the components of the
mixture to be separated. Freqnently, I'eln,i;ive volatilities arc not known,
although differeueefi in hoiling points are known. In either case, Table
XVIII of Chaptpl' I will be n helpful guide in (\hoosing an adequat.e column.
It shonlc[ be bOJ'lle in mind, when considering differences in boiling points,
1hnJ relative volatilities may tlhift markedly ill a way not ind catcd by dif-
rerCll(·l~fi in boiling poinb;. For example, in a mixture of toluene and
methylcyelohexane, the eomponcnts of which boil lOoC. apart, the x, Y
diagmm shows a sharp "pinch" in the vapor composition curve for those
mixtures eontaining lnrge mole fractions of methylcyclohexHne. Separa-
tion of these components would accordingly require grea.ter column ef-
ficieney than would he expeetell for component.s of more normal behavior
boiling lOoC. apart.
TABLE XXIX
lldiux l'aLill
- - - - - - .... ----~-
28 30 37
go:! 24 24 30
40:1 22 22 25
10:1 10 14 15
4:1 ti 8 7
---~------
Inasmuch as column efficiency is ordinarily measured under conditions

of total reflux and is accordingly not strictly a measure of the working ef-
ficieney, or efficiency when product is being withdrawn, due allowance for
this change in efIieieney must be made in l·.hoosing a column for a given
separation. This point has been discussed in Chapter I, and experimentally
studied by Goldsbarry and Ackevold. 200
Table XXIX gives some results derived from graphs published by these
200 GoldHbarry !tud Aekevold, Symposium on High-Tcmpemt,ure Aualytical Distilla-
tion, Amel'iean Petroleum Institute, Meeting at ChiclLgc), Illinois, Nov. 11, 19'16.
298 F. E. WILLIAMS
workers. Column 1 describes performance of an Oldershaw perforated-

plate column; column 2, a helix-packed column; and column 3, a Heli-
Grid packed column-each being 1 in. in diameter with 4 ft. of packed
section. Data for only one value of liquid-return rate, 2000 mI. pel' hour,
are given. Liquid-return rate is defined as the volume of reflux returning
to the column plus the internal reflux resulting from imperfect insulation.
Plate equivalents are defined as a quantity numerically equal to the number
of theoretical plates, operating at totall'efillX, which would give the separa-
tion obtained when the colullln is operating at a finite reflux ratio.
The data of table XXIX apply only to a test mixture of methylcyclo-
hexane find n-heptane, or other binary mixtures of comparable relative
volatility (0: = 1.07). At the reflux ratio of 4: 1, the very low plate equiva-
lents for the eolumns indicate that the eontrolling factor in the separation
attained was the reflux ratio, and not the number of plates.
Of consicll't'ttble interest and usefulness is the faet that column 3 of
Table XXIX had a plate equivalence of 75 at a liquid return rate of 600
m!. per hour and a reflux ratio of 40: 1, 01' three times the efficiency that
the column had at 2000 mI. per hour and a reflux ratio of 40: 1. In practical
work, thjs means that the column could make an excellent separation on a
close-boiling mixture at a low liquid-return rate, while at a much higher
liquid-return rate, the same column would make only a fair separation.
Vacuum-jacketed columns packed with metal helices have shown similar
behavior in the author's hboratory,
Reliable data OIl column efficiency and the effect of boil-up rate on the ef-
ficiency can be obtained only if the utmost care is used in the purification of
the components of the binary pair used in the test mixture. For the test
mixture of n-heptane and methylcyclohexane, the heptane should be ob-
tained from a source other than petroleum to avoid possible presence of
isomeric hydrocarbons. The n-heptane should be fractionated through a
column of at least 100 theoretical plates at a high reflux ratio; only those
cuts having the correct refractive index and correct freezing point should
be used for the test mixture. Methylcyclohexane prepared by hydrogena-
tion of pure'toluene is preferred to that isolated from petroleum. This
material, before use, should be washed with several small portions of con-
centrated sulfuric acid until the acid layer is no longer colored. Following
this, the hydrocarbon should be washed with water, dilute sodium hydrox-
ide, and finally again with water, and then dried. The final step is to frac-
tionate the washed product through a column of at least 100 plates at a high
reflux ratio. Only those fractions having the proper refractive index and
freezing point should be used in the test mixture. Unless these precautions
are taken, efficiency tests, particularly of columns of high efficiency, may
prove erroneous.
II. ORDINAIW ]'RAC'l'IONAL DISTILLA'l'ION 299
Choice of packing material is determined by the cha.racteriBtics of the

material to be distilled. If this is not corrosive, metal packing Inl1,y be satis-
factory. Organic halides, acids, anhydrides, sulfur compounds, and certain
phenols corroele metal packing to a greater or les~er degrec, Hncl penna-
llcntly reduce its efficiency; glass or other inert packing is essential when
working with these types of compounds. Alcohols are dangerously corro-
sive to aluminum plwking. If the material being fractionated has it tend-
ency to polymerize and form an insoluble coating on the packing, glass or
other inert packing is advisable inasmuch as it can bc cleaned by usc of
drastic reagents.
If a distillation is to be successful, the thermal and chemical stability of
the materials in a charge require careful consideration. Polymerization,
cleavage, dehydration of materials containing hydro}.'Y compounds, chemi-
cal reaction between components of the distilland, and oxidation through
leakage of air into the still may change the composition of a distilland and
vitiate results. Reduction of the operating pressure to lower the distilla-
tion temperature will usually reduce or eliminate several of these effects;
air leakage into the still may increase at reduced pressure, but this can
easily be remedied by eliminating the joint between still-pot and column.
To diminish still-pot temperatures significantly, operating pressures
should be reduced from atmospheric to 100 or 50 millimeters. Materials
boiling at 150°C. at 760 mm., for example, will boil at about 90 and 70°,
respectively, at the above reduced pressures; even lower pressures may be
necessary in some instances. Pressure drop across the column will increase
as the operating pressure is lowered if a given boil-up rate is maintained.
Generally, however, the permissible boil-up rate is much less at greatly re-
duced pressures than at atmospheric pressure. A rough guide for predict-
ing permissible boil-up rate when the boil-up rate at atmospheric pressure
is known is to multiply the boil-up rate at atmospheric pressure by the
square root of the pressure ratio. For example, if a given mixture in 11 given
column can be evaporated at a rate of 1000 ml. per hour at atmospheric
pressure, the rate at 76 mm. will be 100ov6}o = 310 m!. per hour. Per-
missible boil-up rate is likely to be somewhat less than this as a result of
increased pressure drop across the column; with tall columns; the square-
root rule is more adversely affected than with short columns. With some
mixtures, low-pressure operation may introduce a problem in condensation;
circulation of ice water or acetone cooled by dry ice, through the reflux
condenser, will usually ensure satisfactory condensation.
Some mixtures may give distillates which solidify in the condenser sec-
tion. If their melting points are substantially below their boiling points
at the operating pressure, proper control of the temperature of the con-·
denser will allow the distillation to proceed normally. If the solid com-
300 F. E. WII,T,TAMS
poncnts boil below their melting point (HubliulP), tllP diHtillation will hnvr
to be condlleted azeotropically to remove Rueh materialH as liqllirl fLzeo-
tropes.
IV. THE nISTILLAND

1. Pretreatment
Hyclroperoxides and peroxides may be present in large m small amonntll
in many mixtureR, plU'ticularly hydrocarbons, aldehyde:,;, and ethers. To
avoid the pORsibility of an explosion during the distillation, large amountH
of hydroperoxides should be removed by suitable pretreatment; nnd small
amounts should be eliminated to prevent pcrnxide-initiated polymel'izlttion
of other components of a disWland.
Hydropel'oxidefl may be detected by shaking t.he sllspected material
with potassium iodide solution acidified with acetie acid. (HOlne materials
having very low soluhility in water may require glaei~tl acetic acid solution
to respond to the iodide tec;t.) If the material relelUles iodine, the hydro-
peroxide may he decomposed by :.;haking the material with several batches
of ferrous Rldfate, alkaline Hmlium sulfide, or sodium hisulfite imlution until
the test with acidified iodide is negative. Raney nickel may also be user!
to decompose hydro peroxides and pp,roxilies. Judicious heating of watm'-
soluble matel'iab containing peroxides with sodium hydroxide solution i:-;
also effeetive in destroying these compounds. Possible undesirahle effect.s
of any reagent on the other components of a eharge should be consider('(l
before the reagent is used to remove hydroperoxi(les and peroxides.
In general, peroxides a.re present to f:4ome extent whenever hydropel'OxicleH
are cIete<)ted; since the former ~Lre not, so readily decomposed as hydroper-
oxides, they are likely to reach a dangerous state of concentration in the still
pot as the distillation progreRRes. By the addition of a large quantity of
high-boiling chaser to the still pot, the eoncentration of the peroxide is kepi'
low, and hence it cannot attain an explosive eoncentration. The. progres-
sively higher tempemture of the (\Qutents of the still pot as the distillation
proceeds gradually decomposes the peroxide into harmless produets. Any
pretreatment to remove hy;dl'Operoxides from a still charge will produee
alcohols or ketones in the distilland. These may distil as sueh, or they may
form :1zeot.ropes with other eomponents present. This possibility sh[)ul~l
be recognized when interpreting the distillation elata.
When working wit,h easily dehydrated hydroxy compoumls, such as ter-
tiary alcohols, slight traces of aeidic mat.erial may be present in the still
pot or column; if present, those may catalyze dehydration and ehange the
composit.ion of the charge. A dilute solution of sodium methoxide in
methanol poured. through the column, followed by removal of the methanol
II. ORDINAltY ],RAC'l'lONAL DIS'i'ILIA'l'ION 301
by 1:1 t:!tream of dry nitrogen, ,'lill frequently condition the column to obviate
dehydration. However, in such cases, due con~idemtioll must be given to
the pOHl'iibility of iuiLiating con(lemmtion reactionH by [LIlY sodium meth-
oxide adhering to the packing. Exc:eHS Hodium methoxide solution should
be removed from the Htill pot before introducing the charge.
Occasionally a fractionation has to be performed 011 a mixture which
eOlltaim; components ,vith a pronoulleed tenderH:Y to polymerize either
slowly or rapidly. Suitahle inhibitors may be added to the distilland at the
beginning of the distillation to prevent or minimize polymerization. The
propel' inhibitor to mle will depend on the ehemical ehamcter of the c:om-
pound or eompounds in the distillaml, and henee will be a specific problem
in each instance.
It, should be pointt~d out, that. an illhibitlll' will ordinarily be effective
only in the l'lt.ill pot and, possibly, to a slight extent in the I()IVeHt part of
the column, depending on the volatility of the inhibitol', which is usually
luw. To ein:ulllvent tIliH limitatioIl, a solution of a suitable inhibitor in it
part of the distillate may be added dropwise throllgh the reflux condenser
into the reHux Htream to provide inhihitor throughout the length of the
column. Ina:;much (t:-J many polymerizations are initiated by small traceH
of peroxidE', the impol'tanee of rigidly excluding [LIl leabge of <.til' into the
lower part of eoluUln through a still-pot joint is apparellt.
All charge:; to a fraetionating (:olumn should be single phaHc with respect
to physieal state exe(~pt for two-phase sYl-lterntl, which become single phase
by the time the nharge reaches boiling temperature. A system which re-
mains two phaHe at its hoiling point will probably bump severely when boiled
and pnwmlt KatiKfaetol'Y adjustment of the boil-up raLe. Even if the two-
phaHc sYRtem beeomes single pha..,e at its boiling temperature, [LS sometimes
O('·(·,lU·H in uzentropie distillationK, two ph!ltlCH may form in the reflux eOIl-
denser, making it cliffieul1. to maintain a uniform pl'f)(hwt rate; if t.he pl'od-
net. iK takon off intermittently by the UHC of Kpeeial heu(lK, sneh aK th£1
tilting-funnel type, tL fairly unifurm pl'mlud will be {lolleeted, (WClll though
[,wo phUHC.
In ,york with hydroearbons, water iH llKually pl'm;ent to some ext,ent.; a
water layer can be mechanically Hepal'1:1ted, and KUHpended and dissolved
watf1l' removed with a small amount of deHiecant. Anhydrous potassium
earbonate, caleium chloride, malJ;nesium sulfate, calcium sulfate, and copp~r
Hnlfate are suitable fol' drying hydrocarbonH. If alcohols llre preHellt, eal-
cium llhloride Hllould Hot be Ilsm! bemulse the alcoholH forlh Itddition prod-
ucts with it. If kotones nrc present, potassiuIll carbonute may eau~~
eondensation reaetiuns. In doubtful eases, anhydroLls calcilUll sulfate ii:l
least likely to cause undesirable reactions in !L mixture, In instances ill
whieh none of the hydrocarbons present form a~eotropes with isopropyl
302
aleuhol, this compouIlll may be added to remove the water as an azeotrope.

Thi:;; pl'oeedure has the advantage that at !Lny given pressure the azeotrope
is of definite compositjon and boiling point and can be withdrawn as a defin"
ite product of normal behavior. When isopropyl alcohol is not; or may not be
used, a watm'-hydrocfl.l'bon azeotrope may condense in the reflux condenser.
In such a ease, the water adheres in druplets and is slowly eliminated with
the product, meanwhile causing wide fluctuations in vapor temperature
and difficulties in regulating the product rate. When water does collect
in t110 conden~er, it may he eliminated by shutting off the coolant for a
short time tu allow the condenser temperature to rise. This proeedure may
have to be l'epentexl several times to eliminate all the water. The two-
phase hydrocarbon mixture in the receiver is then separated mechanically,
and the hycil'oearbon layer i~ dried with a suita,blc reagent andretul'ned to
the still. A small chamber connected to t.he top of the reflux condenser
through a valve, 01' a separat.ory funnel tLttached to the charge tube of the
still, can serve for this purpose, and for other purposes mentioned in Sec-
tion V-2.
Mixtures of water-soluble and -insoluble compounds, such as reaction
products which contain eonsiderable amounts of dissolved water, may be
difficult to distil, particularly if the \\'teter-soluble material is considerably
lower boiling than water. When the lo\v-boiling material has been elimi-
nated, a two"phase system remains in the still pot and will bump; fre-
quently, an azeotrope of water and other material will distil. Use of iso-
propyl alcohol, as explained above, will eliminate trouble during the distil-
lation. Due consideration must be given to possible formation of a ter-
nmy azeotrope of water, alcohol, and a third component, and of a binary
azeotrope of (lloohol and another component. A compilation of binary
and ternary azeotropes 201 will be found useful in this connection.
2. Size of Batches. Chasers. Charging
As was discussed in Chapter I, the proper size of a charoc in comparison

to column holdup is still an unsettled question. Until the problem has
been more completely investigated, the standard pnwtice of maintaining
the ratio of charge to dynamic holdup at 10 or above should be followed for
satisfactory results in high-temperature distillations. If the volume of
sample is only n few milliliters, a special eohlmn of low holdup will be
needed (see Part 1). A series of eolumns of a given diameter, type, and
size of packing will have efficiencies and holdups roughly proportional to
the height of the paekings; aceorciingly, a distillationl'eqniring the highest
201 Horsley, A11al. ('hem., 19, 508 (1947).

II. OHDINARY FHACTIONAL DISTILL.\TIO)1 303
efficiency will require the largest sample, and one requiring; the lowest ef-
ficiency will require the smallest sample if the above rule for sample size is
followed.
In any batch fractionation, assuming that the chttrgc is completely dis-
tillable, all the charge can never be collected as distillate. The uncollect-
able material is equal to the dynamic: holdup of the column if the still pot
is ttlken to dryness; but this is a practice not to be rec:ommended with any
type of still, especially if an internal electric heater is used in the still pot.
The column holdup plus any volatile materitd. remaining in the still pot.
may be dist.illed out if l1 chaser' is added to the charge to drive them into
the receiver. Any chaser used should be nonreactive with the material
being distilled, should not form azeotropes with it, and should not attack
the column or packing. Its boiling point should be considerably higher
than the boiling point of the highest component of the charge in order that
any of the vttporized chaser may be easily fractionated out of the vapor
stream entering the bottom of the column. For laboratory distillations,
a compound devoid of impurities boiling at or near any component of the
charge is highly desirable as a chaser so that complete removal of the charge
from the column may be shown by determination of the boiling point or re-
fractive index of the chaser. Toluene, tetralin, and diphenyl ethel' are
often satisfactory chasers.
When the still pot contents have become free of the charge (the condi-
tion prevailing when the still would become dry were no chaser present),
chaser vapor gradually fills the column to an increasing degree until it ap~
pears in essentially pure condition at the head of the column. In the mean-
time, the column continues to fractionate the last components of the charge
insofar as the decreasing effeetive length of the column will permit. If the
holdup is a single component from the original eharge, and the chaser is
considerably higher boiling, the separation will be quite satisfactorily
sharp. When the eolumn holdup consists of two or more components, the
separation aehieved becomes increasingly poorer as the column becomes
less efficient so far as those components are concerned. If a nonvolatile
residue is known to be present in a eharge, a chaser may be used to push
all volatile material in the charge to the head of the eolumn, the conditions
being as noted above. The nonvolatile residue t.hen remains as a solu-
tion in the chaser if it is soluble in the chaser. If a residue of this sort is t.o
be studied, the use of a ch!.Lsel' is inadvisable unless the two can be satis-
factorily separated.
When a chaser is not used, the column holdup and still residue may be
recovered by refluxing a low-boiling solvent in t.he· column; and the re-
sulting solution may then be fractionated in a small column of low holdup
and adequate efficiency. This slightly complex: procedure will allow nearly
30-b I'" Ii}. WU,LIAMS
100% uf a large ehfLl'ge to lw fraetionated without the neeessity of using a

low-capacity, low-holdup oolumn for the entire fractionation.
Clw1'(fing a colwnn consists of pouring the material to be fractionated into
the still pot and adding a volume of chaser equal to about twico tho holdup
of the column. Still pots ut\ing bare-wire internal ele(~tric heaters may
need an amount of ehnser greator than twice the column holdup in order
to ensure that the heater is completely suhmerged in liquid when the end
of the distillation is reached.
The charge may be introduced either on a weight basis, or a volume basis
referred to some stl\mlard temperature. For general laboratory fractiona-
tions, cha.rging on a weight haRis is probably more satisfactory. Similarly,
fmctions can be collected on a weight basis. If final results are desired on
a volume basis, a simple mathematienl conversion suffices to give this.
For operating; on a weight basis, no equipment to maintain a eonRtant re-
ceiver temperature is required. Much grm1ter precision ean be expeeted
by weighing than ean normally be attained in measuring volume. Ntill
residues can be weighed directly if a detachable still pot is used, whereas
if the volume is mcasUl'ed, considemhh~ drainage errol' may occur, particu-
larly if the residue is viseous. In case of pel'mmlently attached stilI pots,
one error is introduced by incomplete removal of residue from the still
pots by suction, and it Keeond error is made through ine,omp]pte drainage of
the suetion flask if the volume is measured. A torsion balan(~e202 of 120-g.
capacity, weighing directly to 0.01 g., offers a rapid means of weighing
'fractions taken from a ('.o1Ilmn. Larger frantions may he wpighecl on m;
ordinary trip h!1Janr~e.
v. OPERATION
1. Beginning of Operation
After the still pot has been charged and attadH~d to the column, coolant
is started through the eondenser and the system is brought to the desired
operating preStml'e. Preferred practice iA to have one exit from the column,
usually Itt the top of the retiux condenser, connected to H cold trap, to con-
dense smull amounts of low-boiling prodnrts not liquefied by the condense!',
and thence to the pressure-control system. A cold trap is especially im-
portant in low-pressme operation to keep condensable vapors out of the
surge tanh;, manostats, and pumps. The use of powdered dry ice around
the cold trap without liquid is entirely adequate, and obviates loss of liquid
by foaming that frequently occurs when dry ice is added to a liquid in a
Dewar flask surrounding the trap.203
202The Torsion Balance Company, New York.
203Hall and Jouach, SymposIUm on High-Temperature Analytical DiH1',ilbtion, Ameri-
can Petroleum Institute, Meeting at Chicago, Illinois, Nov. 11, 1946.
II. ORDINARY FRAC'l'IONAL nISTILLA'l'ION 305
When starting a distillation, time can be saved if the total avt1ihble heat
is applied to the still pot to bring thc distillt111d to boiling as rapidly as pos-
sible. With low-boiling distillands, the back pressure should be carefully
watched to detect the start of boiling; otherwise, the column may flood
and force large amounts of distillate into the receiver and cold trap. Once
boiling has started, the heat should be reduced to approximately the value
required for the desired back pressure and boil-up rate. When the column
uses heated jacket seetions to eliminate heat loss, proper adjustment of
the power input to the jackets should be made. This adjustment can be
guided only by experience. Except when the reflux has high solubility for
water, drops of water will usually appear in the condenser. If the amount
appears significant, it should be eliminated as described in Section IV-I.
Once the water is eliminated, the column may be preflooded.
2. Preflooding
Preflooding consists in increasing the boil-up rate to such a degree that

the reflux cannot return to the still pot, but fills or nearly fills the column
with liquid; in so doing, the packing is thoroughly wetted wi th the liquid
being distilled, and hence is in the condition of maximum efficiency. Only
packed columns need be flooded; the efficiency of bubble-cap and perfor-
ated plate columns is not improved by proflooding.
Several details are of great importance in the tIooding operation if the
column is to be conditioned to a state of maximum efficiency. No attempt
should be made to flood the coi1imn rapidly; if excessive heat is applied to
the stilI pot to cause rapid flooding, the insulation around the stiLI pqt
reaches a high temperature. This stored heat will cause excessive flooding
and force condensate into the reeeiver and cold trap. If the flood is brought
on gradually by incronsing the heat over a period of 1.5 minutes to 2 hOUTS
or more, depending on the size of the charge, the heater and still-pot insula-
tion reach the temperature necessary to give a vaporization rate satisfac-
tory for a controllable flood. At this point retiuction of the heat input may
commence and the flood will slowly subside. As the liquid in the column
slowly returnf-l to the still pot, it thoroughly wets the packing.
The flood stage should be maintained until liquid fills the column and
the still head a short distance up the condenser tube. If the column is
equipped with a back-pressure manometer, the flooding may be continued
until the back pressure equals the static pressure that would be produced if
the columns were filled with the liquid (at the prevailing column tempera-
ture). As soon as the flooding has stopped and l'efluxing is normal, the
column should be flooded a second time, and if necessary a third time, or
until the back pressure reaches a maximum during the flooding operation.
306 F. E. WILLIAMS
In general, the second and third flooding will not require quite as much heat
on the still pot because the flrst flooding canses a reduction of column capac-
ity. Under no circumstance should refiuxing stop entirely. If it does, the
favorable effect of prefloocling will be partially or totally lost. Sometimes
significant amounts of product will be fOl'ced through the reflux condenser
into the receiver during the fiooding operation. This material can be re-
turned to the still as outlined in Section IV-I. After preflooding, the
jackets are balanced and the boil-up rate adjusted by reference to back-
pressure reading, or by other suitable means.
Back-pressure readings aTe not absolute criteria of boil-up rate. In a
given column, under one set, of operating conditions, a particular back
pressure will indicate a certain boil-up rate for one mixture and another
boil-up rate for another mixture. Fur example, back pressures in milli-
meters mercury for one type of column at 150 ml. per hour boil-up are as
follows 203 : Cs alkylate, 2.5; C12 alkylate, 3.5 j benzene, 3.6 j toluene,
3.9; and 1.'}'lene, 5.
Back pressure in a perforated-plate column was constant in the region of
1100-2200 ml. per hour boil-up, but below and above this range showed
variation with boil-up.20o
3. Establishing Primary Equilibrium
In normal operation the column is kept at total reflux for a time to allow
a primary equilibrium to be established. Time on total reflux is generally
greater for columns of large dynamic holdup than for columns of low dy-
namic holdup. No exhaustive study of the facto)' has been made. High-
efficiency columns (100 or more theoretical plates) are kept at total reflux
for a period of 24 to 38 hours20'lj 3 to 4 hours has been found satisfactory
on small SO-plate analytical columns 203 ; 7 hours at totall'efiux is required
to reach equilibrium in a Hypel'~Cal column of 85 theOl'etical plates. 2oo A
practical guide is to hold the column at total reflux until the head tempera-
ture reaches a minimum after the column has been flooded and is operating
smoothly. Precise control of operating pressure and means of detecting
small changes in head temperature are necessary if this method is to be used.
'Small amounts of material in the distilland boiling much lower than the
next higher boiling component may falsely indicttte equilibrium bofore true
equilibrium is established in the column.
Willingham and Rossini, J. Re8earch Natl. Bur. Standard8, 37, 15 (1946).

204
Brandt, Perkins, and Halverson, Symposium on High~Tempel'ature Analytical
206
Distillation, American Petroleum Institute, Meeting at Chjol~go, Illinois, Nov. 11, 1946.
II. OHDINAHY FRAC'l'IONAL DIS'l'ILLA'l'ION 307
4. Product Collection
Once the column has been brought to its primary equilibrium, collec-
tion of product is started. Mechanically this consists of putting into opera-
tion whatever device is used to permit wi1;hdrawal of a part of the distillate
from the column at a uniform rate, the balance .being condensed and re-
turned to the column as reflux. The ratio of these two quantities is the re-
flux ratio; the larger numericul value conventionally being in the numel'a-
tol'. The reflux ratio must be fixed at (L value as high as necessary to give
the desired purity in the distillate; any higher value wastes time. The
actual reflux mtio to be used can only be npproximated. For colLlmns with
low boil-up rate, the mtmbm' of drops hlling off the cold finger of the reflux
condenser pOL' unit time can be counted and the product rate adjnsted to a
fraction of this number to give a prescribed reflux l'I1tio. Vapor and liquid
dividing heads, actuated by a timing mechanism, may be used to collect
all the distillate for a fraction of the time, and thus establish the reflux
ratio. None of these methods is precise because side-wall condensation,
existing to some degree in all columns, is 110t included in the estimate of the
reflux ratio.
A fairly sound empiriefLl method for setting the reflux ratio consists in
establishing a product rate that will give a product of the desired purity as
measured by boiling point, refractive index, or other suitable m.eans. If
boiling point is to be used as a criterion of purity, the temperatul'e~measnr~
ing element must be protected from extraneous effects, such as drafts and
splashing of cold reflux, that would cause changes in temperature readings
unconnected with real variations in vapor temperature. Significant varia-
tions in operating pressure should also be excluded when vapor temperature
is used as a measure of product purity. Elimination of these factors en-
sures that a temperature reading has realllleaning; how precisely a tem-
perature can be read is the limiting factor in the use of vapor temperature
as a measure of produet purity. Several temperature-measuring; elements
are discussed in Part 1. It suffices here to state that. vapor temperatures
should be readable to at least 0,1 DC. and preferably to 0.05°C., or less,
Refractive index is frequently a much more accurate indica.tor of product
purity than is boiling point. When sufficient difference exists in refrac-
tive indices of two successively dist.illing components of a mixture, this
measurement is more reliable tha,n vapor temperature in that it is not sub-
ject to the variations associated with vapor-temperature readings.
In establishing a suitable rcflux ratio, no effort should be made to es~
tablish the bare-minimum vlJ,lne because the minimum value suffices only
for a short interval when the still composition has a certaill value. Removal
of product changes the still composition, and hence requires a new and
308 F. E. WILLIAMS
higher reflux ratio. Accordingly, the more practical procedure is to set the
reflux ratio higher than is actually required at the beginning of a distilla-
tion, and then modify the reflux ratio as required by subsequcnt changes in
the composition of the material in the still. If time permits, the simplest
procedure is to set the reflux ratio at a value equivalent to or somewhat
above the value equivnlent to the number of theoretical plntes in the column
(as measured under totnl reflux). See Chapter I, Section VI.
Some operators reduce the reflux ratio when a plateau in the distillation
curve has been reached. This practice is permissible if the reflux ratio used
just prior to the advent of the plateau has been deliberately increased to
sharpen the brenk between the two components under immediate considera-
tion, and if thc reflux ratio used during the break is higher than is necessary
to give the desired separation of the next higher boiling component from the
material in the plateau rcgion. That is, the plateau exists solely because
the refiux ratio is adequate, and any significant reduction of the reflux ratio
may reduce the separating power of the column for the particular mixture
being distillecl. IVfjxtures with which an operator is familiar may allow
marked changes in refiux ratio to be made as the several components come
into product range; but with unknown mixtures, the permissiLle variation
of reflux ratio is mueh narrower unless there is no objection to a certain
amount of redistillation. Without prior knowledge, there is eertainly no
justification for reducing a refilL,,{ ratio simply because a plateau appears in
a distillation curve.
5. Size of Fractions
In amlytical distillations when a detfdled picture of the composition of a
mixture is to be obtt'.ined, small fractions of the order of 1% of the charge
are satisfactory. Smaller cuts during the transition from one component
to the Jlext may be desirable under some cil'cmnstances. When a distilla-
tion is being performed to isolate one or more components in sizable quan-
tities lwd prior knowledge of composition justifies it, much larger fractions
may be taken if larger receivers are used. A certain amount of caution
should be used under this proeedure to avoid contaminating a pure distil-
late with the next l~igher boiling component should a transition occur un-
eXpectedly.
During a fractionation, a definite sehedule for securing the distillation
data is desirable to give reliable results; if several columns are attended
by one operatSlr, a sehedule is imperative. When operating high-efficiency
columns where the percentage of product removed pel' unit time is nor··
mally small, readings should be taken hourly if the charge is complex 01' of
unknown complexity. These readings include vapor temperature, still
temperature, condition of heat balance in the jacketed sections of the
II. OlW1NAHY FHAU'l'IONAL IH:-l'l'ILLA'rION 309
column, back pressure, if this is used as a measure of boil-up rate, or deter-

mination of thc number of drops of condemate falling from the reflux con-
denser per minute. Appropriate adjustment of variable transformers COIl-
trolling the several heaters are made at this time. All temperature readings
and voltage readings of the variable transformers are recorded primarily
for check purposes and also as a guide for future work on the column. The
vapor temperature should also be determined at the time a fraction is taken.
A check on product rate should be made occasionally either by counting
drops or by noting the volume of product collected in a given time. With
simpler mixtures, less frequent checking of column behavior is required,
particularly if the composition is known approximately, a.nd only certain
major components are of interest.
Less efficient columns are normally opera tell with greater product rate,
and will accordingly require more frequent ttt[;ention if fractionB of the same
relative size are being collected; however, the essentials are as outlined
above.
When the distillation is nearing completion, the boil-up rate will drop
rather rapidly if no chaser iii used. If a chaser is used, the approaching end
of the distillation will be indicated fin-it by a rapid increase in still-pot tem-
perature, and a large inerease in the heat required by the column jaekets.
When the vapor temperature rises to the boiling I)oint of the ehaser, the
distillation is finished. At this point the still is disconnected from the pres-
sure-control system, the electric pmVer tUl'llcd off, and the column im-
mediately brought to atmospheric pressure with oxygen-free nitrogen 01'
earbon dioxide. A positive pl'esmre of one to two pounds of the inert gas
·will prevent oxidation of liquid adhering to the packing (luring clra,illage
and woling. After eooling to a suitable lower temperaturc, the still pot
may be disconnected, weighed, eleaned, [1mI rewcighe(l to (letcrrnine tllf~
residue. If the still pot is permarwntly sealed to the eolumn, the residue
may be withdrawn by suc:tion into It tared fil1Sk and weighed. The total
weight, of material recovered as distillate and resi(lue will he lei:ls than the
charge weight. The difference represents the statie holdup of the column
and material adhering to the still pot. This material lUt1Y be .l'ecoverecl
by refluxing an appropriate low-boiling solvent in the column, withdrawing
the solution, and distilling off the solvent in n suitable column of low holdup.
By this procedure recoveries of 98-100% arc easily possible if no leakage
has oceurrecL during the distillatioll and nffl('.i(~nt eolrl tmps are used to trap
vapors passing through the l'llHux condell~eJ'.
After the static holdup of the column hILS been removed, methyl or
ethyl Cellosolve or other solvent having a high solvent power should be 1'e-
fluxed in the column to remove tmces of tarry and polymeric material
which may have formed in the packing. Following this, acetone refluxed
310 F. E. WILLIAMS
in the column will remove the higher-boiling cleaning agent and simplify
drying of the packing with n, stream of carbon dioxide. A little heat ap-
plied to the jackets will hasten the drying process except in vacuum-
jacketed columns. The column should then be sealed to prevent aceess
of ail' until the next distillation is to be made. This method of cleaning a
column will ensure years of satisfactory service.
6. Interpretation of Results
A normal fl'actionatioll can provide (1) a set of boiling points, 01' vapor
temperatures, for the fractions collected during the distillation, (2) the
IO~---------------------------------------------'
9
Bromotoluene (b.p.ISI"C.l -------1
a
'"
E
7
W 6
a:
:J
~ 5
a:
UJ
~ 4
IJJ
I- :l
2b;.o..o.e-........_ Carbon disulfide (b.p. 46.e·C.)
o 5 10 15 20 25 30 35 40 45 50 55 60 65 10 15
DISTILLATE, ml.
Fig. 60. Distillation curve of a. simple synthetic mixture.
volume or weigllt percentage of the charge distilled, and (3) a group of frac-
tions of greater 01' lesser purity depending upon the working efficiency of
the column, the reflux ratio, and the complexity of the distilland.
The first step in the interpretation of these data consists in plotting
vapor te1npel'alw'88 as ordinates and per. cent distilled as abscissa on rec-
tangular coordinate paper. Provided that the column was of adequate
efficiency and was so operated as to separate each component in substan-
tially pure condition from all other components in the charge, and provided
the holdup was small in comparison to the amount of any component pres-
ent, the resulting curve of boiling points versus per cent distilled will con~
sist of a series of horizontal lines or plateaus parallel to the abscissas joined
by sigmoid lines, as shown in Figure 66. Each plateau may represent a
II. ORDINARY FRACTIONAL DIS'l'ILLATION 311
.....<l)
c;
-s:..
~
'~ofia Ht tv X3GNI 3"'l~>;tK:J3~ ?
o
-.:I; 0
._
o 0
o ~
,9
..,
0
0
0
".::
0
o '.;5
m '" " c;
,:::
......,
<l)
S
0
0 ,:::
0
0 '"C
<l)
o
....
<t :> >:1
'r:: §
'"
'"Co
OJ
f!l
<:.J Il)
~d
III
Cl
0==
0
..i ,_' g
.::~
Z oz ~o
e'l
:::> fC") - 00-(
== ~...
Cl
,..o
a. IJJ "
..cI
o 'O...J 0~
:.
o
'"a:
:::>
:>"C..a
§...,
a. 0
0
"'-:;;
u. "'0
o 0
E~
3H'VJ,N3dlA.H.1.3t'lII:U""'2 ~-
1
r---"'--1
,..~
'" "::C:
E3~
o
"' 0.: ~
ai
0
'"
~
• <l)
"'"
o 0
0
.~...
'" 0 :;9
~<ll
'"C
~~
"T ...
0
oL--J--~--~--~~~-o~~~~--~--o=--J--~--~--=-~~~~~~~ 0
g ~ ~ 0
0
(.\J ~
po.
l'6H'I!';~ 56! 111 'd'e) '~. NI 3~nlV~3dIr'l3.l.
r:..:
(p
t>O
ii:
312 F. I'~. WIT,LI.\MS
pure component, and the amn1lnt of the pmc component separated by the
column wn be determined by inspection. The total amount of any com-
pOllent in a mixture will indllde dw material in the breaks indicated on the
curve by the sigmoids l1t each cnd of a plateau. As a first approximation,
the amount of a component in a break can be estimated by drawing a line
vertical to the abscissas at a point halfway behveen the beginning of one
plateau and the end of the adjacent pl:1tcau representing the next lower
and next higher boiling components. For many praetieal purposes analyti-
cal results obtained by this method are of sufIicient accuracy. If the
analytical requirements are mure stringent, other means are advisable for
detcrmining the composition of the breaks. These methods will be dis-
eussed httct'.
Although Figure (iC) represlmts a coneeivable distillation, it is by no
means typical of the majority of caseH. An extreme ease of complexity is
shown in Figure 67, which represents an analytical distillation of C4 alkyl-
ate (H 2S0 4) as carried out in a column of approximately 200 theoretical
plates at the National Bureau of Stanclal'cis. 20 'i
Inspection of Figure G7 reveals seven platen,UB in the hoiling-point
curve, only one of which is sufficiently inelined to arouse suspicion of in-
homogencity. Within this region of 87% distilled,. only six components
are indicated, while the rising trend in the region of 54 to 83% distilled
indicates two or more eomponents. Actually, a more comprehensive
analysis by infrared techniques showed the presence of H) components in
the first 87% distilled; one of the plateaus eiter! ahnve included three of the
19 components.
Hardy, if ever, is n distillation interpreted on the basis of vapor tempera-
ture only. The common pmetiee of determining the refractive index of the
fractions as they are collected from a column permits an independent and
mmally more precise measnre of p]'or!uC'.t plll'ity to be made quickly with
only 11 very small alllount of sample. Once the refractive indices of the
fl'!Lction~ have been measured, a emvl' of refractive index versus per cent
distilled can be drawn, preferably on the same sheet. of co-ordinate paper
containing the vapor-temperature curve, as shown in Figure G7 by the
broken-line curve. The combination of vapor-temperature and refractive-
index data provides a sounder basis for interpretation of the fractionation.
In the pa,rticular
I
case depicted in Figure f)7, only three
..
of the seven plateaus
of the vapor-temperature curve show eorresponding plateaus in thererrac-
tive-index curve; all other seetions of the refractive-index curve Showed
wide deviation from flatness. Examination by infrared analysis of those
fractions whose vapor-temperature enrves and l'efraetive-index· )curves
were parallel confirmed the homogeneity of the samples; on thtt other
hand infrared analysis on fractions whose curves of vapor temperature
j ,
II. ORDINARY FRAC'l'IONAJ, DISTII,LATION 313
and refractive index deviated showed the presence of Hcveral components.

The distillation was immediately effective only for clctermininO' and iso-
lating 2,3-dimethylbutane, 2,2,4-tl'imetbylpentane, [lnd 2,2,.5~~rimethyl
hexane; probably of equal importance was the isolation of greatly simpli-
fied mixtures of components whose analysis by infrared spectra was more
easily and precisely accomplished than would have been possible with the
original alkylate.
In general, a group of fractions showing constant vapor temperature and
constant refractive index during distillation may be presumed to be of uni-
form composition and substantially a pure compound, unless an azeotrope
has been formed. Greater significance attaches to constancy of refractive
index than to constancy of boiling point inasmuch as slight mndom varia-
tions in boiling point frequently occur even when highly purified compounds
are distilled.
As was mentioned earlier, Figure 67 represents a fractionation much
more complex than those normally encountered in many fields of organic
research. For these simpler mixtures vapor-temperature and refractive-
index curves of the fractions may be drawn as explained above, and a first
approximation of the composition of the distilland determined. Further
refinement of the analysis is then possible by combining the fractions taken
during the breaks, determining their refractive index, and cnlculating their
composition from the known refractive index of the two components be-
tween which the breaks are collected. Whether. this procedure is permis-
sible depends in general on t,he character of the refraetive-index curve in
the region of the breaks. If the curve shows a smooth transition from the
refractive index of one plateau to the next plateau without dips or peaks,
the break cuts can be considered to be a simple binary mixture of the two
adjacent components. In calculating the composition of the combined
break cuts, a linear relationship between refractive index and composition
is assumed; If there are no pronounced volume changes or temperature
changes when the pure components are mixed, departures from linearity
are small; inasmuch as the total break fractions represent a relatively
small proportion of the totaJ fractions between which the break occurs, small
departures of the refractive index of the mixture from strict linearity will
not greatly affect the composition so established. Qualitative information
about the composition of the distill and may disclose the presence of small
amounts of materials distilling in the region of the hreak between key com-
ponents. If this situation exists, other means of analyzing the bren], frac-
ti0118 will be necessary. If the difference in refractive index of two com~
ponents in the break fractions is small, the use of refractive index for
analy§Jing these fractions may not be so effective as an estimate made by
inspe~tion of the vapor-temperature curve mentioned on page 310.
314 1!'. E. WILLIAMS
Although 20°C. is the temperature commonly used for measuring refrac-

tive indices, there is some trend toward 25 0 and a temperature of 30 0 would
be more satisfactory in many laboratories. Frequently during hot days of
summer, when the relative humidity is high, refractometer prisms held at
20° will condense sufficient moisture to mltke them unusable. At 30 0 no
trouble will be encountered. There is the further advantage that, whereas
a constant-temperature bath for a refractometer may require refrigerated
water for 20 0 operation, ordinary tap water may he satisfactory when the
bath is held at 30°.
There is the objection that much of the literature reports refractive in-
dices at 20 0 • For a given laboratory, where the variety of compounds is
probably not large, this objection can be overcome by measuring the refrac-
tive index of the most commonly used compounds at both 20 0 and 30 0 to
establish the temperature cocffieient of the series, and then adopting the
higher temperature for practical use.
In addition to vapor tempehl,ture and refraetive index eommonly utled in
following the course of a distillation and as a means of interpreting a dis-
tillation, other physical properties are used to gain a more complete picture
of the mixture being investigated. These include determination of density,
viscosity, optical rotation, and melting point. Generally these methods
are used only when refractive index or boiling point, or both, give an ambig-
uous answer. Optical rotation is used with such naturally occurring
products as the terpene hydrocarbons and their derivatives. Melting and
freezing points may have wider applieation, particularly as a criterion of
purity. Use of freezing points has been greatly extended by recent inves-
tigations of low-melting hydrocarbons. 20B Proeedures for making these
physical measurements may be found in books on physicochemical meth-
OclS 207 or in the original literature. Ultraviolet, infrared, Raman, and mass
spectral analyses are being more widely used both for qualitative and quan-
titative analyses.
A number of chem'ical methods are available to help in analyzing and in-
terpreting distillation data. These include determination of bromine
number, acid number, saponifieation number, iodine number, and carbonyl
value, used, respectively, in studying unsaturation of hydi'ocarbons, iclen~
tification of acids, identification of esters, unsaturation of fatty acids
and esters, and determination of ketones and aldehydes.
Specific problems of interpretation may require the use of catalytic hy-
drogenation to determine unsaturation, the Zel'ewitinoff method for de-
208 Glasgow, Streiff, and Rossini, Paper for Division of Pctrolcum Chemistry, American
Chemical Society, Meeting-in-Print, September, 1945.
207 Weissberger, ed., Physical Methods oj Organic Chemistry (Volume I, this series),
2nd ed., Interscience, New York, Part 1-1949, Part U-1950.
II. ORDINAHY FHAU'rIONAL IHS'l'ILLATION 315
termining active hydrogen, the acetyl method for hydroxyl groups, and the
direct determination of o:ll.'Ygen.20S,2nU In addition to these, the methods for
determining cm'bon, hydrogen, methoxyl, halogens, nitrogen, phosphorus,
sulfur, and metals in volatile metallo-organic compounds contribute infor-
mation beyond th(~ limit of fradionaL distillation and. simple physical
measurements.
Certain unavoidable losses occur in nearly every distillation with the re-
i·mlt that, the total weight or volnme of distillate, residue, and static holdup
does not eqnal the weight or volume of the distilland. When the equip-
ment is propeL"ly eonstrllcted [1nd operated, total losses should not exceed
8% of the charge, and may be reported as "loss," or may be distributed
over the sevcml fractions in the ratio of' the weights or volumes of the fra(~
tion to th(1 weight or volume of the charge. This refinement is scarcely
justified in ordinaJ'y fractionntions yielding 1-1 great many fractioll~. When
the residue is essentially nonvolatile compared to the distillate, the loss
should be distributed mnong the fractious of distillate.
I,argel' losses ranging up to 1.5 to 20% of a Chitrge may be incurred
through leakage at the still-pot joint, or thmugh inadequate traps at the
condenser exit, particularly when distilling material containing low-boiling
components. Operation of a column at a pre::;SUl'e slightly below atmos-
pheric so that the still-pot pre::;sure is less than the prevailing atmospheric
pressure will prevent loss of vapor from a still-pot joint. If leakage at the
joint should oocmr, a drop in vapor-liquid equilibriulll temperature will
take place; the column may then be shut down and the leak eliminated.
At greatly reduced pressures, the same conditions prevail to a greater ex-
tent. The air drawn through the still pot [tnd column becomes saturated
with vapor, and any vapor not removed from the gas stream beeomes a part
of the total losses.
When distillation losseH nre large, as they may be through failure of the
equipment as described ttbove, proper distribution of the losses among the
different fractions can be only [1 poor estimate; in fact there is no solution
to the problem except to build the equipment and conduct the distillation
to avoid all but. nominal losses; in other ,vords, the still pot should be
sealed to the column with a glass-to-glass seal, and the charge line closed
with a lubricated plug whieh is in a region of low vapor and liquid concen-
tration.
208 MarkH, Inri. gila. elWin., ilnal. Ed" 7, 102 (1!l3fi).
200 Alui~D, Hall, Stau18, [Lm! Bedwl', hul. Rny. Ulwin., Anal. Rd., 19, 3'17 (1047).
31G ]'. B. WILLIAMS
General References
AmeriCfLIl I llstitute of Physics, Tcmperalurc-l t~ M easnrcmcnl and Control in
Sctonce and Industry. Reinhold, New York, 19·11.
American Petroleulll Institute, Symposium on High-Temperature A.nalytical
Distillation. Meeting at Chicago, Illinois, Nov. 11, 1946.
MortOIl, Laboratory Techniques in Organic Chem·istI'Y. 1st 0(1., McGraw-Hill,
New York, 1938.
Ward, "Review of the Literature on the Con~trllctiun, Testing and Operation
of Laboratory Fractionating Columns," U. S. Bw·. Mines Tech. Paper 600, 1939.
Weber, Temperature },{ea8urement and Contl'ol. Blakiston, Philadelphia, 1941.
Willingham ancl Rossini, "Assembly, Testing, and Operation of Laboratory
Distilling Columns of High Efliciency, J. Rcseareh Natl. Bur. Standards, 37, 15-20
(HH6); Research Paper No. 1724.
CHAFl'EH III
EXTRACTIVE AND AZEOTROPIC DISTILLATION

CARL S. CARLSON, Standard Oil Development Company
I. INTRODUCTION AND DEFINITIONS
In laboratory separations of volatile-liquid mixtures and frequently in

industrial separations, the goal is to obtain the components of the mixture
in pure form or to recover at least one of the components virtually pure.
The first and frequently the only step in such separations is fnwtional dis-
tillation. In certain cases, however, complete sepamtion by fractional
distillation is impossible or impractical. A concentrate prepared by ef-
ficient fractional distilla:tion can often be further separated into fractions
of different molecular type by extractive or u.zeotropic distillation.
As discussed in Chapter I, the separation of mixtures by fra['tional dis-
tillation depends on differences in volatility of the materials to be 8eparated.
In general, the volatility of a compound (which in ideal solutiolls is the
vapor pressure of the pure material in a homologous series) will l)e inversely
proportional to its molecular weight and boiling point.. As the molecular
weight of members of an organic series increases, the number of possible
compounds of similar volatility increases exponentially and the complete
separation of the liquid mixture becomes increasingly difficult, if not im-
possible, by fractional distillation .. Added to this is the complication that
eompounds of different molecular types may have nearly the same vola-
tility or boiling point. For example, the six-carbon hydrocarbons, ben-
zene and cyclohexallc, boil at 80.1 and SO.soC., respc('.tively.l As separa-
t.ions by distillation beeame more difficult, the efficiency and total separat-
ing power of laboratory fractionating columns were jncl'(~ased. Laboratory
columns with a separating power of 100 theoretieal phLtes ~Lre not uneom-
mon,2-4 while columns of several hundred theoretical plates have been
buiW for special applications. Even with columns of this separating
I Handbook of Chemisl1'Y and Ph1ldcf!. 2(Jth erL, Chemical Ruhber Puhlishing Co.,
Cleveland, 1942.
2 Fenske, Tongberg, Quiggle, and Oryder, Ind. Eng. Chem., 28,644 (l\lS(:i).
a Tongberg, Lawroski, and Fenske, Ind. Eng. Chem., 29, 957 (1037).
4 Willingham and Rossini, Paper presented before the Petroleum Division, Am. Chem.
Soc. Atlantic Oity Meeting, 1946.
6 Huffman and Urey, Ind. Eng. Chern., 29,531 (1937).
317
318
power, a diffel'enee ill tho volatilities of the materials to he separated Illllst

exist for any separation to be achieved.
A logical development in the application of frnctional distillation is t.hc
modification of the volaWities of the components of the mixture to be sep-
arated. In extractive distillation and in nzeotropic distillatiun this is ac-
complished by the addition of certain solvents. Another method is to
carry out the distillation under reduced pressure (Chapters V and VI).
Before discussing extractive and azeotropic distillation, fl, few terms will be
defined and explained.
Volatility, v, is a measure of the tendency of a compound to enter the
vapor phase. For a pure compound, the vapor pressure at l), given tem-
perature or the boiling point at atmospheric pressure is indicative of its
volatility. In a solution the tendency of the compound to vaporize may be
altered by the other compounds present, Since we arc interested in the
recovery of the compound from solution, the volatility, having varying
numerical values in different solutions, must be defined for the material
in solution:
v = V/:t: (1)
where v is the volatility, y is the mole fraction in the vapor phase at equilib-
rium, and x is the mole fraction of the same compound in the liquid phase
at equilibrium.
Relative Volatility, DI. The ease of separation by fractional distilla"
tion depends upon the volatility of one eompound relative to that of the
others present in the system. For simplicity, consider a binary system
of A and B. The relative volatility, DIAn, is defined as:
(2)
Again, for an ideal solution, (~ is equal to the ratio of the vapor pressures
of the compounds at the temperature of the boiling liquid (see equation
11). In general, in the absence of specific experimental data, the vapor
pressure rat.io is used as a rough index of ease of separation. In multi-
component systems, the relative volatilities arc usually referred to the least
volatile compound and when written, for example, in a four-component
system, they are designated as DlA-D, or DlB-D, etc., where A is the most vola-
tile and D is the least volatile component.
Modified Relative Volatility, DIS' As was implied above, it is proposed
to modify the normal volatility relationships of a mixture by the addition
of a solvent. In the presence of a solvent, this modified relative 'Vola-
tility will be designated as DIS:
(3)
III. EXTRAC1'IVE AND AZEO'l'ROPIC DHnU,Lll!rION 319
where VAS, XAS, YBS, and XBS are the mole fractions of A and B in the vapor
phase (y) and in the liquid phase (x), respectively, all in the presence of a
solvent at equilibrium. The ahove concentrations may be expressed 011 a
multi component basis or on a binary basis, as long as the basis is consistent.
Extractive Distillation. Vapor-liquid extraction, or extractive dis-
tillation, involves distillation in the presence of a substance which is
relatively nonvolatile compared to the components to be separated and
which is selected to enhance the relative volatility of the components to be
separated. It is added continuously near the top of the distilling column
so that un appreciable concentration is maintained in the liquid phase
throughout the column. 6,7 In the proportions used, the solvent is mis-
cible with the mixture to be separated at the temperature of the distillation.
Azeotropism. The property by virtue of which certain liquid mixtures
distil at a constant temperature under a constant preSHure without ehange
in (Jomposit,ion is eallcd azeotropiHlll. 8
Azeotropic Mixture or Azeotrope. An azcotropie mixture or azeotrope
boils or distils without ehange in compoHitioll, and in general it has a
boiling point higher or lower than tha~ of any of its pure cOllstituents.o
A constant-boiling mixture, C.B.M., is another clesignation for an azeo-
tropic mixture.
Azeotropic Distillatioll. A distillation in which one of the products
is obtained as an azeotropie mixture is known as azeotropic distillation
in its broad sense. In the present discussion, however, azeotropic dis-
tillation is defined as distillation in the presence of a volatile component
added to facilitate sepamtion by distillation when one or more of the eom-
ponents of the original mixture is obtained in a fixed ratio to the added
component. Referring to a binary mixture for convenience, the addition
to the feed of a third eomponent, S, capable of forming a minimum-boiling
azeotrope with all of one component, A, will yield a distillate containing
A and S, in a fixed ratio. This will be true as long as sufficient A and S are
present.
Homogeneous Azeotrope. A single liquid-phase mixture whieh has
the same eomposition as the vapor in equilibrium with it is calleel a homo-
geneous azeotrope or constant-boiling mixture, C.B.M. For binary mix-
tures, vapor and liquid eompositions are the same and the binary azeo-
tropes have boiling points different from those of either of the pure ma-
terials making up the mixture.
8 Benedict and Rubin, Trans. Am. Inst. Chem. En(/r.~., 41, 353 (1945).
7 Dunn, Millar, Pierotti, Shims, amI Souders, Jr., Trans. A.II~. Inst. Chem. Engrs.,
41,631 (1945).
- 8 Sunier and Rosenblum, Ind. Eng. Ohento, Anal. BIZ., 2, 109 (H)30).
D Fleer, J. Chern. Education, 22, 588 (lD45).
320 (.!All!~ S. CAHLSON
Heterogeneous Azeotrope. The term azeotropic mixture has also

been applied to the distillate obtained hy the dist,iilation of a binary mix-
ture having two liquid phases. As long as two liquid phases exist the
vapor will be of constant composition. A common example is steam dis-
tillation with a water phase present.
Solvent. The separating aid used in extractive distillation is usually
referred to as a solvent. It is usually much higher boiling, by 50 to
100 0 0., than the mixture to be separated.
Entrainer. Sinee most azeotropes have minimulll boiling points the
agent added to form the azeotrope usually appears in the distillate.
Hence it is commonly called an entrainer. Its boiling point is generally
close to that of the mixture to be separated.
In extmetive distillation, a solvent is added continuously near the top of
the distilhtion column to maintain a high solvent concentration in the
liquid pha:.;e. In azeotl'opic distillation, a solvent is added with the charge
to be distilled to produce an azeotl'opic mixture with one or more of the
components of the mixture.
n. EXTRACTIVE DISTILLATION
1. Scope
Extractive distillation is frequently helpful even when a difference in
volatilities exists in the absence of a solvent, because an appreciable reduc-
tion in the theoretical-plate requirements or reflux ratio can be obtained.
The relationship between relative volatility, ct, and the number of theoret-
ical plates required for a given separation is illustrated in Table 1. This
table 10 is based on the separation of a binary hydrocarbon mixture into a
TABLE I
RELATION BETWE!;N RELATIVE VOLA'l'ILITY AND NUMBER OF PERF]JCT PLAT]JS
Boiling
}joint Number of
differenco. Relative theoretical
'C. voilltility* plates needed
__ ,._,---- 'r'_'~_ ._-_._
7.0 1.3 22
ti, 0 1.25 26
5.0 1.20 32
4.0 1.15 41
2.75 1.10 60
1.5 1.05 110
0 1.00 Inlinito
* Assuming an ideal solution and the perfect gas laws:
lQ Fenske, ill Science of Petroleum, Dunstan, Nash, Brooks, and Tizard, eds. Vol. II,
Oxford Uuiv. Press, New York, 1938, p. 1631.
III. EX'l'HAC'l'IVE AND AZEO'L'HOl'IC DIS'l'ILLA'l'ION 321
distillate containing 95 mole per cent of the more volatile or lower-boiling

material and a residue eontaining ouly 5 mole pel' cent of the sa~e compo-
nent, the separation being made in a batch fraetional-distillation column
opemting under conditions of total reHux.
Extractive distillation may be m,ed in the bboratol'Y to facilitate separa-
tions of three main types:
Virtually Ideal Solutions of Close-Boiling Components. An example
of this type is the binary mixture n-heptane (b.p. 98.5 DC.) and rnethyl-
cyclohexane (b.p. 100.8°). This mixture, which has a relative volatility
of 1.07 at atmospheric pre~sure, is diffieult to separate by conventional
fractional distillation. When it is extractively distilled with 70 weight pel'
cent aniline as a solvent, the relative volatility is increased to 1.3, and the
theoretical-plate requirement for l1 given separation is reduced by 75%.11
, N onideal Mixtures. An example of this type is the binary mixture
methylcyclohexane (b.p. 100.8°C.) and toluene (b.p. 110.8°). The rela-
tive volatility of this mixture decreases with increasing methylcyclohexane
concentmtion; it is only 1.07 at 90 mole per cent methylcyclohexane.
Therefore it is very difficult to free methylcyclohexane of toluene. The
separation is greatly enlmnced by an extractive distillation in the presenee
of a polar solvent suell as 11niline. l l This is further illustl'l1ted by the X,Y
diagram for this system in Figure 3.
Azeotropes. Constant-boiling mixture:; or azeotropes arc a special
form of nonideal solution. Since they usually eonsist of' dissimilal' moleeu-
lar tYP(lS many of them can be separa,t,ed readily by extractive distilla-
tion. For example, cyclohexane-benzen8, whieh at atmospheric pressure
forms fl11 azeotrope containing 46.3 mole pel' cent cyclohexane, wus extrac-
tively distilled with aniline as a solvent. Virtually pure cyelohexane was
obtained atl a top product with the hydrocarbon portion of the bottoms
containing 22.4 mole per cent cyclohexane 12 (compositions expressed on a
solvent-free basis). The composition of the hydrocarbon mixture in the
column obviously passed through what would normally be the uzeotropie
composition.
Complex fraetions that boil over a range of 20°C. 01' less and which eOll-
tain ('umpounds of different molecular types can be separated. Fol' ex-
ample, a narrow boiling range fraction of petroleum naphtha was ext me·
tively distilled in continuous units during World War II to separate toluene
from nonaromatic substances. 13 Other similar applications are obviously
possible.
11 Fenske, Cll.rlson, and Quiggle, Ind. Eng. Cham., 39, 1322 (H147),
12 Dicks and Carlson, Trans. Am. Inst. Chem. EngT!!;., 41,789 (1045).
13 Dunn, Milltll' , Pierotti, Shil'as, and Souders, Jr., ']'l'ans. Am. Inst. Chem. EngI'8.,
41,631 (1\145).
322 CARL S. CARLSON
2. Theory
III extractive distillation, a non ideal sYRtem is deliberately created by
the addition of a solvent. In the resulting solution, the partial-pressure
relationships, and consequently the relative volatilities of the components
to be separated, will be different. These effects and their magnitudes can
best be understood by reference to the following equations and discussion:
In an ideal solution of two components A and S, the vapor pressure of
each can be expressed by Raoult's bw:
PA = PAXA (4)
Ps = PSXs (5)
where p is the vapor pressure of the pure compound, p the partial pressure,
x the mole fraction, and A and S denote solute and solvent, respectively.
The total pressure P above the solution is given in Dalton's law:
P=PA+Ps (6)
The relative volatility, aAS, of a compound in an ideal binary solution is
obtained as follows: From Avagadro's law:
YA = PA/P (7)
Ys = Ps/p (8)
where YA and Ys are the mole fractions of A and S, respectively, in the vapor.
Combining these equations with (4) and (5) we obtain:
YA = PAXA/P (9)
Ys = Psxs/p (10)
Dividing (9) by (10) we obtain:
(11)
Ys PSXS
i.e., the relative volatility, d!AS, of a compound in an ideal solution is equal

to the vapor-pressure ratio of the pure components at the temperature of
the boiling mixture.
In a nonideal solution the partial pressures and total pressure over a
solution at a given temperature differ from those predicted by Raoult's
law. The deviations may be positive or negative. Since the vast majority
is positive, our d'scLlssion will be limited to that case, although the prinei-
pIes hold equally well for negative deviations from Raoult's law. To allow
for deviations from Raoult's law, a correction factor, 'Y, has been introduced
III. EXTHACl'IVE AND AZEO'l'HOPIC mS'l'ILLA'l'ION
which is actually an activity coefficient. With thiH factor eqnations for

Ilonideal solutions may be written as follows:
PA = 'YAPAXA (12)
Ps = 'YsPsXs (13)
P = PA + Ps (14)
The coefficient 'Y is not a constnnt. Its numerical value is a function of all
the components in the solution and of their concentrations. Referring to
1 0 , - - - - - - - -_ _ _-.., 10
)0
l<> 8
..:z f-" ACTIVITY COEFFICIENT A ;.:
II
w z ,....
w
!2 4 U 6 :i
u..
u.. u:
u..
f::
<!
~ :3 w
0
RELATIVE: VOLATILITY ...I
0
u u AlB :>
4
w
~ >-
f- :>
:;: 2 :; f::
;: ;: 2 <!
u U ...I
W
<! <! 0:
ACTIVITY COEFFICIENT B
1.0 o 0.2 0.4 0.6 0.8 1.0
MOLE: FRACTION ACETONE IN UQUIO(xA) MOLE FRACTION SOLVENT IN LIQUID
Fig. 1. Activity coefficients for binary Fig. 2. Effect of solvent concentmtion
system acetone-water at 1 atmosphere.H on activity coefficients and relative vola-
tilities of a binary mixture with equi·'
molal proportions of A to B.
component A, in a binary system, 'Y A = 1 when XA. = 1, since pure liquids

are defined as ideal solutions. As the concentration of XA decreases, 'YA
increases and reaches a maximum at XA = O. The same relations hold for
the other component, S. rrhese relationships are illustrated in Figure 1 for
acetone and watel'.H
An equation for relative volatility similar to equation (11) may be written
for nonideal solutions:
(15)
The relative volatility, ey_' AS, is equal to the ratio of vapor pressures and
activity coefficients.
14 Brunjes and Bogart, Ind. Eng. Chern., 35, 255 (1943).
32-1: (!AHL R. CAHLSON
The effect of a solvent and its concentration on the activity coefficients

of the two components of a binary system in equimola.l mt.io is shown in
Figlll'e 2. The activity copflicient of each component increases as the
solvent cOIH'{mtratioIl increascH from 0 to 100%. The a,etivity coefficient
of component A increases much more rapidly than that of component B.
This is reflected in the vt1l'iation in the relative volatility with solvent con-
centmtion, whic·h is also shown in Figure 2. To realize a relative volatility
of 2 or higher, nt least 50 mole per cent solvent must he pl'e.~ent in the
liquid.
~ 1.0..-----------"7?]
~ 55 MOLE PER CENT
~ AN ILiNE IN L1QUIO
g 0.8
u
>-
':l
~~ 0.6
f-Il..
w«
:1:>
z~ 0.4
o
i=
u
«
.a:
u.
w
_J
o::;
0.2 0.4 0.6 0.8 1.0
MOLE FRACTION METHYLCYCLOHEXANE
IN LIQUID
l<'ig. 3. Methylcyclohexane-tohmne vflpm:-Jiquid (Oquilibl'ia at

1 at,mosplwl'e, tlXjll'8HSpd Oll lL solvent-free basis.
The effect of solvent concentration is also illustrated in Figure 3, where

the x-y diagmm for methylcyclohexane-toluene is shown on a solvent-free
basis for two solvent coneentratiol1s. 11 For thiH system the effect of solvent
is most marked in the high-methylcyelohexane region since the binary
system (zero solvent concentration) vapor-liquid equilibrium curve pinches l
approaching the ;I; = lIline, as the mole imction of methylcyclohexane ap-
proaches 1.
The foregoing discussion illustrates the effect of solvents on the relative
volatilities of binary mixtures. High flolvent concentrations are necessary
for a maximum effect on the relative volatilit,y. Colburn .and Schoenborn 15
have shovi'll by a study of a number of nonideal solutions that this is a gen-
81'al requirement fot' extradive distillation.
15 Colbul'n and Schoenborn, '['rans. Am. Inst. Chem. Engrs., 41, 422 (1945).
III. EXTRACTIVE AND AZEOTROPIC DIS'rILLA. TION 325
3. Selection of Solvent
A. GENERAL CONSIDIGRATIONS
The suitability of a solvent for extractive distillation depends on its ef-

fect on the volatility of the components to be separated and on certain
other factors mentioned below. It is usually advantageous to select a
solvent which augments the normal vapor-pressure relationships and 'which
forms a highly nonideal system. In addition, the solvent should be suf-
ficiently high boiling so that the components obtained in the solvent phase
can be easily recovered by fractional distillation. It should also be a good
solvent and be soluble itself so that excessively high solvent:mixture ratios
are not necessary and solvent-phase separation will not oecul' during the dis-
tillation. Should phase separation occur, much of the relative-volatility
enhancement will be lost. The solvent should be thermally stable so that
no decomposition occurs during extractive distillation or subsequent frac-
tional distillation for removal of dissolved components from the solvent.
For ease of lmndling, the solvent should be nontoxic. It should not react
with the components of the mixture; formation of stable chemical com-
pounds or of azeotropes with the solvent during extractive distillation is
undesirable and may prevent the desired separation. If the solvent boils
more than 50°C. above the mixture, the danger of azeotrope formation is
negligible (see Sect. III). A broad range of application is desirable but
not essential in a solvent for laboratory use.
Since laboratory Reparations frequently form the basis of new industrial
separations, a few additional considerations in choosing a solvent are pre-
sented. The solvent should be inexpensive and readily available. While a
high-boiling solvent is necessary for proper operation of the extractive dis-
tillation, it should not be so high boiling as to require excessively high tem-
perature levels to regenerate the solvent. Corrosiveness becomes a very
important factor in industrial applications when carbon steel is the pre-
ferred material of construction. From a laboratory standpoint, eorrosive-
ness is not so important a factor as in industry, since most laboratory
columns are of glass, and metal parts are usually stainless steel.
B. VAPOR:-I,IQUID EQun,IBRIA AND EQUILIBRIUM STILLs
The sclcetion of a solvent for use in extractive distillation is best made on

the basis of vapor-liquid equilibria data obtained in the presence of a sol-
vent. This is particularly true if much work is planned with the solvent
selected. An equilibrium still is an apparatus so constructed and operated
that equilibrium between vapor phase and liquid phase equivalent to one
theoretical plate is attained. Provision is made to sample each phase
32li CAUL S. CARLSON
(liquid and vapor) for ttllalYHis. In this way equilibrium vapor and liquid
compositions are established.
·When ternary systems are inve:-;tigated, pn,rticularly those in which one
component has a marked influence upon the relative volatility of the other
t\VO components of the mixture, it is essential that a standardized procedure
be used. A preliminary evaluation of possible solvents is made by inves-
t5gating a mixture containing the two components to be separated at a 1:1
ratio (mole, weight, ur volume) and a eonstant proportion of solvent (usu-
aJly one to three times the binary mixture on a mole basis). One or more
determinations may be made on a given solvent in initial scanning. An
example is given below. In making a solvent survey, difficult analytical
problems are frequently encountered because of the presence of the solvent.
In such cases short-cut evaluntion methods such as those described below
may be Ilsed (see page 333). Any solvent selected by these methods should
be checked in an equilibrium still.
The Othmer circulating stilP6 shown in Figure 4 is satisfactory for most
solvents. The vapors formed in the still, A, rise through the central vapor
tube, B, nnd are condensed in the external condenser, C, which may be of
the bulb type shown, or of a spiral or Liebig type. A displacement distillate
receiver, D, continuously returns the distillate to the still. Siphoning is
prevented by vent E. At equilibrium, distillation is discontinued and a
liquid sample of the vapor nnd of the residue, or still bottoms, is withdrawn
for analysis through stopcocks F and G, respectively. For accurate data,
from this or any other equilibrium still, retluxing in the vapor phase of the
still must be eliminated by external-heater windings or hot-gas jackets not
shown in the figure. In the still shown in Figure 4, the liquid is heated by
an internal immersion resistance wire, H. The same result can be ob-
tained by a wire wound around the outside of the lower part of the still or
by the use of a ring heater under A. In the latter case, the boiling tube .J
would be removed, leaving only a simple stopcock. Certain solvents, such
as aniline in hydrocarbon systems, separate as a second liquid phase on
cooling. Such separation can sometimes be prevented by heating the dis-
tillate receiver of the Othmer still.
Another type of circulating equilibrium still has been described 17 which
was developed for use with mixtures having critical solution temperature
below the boiling temperature. The novel features of this equilibrium
still, shown in Figure 5, are the internal distillate receiver, B, bathed in the
ascending vapors and a Cottrell pump, A, for the boiling tube permitting
simultaneous boiling-point and vapor-liquid equilibrium determinations.
It was found that equilibrium was attained in less than 3 hours when a
IS Othmer, Ind. Eng. Chern., 20,743 (1928).
11 Fenske, Cal'lson, and Quiggle, Ind. Eng. Chem., 39, 1322 (1947).
IlL. EX'l'RACTIVE AND AZEOTROPIC DISTII,LA1'ION 327
solvent boiling approximately 80°C. above the binary mixtme was used
and 12% of the charge was held up as distillate.
Both these stills have a tendency to give diffinulty when the distillate
returned to the still boils apprecilLbly, i.e., 50°C., below the mixture in the
still. In such cases the returning distillate tends to flash into the vapor,
thus upsetting true equilibrium. This is sometimes encountered in testing
,
"
"
"
"'I
II
"
I,
"
""
"
"
"
"
I,
"
"
,I
"
Fig. '1. Ot.hmcr-type equilibrium Htill. Fig. 5. Penu-State equilibrium still."
high-boiling extractive-distillation Holvents. An equilibrium still which ob-

viates this difficulty has been described. IS This is a flow-type equilibrium
still which usually requires relatively large quantities of test mixtures. It
has the advantage that it may be used for mixtures having two liquid
phases in either the boiling liquid or the distillate, or both.
After a preliminary exploration of possible solvents has been made, more
18 Colburn, Schoenborn, anel Schilling, Ind. Eng. Chem.,' 3S, 1250 (1943). Mertes
and Colburn, ibid., 39, 787 (1947).
328 CARL S. CARLSON
complete data on the solvent selected may be desired. In this case, the
ratio of the two components in the mixture under study is varied, and the
vapor-liquid equilibrium is determined with the same mole pel' cent of
solvent. This procedure may then be repeated at one or more solvent-
mixture ratios. The data can be expressed in terms of modified relative
volatilities, O's, or as x-y data on a solvent-free basis.
As an example of the evaluation of extractive-distillation solvents, we
may cite the effect of a series of solvents upon the separation of the binary
mixture methylcyclohexane-toluene by extractive distillation as determined
in the internal-receiver equilibrium stilI described above. In each case the
still was charged with the same total number of moles of solution contain-
ing 75 mole per cent solvent and 25 mole per cent hydrocarbon. The latter
consisted of methylcyclohexane and toluene in a 1: 1 mole ratio. Distillate
and residue were freed of solvent by washing with acid, alkali, or water as
required, and the resulting binary hydrocarbon mixtures were analyzed by
refractive index using the data of Quiggle and Fenske. 19 From the analyses
of the hydrocarbon portion of distillate and residue, relative volatilities
were calculated and expressed as as (see equation 3). In order to avoid
misleading conclusions, two additional determinations were made with
each solvent, keeping the total number of moles charged to the still and the
hydrocal'bon:solvent mole ratio the same, but changing the hydrocarbon so
that it contained 25 mole pel' cent, and 75 mole per cent of methylcycIo-
hexane, respectively.
Of the solvents investigated,J7 aniline gave the highest values of as.
For an extended study with this solvent an analytical procedure was de-
veloped permitting the determination of the composition of the ternary
system. 20 A series of five samples was made up. All contained the same
total number of moles with aniline and hydrocarbon in a 1: 1 mole ratio.
The hydrocarbon portion of the samples contained 0) 25, 50) 75) and 100
mole per cent methylcyclohexane. Each sample was charged to the
equilibrium still and distilled for at least 3 hours to establh,h equilibrium.
At the end of that time heating was stopped and, after distillation ceased,
the still was completely drained of distillate and residue for analysis.
Typical data for one run, with all analyses expressed in weight pel' cent,17
are listed below.
Component Charge Distillate Residue

Methylcyclohexane ......... 26.1 60.0 21.'1
Toluene ...... , .... , .. " ... 24.4 31.0 24.2
Aniline ....... , , ........... 49.5 0.0 54.5
19 Quiggle and Fenske, J. Am. Dhem. Soc., 59,1829 (1937).

20 Carlson, Schubert, and Fenske, Ind. Ellg. Chem" Anal, Ed., 18, 109 (HJ46).
III. EX'l'RACTIVE AND AZEO'l'ROl'IC DISTILLA'l'ION 329
From the results for the hydrocarbon portion as can be calculated by

equation (3):
as = (60.0/31.0)/(21.4/24.2) = 2.19
The dnta obtained for the five samples served to establish the relative vola-
tility-compoHition CUI've. The x-y diagram shown'in Figure 3 waR calcu-
lated by meam; of eqnation (8) arranged for a binary ~w:,;teIll to the form:
asx
y = ------- (16)
1 + (as - l)x
where y and x ,Lre the mole fmctions of the more \Tolatile eomponent ill vapur
and liquid, and as is the relative volatility of the more volatile hydrocarbon
with respeet to the other hydrocarbon in the presence of the solvent, ani-
line. Calculation of the equilibrium vapor compositions for various liquid
compositions defines the x-y curve. (Obviously the experimental points
expressed on a binary ba.sis as mole fractions will also define the curve.)
For x = 0.20 and using as = 2.19, as read from the relative volatility-
composition curve just established:
1 = (2.19)~0.20) = 0.353
y 1 + (1.19) (0.20)
TABLE II
n-HJo;p'!'ANE-NIBTHYLCYCLOHBXANE R,BLA'l'IVE VOLA'l'ILI'I'Y IN PRl,SENCE OF VARIOUR
ROLVENrrs
Mole ' Irnprov'e-

Der cent Av. reI. ment
in liquid T,oC. volatility. fRctor, Ref.
Solvent phase (av.) "8 "slot. No.
Aniline ...... " , ... , .. \)2 l3\) 1.52 1. 42 21

78 121 1.40 1.31 21
70 110 1.27 1.19 22
58 113 1.26 1.18 21
Furfural .............. 79 1.35 1.26 21
Phenol ............ , .. , 81 1.31 l.24 21
Nitrobenzene. , ..... , .. 82 1.31 1.24 21
Dichlorodiethyl ether ... 81 1.28 1.20 21
Aminocyclohexane ..... 713 1.16 1.08 21
Pyridine •............. 70 1.4 1.31 22
Ethanol .............. 70 1.3 1.21 22
n-Butanol, ....... , .... 70 1.3 1.21 22
tert-Butu,nol . . ' , , .. , , .. 70 1.25 1.17 22
Acetic acid ... ~ ... , .. , . 70 1.27 1.19 22
None, , .. , , .. , ........ 1.07 1.00 22
21 Griswold, Andres, Van Br.rg, !mci Kasch, Ind. Eng. Chem., 38, 66 (1946).
22 Fenske, Carlson, and Quiggle, Ind. Eng. Chern., 39,1322 (1947).
330 CARL S. CARLSON
TABLE III
i\JE'l'HYLCYCLOHlcXANE- TOI~FJcNIC n"LATrvl~ VOr,ATrr,!'l'Y IN PRESJ'WCE OF VARIOUS
SOLVENTS
Mole Improve-
per cent Av. reI. ment
in T. "C. volatility. factor Ref.
Solvcnt charge (av.) as "s/a No.
Acetic acid .............. . 67.9 95 2.2 1.6 23

70" 2.1 1.0 22
Methanol ............... . !l7.3 131. 1. 72 1.3 23
Ethanol ................ . 66.9 70 1. 76 1.3 23
Ethanol ................ . 70" 1.0 1.4 22
n-Propanol .............. . 67.5 87 1.8 1.3 23
Isopropanol ............. . 62.5 73 1. 75 1.3 23
n-Butanol ............... . 70" 1.55 1.1 22
tert-Butanol. ............ . 70· 1.6 1.2 22
Acetone ................. . 30.1 70 2.1 l.G 23
·19.4 66 2.4 1.8 23
58.3 132 3.15 2.3 23
01.0 62 2.9 2.1 23
115.0 61 2.6 l,!l 23
68.1 50 3.3 2.4 23
Methyl ethyl ketone ...... . 66.2 80 2.3 1.7 23
Piperidine ............... . 65.8 97 1.7 1.3 23
Pyridine ................ . 33 102 1.!l 1.4 23
4fl 101 2.2 1.0 23
57.5 101 2.3 1.7 23
70" 2.4 1.8 22
06.7 100 2.5 1.9 23
Aniline ................. . 32.9 116 2.05 1.5 23
50.0 115 2.4 1.8 23
50.8 120 2.6 1.9 23
60.2 125 2.7 2.0 23
65.8 121 2.6 1.G 23
43.2" 113.6 l.OG 1.'15 22
54.4" 117.5 2.HJ 1.62 22
6G.7" 123.0 2.37 1. 7G 22
77.7" 132.5 2.54 1.88 22
83.4" 139.8 2.59 1.92 22
Nitl'ornethanc ........... . 32.4 88 2.3 1.71 23
58.8 87 4.24 3.14 23
Acetonitrile .............. . 68.3 73 4 3 23
Phenol ................. . 34.7 11.0 2.0 1.5 23
52.1 113 2.3 1.7 23
59.8 118 2.2 1.6 23
65.8 119 2.7 2.0 23
Triethyl borate .......... . 65.5 112 1.24 0.9 23
Table continued
23 Updike, Langdon, and Keyes, Trans. Am. Inst. Chem. Engrs., 41,717 (1945).
III. EX'CltAC'l'IVE AND AZEO'fROPIC DIR'l'ILLA'frON 331
TABLE III (Concl1((le.rl)
Mole Improve-
pm ce'nt Av. reI. )Jl~nt.
in '1'. 'C. volatility, factor, Hel.
t)olvent charge (av.) "13 aRIa No.
Diethylene glycol mono-

ethyl ether. . . . . . . . . . . . . 70" 2.0 1.5 22
Diethylene glycol mono-
butyl ethul'. . . . . 70" 1.3 1.0 22
Ethyl acuLate. . . . . . . . . . . . . 70" 1.8 1.3 22
Ethylene glycol mOllo-
methyl ethel'. . . . . . . . . . . 70" 2.2 1.0 22
Ethylene glycol mono-
ethyl ether. . . . . . . . . . . . . 70" U) 1.4 22
Ethylene glyeol mono-
butyl ether... .......... 70" 1.4 1.1 22
None ................ ; .. . 1. 1-1. 6 22
lW. 1.3,)
a Mole pel' CBllt aniline in liquid phase.
TABLE IV
CYCLOlnlXANI"-BENZENE RELA'l'IVE Vor"I,TILITY IN Pm:Sl,;NCE OF VAUrous SOLVENTS21
Mole Improve-
per cent Relative ment
in volatilitJ', factor!
Solvent charge T,oC. as CIS/a
Acetic acid .................... li9.0 84 1. 75 1.78

Methanol ..................... 67.3 53 1.58 1. 61
Ethanol ....................... 67.3 65 1.36 l.38
n-Propanol .................... 70.5 79 1.26 l. 28
Isopropanol ................... 07.9 70 1.22 1.24
Dioxane ...................... n7.4 86 1. 75 1. 78
Chlorex (dichlorodiethyl etlwl') .. 67.5 105 2.31 2.36
Methyl CeJlosolve .............. 66.7 85 1.84 1.88
OelloBolve ..................... 67.5 95 1. 58 1.01
Oarhitol ...................... 66.8 87 1. 99 2.03
Acetone ...................... 06.3 55 2.03 2.07
Methyl ethyl ketone ........... 65.1 72 1. 78 1.81
Diacetone ..................... 67.3 80 1.82 1.85
Pyridine .............. , ....... 66.9 93 1.83 1.86
Aniline ....................... 6n.8 93 2.11 2.16
Nitromethane ................. 67.8 74 3.00 3.06
Nitrobenzene .................. 68.2 lO2 2.25 2.30
Acetonitrile .................... 67,3 65 2.85 2.92
Furfural ....................... 67.1 79 3.10 3.16
Phenol ....................... 66.8 92 2.01 2.05
24 Updike, Langdon, and Keyes, T7·ans. Am. Inst. Chern. Engrs., 41, 717 (1945),
332 CARL S. CARLSON
TABLE V
N O'NTOI,UENE-ToI,UENE a VOLATILI'l'Y RA'froS IN PnESMNCE OF VARIOUS SOfNENTS
-------_-"_- Improve-
Wt. Volatility ment
H.P .. per ceut rat.io, factor, Ref.
Solvent °C. solvont b "8 "s/a No.
Antimony trichloride .............. 277 50 3.20 2.4 25

Sulf01ane ........................ 287 75 5.60 4.2 25
Dimethylsulfolane ................ 281 75 3.23 2.4 25
281 50 2.76 2.1 25
Dimethylsulfolune-cumene (85-
IS, wt. %) ..................... 202 50 2.41 1.8 25
Furfural ......................... 153 50 2.30 1.7 26
Acetonyl acet.one ................. 188 50 2.20 1.7 26
Nitrobenzene ..................... 211 50 2.16 1.6 26
Nitrotolurme ..................... 222 50 2.Hi 1.6 2(j
Phenol .......... , ., .............. ]82 75 2.78 2.1 25
Phenol. .......................... 182 50 2.10 1.6 26
Aniline ...................... , ... 183 75 2.75 2.1 25
183 50 2.08 1.6 26
Chlorex (dichlorodiethyl ether) ..... 178 50 2.09 1.6 26
Phenyl Cellosolve ................. 240 50 2.01 1.5 26
Phenol-cresol (60-40) ............. 192 50 1.98 1.5 26
(40--60) .............. 195 50 1.95 1.5 26
Acetophenone .................... 202 50 1.95 1.5 26
I-Methoxy-2-hyclroxy-3-phenoxy-
propane ........................ 278 50 1.92 1.5 26
Ethyl Oarbitol .................... 202 50 1.85 1.4 26
In- or p-Crc601 .................... 202 50 1.85 1.4 26
l-Ethoxy-2,3-dihydroxypropane ..... 222 50 1.80 1.4 26
l-Isopropoxy-2,3-dihydroxypropanc. 226 50 1.70 1.3 26
Diacetone alcohol ................ 190 50 1.64 1.2 26
1-Methoxy-2,3-dihydroxypropane
(2 phases) .......... , ........... 220 50 1.46 1.1 26
No solvent ....................... 1.32 1.0 25
~ Nontoluene fraction was Il. dell.l'Omatizee:! straight-run petroleum fraction boiling
between 210 and 235 cr,'. (true boiling point 5 and 95% points, respectively).
/, Weight per cent, solvent in charge to equilibriuTil still. The hydrocarbon mixture
WItS a 50-50 mixtul'fl of ltrollll1tic nne:! nonaromatic hydrocllrhons. .
Since the analyses were obtained 011 a ternary basi.s, the. solvent concentra-
tion in the liquid phase at equilibrium is known as well as t.he solvent con-
centration in the vapor phase. This is important for design purposes and
for the operation of a laboratory column employing cxtnwtive distillation.
Sets of five samples each were made up at solvent. f'oll('entrations of 40,60,
~ll
Staaterman, Morris, Stager, and Pierotti, Chern. Eng. PI'()grp.~.~, 43, 148 (1947).
26DUnn, Millar, Pierotti, Shiras, and Souders, Jr., Trans. Am. Inst. Chem. En(J'r8., 41,
631 (1945).
III. EX'!'RAC'l'IVE AND AZEO'rROPIC DISTILLA'l'ION 333
70, und 75 mole per cent aniline in each charge. The experimental pro-
cedure was then repeated.
Relatively few data are yet available on the vapor-liquid equilibria of
binary mixtures in the presence of a solvent: very few were published prior
to June, 1945. Pertinent data arc summarized in Tables II-VI (pp. 329fL).
When the data are known, hoth the liquid and the vapor compositions OIl a
solvent-free basis are given, as well as modified relative volatilities, as, and
enrichment ratio, i.e., the ratio of as:a which indicates the effectivenesi:i of
the solvent. It should be remembered that as depends on the solvent COll-
centration as discussed in detail on page 324.
TABLE VI
RELATIVE VOLATILITIES OF BINARY MIXTUREI;l IN PRESENCE OF VARIOUS SOLVENTS'~
Mole
por cent Rel",tive
in . volatility,
Solvent liquid
n-Octane - Toluene Mixtures

Aniline .•.............. , 70 1.6-1. 7
Pyridine .............. . 70 1.4-1.6
None ................. . 0.55--0.83
Methylcyclohexane-n-Octane Mixtures
Aniline ............... .. 70 1.4-1.5
None ................. . 1.6--1.8
C. ESTIMA'1'lON OF EFFECTIVENESS OF SOLVENT
If the neCCS;;!1l'Y data Oll the respective vapor-liquid eqUilibria are not
known and lack of time or experimentul equipment do not permit an inves-
tigation of these equilibria, !Ul estimate of the potential applicability of It
solvent is desirable. Several methods for such an estimate are described
in the following:
Boiling-Point Deviation. The basis for separation by extractive dis-
tillation is the deviation from ideality of the components to be separated
when in the presence of the solvent, the deviation of one component being
much greater than that of another. Thus, a comparison of the actual
hoiling points of solutions of the individual components with the boiling
points predicted for ideal solutions will give a rough index of the separating
effieacy of the solvent. 'rhis test can be eal'l'ied out as follows: Equal
volumes of one component and of the solvent are mixed, and the boiling
point of the mixture is determined, This is repeated with the next com-
ponent. The difference in the observed boiling points is noted. The
334 CAUL S. CAULSON
theoretical boiling point of each mixture is calculated from its composition

(mole basis) and the boiling points of the pure compounds, assuming a
linear variation of boiling point with composition. The difference between
the calculated boiling points is compared with the experimentally deter-
mined difference. If the latter is appreciably greater than the calculated
value, the solvent should give improved separation when used in extractive
distillation. The exact magnitude of the improvement, however, requires
experimental verification in an equilibrium still.
Another method using boiling-point determinations has been applied
to the separation of diolefin, olefin, and paraffin hydrocarbons with similar
boiling points and in particular to solutions of butadiene, butylenes, and
butanes. 27 The test consists of determining the boiling points of mixtures
of 10 g. of proposed solvent, such as 1,4-dioxane or ethylene dichloride,
with 1 g. of n-butalle (b.p. -5.5°C.) and of 10 g. of proposed solvent with 1
g. of butadiene-1,3 (b.p. -4.5°). If the boiling point of the latter mixture
is 10° higher than that of the former mixture, the solvent is considered suf-
ficiently selective to be used. A difference of 20° or more is preferred.
Critical Solution Temperature. Many solvents useful in extractive
distillation are not completely miscible at room temperature with one or
both of the components to be separated. The maximum temperature
at which any mixture of component and solvent, for example, a hydrocar-
bon in nitrobenzene, can exist as two liquid phases is defined as the critical
solution temperature (C.S.T.) of the mixture. These values have been rc-
ported for illany binary mixtures. It appears that the greater the differ-
ence in C.S.T. of the components to be separated in the solvent, the better
the solvent is suited as a separating aiel in extractive distillation. For ex-
ample, n-heptane and methylcyclohexane, with critical solution tempera-
tures in aniline of 70 and 41°C., respectively,28 have been separated by ex-
tractive distillation using aniline as a solvent. With a 2:1 aniline to hy~
drocarbon weight ratio, the separation was more than threefold that at-
tainable by fractional distillation in the same column. 29
Critical solution temperatures of seven selected nonaromatic hydrocar-
bons and three nonaromatic oil fractions in a number of solvents have been
compiled by Francis. 28 A similar study of 38 cyclic hydrocarbons in a
number of solvents has been reported by the same investigator. 30
In general, aromatic hydrocarbons have been found to be much more
soluble than nonaromatic, as indicated by lower critical solution tempera-
ture. The latter are frequently completely immiscible with a solvent in
21 Young !Ind Perkins, U. S. Pat. 1,948,777 (1934).
IS Francis, Ind. Eng. G'hem., 36, 764 (19'14).
!9 Dicks and em'lson, 'i'rans. A.m. [nst. Chem. Engrs" 41, 789 (1945).
!O Francis, Ind. Eng. G'hem., 36, 1096 (1944).
III. EXTRACTIVE AND AZEO'l'ROPIC nrSTILLATION 335
which aromatic hydrocarbons have fairly low critical solution temperatures

and aromatics are often completely soluble in solvcnts in which the non-
aromatic hydrocarbons have low critical solution tcmperature. In either
case such a solvent should facilitate separation by extractive distillation ..
An example of the latter type is the use of aniline to facilitate the separation
of methylcyclohexane (C.S.T. 41°C.) from toluene (miscible down to the
freezing point of aniline). By extractive distillntion using 40 weight pCI'
cent of aniline, toluene-free methylcyclohexane can
be obtained as distillate with only 1.6 mole pel' cent o
of methylcyclohexane in the still. In the same
column undcr fractional-distillation operation at total
reflux with 5.5 mole per cent methylcyclohexane in
the still, the overhead product contained only 82
mole per cent methylcyclohexane. 2u
Another form of C.S.T. frequently reported is thc
aniline point (A. Pt.). The aniline point is the
highest tempemture at which a 1: 1 volulUc mixture
of aniline and a givcn hydrocarbon exists as two
liquid phases. The aniline point is widely used in
the petroleum industry where a standard apparatus
and test procedure have been proposed.31 A modi-
fication of this A.S.T.M. test apparatus is shown in
Figure 6. It consists of a jacketed test tube con-
taining a looped wire for agitation and a thermometer.
Equal volumes of component and solvent, usually 10
ml., are introduced and heated slowly with agitation
until completely miscible, when the temperature of
the solution is noted. The solution is then cooled
"lowly with agitation until cloudiness indicates phase
separation. The temperature at which the cloud first
forms is noted. This is repeated several times and Fig. 6. Aniline-point
the average temperature is taken ItS the aniline apparatus.
point.
The same apparatus can be used to determine edtical solution tempera-
tures, the only variation being that the ratio of solvent to component is
varied and n number of determinations is made so that a solution tempera-
ture-liquid composition curve can be plotted. The maximum of this
curve is taken as t.he critical temperature of solution. Since this curve is
usually flat in the range of 40 to 60 volume per cent solvent, values for criti-
cal solution temperature and aniline point are frequently very close. Ani-
31 A.. S. T.M. Standards on Petroleum Products and Lubricants, D61l-44T, P41-43.
American Society for Testing Ma.terials, Philadelphia, 1945.
336 CARL S. CARLSON
TABLE VII
TERNARY SYS'l'EMS WITH W A'rER AS Orn~ COMPONJ~N'r33
Temp.!
Components °C. Component~
Aeeti(J acid- lsonmyl ether .. , . , . , . , .... . o

Benzene .. , , , . , ... , . , ..... , 25 Benzaldehyde. , .. , .. , ..... .
Benzene, ............. , ... ,
Benzene ............ " .. , ..
25,35
20,25
Benzene, .. , , ... , , . , ... , .. .
Benzene. , . , . , , ..... , ..... .
°
25
26,50
Chloroform ........... , . , .. 18 Benzene. , , . , .... , ....... , . 20
Chloroform ........... , , .. . 20,25 Benzene .. , ... , .. , , .. , . , . , , 25
Epichlorohydrin ... , ....... . 10 Benzene ........... , , . '. , .. , 25
Toluene .... , ........ , .... . 25 Benzyl acetate. , . , , . , . , , ... o
Acetone-
Benzyl aleohol .... , . , , , , , .. o
Benzyl ethyl et.her, , , .. , , _. _ o
Bromo benzene ............ .
Chloroform ............... . ° 25 Bromobenzcne . , , ... , .. , ...
Bromotoluene.,., .. , ... " ..
o
o
Chloroform ....... , ... , , .. .
Furfural. , ... , , , ..... , .... . ° 25
Isobutl1nol. , , , , , ... , . , . , . , .
Isobutyl bromide, .... , . , , , .
o
Phenol. .. , .. , ........ , . , . , 56,5 Carbon tetrachloride, . , , . , . , °
°° °°
Pot!tssium hydroxide .. , , . , .. Chloroform .. , . , . , , , , , .... .
Sodium hydroxide. , . , , , .. , , Cyciohexane, .. , .. , . , ... _.. 25
Aniline- Cyclohexane ... , . , , . , . , . _.. 20,25
Formic acid .............. . 0.20 Cyclohexallc. , . , , , , , , ..... , 25
Propionic [I,deL ........... . 0.20 Cyelohexene, . _ , , . , . , , _.. , . 25
Ethyl aeetate, ... , ..... , .. . 0,20
Benzene- Ethyl acetate, ..... _..... , .
Acetic acid .. , ....... , , ... ,
Acetic acid ... , , ..... , , ... .
Acetic acid ...... , .. , , ..... ,
25
25,35
20,25
Ethyl butY1'l1te. , , ....... , , .
Ethyl ether, ...... , ..... ,.,
°o
25
25 Ethyl ether, , , . _.. , ...... , , 0,25
Ethanol. . , ... , ........... .
Ethanol. , .... , ........... , 25,50 Ethyl ethOJ'. , .. , , , ..... , , , . 25
Ethanol.. , ............... . 20 Ethyl ether ...... , .. , , . , .. . 0,25
Ethyl propionate .... , , .... .
Ethanol ...... , ........... .
Ethanol. ................ ..
20,25
25 Ethyl chloride, ... , , , ..... .
Ethylidene chloride ..... , .. .
°oo
Ethanol., ................ . 25
Pyridine .......... , ...... . 25 Hexane.,." .. ,." ....... . o
Hexane ...... , , , , , , ' . _. , . , 25
Isobutanol- Mesitylcne, . , , , . , .. , , _. , , . o
Hydrobromic acid ......... . 25 Methylaniline. , , ... , ...... .
Hydrochloric acid ......... .
Hydriodie acid ............ .
25
25
Nitrobenzene .. , , .. , . , .... .
IJ-Nitrotoluene ... , .. , .. , .. .
°°o
Phenetole ....... _, ..... , , . o
n-Butanol-methanol ......... . 0,15,30 Pinene .. , . , . , ...... , , .... .
tert-Butanol-ethyl. a(:etate .... . 0,20 Propyl bromide ..... , , , , . , .
Toluene ... , . , .. , , , .. , , . , ..
°o
20
CS2-acetic anhydride ........ . 0, IS Toluene, .. , ... , , , .. , ... , .. 20,25
Carbon tetraehloride- Toluene ... , . , . , .. , , , .... , , 25
Ethanol, ........... , ..... , rn-Xylene, . , , . , , .. , ...... , 0,50
Methanol, ............... .
n-Propanol ... , , , , ...... , , ,
°° o
o-Xylene ... , , , ... , , . , , , . , .
p-Xylcnl!, . , . , . , , , .... , , , . ,
o
o
Ethyl acebtte-
Chloroform- tart-Butanol. , . , , , .. , , . ' , .. 0, 20
Acetic acid, , .. , . , , , . , .. , .. 20,2,5 Ethanol. ... , .. , , , , . , .. , .. _ 0, 20
Acetic acid, ... , , , .. , ..... . 18 Furfural ... , . , , .. , , .. , ... . 25
Acetone .... , , , , , , , , , . , , , .. 25 Methanol. , , , . , . , , ....... . 0,20
Acetone, , .. , , , ... , ....... .
°°°
Isopropanol ... , , , , , ..... , . 0,20
Ethanol ... , ..... , .. , , , ... , n-Propanol, .. , , , , , , . , ... , , 0,20
Methanol ... , .. , .... , .... .
Ethyl ether-
Ethanol- Ethanol. , , , , . , .. , , , , .. , __ , 26
Isoamyl alcohol. ......... .. 15,5,28 Ethanol. . , . ' , ... , , , , , .... . 0,25
Isoamyl bromide .... , ... , , , o Ethanol. . , , , . , , .. , , ...... , 25
Table conlinusa
III. EXTRACTIVE AND AZEOTROPIC DISTILLA'l'ION 337
TABLE VII (Concluded)
Temp.,
Component!: ·C. Components
Ethanol .................. . 0,25 Nit.rohenZClle-

Triethylamine ............ . 0, 12.4, Ethanol. .......... " ..... . o
30.5 Sulfuric acid .............. . 22
li'urfural- Phenol-
Acetone .................. . 25 Acetone. . . . . . . . . . . . . . . . . . . 56 . 5
Isoamyl !lcetate ........... . 25 Hydrochloric acid. . . . . . . . . . 12
Ethyl acetate ............. . 25 Sodium hydroxide ......... .
Hydrobromie aeid- Potassium hydroxide ....... .
Isoamyl aloohol ........... . 25 Triethylamine ............ -2,7,10
Isobutanol ................ . 25 57,75 '
Hydrochloric acid- Isopropanol-
Isoamyl alcohol. .......... . 25 Cyclohexane .............. . 15,35
Isobutanol ................ . 25 Ethyl acetate ............. . 0,20
Cyclohexallonc ............ . 25 Toluenc .................. . 25
Phenol ................... . 12 n-l'l'opanol-
Hydriodie lLcid- Isoamyl !tlllohol ........... . 25
Isoamyl ulcohol ........... . 25 Bromotoluelle ............. . o
Isobutanol ............... . 25 C[!'l'bon tetrachloride, ...... . o
Ethyl [!,cetate ............. . 0,20
Methanol- Propionic acid-
Isoamyl alcohol ........... . 28 Aniline ................... ,
Bromobenzene ............ . o 0,30
n-Butltnol ................ . 0,15,30 a-Toluidine ............... , 0,20
Carbon tetrachloride ....... . o Toluene-
Chloroform ............... . o Acetic acid ............... . 25
Cyclohexane .............. . 25 EthanoL ................. . 20
Cyclohexene .............. . 2,1} Ethanol. ................. . 20,25
Ethyl acetate ............. . 0,20 Ethanol. .................• 25
Ethyl bromide ............ . o Methanol .... , .. , ........ . 25
Toluene ............. , .... . 25 Isopropanol. .............. . 25
TABLE VIII
NONAQUJilOUS TERNARY SYSTJilMS"
Temperuture l
Component ac,
--------,------ --------------_--
Acetone-glY(101
Benzene..... ............................... 27
Bromobenz[)ll{). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
Chlorobenzcnc. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
Nitrobenzene ... -. . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
Toluene ........................ " .. , ....... 27
Xylene ........ , . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
Aniline- n-heptane-methyleye]ohext\ne. . . . . . . . . . . 25
Ethanol-henzene-glycerol. . . . . . . . . . . . . . . . . . . . . . 25
line points are sometimes incorrectly reported in the literature as critical

solution temperatures. In no case can the aniline point exceed the critical
solution temperature. Discrepancies in literature values stem from four
sources, which are listed in order of decreasing; probable magnitUde: (1)
338 CARL S. CARLSON
Impurities in sample or solvent. (2) Too rapid heating and cooling during
tl.tl:l determination. (3) Error in calibration of thermomoter. (4) Heat
piclcup from unshielded lights llear app~tratus. Hygroscopic solvents such
as aniline must be carefully dried just before use, and a nonaqueous heating
and cooling medium must be provided. An extensive survey of aniline
points and critical solution temperatures of hydrocarbons in aniline has been
made by Ball. 32
Liquid-Extraction Data. Prior to the development of extractive dis-
tillation, separation of narrow boiling-range fractiolls according to chemi-
cal type was effected chiefly by liquid-liquid extractioll. Phase-equilib-
rium studies of !1 number of systems with partial miscibility have been
reported in the literature. Solvents suitable for liquid-liquid extractions
can frequently be uscd in extractive distillation provided that they meet the
other qualifications mentioned above, in particular that of having rela-
tively high boiliIlg points. As an aid in selecting a solvent based on such
information, a .summal'y of ternary extraction systems prepared by J. C.
Smith 33 is reproduced in Tables VII-VIII (pp. 33G-337). Note that the
ternary system aniline-n-heptane-methylcyclohexane is listed which can
be separated by liquid-liquid extraction with aniline. 34 This system has
also been separated by extractive distillation 29 using aniline as a solvent.
Chemical Nature of Solvent. It waS mentioned above that solvents
which form highly nonidcal solutions arc usually advantageous. Inas-
much as deviation fr0111 ideality is caused by interaction between solute
and solvent, the solvcnt should be selected with reference to the nature of
the mixture to be separated. The interaction between solute and solvent
may vary from very loose solvation, e.g., by dipole interaction, to fairly
stable complexes involving hydrogen bonds and other co-ordinative link-
ages. The interaction should takc place with the less volatile component
of the distilland in order to enhance the normal volatility relationships.
The relative polarity of a compound has hcen used in the past to cxplain
its ideal or nOllideal behavior in solution. Liquids of similar polarity when
mixed are expected to form nearly ideal solutions, while those of different
polarity are expected to dcviate from ideality in proportion to their differ-
ence. Ewell et al., 35 among others, disagree with this concept in connection
with azeotropic distillation, and point out certain exceptions to the above
generalization. Hildebrand 36 has recently introduced a modification of
the polarity {Joncept to make allowance for sterie hindrance existing in some
, 32 Ball, U. S. B1tT. Mines Rpts. Invest., 3721 (1943) .
• 33 Smith, Inil,; Eng. Chem., 34,234 (1942).
I ,.3!VarteressiILn and Fenske, Ind. Eng. Chem., 29, 270 (1!l37) .
.. n Ewell, 'Hll.rrieon, and Berg, Ind. Eng. Chern., 36, 871 (1944).
, .0 fiilctcbr&nO,,§olubi(itics. 2nd ed., Reinhold, New York, 1936.
III. EX'rUACTIVE AND MmO'rUOPIC DISTILr,ATION 339
molecules having high dipole moment, whicb in effect tends to neutralize

the polarity of the compound. Bearing in mind that there are exceptions,
we shall use the concept of polarity as the basis for discussion.
The types of mixtures to be separated by extractive distillation can be
classified as follows: Ca) highly polar; (b) nonpolar; (c) one component
polar, the other nonpolar; and (cl) both components of intermediate polar-
ity.
For a highly polar mixture such as acetic acid and water, a nonpolar
solvent such as a high-boiling oil fraction should be helpfupr by increasing
the volatility of wat,er. If the mixture is nonpolar like cyelohexane-ben-
zene, a polar solvent sueh as aniline will ffwiiitate the separation Z9 by render-
ing cyclohexane more volatile.
For the case in which one constituent is nonpolar and the other polar, one
might expect either a polar or a nonpolar solvent to be satisfactory. In
practice, the solvent which augments the llormall'elative volatilities is pre-
ferred. For examplc, if the polar component is less volatile, a polar solvent
will augment the normal volatilities while a nonpolar solvent willl'cverse
them. Other things being equal, however, a polar sobrcnt will be prefer-
able since a more highly llonideal solution will usually be obtained.
For a mixture of intermediate polarity one would expect either H highly
polar or a nonpolar solvent to be satisfactory. The effect of the solvent on
normal volatilitiec: is not easy to predict in this case but a polar solvent
would be recommended for an initial trial.
The points to be considered in'selecting a solvent for enhancing relative
volatilities in extractive distillation are:
(a) Binary mixture, highly polar: Select a polar solvent or a nonpolar solvent
more soluble with the higher boiling component.
(b) Binary mixture, nonpolar: Select a polar solvent more soluble with the
higher boiling component.
(c) Binary mixture, the more volatile component polar, the other nonpolar:
Select a nonpolar solvent or a polar solvent. The latter may reverse relative vola-
tilities.
(d) Binary mixture, the more volatile component nonpolar, tIle other polar:
Select a polar solvent.
(e) Binary mixture, moderately polar: Followa.
In Table IX are listed the polarities of !1 few t.ypical organic compounds
reported by Hildebrand,36 expressed as dipole motnents.;18
It was shown in Figure 2 that for a given solvent the enhancement in
relative volatility depends upon the solvent concentration, the maximum
37 Riegel, lndusirial Chemistry. 4th mI., Reinhold, New York, 1942, p. 283.
38 For an exphnation of dipole moment, see Smyth in Weissllerger, ed., Pl/.1jsicai
Methods of Organic Chemiatry. 2nd ed. Intersciellce, New York, l)al't II, 1950. .
340 CAUL S. CARLSON
TABLE IX
POLARITIES EXPU!;:sSED AS DIPOL~: MOMENTS
. p. X 10" e.s.u.
Compound at 20°C.
H,O Water ................... ·.······:·· 1.85

CS,. Carbon disulfide ..................... . o
S02 Sulfur dioxide ....................... . 1.61
CCI, Carbon tetraehloride ................. . o
n-C.H14 Hexane ............................ . o
n-C 1 lI16 Heptane ......... , ................. . o
n-CsH" Oetane ................... ·········· . o
C,H 12 Cyclohex(Lne ........................ . o
CsH6 Benzene .............................. . o
C,H.CH a Toluene ............. ·····.······.·· . 0.4
p-C.I-I.(CH,), p-Xylenc ........................... . o
CloHa Naphthalene ................ , ....... . o
CHCla Chloroform ............. , . " , , ...... . 1.05
CHB r 3 Bromoform ........... , ......... , , .. . 1.0
C,IInBr Ethyl bromide ...................... . 1.83
CiH,I Ethyl iodide, ....................... . 1.66
CJ-I,CI Chlorobellzene ...................... . 1.56
C,H,BI' Bromobenzene ...................... . 1.4\)
ColI.I Iodobenzene ...................... , .. 1.30
o-C.H.CI, o-Dichlorobenzene ... , , .............. . 2.24
m-C 6lI,Ch tn-Dichlorobenzene ... , ....... , ...... . 1.48
p-C,fI,CI. p-Diehlorobemene ................... . o
CH,OH Met,hyl alcohol ...................... . 1.68
C,H,OH Ethyl alcohol ....................... . 1. 70
CsH17 0H Octyl alcohol ...... , ................ . 1.68
C,H,OH Phenol .... , .................... , ... . 1. 70
(CH 3 ),O Dimet.hyl ether ...................... . 1.32
(C.H 5),O Diphenyl ether .................... , .. 1.14
(CH 3 ),CO Acetone ............................ . 2.8
CH,COOC.H, Ethyl acetate, . , .................... . 1.86
CH3CHO Acetaldehyde ........... , ........... . 2.4G
CH,CN Acetonitrile ......................... . 3.2
CHaNO. Nitromethano ....................... . 3.42
C.H,NO, Nitrobenzene ....................... . 4.19
C,H,NH. Ethylamine ......................... . 1.3
C,H,NH2 Aniline ............................. . 1. 51
value of relative volatility being realized fur the limiting case of 100%
solvent. In practice, a maximum enhancement in relative volatility is not
necessary, and a value of 1.5 to 2.0 is in general adequate for good separa-
tion by batch distillation in a moderately efficient column. For example
a mixture with a relative volatility of 1.5 at 10: 1 reflux ratio in a15 theoret-
ical plate column will give an overhead product containing 90 mole per
cent of the more volatile component, with only 20 mole per cent in the still.
III. EX'l'RAC'l'IVE AND AZEO'l'I{QPrG DIWrILLA'l'ION 341
This is very dose to the maximum separation at 10: 1 reflux ratio, as can be
seen in Figure 7:
Ct:
a
n.
~ 0.8
i:
u
~ 0.15
2:
o
ti 0.4
~
lL
~ 0.2
a
::;;
0.2 0.4 0.6 0.8 1.0

MOLE FRACTION MVC IN LlQUIO
Fig. 7. Th~'O!'etiC[tl-platerequirements for batch distillation at
10: 1 reflm: ratio with 1.5 relative volatilit.y.
4. Laboratory Apparatus and 'Procedures

The apparatus and procedure for laboratory extractive distillation can
be very simple or quite elaborate, depending upon the frequency of such
distillations aqd the number of independent operations the operator is will-
ing to make.
A. BATCH EXTRACTIVE DISTU,LATION WITHOUT SOLVENT RECYCLE
The simplest laboratory application of extractive distillation involves

merely a modi(ication of the operation of a conventional batch laboratory
fractional-distillation column. During distillation, solvent is added con-
tinuously at the head of the column through the thermometer OJ' thermo-
couple well or at the condenser. The product withdrawn at the top of the
column contains solvent, which can be removed by a separate operation,
and solvent builds up in the still pot which should be of adequate size to
ha.ndle the quantity of solvent added. Bottoms product and solvent can be
separated after the extractive distillatiun by running the column as a con-
ventional fractional-distillation column. This type of operation is particu-
larly suitable for the purification of volatile compounds used in other physi-
calor chemical studies, particularly when the impurities are present in low
concentrations (5% or lesl:l).
An illustration of such a laboratory setup is shown in Figure 8. A typi-
cal packed laboratory column has been selected, which for purposes of dis-
342 CARL S. CARLSON
cussion has a column scction, A, 12 mm. I.D. packed for 50 cm. with l/a-
in. stainless-steel helices which will give 30-40 theoretical plates. When
operated under adiabatic conditions of total reflux with a throughly wetted
packing, this column will have a toluene boil-up rate of at least 500 ml. per
hour. The column is rendered adiabatic by an electrically wound glass
jacket, .J, which is surrounded by a second glass jacket not shown. 'The top
of the column consists of a conventional head which provides a cold-finger
concienser, C, and product take-off, D, through stopcock E and vent G.
The thermowell, H, has been modified by the addition of a dropping funnel,
P, by means of which solvent can be added continuously at a controlled
rate through stopcock J( to the liquid reflux in the column. A 500-m!.
funnel is adequate, Rince it can be refilled as needed. A graduated funnel
is recommended to facilitate checking the rate of solvent addition. An
enlarged still pot iK shown to provide capacity for solvent build-up.
The purification by extractive distillation of a liter of benzene containing
5 volume per cent cyclohexane as an impurity would require taking over-
lwad slightly more than 50 mI. of hydrocarbon (5% of 1000 m1.) plus a
volume equivalent, to the column holdup of 7 ml. to allow for the concentra-
tion gradient in the column during the transition from cyclohexane to ben-
zene. If aniline were used the distillate would contain between 10 and 20
volume per cent solvent. In this case the minimum volume of distillate
necessary during the extractive distillation would be approximately 80 ml.,
although it would be well to take over n.t least 100 or 150 ml. making ap-
propriate cuts, particularly at the expected transition from cyclohexane to
benzene. It is good practice to take volume or weight fractions as in any
analytical fractional distillation and to test the purity of the solvent-free
products.
A hydrocarbon reflux ratio of 10: 1 is recommended for good separation
and clean-up, while a solvent-liquid hydrocarbon ratio of 1: 1 should be
satisfactory. At 500-ml. boil-up rate and 10: 1 reflux ratio it will take
slightly over 2 hours to collect 100 m!. of product. Allowing 2.5 hours for
product take-off and an hour for column equilibration, extractive distilla-
tion will continue for 3.5 hours. This will require 450 m!. solvent per hour
or 1575 ml. total. With an initial still charge of 1000 mI., a 5-liter flask
will be required for B in Figure 8.
In general, the data on the system to be separated are not as complete
as in the above illustration. In such cases a hydrocarbon reflux ratio of
10: 1 and a solvent-hydrocarbon ratio of 1: 1 are selected and the equivalent
of an analytical distillation is carried out. When apparatus such as that
shown in Figure 8 is used; the extractive distillation can be carried out in
several successive steps jf necessary. First, solvent is added during extrac-
tive distillation to the limit of the still capacity; next, solvent addition is
!II. EX'I'RACTIVE AND AZEOTROPlC DIS'l'ILI, .... TION
discontinued and the unknown mixture is stripped from the solvent by con-
ventional distillation; then the unknown is recharged to the still after re-
moval of regenerated solvent, which is added at the head of the column for
continued extractive distillation. A procedure such as that just described
would have been necessary in the above example jf the mixture had con-
Fig. 8. Modification of labOr!Ltory frac- F'ig. O. Lllbomtol'Y ext,motive-distilla-

tional-distillation appltl'll.tus for extmctive tion apparatus with solvent recycle, single
distillation. control point.
tained 5% benzene and 95% cyclohexane, since it would have been neces-
sary to take all the cyclohexane overhead as distillate.
It should be noted here that if the solvent is not preheated but is at room
temperature, as will generally be the case in apparatus such as Figure 8, it.
will tend to condense some vapors so that the hydrocarbon reflux ratio will
be higher than that observed at the condenser. It is usually necessary to
warm the solvent to ensure that vapors roach the condenser.
344 CARL S. CARLSON
B. BATCH EX'l'HAC'l'IVE DISTILLA'l'ION WITH SOLVEN'l' HE CYCLE
Where extractive distillations are carried out with any frequency, a

batch laboratory unit designed for solvent recycle will be desirable. The es-
sentials of such an apparatus are shown in Figure 9 on page 343. It con-
sists of two fractional distillation columns. Qne contains the extractive-
distillation section, A, the enriching section, B, for solvent removal from
the distillate which is recovered at F, and an accumulator, C, for solvent
and mixture. The second column contains a solvent-stripping section, D,
for solvent regeneration and to provide vapors for the extractive-distilla-
tion section, A. It has a heated still pot, E. 'If no heat losses were incur-
red this would be the sole source of heat in the unit. In laboratory equip-
ment, where heat losses can be a relatively large portion of the total heat,
it is usually necessary to heat accumulator C to ensure adequate vapor for
the extractive-distillation zone. Regenerated solvent from E is pumped
continuously to the extractive-distillf1tion column. For smootlmess of con-
trol, intermediate storage of solvent at G, ahead of the pump, is recom-
mended. A mechanical pump, L, is shown, but a thermal pump (total
vaporizer) with a condenser at point H would be equally satisfactory for
heat-stable solvents. There are several points of control in such a unit in
addition to the common ones of reflux ratio and boil-up rate in fractional
distil1ation. The liquid level in accumulator C is held constant by over-
flow J. Solvent withdrawal from E is controlled by a similar overflow, K.
A swinging-bucket type of column head, N, is used with automatic reflux
control (see Chapter II). The pump, L, can be set to deliver solvent at a
predetermined rate. The heat input at E and C can be set at a safe value,
based on a little operating experience, and readjusted gradually as the sys~
tem is depleted in volatile components during the extractive distillation.
The boil-up rates at C and particularly at E become the control points. To
he sme the solvent in still pot E is completely stripped of volatiles, the
temperature in E should be that of boiling solvent. This can be checked by
thermometer or thermocouple in a thermowell in still pot E. Stopcocks
S1, S2, Sa permit complete draining of the apparatus. The intermediate
solvent. storage vessel, G, is vented to the atmosphere at P, which may also
serve as a solvent charge line. A large flask or beaker open to the atmos-
phere makes a good solvent storage vessel. It will usually be necessary to
cool the overflow from E and to preheat the solvent at H for satisfactory
operation.
Suitable dimensions for the component parts of the apparatus shown in
Figure 9 are as follows:
Rectifying scction, B, 15 to 25 mm. a.D. X 25 cm. long.
Extractive distillation section, A, 15 to 25 mm. O.D. X 60 cm. long.
III. EXTRACTIVE AND AZEOTHOPIC DIS'l'ILLATION 345
Accumulator, C, 1 to 3 liters.
Overflow, J, 8 to 10 mIll. O.D. 1.1ll(1 10-20 cm. long with at least 3 cm. Lliffer-
ence in verticallcmgth of arms.
Vapor line, 0, 8 to 12 mm. OJ).
Stripping column, D, 25 to 35 mIll. O.D.
Still pot, E, 2 to 3 liters.
Overflow, K, 8 to 10 mm. O.D. and 10-2U em. long with at leaRt 3 (~Ill. differ-
ence in vertical length of arlllS.
Vent line, Q, 2-mm. capillary.
Solvent reservoir, G, 500 to 1000 ml.
Stopcocks, S1, 8 2, Sa, 2 mlll. oblique bore, lubricated with high-temperature
stopcock grease.
Solvent feed line, H, 8 to 10 mm. O.D. A small rotameter in this line to incli-
cate flow rates would be very desirable.
Pump, L, can be a small centrifugal pump such as those available from Eastern
Engineering Company provided with a bYP~RS, shown dotted, for flow con-
trol.
Other laboratory variable rate pumps which may be used, but which are
leRs desirable because of their pulsating flow, are the positive-displacement
pumps, such as those availahle from Hills-McCanna and Proportioneers,
Inc., or pumps such as those available from Wilson Pulsafeedcrs, Inc.
Equipment not shown in Figure g:
Heater for still pot E, 450--550 watts.
Heater for accumulator C, 450--550 watts.
Heater control: jor alternating current, auto transformerR such as Variac (Gen-
eral Radio Company) or Varitran (United Transformer Company); for
direct wrrcnt, slide-wire rheostats, 90 to 180 ohms.
Insulation, various altematives: (11,) electrioal windings, 10 watts pel' oenti~
meter length of column to he wound, maximum; (b) vacuum jackets;
(c) vapor bath in jackets.
Timer and solenoid for swinging-funnel reflux divider such as those available
from Eagle Signal Company, Scientific Glass Apparatus Company, Ace
Glass Company, General Electric Company, or Glass Engineering Labora-
tories. '
Internal construction of columns, various alternatives: (a) Packed columns.
Single-turn wire helices, Berl Radclles, Raschig rings, or similar small lab-
oratory packing. Nominal diameter of packing should be less than one~
fifth inside diameter of column. For example, in a 25-mm. J.D. column,
single-turn wire helices should bo less than 5 mm. (approximately 3/ 15
in.) in diameter, in this case 3/1dn. helices or smaller would be recom-
mended. (b) Perforated plates with overflows such as the Oldcrshaw
columns. so (c) Bubble-cap columns.. (d) Any suitable vapor-liquid
contacting arrangement used in fractioual-distillationlaboratory columns.
39 Oldershaw, Ind. Eng. Ghem., Anal. Ed., 13, 205 (1941). Manufaotured by Glass
Engineering Laboratories, Belmont, California.
34(} CARL S. CARLSON
The practicability of a batch laboratory apparatus with solvent recircu-

lation has been demonstrated. 40 The apparatus, shown schematically in
Figure 10, was designed for the specific purpose of determining the effec-
tiveness of extractive distillation in a packed laboratory column as com-
pared with fractional distillation. It can be used as shown for extractive
distillation or with minor modifications to
produce a solvent-free distillate. It contains
several uniquefeatures: (1) A thermal pump
to circulate solvent. (2) The solvent-stripping
column and the extractive-distillation column
as one vertical unit. (3) Solvent and hydro-
carbon reflux mix as vapors at the condenser.
H
The apparatus consists of the following main
parts:
J A vapor-liquid extractive-distillation contacting
section, B; a stripping column, E, for purifying
o high-boiling solvent; a means, G, H, for recycling
solvent to the top of the extractive-distillation
section; a condenser, A; a means, E, J, for in-
troducing hydrocarbon and solvent vapors to the
bottom of the extractive-distillation section; and
a reservoir or accumulator, D, for the mixture being
extractively distilled.
F Detailed dimensions of the apparatus shown
in Figure 10 are given in the original publica-
tion:lO Below the condenser is suspended [L
positive-displacement distillate receiver. The
/G condenser is of the cold-finger type but hollow
in the center to permit distillate withdrawal
through a siphon tube, 3 mm. O.D. The ex-
tractive distillation section, B, is packed with
Fig. 10. Batuh laboratory
extractive-distillation applll"ll- 3/16- 1n. single-turn wire helices made from No.
tus with solvent recycle. 26 B & S gage stainless-steel wire. The same
packing is used in the stripping column, E.
A vapor jacket surrounds the extractive-distillation section, E, for heat
control but electricall'esistance windings would provide greater flexibility.
The mixture to be extractively distilled, which for ease of reference will Le
called a hydrocarbon mixture, is retained in reservoir D and contains solvent
in the ratio being circulated. This ratio automatically adjusts itself. In the
center of the reservoir is a vertical 25-mm. LD. tube, open at the top, which
10 Dicks and Carlson, Trans. Am. Inst. Chem. Engl's., 41, 789 (1945),
III. EX'l'RAC'l'IVE AND AZEO'l'ltQPIC DISTILLATION 3"17
is really the top of the solvent-stripping column returning hydrocarbon

vapors with some solvent to the extractive-distillation zone. The liquid
level in the reservoir is always below the top of this tube. Hydrocarbon
and solvent are fed continuously to the solvent-stripping column through
the stopcock (K, Fig. 10). Complete stripping of hydrocarbon from the
solvent is achieved in stripping column E. This is wound with electrical
resistance wire to prevent heat loss. From the bottom of the reb oiler, F,
~30lvent flows continuously to the vaporizer, G, which serves as a thermal
pump and as a solvent reservoir. In working with this apparatus, aniline
was used HS the solvent and any decomposition products accumulated in
the vaporizer, G, so that only clean solvent was recirculated. Solvent
vapors rose through the electrically heated riser, II, and condensed with
hydrocarbon vapors on the condenser, A, thus ensuring excellent mixing of
solvent and hydrocarbon reflux.
The hydroearbon reservoir, D, was insulated both with asbestos and elec-
tl'icall'esistance windings. It was kept warm during operation but no boil-
ing occurred here. In the course of an extractive distillation, the volume of
liquid illreSCrYoir, D, would gradually decrease as hydrocarbon and solvent
was withdrawn as distillate. If a solvent-free distillate is withdrawn, the
volume of solvent in the l'eboiler, F, and vaporizer, G, will increase if' a
constant solvent: hydrocarbon ratio is maintained in accumulator D.
Should the solvent volume in F and G become too great, it can be withdrawn
through a drain L.'
Three obvious modifications of this apparatus for routine extractive dis-
tillations are: (a) Addition of a rectifying section, 40 cm. long, above the
extractive-distillation zone to give a solvent-free distillate. (b) Condensa-
tion of solvent prior to its addition to the extractive-distillation zone if
modification a is made. This is also a desirable change without modifica-
tion a in order to avoid dilution of hydrocarbon distillate with recycled
solvent. (c) An increase to 40 mm. of the diameter of the stripping col-
umn, which now limits the distill£Ltion rates.
When employing solvent recycle, the solvent must be completely
stripped of hydrocarbon or the desired separation will be defeated by intro-
ducing material of still-pot composition at the condenser.
C. CONTINUOUS EXTRACTIVE DlSTILLA'l'ION WITHOUT SOLVEN'r RECYCLE

I
A more elaborate device for laboratory extractive distillation is shown

in Figure 11, where solvent and feed are added continuously to the ex-
tractive distillation column, a distillate product, D, being withdrawn con-
tinuously at t.he top of the column. A still is provided at the bottom of the
column to furnish vapors to the extractive-distillation column. Less vola-
348 CAIlL S. OARIJSON
tile components and solvent which accumulate in the still are continuously
drawn off and sent to storage. Solvent and less volatile components are
recovered in a separate step.
The feed, F, is introduced at a point intermediate between solvent feed,
8, and the reboilel' to ensure complete stripping of more volatile material
from the solvent-hydrocarbon mixture withdrawn to storage. The solvent
is shown entering below the reflux return to give a solvent-free distillate, D,
[til overhead product.
o
5 5
T-I w
F
F
PUMP
Fig. 11. UoutillUfJUA F'ig. 12. Continuous extl'active-distilbtioll

extractive-distillation anl- appltmtus with solvent recycle.
limn without solvent re-
covery and recycle.
Griswold et al. 41 described a laboratory appnratus operating in a manner

similar to that shown in Fig'nre 11. It was designed to operate on hydro-
carbon fractions boiling in the C'e-Cs range with aniline as a solvent. The
column consisted of a I-in. standard iron pipe packed for 48 in. with lis-in.
metal helices. Feed was pumped directly to a hydrocarbon vaporizer, at
the bottom of the column, while solvent and hydrocarbon were withdrawn
as a liquid jllst above the feed vaporizer H,nd sent to RtOl'age. Solvent was
pumped to the top of the column, mixed with hydrocarbon reflux, and
heated to ensure complete miscibility. 1'he extractive-distillation column
U Griswold, Andres, Van Berg, and Kasch, Ind. Eng. Chr;m., 38, 65 (1946).
III. EXTRACTIVE AND AZEO'rROPIC DISTILLATION 349
was lagged with 80% magnesia and wound with resistance wire in fonr sec-
't,ions of 150 watts each. The hydrocarbon vaporizer and the solvent-refiux
preheator each containecla 600-watt cartridge heater.
D. CONTINUOUS EXTRACTIVlC DISTILLA'.rION ,VITH SOINENT RECYCLE
The arrangement of equipment for continuous extractive distillation with

solvent recyde is shown in Figure 12, on page 3·18. It consists of two
complete fractionating columns. The extractive distillation column, '['-I,
is a replica of that shown in Figure 11. Column T-2 is the solvent-stripping
column, which produces stripped solvent as a bottoms product and an over-
head product, W, which is the less volatile material in the feed, F. In
ordinary fractional distillation it would havc been the bottoms product
(assuming it is the high-boiling material in the absence of solvent).
To carry out this operation in the laboratory, the incorporation of various
arrangements such as those discussed for Figure 9 to decrease the number of
control points is essential. The schematic arrangement shown in Figure
12 requires eight points of control:
Reflux ratio for each column (2).
Still-pot boil-up rate for each column (2).
Liquid levels in P-l and P-2 (2).
Solvent rate.
Feed rate.
By mounting column T-l at a higher level than T-2 a gravity overflow sys-
tem from still pot P-l can be used for column T-l as well as for T-2. Reflux
ratios can be controlled automatically by the use of a swinging-funnel still
head. Heat input to P-l and P-2 needs little adjustment after some experi-
ence has been gained in operating such apparatus. Boil-up rate can be con-
trolled automatically if desired by means of it manometer measuring pres-
sure drop across the packing of the column, since pressure drop increases
·with dist.illation rate. This leaves two critical points of control, feed and
solvent rates. These can be cont.rolled manually or automat.ically by means
of flow control devices, depending upon the elaborateness of the laboratory
~etup. In any ease, a means of indicating flow rates of these streams to T-1
is essential. Such devices as l'otametcrs, orifices, or capillary constrictions
lLre suitable.
Griswold and Van Berg42 have described an apparatus similar to that
sketched in Figure 12. The extractive-distillation column was made from
2-in. standard iron pipe, and had 150 screen plates P/s in. apart. It was
wound with six sections of electric!.l resistance wire, 200 watt.s each, con-
42 Griswold and Van Berg, Ind. Eng. Chem., 38,170 (1946).
350 CARL S. CARLSON
trolled by individual Variacs. Tho hydrocarbon vaporizer and SOIVOllt-

reflux preheater were heated with steam.
The solvent-stripping column was also made from 2-in. standard pipe
but packed for 4 ft. with liB-in. carding teeth. It was wound with a 300-
watt heater element to render it adiabatic. A 2-liter capacity reb oiler was
provided with a 660-watt heater. Gear pumps were used to control the
flow rate of all the liquid streams. A liquid-level controller was used for
the bottom of the extractive-distillate column, while the level in the 1'e-
boiler of the stripping column was manually controlled.
E. CONTROL OF OPERATION IN EXTRAcnVE DISTILLATION
The major factors of control have already been discussed in connection

with specific apparatus. One point is so important that it will be repeated
for emphasis: the separation realized in extractive distillations results
from the presence of a high concentration of solvent in the liquid phase.
This must be provided. In those cases in which solvent is recirculated, it
must be completely stripped of feed components or the desired separation
will not be realized. Solvent preheating is also vel'Y important since it will
influence markedly the vapor load at the condenser. The solvent should
be preheated to the proper temperatUl'e and held at this temperature dur-
ing any distillation.
Since, in general, the solvent used boils 50 to lOOce. above the tempera-
ture of the mixture being separated, the most satisfactory way of following
the extractive distillation and solvent stripping is by temperature readings.
Whenever possible, thermometers or thermocouples should be installed in
the apparatus to permit temperature readings at the following points:
Overhead vapor.
Solvent just before entering column.
Just below solvent feed point.
At bottom of extractive-distillation section A (see Fig. 9).
In accumulator, C.
At top, middle, and bottom of stripping column, D.
In still pot, E.
The temperature at the bottom of the packed section of stripping column

D should be that of pure boiling solvent.
Heat losses should be minimized by adequate use of lagging and heating
elements in order to avoid excess condensation of vapor at various points
in the apparatus.
III. EXTRACTIVE: AND AZEOTROPIC DISTILI,ATION 351
5. Calculation of Extractive-Distillation Separations

Since it has been the practice to express vapor-liquid equilibria in the
pI'esence of n solvent, as a modified relative volatility C\'s, indicating the
separation realized in a one~plate still, these values can be used directly
fo!' extractive distillation calculations in the same manner that a has been
used. This is illustrated in the following mmmple:
(1) Dato.
Mixture to bc separated: ,n~heptane-methylcyclohexane.
Solvent: Aniline.
Charge: 50 mole per cent n-heptane in methylcyclohexane.
Overhead product: 90 mole per cent n-hoptane on a solvent-free basis.
Aniline, molecular weight 93, dill 1.02.
n-Heptalle, molecular weight 100, rl~1I 0.684, latent heat 76.1 cal./g.
Mothylcyclohexane, molecular weight 98, d~1I 0.769, latent heat 76.\) cal./g.
Column (see Fig. U) :
Section Ii, 2.5 CIll. LD. X 150 cm.
Section B, 2.5 cm. LD. X :35 cm.
Section D, 3.5 em. LD. X 50 em.
All sectiOllH packed with a/lo-in. sillgI()~turn helices.
H.E.T.P. 7.1 om,·10
Packing throughput, 2800 Ill!. liquid pel' hour. 40
(2) Solvent-h1frl1'Ocm'bon l'atio. From the equilibrium data for this system in the
presence of aniline 41 given in Table II it can be seen t,hat t,his Rystem. is difficult to
separate. In order to realize a relative volatility of 1.4 to 1.5 a solvent concentra-
tion between 80 and 90 mole pel' cent in the liquid is necessary. As an initial as-
sumption consider that this corresponds to a 3: 1 liquid volume ratio of aniline to
n~heptalle (initial overhead product):
Basis 100~ml. liquid mixture:

Aniline (75) (1.02)/98 = 0.822 mole (82.8 mole per cent).
n-Helltanc (25) (O,(}84) /100 = 0.171 mole (17.2 mole per cent).
Thus a 3: 1 solvent: hydrocarbon liquid volume ratio corresponds to approximately

83 mole pel' cent ~1nUine in the liquid phase. This should give a relative volatility
of 1.44.
(3) Reflux ratio at condenser. The minimum reflux ratio for a binary mixture can
he readily calculated using the Fenske cquation:13
(17)
13 Fenske, Ind. Eng. Cham., 24, 482 (HI32).

352 CAHL S. CARLSON
where HIllin. is minimum ratio of reflux to (listillute; '" is 1.44 for n-heptane to
lUethylcyclohexane in the presenee of a volumes aniline pel' volume hydrocarbon;
:rDl! is mole fraction ll-heptane in distillate, 0.90; XDm if'. mole fr:1dion methylcyclo-
hexane in distillate, 0.10; and x" is mole fraction at pineh. in this cuse considered
equal to hydrocarboll compm:ition in still pot, C, which will change all the mixture
b depleted in n-heptane. At the start of the distillatioll:
it· =
"'<n.
-
1.44 - 1
-~-.-~~ [9..:..~
0.5
- 1M (~~)J
D,D
= a.4
During the extraetive di~tilbtion the minimulll rDflux requirements inlJreliHe as

follows, expressing composition on a solvent-froe basis:
Mole fraetion
,,-hoptane in Bti II Rutin. Ractuul
0.50 3.4 5.1

0.'10 4.1 6.1
0.30 5.7 8. f$
0.20 8.8 13.2
0.15 13.2 20
0.10 20 30
A useful rule of thumb is to operate at 1.5 times the minimum reflux ratio for the
desired separation. During a batch distillation the reflux ratio c~m be incrcased
gradually but it is uSll:111y easier to operate at one setting. At 20: 1 reflux ratio,
the n-heptune in the still can be reduced to a mole fraotion of 0.15 (solvent-frec
basis).
This reflux ratio is more than adequate for virtually complete removal of solvent
aniline from the overhead distillate since the relative volatility of hydl'Oo!1l'boIis
to aniline is approximately 8.5.
(4) Theoretical plates in extractive distillation section. The final separation de-
sired corresponds to an enrichment ratio H of 51. With a relative volatility of 1.44
a minimum of 10 theoretical plates will be necessary. With an H.E.T.P. (height
equivalent to a theoretical plate; see definition, Chapter I) of 7.1 cm. for this pack-
ing, a packed length of at least 71 em. will be required. The 150 CIllo of packed
length actually provided should be adequate. If this is not the case, the hydro-
carbon reflux ratiD ancl the solvent: hydrocarbon ratio nee(l only be increased
slightly. It should be Iloted that the reflux ratio and solvent rate are related if a
constant solvent concentration in the liquid is to be maintained. Another useful
rule of thumb is that at 1.5 times the minimum reflux ratio 1.5 times the minimum
number of theoretical plates necessary should give the desired separation. This
holds only over the common operating ranges, 5 mole pel' ccnt Illore volatile COI11-
H The ratio of conpentmtions ill the distillate divided by the ratio of coneentrtttions
in the still, E.R. = (yt!Y2)D(XI!X2).. See Fenske ill Science of Petroleum, Vo). II, Oxford
Dniv. Press, New York, 1\)38,11. 1631.
III. EX'l'RACTIVE AND AZEO'l'ROPIC DIS'~'ILLA'rION 353
ponent in the still to 95 mole per cent more volatile cmllponcnt in the distillate, and
for mixtures with regular X,y diagrams showing no pinches.
It should be noted that, in the extractive-distillation section, the algebraic cal-
culations for theoretical plates are made using relative volatilities in the presence
of solvent. The same calculations could be made graphically nsing a McCabe
and Thiele diagram drawn on a solvent-free basis.
(5) Theoretical plates in enriching section B and stripping section D. Taking ex as

8.5 for hydrocarbom; with respect to aniline, the minimum number of theoretical
plates will be less than 3; thus the 35-cm. enriching section, B, should be adequate.
For complete removal of hydrocarbon from solvent, a minimum of approximately
3 theoretical plates will be necessary. At 1.5 times this or 4.5 theoretical plates, a
packed length of 32 cm. will be needed ill the stripping column, D. The 50 em. pro-
vided should be more than adequate.
(6') Yapor rate in ea;tractive-distillation column. The column, 2.5 em. LD., packed
with 3/ IB-in. single-turn helices, will have a distillation rate without solvent of at
least 2800 mI. liquid per hour. The addition of solvent will materblly (iecrease
this. In the absence of ennctete data it will be assumed the distillation rate will be
decreased 50% to 1400 ml. liquid per hour.
(7) Solvent rate. Since 1400 ml. hydrocarbon 18 liquefied per hour at the con-
denser and a reflux ratio of 20:1 win be used:
(1400)(1/21) = 67 ml. hydrocarbon distillate per hour

1400 - 67 = 1333 mI. hydroCl1rbon reflux per hour
At a 3:1 volume ratio of solvent to hydroearboll in the reflux, 3 X. 1333, or 3999,

ml. solvent per hour must be circulated. The hydrocarbon vapors leaving the ex-
tractive distillation section will contain 10% aniline. 45 Since this will be refluxed
back, an equivalent reduction in solvent requirements of 133 ml. can be made if
desired. Thus solvent circulation rate of 3999 - 133, or 3866, ml./hr. may be used,
(8) Heat requirements. Assuming no heltt losses, it will be necessary to vaporize

only 1400 1111. per hOllr of hydrocarbon:
n-Heptane: (1400)(0.68'1)(76.1) = 73,000 cal. per hour

Methylcyclohexane: (HOO)(0.769)(76.9) = 82,500 cal. per hour
(73,000)/860 = 85 watts
This will require a minimum of 85 watts input. Experience has shown that at least
five times this amount should be provided if an external heater is used.
During the course of the distillation under discussion, the hydrocarbon mixture
will gradually be impoverished in n-heptl1Uc as it is removed as distillate.
4. Carlson, Ph.D. Thesis, Pennsylvania State Collcgc, 1939.

354 CARL S. CARLSON
(.9) Summary of example.

Equipment requirements:
At finite Available
Theoretical plateR l\'Iinimnm refiux: in column
Section Ji 10 15 21.
Section B <3 <4.5 5
Section D 3 4.5 7.0
Reflux ratio at cOJlclensel':

mininulm ............................................... . 13.2
at operating conditions ................................... . 20
Solvent: hydrocarbon volume ratio ............................. . 3:1
Distillate rate, Illl. per hour ................................... . 67
Solvent rate, Illl. per holll' .................................... . 3866
Solvent concentration in liquid, mole per cent ................... . 83
Relative volatility, n-heptane to methylcyclohexane .............. . 1.44
. Relative volatility, hydrocarbon to aniline ...................... . 8.5
Heat input to still pot:
Il1ll1lmUIl1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 85 watts
recommended ............................................ 425 watts
6. Equipment Performance
A. PACKED LABORA'l'OHY COLUMNS
Few data are available on the effect of a high-boiling solvent on the

H.E.T.P. of a column packing or on the efficiency of industrial plate col-
umns. Detailed data on the operation of a packed column during extrac-
tive distillation of n-heptane and methylcyclohexane with aniline as a
solvent have been reported. 46 Similar data were reported for operation in
the absence of a solvent. From this information and the vapor-liquid
equilibria data of Griswold et al. U it htl,s been possible to calculate packing
efficiencies; they averaged 88%. A correlation of plate efficiencies as a
function of liquid viscosity, based on commercial towers,47 would have pre-
dicted about 70% effieiency.
B. COMMERCIAL BUBBLE-PLATE TOWERS
Test data on a commercial toluene extractive-distilla.tion unit using

phenol as a solvent have been reported by Drickamer and Hummel. 48
An over-all avel'age plate efficiency of 50% was found above the solvent-
feed tray when 5 mole per cent phenol was present. An average plate ef-
46 Dicks aud Carlson, Trans. Am. Inst. Chern. Enor.~., 41, 789 (1945).
47 Drickamol' and Bradford, Trans. Am. Inst. Chern. Engrs., 39, 319 (1943).
48 Drickamer !lnd Hummel, Trans. Am. Inst. Chern. Engrs., 41, 607 (1945).
III. l'JX'l'HAC'rIVE AND AZE01'ROPIC DIS'l'U,L,\'l'ION 355
ficiency of 56% was fOllnd helow the solvent-feed t.my when 55 mole 1)er
eent phenol was present. If the solvent had allY effect, it was the reverse
of what one would expect, since the viscosity-plate effipiency correlation'17
predicts 60 and 40% for these two tower sections, respectively.
Test datu. have also been reported for several towers of another com-
mercial extractive-distillation unit 4n using furfural containing 4-6 % water
(by weight) as a mixed solvent for the separation of butanes, butenes, and
butadiene. In the n-but~ne tower, n-butane is separated from butene-2.
In the isobutane tower, isobutane is removed from butene-1 and butadiene.
In the butadiene tower, separation is made between butene-1 ane! butltdiene.
In each tower, the first-named component is rejected as overhead product
using aqueous furfural as an extractive-distillation solvent. The plate ef-
ficiencies .of the individual towers, based on liquid-tray analyses were:
Murphree plato
Tower officiency. %
n-Butane .. , . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
Isobutalle. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 28
Butadiene .................... , . . . . . . . . .. 20
'Based on additional data not given, an average tray efficiency of 25% is rec-
ommended for these towers. The tray efficiency predicted by the cor-
relation of Drickamer and Bradford 47 is also approximately this figure.
In the absence of data other than those presented above, one is forced to
conclude that the effect of the solvent on the efficiency of a packing in a
laboratory column is slight. In commercial equipment the effect of the
solvent is either slight or in the direction of the viscosity-plate efficiency
correlation,47 where decreasing pIa to efficiency is predicted for increasing
viscosity of the liquid on the plate. Additional data on this score would be
desirable.
7. Typical Industrial Application of Extractive Distillation

'l'oluene from Peiroleum, Shell Process, The recovery of toluene from a
complex petroleum fraction by extractive distillation is one of the striking
applications of this process. A number of plants were producing nitration-
grade or aviation-grade toluene in this manner at the end of World War II.
Such a plant operated by Pan American Refining Corporation at Texas
City, Texas, has been described. 48 A diagram of this plant is shown in Fig-
ure 13. It consists of an extractive-distillation or toluene-recovery towel'
and a solvent-stripping tower. Not shown are the feed-preparation towers,
"Happel, Cornell, Eastman, Fowle, Porter, and Schutte, Trans. Am. Inst. Chem.
Engrs., 42,189 (946).
35G CARL S. CARLSON
which feed the narrow hailing-range fraction eOlltinuously to the toluene-

recovery tower. The toluene-recovery tower is 7 ft. in diameter with 65
trays, while the stripper is 4 ft. in diameter with 30 trays. Toluene of 99.4
mole per cent purity was produced from a. feed containing 63.4 mole per
eent, with phenol as the solvent.
PHENOL
CHARGE
ffi NON-AROMATICS
w3: TO FIELD
z~
w)-
:lit:
..J
0>
w
1-0
u
w
It:
Fig. 13. Shell toluene-reeovery proeess. 18
III. AZEOTROPIC DISTILLATION

In azeotropic distillation a solvent, is selected which forms a constant-
boiling mixture, or azeotrope, with one or more of the components in the
feed. The solvent is generally added with the feed to the column. Either
minimum-boiling: or maximum-boiling azeotropes may be obtained, but
Lecat in his first compilation of several thousand azeotropes found lesH
than 10% formed maximum cOIlHtant-boiling mixtureH. 6o For this reasoIi,
discussion in the following pages will be in terms of minimum constant-
boiling mixtures which are obtained as distillate.
~o Lecat, La ten.~ion dB. vapeur des melanges de liquides. L'Azeotropisme. Lamertin,
Brussels, 1918.
III. EXTRACTIVE AND AZEO'l'ROPIC DISTILT,ATION 357
1. Scope
The application of azeotropic disWlntion, like that of extractive distilla-
tion, is usually limited to close-boiling mixtures not readily separable by
conventional effieient fl'aetional distillation. A concentrate prepared by
efficient fractional distillation often can be further separated by an azeo-
tropic distillation in the presence of a component whieh forms a minimum
or maximum constant-boiling mixture with one of the materials to be sepa-
rated. In general, the mixture to be separated will boil over a temperature
range of less than 20 ac. and may even itself be an azeotrope.
2. Theory
Azeotropes exist in solution because of molecular dissimilarity of the
components of the mixture. In binary mixtures exhibiting minimum-
boiling azeotropes, a lack of compatibility is evident in view of the large
VAPOR PRESSURE OF A + B
W
II: VAPOR PRESSURE OF A
::0
VI
VI
W
II:
Q_
VAPOR PRESSURE OF B
o 1.0
MOLE FRACTION OF A
Fig. 14. Vapor pressure - composition relationships for
completely immiscible binary mixture.
positive deviation from Raonlt's law. The ultimate in incompatibility is

shown by completely immiscible liquids. When present ill the same vessel
and boiled, each exhibits its own vapor pressure und a vapor of constant
composition is obtained. The distillate from a system such as this has been
defined as a heterogeneous azeotrope. Figme 14: shows the vapor-pressure
relationships in a bin-ary mixture of this type. The vapor pressures of
pure A and pure B are shown as dotted horizontal lines. The vapor pres-
sure above the solution, as long as two liquid phases are present, is the sum
358 CARL S. CARLSON
of the two vapor prC<isnres as shown by the solid horizontal line. The
composition ratio of the vapor equals the ratio of the vapor pressures. The
boiling temperature of this system at any pressure is eOllstant. This can be
seen from the phase rule:
p+V=C+2
where P = 3 phases (2 liquid, 1 vapor), C = 2 components, A and B, and
V = 1 degree of freedom, which in this ease is pressure, fixed at one at-
mosphere. Hydrocarbons and water are good illustrations of mutually
incompatible components. Azeotropic distillation of this type is fre-
quently used both in the laboratory and in industry, and has been given
the special name of "steam distillation."
An intermediate case of incompatibility is the system n-butanol-water.
Two liquid phases exist at the boiling point when the over-all composition
of the liquid is between 65 and 97 mole per cent water. Under these con-
ditions one layer is rich in n-butanol while the other layer is rich in water.
The vapors evolved when two phases are present are of a constant composi-
tion.
In contrast with the lack of mutual compatibility which results in mini-
mum-boiling azeotropes is the case of high compatibility, which results in
maximum-boiling azeotropes. One of the best known examples is the
system hydrogen chloride-water. Hydrogen chloride, normally a gas, is
very soluble in water and forms a maximum constant-boiling mixture con-
taining 11.1 mole per cent hydrogen chloride (20.2 weight per cent) boiling
at 110 e C .
. An azeotrope is not necessarily a definite chemical compound although it
is frequently treated as one. This was demollstrated in 1861 by Roscoe,61
who showed that the composition of the azeotrope varied with the pressure
at which the distillation was carried out, and was further verified by the
work of Merriman52 on the system ethanol-wl1ter, data from which are re-
produced in Table X. It will be noted that no azeotrope exists at distilla-
tion pressures below 70 mm. of mercury. The manner in which azeotropic
composition is infl.uenced by pressure will vary depending upon the slopes
of the vapor-pressure eurves of the individual components of the mixture.
Britton and co-workers53 have found that, in methanol-ketone systems
which normally form azeotropes, the azeotrope can be eliminated by in-
creasing the pressure on the system.
The existence of polycomponent systems showing anomalous distillation
behavior is mentioned only briefly. Their occurrence, however, is by no
51 Roscoe, Trans. Chem. Soc., 13, 146 (1861); 15,270 (1862).
n Merriman, J. Chem. Soc., 103,628 (1913).
II Britton, Nutting, and Horsley, Anal. Chern., 19,601 (1947).
111. EXTRACTIVE AND AZEOTROPIC DISTILLATION 359
TABLE X
COMPOSI'l'ION OF E'rtIANOL-\VATillR AZillOTROPillS A'l' Dn'FEREN1' PItillSSURES
Boiling point
Pressure, '\\' a ter in for
n.bs., nzeotl'ope. azeotroIljr
nun, Hg::_,_ _ _l_lll_,l_e:_pe_r_ce_n_t_ _ mixture/tiC.
760.0 4.4 78.15

404.6 3.75 63.04
198.4 2.7 47.63
129.7 1.3 39.20
94.9 0.5 33.35
70.0 0.0
means unusual. The investigation of Huch systems is time consuming and,

in general, undertaken ollly when essential to the solution of n, specific prob-
lem. For this reason, little published information is available. Not only
are ternary azeotropes quite common but, even in the absence of ternary
azeotropes, the behavior of complex systems containing binary azeotropes
may be quito unpredictable. It is not within the scope of this chapter to
include any extended discUf-iSion of such systems, but their existence should
not be overlooked.
3. Selection of an Entrainer
A. CHEMICAL CONsIDEHA'rIONS
Ewell et al. 54 have published an interesting approach to the problem of

predicting azeotrope formation based on the concept of hydrogen bonds.
The deviation from ideality which results in azeotropes is attributed to
hydrogen bonds or internal pressure, the former being the more important.
The hydrogen-bond concept is stated54 as follows: "Hydrogen can coordi-
nate between two molecules of oxygen, nitrogen, or fluorine. It can also co-
ordinate between one of these donor atoms (oxygen, nitrogen, fluorine) and
a carbon atom, provided a sufficient number of negative atoms or groups
are attached to the carbon atom. Hydrogen cannot coordinate between
two carbon atoms." Depending upon the atoms between which hydrogen
is co-ordinating, the hydrogen bonds have been classified as follows:
Strong Weak
N _" lIN
\
0 ...... lIO HCCl~
N ->- lIO o _" HCCI-CCI
o _.. lIN N ->- HCN0 2
HCCN
54 Ewell, Harrison, and Berg, Ind. Eng. Chern., 36, 871 (1944).
360 CARL S. CABLSON
Using this classificatioll, organic liquids have been arranged into the five
groups given below:64
"Class I-Liquids capable of forming three dimensional net-works of strong hy-
drogen bonds.
"Class II-Other liquids composed of molecules containing both active hydrogen
atoms and clonor atoms (oxygen, nitrogen, fluorine).
"Class III-Liquids composed of molecules containing donor atoms but no ac-
tive hydrogen atoms.
"Class IV-Liquids composed of molecules containing active hydrogen atoms
but no donor atoms. These are molecules having two or three chlorine atoms Oil
the same carbon as a hydrogen atom, or one chlorine on the same carbon atom and
one or more chlorine atoms on adj!Lcent carbon [Ltoms.
"Class V-All other liquids, i.e., liquids having no hydrogen bond forming capa-
bilities."
Typical examples of liquids falling into the above classification are given
in Table XI.
rrABLE XI
EXAMPLES OE' LIQUIDS OF VAnIOUS CLASSIFICA'l'IONS
CIIlilB I Class II Class III ClasB IV CI"." V

Water Alcohols Ethers Hydrocarbons
Glycol Acids Ketones Carbon disulfide
Glycerol l)hcnols Aldehydes Mercaptuns
Amine alco- Primary Esters Iodine
hols amines Tertiary Phosphorus
Hydroxyl- Secondary amincs Sulfur
amine amines Nitro com- Halohyclrocarbons
Hydroxy Oximes pounds 110 not in Class IV
acids Hydrazine '" Hntoms
Polyphenols Hydrogen Nitl'iles with
Amides fluoride no CI II
Hydrogen atoms
cyanide
Nitro com-
pounds
with a! H
atoms
Nitriles with
IX Hatorns
From the above five classifications Ewell et al. 64 have predicted deviations
from RaouIt's law for mixtures of members of the various classes. For
positive deviations the vapor pressure above the solution is greater than
predicted from Rnoult's hw, and if it reaches a maximum a minimum-
boiling azeotrope exists (see Fig. 22 and 24). This information is sum-
marized in Table XII. In the last grouping of classes in this table, the
quasi-ideal systems may form azeotropes if the boiling points of the eom-
III. EX'l'RACTIVE AND AZEOTROPIC DISI'ILLATION 361
ponents are very close to each other. Maximum. azeotropes have been
classified as follows ;54
"1. '"Vater with strong acids-e.g., water with HCI, HBr, HI, lINDa.
"2. Water with certain aSRociated liquids-e.g., water with formic acid, hydrazine,
ethylenediamine.
+
"a. Donor liquids (Class III) with nOIlas~ociated liquids, e.g., acetolle chloro-
form; cyclohexanone +
bromoform; butyl acetate + 1,2,2-trichloropropane.
"4. Organic acids with amilles-e.g., acetic acid +triethylamine; propionic
aeid + pyridine.
"5. Phenols with l1111ines--e.g., phen()l + aniline, o-Cl'elS()l +
dirnethylnnilinc.
"6. Organic acids with donor liquids containing. oxygen-e.g., formic acid +
+
cliethyl ketone; butyric acid cyc:lohexanone.
"7. l)henols with donor liquids containing oxygen-e.g., phenol + Illethyl-
hexylketone, o-cre~ul + ethyloxalate.
"8. Phenols with alcohols-e.g., phenol +
n-octanol; o-cresol +glycol."
w
~ 18~-----------------------'
0:
I-
[ill- 16
N;Z
<lw
LLZ 14
00
I-Cl.
~~ 12
a: u
~~
::];;1
5:5
CD,
21-
ttli3
;;=0:
~L:5
Z
IIl
WO
.~~
W
II::
UJ
~ 60
Ci BOILING POINT DIFFERENCE OF
HYDROCARBON AND ALCOHOL I ·C.
Fig. 15. Rehltion between boiling-point differences of components
I)f binary mixture and the azeotrope based 011 paraffin hydrocarbons
with methanol and ethanol.
Recently a number of excellent correlations of azeotropic data have been·

published, based largely on existing azeotropic data. 56 These correlations
should help in the selection of an entrainer.
65 Horsley et al., Anal. Chern., 19, 508 (1947). Meissner and Greenfeld, Ind. Eng.
Chern., 40, 438 (1948). Skolnik, ibid., 40, 442 (1948).
362 CA.RL S. CARLSON
~
+
H
III. EXTHACTIVE AND AZEOTROI'IC DIf;'l'n~LA'l'ION 303
For ease of discussion, the separation of n binary mixture will be con-

sidered although the principles involved apply equally to multicomponent
mixtures. An important factor in the selection of an auxilit1ry agent is
that it must have a boiling point relatively close to that of the material to
be separated, if a reasonable difference in boiling point between the azeo-
trope and the other constituents of the mixture is to be realized. (There
are notable exceptions to this rule, such as hydrogen chloride, b.p. -8:3.7°C.,
and water, which form an azeotrope boiling at 110°C.) This is illus-
trated in Figure 15 by data for binary azeotropes of paraffin hydroc:arbons
with methanol and ethanol. The diffcrcnc:e between the boiling points of
the hydrocarbon and alcohol is given on the abscissa while the difference
between the boiling points of the minimum boiling azeotrope and the near-
est-boiling constituent of the mixture is shown as the ordinate. It will
be noted that a maximum boiling-point depression of 16°C. is obtained
when there is no difference between the boiling points of alcohol and hydro-
carbon. The depression decreases rapidly as the difference between the
boiling points of alcohol and hydrocarbon increaseR.
As a rule auxiliary agents are used which boil within 0 to 30°C. of the
material to be separated. For paraffin hydrocarbons, with metl1anol or
ethanol boiling 30 0 e. from the hydrocl1rbon, a 5 ° lowering of boiling point
for the azeotrope should be realized. Different mixtures vary in the boil-
ing-point depression realized with various entrainers. Curves similar to
the one in Figure 15 hELVe heen published for a number of systems. 56
B. SELECTIVE AND NONSELEC'l'IVE ENTRAINING AGENTS
In the separatioll of a binary mixture an auxiliary agent is desirable

which forms an azeotrope with only one of the constituents. Such an
auxiliary agent is called a "selective" azeotrope former or entrainer. En-
trainers of this type are commonly used in processes in which heterogeneous
azeotropes are formed, e.g., in the dehydration of aqueous acetic acid with
diethyl ether 57 as the entrainer. Water and entrainer {<steam distil" as
overhead product, forming two phases on condensing, and the entrainer is
recycled as reflux.
In homogeneous azeotropes, nonselective cntrainers are more common.
For example, methanol and ethanol form azeotropes with successive mem-
bers of the homologous series of paraffin hydrocarbons, illustrated in Figure
15. They act in a similar manner with oletins, naphthenesJ and aromatic
hydrocarbons. The difference in boiling point of the azeotrope and the
hydrocarbons of the same boiling point depends on the chemical type of
56 HOl'sley, AnaL Chem., 19, B03 (1947).

57 Othmer, Tmns. Am. In.qt. Chem. Engrs., 30, 299 (1933).
364 CARL S. CARLSON
the latter. Mair et al. 5B used this fact to isolate pure hydrocarbons from a
straight-run (un cracked) petroleum naphtha, as illustrated by Figure Hi,
taken from their work. Both vapor and liquid composition curves are
shown with the constant-boiling mixture at 50 mole per ceni; entrainer.
A petroleum fradion boiling at 110.0 to 110.5°e. fwd containing paraffins
and naphthenes was mixed with toluene, which boils at 110.6. Using
w
0:
::l
~
0:
w
a.
::E
w
... p
o 0.2 0.4 0.6 0.8 1.0

MOLE FRACTION AZEOTROPE
FOAMING AGENT
Fig. 16. Relation between boiling points of 8z8otropes of various
types of hydrocarbons having the same boiling point in the pure
staGe, with the same entrainer.68 A, aromatic hydrocarbon; N,
naphthene hydrocarboll; P, paraffi11 hydrocarbon. Pressure, 1
a.tmospherll.
acetonitrile as an entrainer, an almost quantitative separation of toluene

from naphthenes and paraffins was obtained, and likewise a considerable,
though not quantitative, separation between the latter.
C. DESIRABLE CHARACTERISTICS OF AZEOTROPE FORMERS
The difference in boiling point caused by the addition of the entrainer

should not be the sole criterion for its selection since the separation is fre-
quently much more complete than can be expected solely from that differ-
ence. This means that the spread between the compositions of vapor and
liquid is much greater than that of an ideal solution with the Same differ-
en!)es in the boiling points of the components. Since we are dealing with
highly nonideal systems this is not surprising.
IS Mair, Glasgow, Jr., and Rossini, J. Research Natl. Bur. Standards, 27, 39 (1941).
III. EXTHACTIVE AND AZ.EO'l'ROl'W DlS'l'ILI,_"'l'lON 3()5
For minimum-boiling azeotropes the ratio of desired component and

j
entrainer in the azeotrope is important. In general, particularly in indus-

...
Q_
a
a::
I-
a
LLI
N
<;(
~
z
a
OJ
a:
<;(
u
a
a::
Q
>-
:r; 40
I-
Z
W
U
...a:
Q_
20
HYDROCARBON
I-
:r;
r<---- LOW-BOILING ----1---- HYDROCARBON
HIGH-BOILING - - - - . . . l
I,!)
jjJ 0
;s 50 40 30 20 10 0 10 20 30 40 50
BOILING-POINT DIFFERENCE aETWEEN HYDROCARBON AND ALCOHOL, ·C.
Fig. 17. Effect, of Loiling-point diff(}l'ence Lntween ethanol anti hydrocarboll 0l1' the
concentration of hydrocarbon in the Ltzeotl'ope.
~ 60
w
N
Z
W
OJ
I- 40
z
w
u
a:
i'l: 20
I-
:z: ____, _ _ BENZENE BENZENE
I,!)
LOW-BOILING - HIGH-BOILING
jjJ
~ ~2~O~-----------1~0-------------O~------------I~O--~--------~20
BOILING-POINT DIFFERENCE BETWEEN BENZENE AND NON -AROMATIC HYDROCARBON, ·C.

Fig. 1 R. Bff€'ei; of Loning-poiut diffel'!Jllee b~tweell bl~IlZ(me and lIoll'Ll'OmatiCl hydro-
carhun Oil the cmwlllltmtioIl of beIlzene ill the ItZHotl'Ope •
.
trial practiee, a high value of this ratio is desired. This ean be achieved
by selecting an entl'ainer which boils higher than the desired component.
366 CARL S. CARLSON
In hbomtory azeotropic sepamtions in batch fractional distillation an-

other consideration becomes important, namely, the concentration and the
amount of the desired component in the distillnnd and the column holdup.
If the amount of the component is small relative to the column holdup,
an en trainer which forms a constant-boiling mixture containing a large
proportion of entminer should be selected. This requires an entminer
which boils lower than the component to be isolated. The effect of the
boiling-point difference between ethanol and hydrocarbons on the concen-
tration of the azeotrope is shown in Figure 17. If the hydrocarbon is the
low-boiling material it forms the major part of the azeotrope. The propor-
tion decreases as the boiling point of the hydrocarbon approaehes the boil-
ing point of the alcohol and if both have the Rame boiling point the azeo-
trope contains 70 weight pel' eent hydrocarbon (55 mole pel' cent of the
hydrocarbon). If the alcohol is the low-boiling material it forms the major
part of the azeotrope. Figure 18 showfl a similar curve for the benzene-
hydrocarbon azeotropes. It will be noted that these exist over a much
narrower temperature range.
The ease of removal of the entminer from the azeotrope is very important
since azeotropic distillation resolves a difficultly separable solution by the
formation of a constant-boiling mixture which in turn cannot be separated
by distillation. The azeotrope can be separn,ted in a number of wayfl,
hO\vever, such as:
(a) Phase separation Oil cooling.
(b) Phase separation by "salting out."
(c) Chemical removal of entrainer.
Cd) Solvent extraction or water washing.
(e) Second azeotl'opic distillation.
The most desirable case is that in which the azeotrope, on cooling, sepa-
rates into two layers, one rich in entrainer, which is refiuxed to the fractional-
distillation column, and the other rich in desired component, which is with-
drawn. The latter is then subjected to a second fractionaJ distillation to
remove residual entrainer as the azeotrope. For a minimum azeotrope this
operation will leave pure product in the still. In industrial practice the
phase separation mentioned above is essential for a successful applicatioll of
azeotropic distillation. Examples of such processes are the production of
anhydrous ethanol using benzene, trichloroethylene, or cyelohexane as the
entrainer,59 and the dehydration of acetic acid with ethers or esters.eo
69 Guinot and Clark, Trans. Inst. Chem. Engl'.~. London, 16, 189 (1933). Keyes, Ind.
Eng. Chem., 21, 998 (1921). Colburn and Phillips, Trans. Am. Inst. Chem. En(lr.~., 40,
333 (1944).
80 Othmer, Trans. Am. [nM. Chem. JiJng?'s., 30, 29\l (H133).
III. EXTRACTIVE AND AZEOTROPIC DlR'l'ILkWION"
Frequently, the addition of another component causes separation of the

el1trainer. For example, the binary azeotrope methyl ethyl ketone and
water, which shows no natural separation into two layers, separates into a
water-rich and a water-poor layer upon addition of a small amount of a
suitable light hydrocarbon. In a similar manner the addition of salts
such as sodium or calcium chloride will frequently cause phase separation
by "salting out."
Phase separation can sometimes be induced by the addition of an excess
of one of the constituents of the azeotrope. For example, the ternary
azeotrope ethyl acetate-ethanol-water is 11, homogeneous liquid at room
temperature. The addition of excess water will give two layers, one essen-
tially water-ethyl acetate and the other ethanol-water saturated with ethyl
acetate. The water-ethyl acetate byer at room temperature contains
half as mnch water as the binary azeot.rope; hence on distillation water-
free ethyl acetate will be obtained as residue, with water being removed
as the low-boiling binary azeotrope.
In laboratory work when recovery for reuse of the entrainer is not essen-
tial, its chemical removal is often possible. For example, ill the uzeotropic
fractional distillation of hydrocarbons with organic acids, the latter can be
removed by washing with alkali, f\,nd amines such as aniline ('.!"Lll be removed
'with mineral acids. If water is a component of the azeotrope it can be
removed by suitable drying agents. In fact the lime distillation process
used in Germany prior to the development of the E:eyes process for the
manufacture of anhydrous ethanol was an application of this principle for
breaking the water-alcohol azeotrope.
In some cases liquid-liquid extraction will be found more suitable for
the removal of the entrainer. In t.he commercial isolation of toluene from
petroleum by azeotropic distillation with the methyl ethyl ketone-water
azeotrope as entrainer for the nonaromatic constituents, t.he entrainer is
recovered by liquid-liquid extntction with water.51 The water phase is
then fractionated to obtain the methyl ethyl ketone-water I1zeotrope. In
ot.her cases liquid-liquid ext.raction of the azeotrope with diethyl ether may
remove the ent.rainer. In the laboratory the removal of entrainer from
hydrocarbons by extraction with \vater can be carried out in separatory
funnels. The heavier water layer is discarded and the extraction repeated
if necessary.
Anot.her method for separating the azeotrope is to subject it to a second
azeotropic distillation with a different entrainer forming an azeotrope
from which it can be recovered more easily.
In addition to forming an azeotrope from which the desired compound is
61 Lake, Trans. Am. Inst. Chem. Engrs., 41, 327 (1945). Hartley, Petrolettm Refiner,
24,519 (1945).
368 CARL S. CARLSON
readily recovered an eut-rainer should preferably have several other desir-

able characteristics. In order to be easily handled it should be noncorro-
sive, nontoxic, and stable to light and heat. It should be cheap and readily
available.
As mentioned above it is desirable, but not essential in laboratory work,
that phase separation of the azeotrope occur on cooling. For hydrocarbon
separations solubility of the entrainer in wat.er is desirable.
For industrial use low latent heat and high ratio of product to elltrainer
are important. The latter point has been discussed above from a labora-
tory viewpoint. The thermal efficiency of an entrainer affects directly
the operating costs. An entrainer of higher thermal efficiency carries over-
head a greater amount of product for a given heat input to the still. Ac-
cording to Othmer,60 in the dehydration of acetic acid with isopropyl ether,
1425 calories of heat are required to vaporize the ether in order to remove
one gram of water, while the similar process with ethylene chloride requires
only 920 calories. These heat quantities are in addition to the 540 calories
needed in each case to vaporize one gram of water.
D. SUMMARY
The discussion on page 359 and a study of the available tables of azeo-
tropes will indicate the chemical types which form azeotropes wit.h the
compounds to be separated. As a guide for furthp,r Heleet.ion the following
general rules are summarized:
(a) Select an entrailler boiling as close to t.he hoiling point of the compound to 1)(>
separated as possible (see Fig. 15).
(b) If there is a choice, use a selectivp entrailler in preference to a nonselective
entrainer.
(c) Choose a chemically stable and heat-stahle entrainer. If any doubt exists,
select an entrainer whose decomposition products will not contaminate the desired
.prodllct.
(d) Other factors being equal, the entrainer with the greatest ease of recovery
Rhoulrl be selected. Possible recovery methods are given above.
E. EXPERIMENTAL VERIFICATION OF EFFECTIVENESS OF AN ENTRAINER
Experimental verification of the effectiveness of an entrainer for azeo-

tropic distillation can be obtained by means of an equilibrium still such as
that described on page 327. It is necessary t.o determine the vapor-liquid
equilibrium relationships for each component with the entrainer to estab-
lish whether an azeotrope is formed with either or both components. The
difference between the boiling points of the azeotrope and the other com-
ponents present is a good indication of the ease of separation of the ?lixture
III. EXTRACTIVE AND AZEOTROPIC DISTILLATION 369
with the azeotrope. It is not an absolute criterion, as mentioned above,

since a better separation is frequently realized upon distillation than would
be predicted from the difference between the boiling points of the azeotrope
3.0~----------------------------------------~------~
~
,:
I-
::;
~..J
o 2.0
> _-
lJJ
>
~
...J
---
lJJ
0:
5 10 15 ;::0
BOILlNG·POINT DIFFERENCE, ·C.
Fig. 19. Relative volatility VB. boiling-point difference.
VAPOR~ .............. _ B
B
w w A
0: _,. .,'. 0:
:::I
;:)
I-
<[
" l-
<t
0: 0:
w LIQUID lJJ
c.. c..
:!: :::E
w ILl
l- I-
o 1.0 o 1.0
M(ll F FRACTION OF B MOLE FRACTION OF B
Fig. 20. Temperature - composition Fig. 21. Tel1lpemtlll'e - composition

f(~lat.i(llJships
for a minimum (lonstant;... relationships for a maximum constant-
boiling mixture. C()n~tant pre~~Ul'e. boiling mixture. Constant pressure.
and the nearest-boiling component. This is illustrated by the data of

Matuszak and Frey 62 on the four-carbon hydrocarbons, butenes and bu-
tanes, with sulfur dioxide as an entrainer. In Figure 19 are plotted rela-
52 Matus~ak and Frey, Ind. Eng. Chem., Anal. Ed., 9, 111 (1937).
370 CARL S. CARLSON
tive volatilities, <x, calculated from experimentally determined vapor pres-

sures of the azeotropes as a function of the difference in boiling points of
the azeotropes. It will be noted that the relative volatilities lie appreci-
ably above those for an ideal solution with the same difference in boiling
point of its constituents.
A quick experimental technique used by early investigators of azeotrop-
ism was to determine the boiling point of a binary mixture at various con-
centrations of one component in the other. A minimum or maximum in the
boiling-paint-composition curve of the mixture \Vas taken as proof of the
existence of an azeotrope. This technique is subject to considerable errol'
B
w w
0: 0:
::J ::J
U) U)
lJ) A U)
w W
0: 0:
a. a.
o 1,0 o
MOLE FRACTION OF A MOLE FRACTION OF A
Fi". 22. VnnOf-pl'eSSlIl'e - com pORi- Fig. 23. V!tpol'-pressure - comjYlsi-

tion m\ationships for a binary mixtul'(1 tion relationships for a binn,ry mixtUl'''
with n minimum constnnt-boiling mix- with a maximum constant-boiling mix-
tUTe. Constnnt temperature. ture. Constrmt tHmpel'atlll'e.
for aZBotropes boiling close to one of the pure components. Figure 20

shows a boiling-point-composition curve for a binary mixture of A and B.
The curve for the liquid goes through a minimum at Z indicating a mini-
mum boiling azeotrope at this temperature and composition. Similar
curves are given in Figure 21, which shows a maximum-boiling azeotrope
atZ.
Another rapid experimental technique is the determination of the vapor
pressure of mixtures of the two components at a constant temperature.
A maxjmum in the vapor-pressure curve of the binary mixture is indicative
of a minimum-boiling azeotrope, while a minimum in the vapor-pressure
III. EX'I'RACTIVm AND AZEO'l.'ROPIC DISTILLA'l'ION 371
curve of the binary mixture is indicative of a maximum COllstant~boiIing

mixture. In Figures 2'2 and 23 are plotted typical curves of the former and
httter typc, rcspectively. The straight lines in each figure are the partial
pressures of each component of the solution calculated from Raoult's law.
The sum of the ordinates of these two lines, shown tiottetl and designated
as A + B, is the vapor pressure of the mixture predicted from Raoult's
law. The actual vapor-pressure curve of the mixture in Figure 22 shows a
maximum at Z} while in Figure 23 the vapor-pressure curve of the binary
mixture shows a minimum at Z. Positive (or negative) deviations from
Raoult's law are frequently observed for binary mixtures. Figure 24
MOLE FRACTiON OF A
Pig. 24. Vapor pressure - composition l'Blationships for a

hinary m.ixture exhibiting; positive devilttion from Ruoult'slaw.
shows a typical example of positive deviations without an azeotrope as the

difference between the dotted line A + B and the curved line. Positive
deviation from Raoult's law is not ill itself proof of azeotl'Ope formation,
which requires that the vapOl'~pressure curve go through it maximum. De-
termination of vapor pressure as a test for azeotropism like the determina-
tion of boiling points becomes difficult for azeotropes boiling close to the
boiling point of one of the pure components of the mixture.
Unless extensive work is planned with a given entrainer, it is frequently
easier to add the en trainer to the mixture and to cafry out a fractional
,azeotropic distillation in a column of known separating power than to de-
termine the fundamental vapor-liquid equilibria. The effect of the en-
trainer on the separation of the mixture will be readily observed..
372 CARL S. CARLSON
4. Laboratory Apparatus and Procedure for Azeotropic Distillation

A. BATCH DIS'fILLA'l'ION
A21eotropic distillation has been llsed extensively in laboratory work be-

cause of its ease of application. Using cOllventional batch fractional-dis-
tillation equipment, entrainer in sufficient quantity or in excess is added to
the still with the mixture to be separated. A regular analytical distillation
is then carried out taking distillate fractions representing 1 or 2% of the
charge. After removal of thc entminel' from thc distillate the usual physi-
cal measurements such a.'S refractive index and density or boiling point can
be made to determine the degree of purity of the entrainer-frce cuts. A
plot of vapor temperature at the still head versus volume pel' cent of charge
distilled will give a good indication of the cut point between different azeo-
tropes and excess entminer. COllventional laboratory distillation col-
umns suitable for this work have been described in Chapter II.
For those distillations in which the entrainer forms a separate phase in
the distillate on cooling, care must be cxerciscd in selecting the type of
column head to be used. Reflux condensers with little liquid holdup and
with provision for intermittent take-off of total distillate are preferred.
In this way both liquid layers will bc obtaincd as product. Examples of
such column heads are the swinging-funnel, liquid-dividing type and the
vapor-dividing head shown in Chapter II. If, on cooling, the entrainer is
concentrated in the upper layer, the simple still head (exclusive of funnel)
shown in Figure 8 may be used. The lower layer, in which the product
will be concentrated, is withdrawn through stopcock E. When the reverse
case obtains, still heads of this kind should bc avoided.
Continuous separation of distillate phases with return of one phase to
the column can be carried out automatieally with the aid of a decanter and
auxiliaries such as those shown in Figure 25. In order to return one phase
to the column by gravity flow, an extension approximately 17 in. long is
added to the top of the column to deliver the distillate vapors to an elevated
condenser. The complete arrangement is shown in Figure 25 with alter-
nate arrangements in Figures 26 ancl27. Thc vapor delivery line is usually
the same diameter as the column and is insulated or heated by electrical
windings to prevent excessive refiuxing. Provision must be made to vent
the column as well as each decanter phase, which in the case of atmospheric-
pressure distillation may be directly, to the atmospherc. These vent
points are designated as A, 13, and 0, respectively, in Figures 26 and 27.
Typical decanter dimensiolls are 15 cm.long X 2.5 cm. O.D. (6 in. X 1 in.),
the delivery and overflow tubes being usually 8-10 mm. O.D. (Graduated
separatory funnels of 125~ and 250-ml. size have been converted into de-
canters by the addition of a side arm near the top.) The interfacial level
III. JilX'l'RAC'l'IVE AND AZEO'l'ROPIC DIS'l'lLLA'l'ION
CONDENSER
8-IOmm,Q,D.
8-10 mm, O.D.
Fig. 25. Continuous uec!1uter and distillation columB.

~
Fig. 2ti. Altern!1te tIJ'l'IIll!!;tllllfut for Fig. 27. Auother niterll>lttl !\1'l'!111ge-
setup in Figure 25. ment for Figllre 25,
374 CAUL S. CAHLSON
in decanter D is determined by the differenee in density of the two phases

and the location of overflow point C. The latter can easily be adjustecl if
flexible tubing such as Tygon, Saran, neoprene, Paracril, or rubber is con-
nected at point F and C. A glass Y- or T-tube is used for convenience at
point C. When the material being handled attacks flexible connections of
the type mentioned above so that n.ll-glass or metal connections are neces-
sary, semispherical ball joints at P, C, and G will permit adjustment of the
level of C. Glass decanters or separators are preferred since the level of
the interface can readily be observed.
In Figure 25 the lower dist.illate phase is being returned through the
thermometer well of a vapor-dividing; head from which the valve rod and
condenser have been removed. A column-feed section can also be used.
Both these pieces of equipment can be purchased from commercial glass
blowers as stock items. The loop in the return line at G provides a liquid
seal so that vapors from the column cannot by-pass the condenser and es-
cape into the atmosphere.
In Figure 26 the decanter is attached to a laboratory column of the type
described in Figure 8 but from which the vent and product draw-off have
been removed. The lower phase is being returned through tlie thermometer
wen at H.
Obviously, either the upper or lower phase may be returned to the
column. In Figure 27 the upper phase is being recycled. Note th~t this
line is vented at point B to permit proper operation of the decanter. The
returning phase in this caSe is introduced through the vapor delivery tube.
Frequently, phase separation of the distillate does not occur except
after considerable cooling. In such cases, it is often preferable to operate
the condenser so as to cause liquefaction but not phase separation. The
distillate, withdrawn while warm, will separate into two phases on further
cooling. In more elaborate arrangements, two condensers are used, the
first for reflux condensation, the second for phase separation of the product
and recycling of the entrainer phase.
Steam Distillation. One very common form of azeotropic distillation
is steam distillation. It is widely used in organic laboratory work, par-
ticularly when dealing with heat-sensitive materials. It is used for the
distillation or purification of water-immiscible or slightly water-miscible
materials such as aniline, glycerol, or fatty acids from cottonseed foots.
It was pointed out in the discussion of the theory of azeotropic distillation
that, when two immiscible liquids are distilled and are present as separate
liquid phases, each liquid will exert its own vapor pressure independent of
the other so that the total pressure above the boiling solution is the sum of
the individual vapor pressures. Thus, in Rteam distillation with a water
phase present, the distillat.ion t.emperature will never exceed the boiling
III. EX'l'RAC'rIVE AND AZEOTROPIC DIS'l'ILIATION 375
point of water at the pressure, usually atmospheric, at whioh the distillation

is conducted. The approximate boiling temperature can be calculated
by employing this principle if the vapor-pressure data are known.
For example, if the vapor-pl'eSHUre curves of toluene and water are
plotted as log p VB. I/ToK, straight lines are obtained. Adding vapor-
pressure readings at several different temperatures and plotting these
points will also result in a straight line. Where this line crosses the 760-
mm. pressure ordinate will be the predicted boiling point. This intersec-
tion occurs at 84.4 °C., at which temperature the vapor pressure of toluene
is 337 and that of water is 423 mm. of mercury. The azeotl'opic boiling
point reported by Lange 63 is 84.1 °e.
The molal composition of the distillate can be calculated from the above
vapor pressures. The weight ratio of the two components in the distillate
is given by the following equation:
(18)
where A refers to the upper-phase material, in this case toluene, and B
refers to the lower-phase material, water. W, P, and JJ!1 are weight, vapor
pressure at distillation temperature, and molecular weight, respectively.
Substituting the vapor pressures given above and the appropriate molecular
weights the following result is obtained:
WA/W B = (337)(92)/(423)(18) = 4.1
Thus one can expect 4.1 g. toluene for each gram of water in the distillate.
The predicted distillate composition is (4.1/5.1)(100) or 80.3 weight per
cent toluene. The azeot.ropic composition reported by Lange sa is 86.5
weight per cent toluene.
A very useful chart for estimating the boiling points of immiscible water
mixtures is given by Badger and McCabe. 64 Vapor pressures are plotted
vs. temperature in the normal manner. The curve for water is plotted as
760 minus the vapor pressure of water' so that the resulting curve intersects
the other vapor-pressure curves at the steam-distillation boiling point at
760 mm. of mercury. Similar curves for water can be drawn for other
pressure levels such as 300 mm. or 100 mm. distillation pressure.
The use of steam distillation as a laboratory technique merely requires a
source of steam, a steam delivery tube, a distillation flask, condenser, and
receiver. Such an apparatus is illustrated in Figure 28. When steam is
available, the steam generator may be dispensed with. Numerous modi-
03 Lange, Handbook of Chemistry. 5th ed., Handbool, Publishers, Sandllilky, 1944, p.
1368.
64 Badger and McCabe, Elements of Chemical Engineerinu. 2nd ed., McGraw-Hill,
New York, 1936, p. 369.
376 CARL S. CARLSON
fications and improvements ean be incorporated in this basic apparatus.

For example, laboratory steam llHuully eontuins an appreciable amount of
condensate which will rapidly fill the distillation flask with water phase.
Much of this water can be eliminated by a trap or knockout flask placed
just ahead of the distillation flask. Provision should be made for inter-
mittent draining of the trap. A trap with automatic drain is described
by Egly in Chapter I, Voilime III, of this series.
With the arrangement shown in Figure 28, all the heat required for
warming the solution and for vaporization of the distillate must be sup-
plied by the steam which will condense in the distillation flask, gradually
STEAM DISTILLATION
GENERATOR FLASK
Fig. 28. Steam IliHti11at.jOIl.
filling it with water phase. The latter ean be reduced intermittently by

draining out part of the water phase. An easier method is to supply heat
to the distillation flask also, either through a steam superheater (see Egly,
Chapter I, Volume III, this series) or by heating the dist.illation flask di-
reetly. Drains for the klloekout and the distillation flask can be provided
as bottom stopcocks or as suction tubes inserted through the flask stopper
and extending nearly to the bottom of the flask. The latter is better
praetice as it is well to place all heated distillation flasks in vessels or pans
of sufficient size to retain all the liquid in the distillation flask in case of
breakage. This greatly reduces fire hazard in the laboratory.
The laboratory procedUre for steam distillation is to charge the organic
material into the distillation flask shown in Figure 28. A small amount
III. EX'rRACTlVE AND AZEOTROPIC DISTILI,,\TION 377
of water may be added if desired, although condensing steam 'will provide

an adequate water phase shortly after starting the distillation. The dis-
tillation flash should be no more than one-third full at the start of the dis-
tillation. A thermometer in the liquid phases in the distillation flask makes
it possible to follow the course of the distillation. The distillation is dis-
continued when negligible amounts of distillate are obtained in a reasonable
time such as an hour. Frequently, organic matter will be dispersed in the
aqueous distillate as evidenced by a haze or white cloud in the water phase.
This can usually be eliminated by adding salts such a.s sodium OJ' potassium
chloride to the aqueous phase.
It should be pointed out that the distillate recovered during steam distil-
lation usually contains much more water than is predicted from azeotropic
compositions. In other words, the steam is less than 100% efficient in
vaporizing the organic components. The relationship between theoretical
and actual distillate ratios is usually referred to as steam-vaporization ef-
ficiency and will depend upon the depth of organic phase above the water
phase and the manner in which steam is dispersed in the distillation flask.
The vaporization efficiency of the steam is increased as the organic phase
is increased in depth to about 50 centimeters. It also increases as the
diameters of the openings for admitting steam into the distilhttion flask
are decreased to one millimeter. If these steps are taken to increase va-
porization efficiency, a greater pressure is required to force steam into the
distillation flask.
One of the simplest methods of employing steam distillation in the labo-
ratory involves placing the organic phase and sufficient water phaRe in a dis-
tillation flask. The heterogeneous mixture is heated externally and the
distillate condensed and collected as above. This technique is, of course,
limited to those cases in which the quantity of water required for the dis-
tillation can be added to the flask at the start of the distillation.
In certain cases only a small amount of volatile material is to be re-
covered from a large amount of relatively nonvolatile materlal. In such
cases an integral steam-distillation apparatus such as that just described
is used but with a vertical condenser. A distillate receiver which has pro-
vision for returning water to the distillation flask is used so that the system
will not be depleted in water. The oil-dilution test employed by the oil
industry, in which small amounts of gasoline are distilled from lubricating
oils, is an example of this type of operation. Distillate receivers suitable
for this purpose may be purchased f1:om commercial glass blowers. Such
receivers are listed under oil-dilution apparatus.
The discussion thus far has been limited to steam distillation with a water
phase present. Steam distillation in the absence of a water phase is
widely used commercially, particularly in the petroleum industry. This
378 CAUL S. CARLSON
type of operation is really a part of a broader technique known as distilla-

tion in the presence of an inert gas. In this case vapor pressures are still
additive, but the distillation temperature will vary as the proportion of
inert carrier is varied. Inert carriers which may be used are any of the
common inert gases such as nitrogen, carbon dioxide, or methane, as well
as steam. The last is preferred because condensation problems are simpli-
fied and water is readily available. See Chapter V.
B. CON'l'INUOUS DIS'l'II,LATION
For the satisfactory use of continuous azeotropic distillation consider-

able information about the system is necessary. This includes feed com-
position, percentage entrail1er in azeotrope, and boiling points of azeotropes
and nonazeotropes. As usually applied industrially, for example, in the
manufacture of absolute ethanol or in acetic acid dehydration, a selective
entrainer is used which forms a minimum constant-boiling mixture. The
separation in the fractionator is between the azeotrope obtained as distillate
and the nonazeotrope obtained essentially pure as bottoms. In such a
case the entrainer is added with the feed in amount just sufficient to form
the azeotrope which is removed as distillate.
An apparatus for continuous azeotropic distillation with a miscible en-
trainer is shown in Figure 29. Feed and entrainer in the correct propor-
tion are introduced at the propel' point in the column. A minimum-boiling
azeotrope will be obtained as distillate and segregated for later separation
of the entrainer from the desired product. When a selective entrainer is
used, tho bottoms should theoretically be free of entrainer. In the practi-
cal case a very slight excess of entrainer ensures obtaining all the desired
product azeotrope. If the entl'ainer boils between the azeotrope and the
bottoms product, some of tho elltrainer may appear in the bottoms if ade-
quate stripping in the lower section of the column is not provided.
An apparatus for continuous azeotl'opic distillation with a partially mis-
cible entl'ainer forming a minimum-boiling azeotrope is shown in Figure 30.
Phase separation of the distillate occurs on cooling, and by means of separa-
tor S the heavier entrainer phase is refluxed to the column through over-
flow H, which is vented at V. A certain proportion of the light phase may
have to be l'efluxed also in order to maintain the necessary reflux ratio.
Sillce most of the entrainer is returned to the column, the feed introduced
contains only enough entrainer to make up for that removed with the dis-
tillate product. Entrainer-free bottoms are withdrawn from the still.
The number of points of control required in the arrangements shown in
Figure 30 can be decreased. A modified still-head arrangement is shown
in Figure 31, in which the condenser and reflux device are inside an ex-
III. gXTHACTIVE AND AZEOTROPIC DIS'l'ILLA'l'ION 379
panded section of the column. The reflux ratio is controlled by a swinging

bucket or funnel, positioned by a solenoid, which is actuated by an electric
timer not shown. The reflux ratio is really a time ratio. The column oper-
ates under eonditions of total reflux, all condensate being returned to the
DISTILLATE
REFLUX
DISTILLATE
(AZEDTROPE)
FEED + FEED +
ENTRAINER ENTRAINER
MAKE-UP
BOTTOMS
BOTTOMS
Fig. 29. Continuous !lzeotl'opic Fig. 30. Continuous azeotl'opie

distiJIntion with miseibJ een- distilllltion with partially miscible
trainer. ent-raine!" •
column through fUllllelH F ttIlcl R, uutil the solenoid is actuated, when the
funnel F is pulled to the right, pivoting on point P. All condensate then
runs. through F and D to the separator, S, where heavy entrainer phase is
returned as reflux through II, and distillate is withdrawn. If line II is of
,~ufficjent diameter for the flow involved it neeel not be vented. With this
380 CARL S. CARLSON
I1rrangemcnt, only the feed and bottoms streams require continuous con-
trol.
DISTILLATE
,..-.,_-.,..
COLUMN
Fig. 31. Automatic column head for llzeotropie distillation

with 1\ partially miscible entraincl'.
C. EQUIPI'ImN'f PlDltFOHMANCE
As in ext,l'active diHtillatioll, little information is available on equipment

performanue. GUillut and Clarke,66 in discussing the continuous produc-
tion of anhydrous ethanol in commereilLl bubble-clLp plate towers, stated
that, in the upper part of the column below the entrainer refiux, two liquid
phases existed on several of the plates, but agitation wal:> so vigorous that
uniform dispersion of the phases was realized and little effect on plate ef-
ficiency was observed, although it was not actually measured. More
recently, Sehoenborn et al. 66 showed that the presenee of a moderate propor-
tion of inl'oluble component in the liquid made no appreeiable change in
plate efficiellcy of a laboratory bubble-cap column.
In laboratory columns of the packed variety, liquid-phase separatiou
66 Guinot and Clark, Trans.lnst. Chem. En(/rs. London, 16, 189 (1933).
66 Schoenborn, Koffolt, and Withrow, Tmns. Am. Inst. Chem. EngTs., 37, 997 (1941).
III. EX'.rRACTIVE AND AZEOTROPIC DISTILLATION 381
may have a deleterious effect on H.E.T.P. if one liquid preferentially wets

the column packing.
5. Calculation of Azeotropic-Distillation Separations

In the calculation of azeotropic-distillation separations, th€ azeotrope
can be treated as an individual component. This is illuRtrated by the work
of Matuszak and Frey,67 who studied the separation of the four-carbon
hydrocarbons into butanes and butenes by azeotropic distillation with sul-
100
63% SOa 63
z 62
0
CD
n:
<I
u 80
0
P' 2.4 atm. p'I.85atm.
:30;
>- ..
:1:'::
0
060
WUl
~~
64
n:w
::>1-
1-<1
<1--'
Ul::! 40
1-1-
z~
wO
U z
n:-
w
u. 20
w
--'
0
::;;
72% SOa
0 20 40 60 80 100
MOLE PER CENT DISTILLED
Fig. 32. Separation of butanes from butenes by azeotropic distillation with

sulfur dioxide in a laboratory vacuum-jacketed glass gas-analytical column.
fur dioxide. Data from one of their runs have been used to plot the curve
in Figure 32. The hydrocarbon mixture ~onsisted of (a) n-butane with a
trace of isobutane and (b) n-butenes with a trace of isohutene. The dis-
tillate composition is expressed on an entrainer-free basis, while the mole
per cent distilled refers to the total distillate including entrailler.
The distillation was carried out isothermally at 3°C. in a laboratory
vacuum-jacketed (ullsilvered) glass analytical column with approximately
10 theoretical plates. sB The hydrocarbon in the first 56% distilled aver-
aged 99.5% saturated hydrocarbons, while the hydrocarbon in the residue
(exclusive of a trace of pentane impurity) averaged 99.4 mole per cent
81 Matuszak and Frey, Ind. Eng. ahem., Anal. Ed., 9, 111 (1937).
18 F. E. Frey, private communication.
382 CARL S. CARLSON
bn tenes ([d' tel' 82.5% distillate was rcmoved). Shown in Figure 32 are
the average sulfur dioxide conccn tl'atiolls in the distillate which correspond
very closely to the azeotropic compositions given in Table XIII. For
example, the first distillate, which contained the saturated hydrocarbons,
contained on the average 63 mole per cent sulfur dioxide, which corre-
sponds closely to the n-butane azeotrope when allowance is made for the
azeotrope of isohutane, which contains only 54.9 mole per cent sulfur di-
oxide. Similarly, a portion of distillate, the hydrocarbon portion of which
contained 95.0 mole per cent n-hutenes, 3.4% isobutene, and 0.6% paraf-
fins, contained 71.8 mole pel' cent sulfur dioxide, very close to the azeotropic
composition listed for 2-butene.
'fABLE XIII
BINARY AZEOTROPIC MIXTURES OF SULFUR DIOXIDE AND FOUH-CARBON HYDROCARBONS
Malo per "cnt

Temp., !)ressurc, Hydro-
Hydrocarhon °0. atm. carbon SO,
Isobutaue ......... 3 3.17 45.1 54.9

n-Butaue ........ " 3 2.65 35.6 (i4.4
Isobutene .. '1 •••••• 3 2.2'1 36 64
1-Butene ........ " 3 2.37 40.7 59.3
2-Butene .......... 3 2.05 27.7 72.3
The separation of an azeotrope from a nonazeotrope or of two azeotropes

. from each other can be calculated by the conventional methods given in
Chapter I, treating the azeotrope as an individual component, provided the
cOlTef:lponding relative volatilities are known. Most of the data on azeo-
tropes consist of their compositions and boiling points. The use of boiling-
point differences to estimate relative volatilities, a, will in general give a
conservative result. The actual separation realized during azeotropic
distillation would indicate a much higher value for a than the calculated
one. This was illustrated in Figure 19 for the four-carbon hydrocarbons
n~butane, isobutane, n-butenes, isobutylenc, with sulfur dioxide.
A. AMOUN'l' OF EN'l.'RAINER REQUIRED
If the composition of the azeotrope or azeotropes is known and the ap-

proximate composition of the charge is known, the entrainer requirements
can readily be estimated. For example, consider the distillation 67 plotted
in Figure 32.
Basis of calculation.
(a) Hydrocarbon charge is 75 mole per cent n-butane and 25 mole per cent 2~
butene,
III. IGXTRAC1'IVE AND AZEOTROPIC DIS'l'ILLA'l'ION 383
(b) Distillation conducted isothermally at 3°C.,

(c) Entrainer to be used, sulfur dioxide.
Data.
n-Butane-sulful' dioxide azeotrope contains 134.4 mole per cent sulfur dioxide.
2-Butene-sulfur dioxide azeotrope contains 72.3 mole per cent sulfur dioxide.
Entminer requirements ]Jel' gaseous liter hydrocarbon charge.
(a) For n-butane: (750)(64.4)/(35.5) 1350 ce. sulfur dioxide.
(b) For 2-butene: (250) (72.3)/(27.?) 650 cc. sulfur dioxide.
Total for one liter of hydrocarbon gas. . . . . . 2000 cc. sulfur dioxide.
Thus for theoretical en trainer requirements the charge should contain

66.7 mole per cent sulfur dioxide and 33.3 mole pel' cent hydrocarbon.
The actual charge contained only 60% sulfur dioxide, which was found to
be insuffieien t. 67
It is of interest to considcr what results should be expeeted in a batch
fractional distillation when various proportions of entrainer are used. For
purposes of discussion a three-component mixture will be used with a high-
boiling entminer added as a fourth component, capable of forming azeo-
tropes with each component.
(a) Insufficient Entrainer Present to Form Azeotropes with All Com-
ponents. When this situation exists, the entrainer will appear in the
distillate with the lowest-boiling azeotropes until the entrainer is ex-
hausted. Separation will then depend upon the normal volatility relation-
ships of the remaining components.
(b) Theoretical Amount of Entrainer Present to Form Azeotropes·
with All Components. In this case the distillation will proceed as though
three components were pres(~nt with volatilities corresponding to those of
the azeotropes. Better separation of the azeotropes will be realized than
for the above case.
(c) Entrainer Present in Excess of the Amount Necessary to Form
Azeotropes with All Components. SiIlce a high~hoiling entrainer was
designated, the distillation will progress with the three azeotropes being
obtained successively as distillate with excess entrainer being distilled
last if the distillation is carried that far. The separation of the azeo-
tropes may be sharper than in b because of the effect of the presence of
excess entrainer.
In brief, in batch azeotropic distillations the entrainer and charge will be
added directly to the still in calculated proportions, and usually with a
slight excess of entrainer. If the charge composition is unknown and in-
sufficient entrainer has been added, the distillation can be stopped and ad-
ditional entrailler added to the still. The distillation can then be con-
tinued after the usual equilibration time under total reflux has been al-
lowed.
384 CAUL S. CARLSON
a:
w
~
..d
'-'
.~
,
'"r::::
1--------------
-, 1:131.'ttM
+ lOHO::ll'tt
'"'"r::::
'"
.0
I ~
~
I
I ..d<.1
.~
I ...
I "''oj""'
I Il=
..J
I 0 ~
~I
2
I w
<:(
::c fi
I
~I
I-
w
I
I
I
COl
it('
LO
(Jl
1
-:S
I
~--- I .g
...'"
OJ
I '"
I
'"'"
~
I
LE--- -
I P..
'"
I '"
;;.,
I ~
~
I toi)
I l£:
<J
I
~
»OW
I
CJ:0,_
"(0
zl= L _ _ _ _ _-I
I
0:0
ww
I-N
I
<:(
W
20
I
WW
~~ I
~ + - - - - - - - - - --~
III. EXTRAC'l'IVl!l AND AZEOTROPIC DISTILLATION 385
6. Example of Commercial Application of Azeotropic Distillation

While extractive distillation appears rather complicated on a laboratory
scale but is relatively simple on a continuous industrial scale, azeotropic
distill[1tion is very simple to use in a laboratory batch distillation. How-
ever, the commercial applications of continuous azeotropic distillation are
rather complicated. This can be seen in Figure 83, illustrating the Keyes
process for the production of absolute ethanol. Ethanol (95%) and ben-
zene cntrainer are fed to column 1, where absolute ethanol is obtained as
bottoms while all the water is carried overhead as a ternary azeotrope with
ethanol and benzene, boiling at 65 DC. The composition (in per cent by
weight) of the ternary azeotrope is as follows:
Benzene. . . . . . . . . . . . . . . . . . . . . . . . . . . 74. 1
Ethanol. ................... , ..... , 18.5
Water............................. 7.4
All the remaining equipment serves to recover benzene and alcohol
carried overhead as the ternary azeotrope in column 1. On condensing,
the ternary azeotrope forms two phases, which are separated in the sepa-
rator. The upper, benzene-rich phase, which at 28 DC. contains 85.6%
(by weight) benzene, 11.6% alcohol, and 2.8% water, passes to the benzene
recovery column 3, where anhydrous benzene is obtained as bottoms and the
ternary azeotrope as overhead; the latter is returned to the separator.
The lower, water-rich Iayer flows to a scrubber, where water is added in
amount sufficient to cause further benzene-phase separation. This ben-
zene phase also goes to the benzene column. The water layer, which at
28 cC. contains 40,6% (by weight) water, 51.8% alcohol, and 8.1% ben-
zene, passes to column 2, where benzene is taken overhead as the ternary
azeotrope and returned to the separator. Alcohol-water bottoms from
column 2 flow to the alcohol recovery column 4, where water is removed
as bottoms, and 95% ethanol is taken overhead for return to alcohol feed
storage.
This process is interesting since it employs the features of distillate phase
separation and phase precipitation by water addition. All heating is done
by closed steam.
General References
Extractive Distillation
THEORY OF EXTRACTIVE DIS'I'ILLA'rION
Benedict and Rubin, "Theoretical aspects of extractive and azeotropic distilla-

tion," Trans. Am. Inst. Chem. Engrs., 41, 353 (1945).
Colburn and Schoenborn, "The Selection of Separating Agents for Azeotropic
386 CARL S. CARLSON
and Extractive Distillation and for Liquid-Liquid ElI.'traction," Trans. Am. Inst.
Chem. Enol's., 41, 421 (1945).
Scheibel, "Principles of Extractive Distillation," Chem. Eng. Progress, 44, 927
(1948).
DUlln, Millar, Pierotti, Shiras, and Souders, Jr., "Toluene Recovery by Extrac-
tive Distillatiol1/' Trans. Am. Inst. Chern. Engl's., 41, 631 (1945). Includes vapor-
liquid equilibria data for nontoluene-toluene systems.
FUNDAIlIEN'l'AL VAPOR-LIQUID EQUILIBlUA
Griswold, Andres, Van Berg, and Kasch, "Pure Hydrocarbons from Petroleum,"
Ind. Eng. Chem., 38, 66 (19'15). n_Heptane-methylcyclohexane-solvent.
Updike, Langdon, and Keyes, "Effect of Added Component on the Relative
Volatility of Two Binary Mixturcs," Trans. A.m. Inst. Chem. En(J/'s., 41, 717 (1945).
Methylcyclohex!ine-toluene; . cyclohcxane-benzene.
Drickamer and Hummel, "Vapor-Liquid Equilibria in Phenol-Hydrocarbon
Systems," 'l'rans. Am.lnst. Chem. Engl's., 41, 55,5 (1945).
Fenske, Carlson, and Quiggle, "Separation of Hydrocarbon Mixtures by Vapor
Extraction," Ind. Eng. Chem., 39, 1822 (1947).
EQUIPMEN'r PERlrORMANCE
Drickamer and Hummel, "Application of Experimental Vapor-Liquid Equilibria

to an Analysis of the Operation of a Commercial Unit for the Purification of Toluene
from Petroleum," Trans. Am. Inst. Chem. Engrs., 41, 607 (1945).
Happel, Cornell, Eastman, Fowle, Porter, and Schutte, "Separation of C4 Hy-
drocarbons Using Furfural," Trans. Am. Inst. Chem. Engrs., 42, 189 (1946).
Dicks and Carlson, "Extractive Distillation of Hydrocarbon Mixtures in a
Packed Column," Trans. Am. Inst. Chem. Engr.~., 41,789 (1945).
Griswold and Van Berg, "Pure Hydrocarbons from Petroleum," Ind. Eng. Chem.,
38,170 (1946).
Azeotropic Distillation
THEORY OF AZEOl']WPIC DIS'rILLA'l'ION
Fleer, "Azeotropism: A Useful Tool Clarified," J. Chem. Education, 22, 588

(945).
Craven, "The Third Component Method of Fractional Distillation," Ind. Chem-
ist, 1933, 414.
Ewell, Harrison, and Berg, "Azeotropic Distillation," Ind. Eng. Chem., 36, 871
(1944) .
Ewell and Welch, "Rectification in Ternary Systems Containing Binary Azeo-
tropes," Ind. Eng. Chem" 37, 1224 (1945).
Horsley, "Graphical Method for Predicting Azeotropism and Effect of Pressure
011 Azeotl'opic Constants," Anal. Chem., 19,603 (1947).
Meissner and Greenfeld, "Correlation of Azeotl'Opie Data/' Ind. Eng. Chem., 40,
438 (1948).
III. EX'l'RA(J'l'[VE AND AZEOTHOPIC DIS'l'II,L:\'l'ION 387
Berg and Harrisou, "Evaluation of AZ8otropic 1Dutraincl's," Chcl/!. En(/. Progress,

43,487 (1947).
DrCUYDRA'1'ION OF AQUgOUS ETHANOL
Keyes, Ind. Eng. Chern., 21, 998 (1929).

Guinot and Clark, Trans. Inst. Chem. Engrs. London, 16, 189 (1933).
Colhurn and Phillips, Trans. Am. Inst. Chern. Engrs., 40, 333 (1944). Labomtory
study using trichloroethylene.
Othmer and Wentworth, Tmns. Am. Inst. Chern. Enars., 36, 785 (HJ40).
DEIIYDllA'l'ION OF' AeE'l'IC ACID
Othmer, Trans. Am. Inst. Chem. Engrs., 30, 299 (1933-1934).

Othmcl', Chem. & Met. Eng., 48,91 (1941).
HYDROCARBON SHPARA'rIONS
Mail', Glasgow, Jr., and RosHini., J. Research Natl. Bur. Standards., 27, 39 (1941).
Lake, Trans. A.m. Inst. Chern. Engrs., 41, 327 (1945). Toluene purification.
Hartley, PetrolewnRefiner, 24, 519 (19 15). Toluene purification.
L
'l'ABI,BS OF AZEOTROPlcS
Horsley, "Table of Azeotropes [tIld N onazeotropes," Anal. Chern., 19, 501H300

(1947); 21,831 (19·19).
Lange, Handbook of Chemistry. 5th ed., Handbook Publishers, Sandusky, 1.944,
pp. 1386-1395. Extensive, well-arranged tables.
Ewell, Harrison, and Berg, Pet1"olemn Engr., 16, October, 255, November, 259,
December, 219 (1944). Excellent compilation of alleotropes involviltg hydrocar-
bons.
Lecat, La tension de vapeur de.~ melanges desliquides. L' Azeotl'opisme. Lamertin,
Brussels, 1918.
Chemical Engineers' Handbook. 2nd ed., McGraw-Hill, New York, 19,11, pp.
1370-1374.
International Critical Tables. Vol. 3, McGraw-Hill, New York, 1928, pp. 318-
324. Largely data of Lccat in more accessible form.
CHAPTEII. IV
DISTILLATION OF LIQUEFIED GASES

AND LOW ·BOILING LIQUIDS
ARTHUR AND ELIZ,\.BETH ROSE, Slale Col/eye, Pennsyit1uniu
When a mixture of gases is to be separated by distillation, it must first

be converted to the liquid state. This may be accomplished in one of
three ways: by increase in pressure if the gas is below its critical tempera-
ture, by a combination of inerease in pressure and cooling, or hy cooling
alone. Little has been published on the lltboratory-scale distillation of
gases liquefied by pressure alone. 1,2 The proeedure involving liquefaction
by cooling alone is by far the most common, I1llCi the subsequent dist.illation
is referred to as low-temperature distillation. Low-temperature distilla-
tion is accepted as dealing with boiling points below about 35 DC. Of room
temperature; high-temperature distillation deals with all boiling points
above. The apparatus ordinarily used is a fractionating column, of the
Podbielniak type or one of its mallY modifications (see Figs. 1 and 2).3
Simple distillation from one bulb to another without a fractionating column
is an alternative procedure of (Jollsi<iel'able interest.
The most frequent. purpose of low-temperature distillation is that of
analysis of a gaseous sample. Other objcetives are the separation, identi-
fication, or purifieation of a Hingle material. The results are usually ex-
pressed as a distillation eurve as in Figure 3. Most of the discussion in
this section relates to analysis, espeeially t.he analysis of hydrocarbon gas
mixt.ures, since in general its requirements are representative and most ex-
tensively used and developed. The appamtus and procedure are quite
similar to those for analytical fractionation at and abov(~ ordinary tempera-
tures. The problems of refrigeration, im;ulation, anel handling of gaseous
sample and product do, however, introeluce differences requiring separate
attention. The industrial distillation of liquefied air and the large-seale
distillation of easily liquefied gases under superatmospheric pressure are
1 Simons, Ind. Eng. Chem., Anal. Ed., 10, 30 (1938).

1 Benoliel, M.S. Thesis, Pennsylvania State College, 1941. See also page 442.
3 Figures 1-9, 11, 12, 14, 15, 17B, 19,23, and 25-28 courtesy Podbielniak, Inc.
389
3HO A. AND E. ROSE
IV. LIQLTEFlIW G..\SES AND LOW-BOILING LIQUIDS 39}
not ineluded in the present, discussion, but references to reeent literature in

these fields may be found in the elassified bibliography OIl page MiO.
By way of defining the field, Table XI4 (pp. 447-456) gives the boiling
Fig. 2. Low-tcmpcl'I1t,ul'e fractional-distillation apparatus (Podbielniak Semi-Robot).

4 Compiled by Art.hur Rose and T. J. WiJliaU1fS.
392 A, AND K ROS}]
o0 0 0 0 0 0 0 0 a a 0·0 0 a a 0 0 0 0 a 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
PreSSure
reducec1
100
o
Vi
W
0:
"'4 PROPYLENE
0:
uJ
0-
~o
I-
REFINERY GAS
PENTANES PLUS
/
50
a ,0 '0
10 20 ~o 40
C) (I 0 0 II 0 0 II iO 0 DISTILLATE COLLECTED .0 0' 0 0 0
Fig, 3. Typical ]O\\,-tetnpl'rat.ul'c <iist.illa.t,ion curves.

1000
BOO , JL
.-.
600 7 ~7 I " 7 I 'I III LU
~ ~ ~~'"
500 1--I-- f-,lul / V
!f 1f--i!1 17/ I :11 J-IL
~
400 1-- I - 1'1
711
~;... -.J
~ ~/ /;
~~r,
>- 300
"'
::!'
iJ lu / II /;11
,r
=>
o0:: 200 I /'"
t.i
!i~Q:-
Q, Q,
o:t~~
Sc}"'I'__
CD:_") ~
f-WL~ II III
",'I V;
~ / I 1/
/ l
It~~IH~~
lJJ
::;; § ~
II 1II1 IV H ,,'W~
, '" Q, "'''
)~
lJ..
o ~ !j ~~ ~~'O:
I IT + __ ~ ~ ~~
r-J ...·-ti~~L"'tt
::E 100
::E BO 'J f7 ~"':$'
I-If- t--
.t-/~
W
~ 60 I II I I I V/ II
I 11 II I I " ) II; / 1/ 11/
/ I'_ ~._
~. 50
~ 40 17 I I VI /1/ II
Q..
II I "/ I/rl
~ {I Iii ~ ..
30
20
{
L! I I / I /' l~ Ii /
_I "I II II /'/Ill 'I;
/
/ ~I
/ it II / It I ~ rl
I oJ
-200 -180 -h,W . '140 -120 -100
r..J][I!J rl I/il V /II ;1/ II/; rl II
17
-80 -60 -40 -20 0 20 40 60
TEMPERATURE, ·C,
Fig, 4, Vt~rWi . pr'l:'sure-tempel'ature relations of volatile hydrocarbons.
IV. LIQUEFIED ClASt',S .\.ND I,OW-BOILING LIQUIDS 393
'rABtE 1. BOIl,ING POINTS OF' Hnmor:.umm; GASES "\:-.ill LNFID;; CO~IMONLY

ENCOUN'fERED IN LOW-'l'I,MPElUl'UllE DIS'rn,[.A'l'JOK
Cumponnd 'rmnpcrature, °0.

-----~.
Nitrogen ..... -19.5.8

Carbon monoxide ..... -102
Oxygen ... . -183
Methnne .. . -1Il1. ·.HI
EthyleJle .. . -]03.71
Ethmll' .... . 88.ti3
Propy]pnlJ .. '17.70
PropllllC .... . '12.07
Isobutmw .................. '" ........ ' 11.73
I~obntel1e ...... . !i.900
Buteu!::-l .. . (i. 20
II-Butane .. 0.50
trans-Butelle-2 .. . O.SS
cis-Butene-2 ... . 3.720
Isopentmw .................. . 27.854
1t-PentallC ..... . 3(:].074
Hexane ... . 08.7'12
TABLE II. RELATIVE VOLA'rILIl'mS 01' SO~f!.l HY!.lllOCAltnON )'IIX'l'{.lRgS
Hutio of
\'apor prc::!surC8
witiwu.y betw(!en
Boiling points h.p. of two
Binnry hydl'oOllrhon rnixtllr£! at. 71:0 Dun. components
------
Methaue-ethylellc .... . (-HH.5) (-103.7) (il.2
Methane-ethane .. . (-WI.5) (-88.6) 107
Ethylene-ethane ..... . (-103.7) (-88.6) 2.3
Ethane-propene .... . (-88.6) (-47.7) 8.1
Etlmne-pl'opane ..... . (-88.0) (-'12.1) \l.3
Pl'Opene-propane .... . (-'17.7) (-42.1) 1.3
Propane-isobutUlJ{' .. (-42.1) (-11.7) 3.3
lsoimtane-n-butunt: .. (-11.7) (-0.5) 1.5
Isubutane-isobutene .. (-11.7) (-6.n) 1.20
lsuimtene-butelle-I ..... . (-5.!!) (-6.3) 1.03
Butene-l-butadienc-I,3 ....... . (-6.3) (-4.5) 1.05
Butadiene-I,3 !lnd C.B.M." ... . (-4 ..5) (-.1.1) 1.02
Butl1t!iene-l,3-n-butllllC ...... . (-4.5) (-0.5) 1 . 1-1
Butene-l- n- butane ... " ... " . (-6.3) (-0.5) 1.23
/1- Bl! t.UIle - trans- butellc-2 . (-0 . .1) (+O.!I) 1.04
n-Butane-cis-butene-2 ... (-0.5) (+3.7) 1.15
trans-Butenp.-2 -cis- butene-2 .. (+0.9) (+3.7) I.O\!
n-But!1ne-isopentune ....... . (-0.5) (27.9) 2.8
Isopentulle-n-Jlentane ...... . (27.9) (30.07) 1.33
n-Pentane-n-hexl1ne .... . (30.07) (68.7) 3.0
n-Pcntalle-di-isopl'opyl ..... . (31i. 07) (58. 1 )
--------.----------__--------
2.15
a Constant-boiling; mixture: butadiene-l,3 79.5 mole per cent, 1,-tll1tane 20.5 molll
per cent.
points of substances with normal boiling points below :~5 °C. Table P
lists the more common hydrocarbon gases and low-Lr:ir>lf.! liquids. Values
5 Compiled from: Sdectr,rl Values of Properties of HlidrOC<il'ilolI", Nationl11 Bureau of
Standards, Circular C461, Wushington, 1947; and Internatir.,:n., Critical 'I'able.~, Mc-
Graw-Hm, New York.
394 .-\, AND g, HORE
for rela.tive volntility of numerous hydroenrbon mixtures are in Table II"

and vapor-pressure curves for common low-boiling hydrocarbons arc in Fig. 4.
1. FRACTIONAL DISTILLATION
1. Apparatus and Accessories
A general knowledge of the apparatuR and procedures for ordinary frac-
tional distillation is assumed. The nee(l for manometers, vacuum pumps,
stopcocks and vltlves, potentiometers, millivoltmeters, and thermoc:ouples
is obvious. Water pumps have been used for evacuH,t.ioll, hut oil pumps are
Buperior. Glass stopeocks must be precision ground, and the grease used
must be chosen to avoid absorption or contaminat.ioll of the sample. Dia-
plmlgm valves of small dead space are avail!thle, and are advantageous.
Ordinary rubber tubing is not ,mtisfiLctol'Y, but neoprene tubing can be
used with Sllecess, (A descriptioll of some of these accessories will be
found in Chap tors V Hud VI.)
The quant.ity of sample necessary or fwailable for dist.illat.ioll varies
gl'ea,tly with eircumHtaneec> so Umt bot,h very fimall and relat.ively large
llpparatus have been described lwcI used. The height [md other charac-
teristics of the apparatns depPlld 011 the degn~e of Hepamtioll desired and
the efficiency of the packing used, Detailed descriptions of various typmi
of apparatus will not. be attemptecl here. Illst8ad, t.he fundioll of each
important part will be indieated, together with various possible construc-
tions and arrangements, following the order listed in the outline.
A. COLUMNS AND PAC KINGS
The most widely used low-temperature columns and contacting devices

are Htraight glass tubes fitted with a specially formed wire packing. Pre-
sumably a Humber of the types of contaet,iug devices developed for high-
temperature distillation could be used for low-temperature distillation.
However, there are no reports in the literature on such work. Indeed,
evell the eommonly used pl1ekillgs have not been ext.ensively t.ested at low
temperatures, The majority of the tests refer to the usual mixtures such
as meihyleyclohexalle-n-heptane (boiling range 98.4-100,8°C.) rather
than to any low-boiling mixture (see Table I, Chap. I).
Buta-Heli-Grid Packing. The Buta-Heli-Grid Lype of packing (Fig,
5), patentecl by Podbielniak, is stated7 to have 200 to 400 theoretical plates
• Compiled from Podbielniak, Analytical Distillation and Its A.1Jplicatlon til the Pe-
Irolell/n Industry (uvailable from the aut.hor), Chicago; IIll!1 Selecled Valucs oj Pl'operlie~
of H1ldrocal'bons.
7 Podbielniak, Analylical Distillation and lt~ Applications to the PetroZGum Inilu8li',II.
IV. J,IQUEFrED GASES ,\ND LOW-BOILING LIQTTIDS 305
in 37 inehes of lwip;ht, 01' an iuclimh·tl H.E-T.P. of appl'(Jximutdy 0.2 to 0.1

ill. This is estimated for specifi(' operating c()llciitioll:-( dmhllJ; ::wpal'M,ioll
of butadiene from Ute butadiene--1I-1mtullc l'ullstallt-hoiling mixture haying
a boiling point ditference of D.H °C. The aetual eun·p Ilf volume versus
composition of distillate for thiN ("1.~e7 appearH in Figure o. Thi,; wa:o pre-
sumably obtained with low vapor veloeity lwd otlwrwiKc f[womble tondi-
tiolls for good fleparation.
The method for tram:latillg this curve into an l'f't·inmt:e of the llnmber of
theoretinll plateH is not de:-;eribed hut is probably that ilillieatecl by Rose. 8
No determination hus been puhli:=;hed of the number of theoretimtl plates
hy analysis of (listillaj.(~ and Aiill eompositiolls. Before suth values could
100 a
97.5
SAMPLE "B" -.----,~
2.5
IMORE N·BUTANE THAN CSM COlolP051TIONI
95
25 75
.,
w
~ 20 80
..
IU
Z
IU
55 6
:Z i!::J
ii-t' IS
.J
o
:l:
10
85
90
.
CD
.J
0
l.ESS N-SUTANE THAN) ::E
( CBf,l COMPOSITION
95
o ~--~~---L----~~~L_ ___J____-L____ ~IOO

400 GOO 800 1000 1200 1400
DISTILLATE IN CC.GAS@2.0·C.S760 mm .
Fig.5. Buta-I-ll'li- Fig. U. DisWlation of hut.tlllierm -Il-hutltn{\ mixture

Grid paeking showing sl'paration of (\ollst.nnt-hoilillg mjxtun~.
he obtainod it would be necessary to determilw the vapor-liquid equilibrium

curve for U sllitaNe dose-boiling, low-boiling billary mixtme. TeRt mix-
tures and proeedures Imve not yet been satisfactorily developed fDr testing
any columns of over 10() theoretic:tll plateR, regardless of t.he temperature.
The diameter of the Buta-Heli-Gl'id wire paeking is 0.25 ill. ane[ the
usual height is 37 in., with eonstmction lL::l RhoWll in Figure .5. There is no
flow of vapor or liquid through tho cent.ral hollow core. The hold\lp of the
packing is reported as 3 to 4 m!. liquid or 750 to 1000 ce. vapor for 11 87-ill.
length. This i:=; referred to as a high holdup paeking, so designed because
of observations tlutt a high mt,io of holdup to charge pfoclueed sharper sepa~
ration. 7 "Vhile tJds is eontrary to eOllventionn.l assumptions, the same ef-
; Rose', .T. Am. Chem. Soc., 62, 7lJ3 (19·10).

39fl ,~:~-. ~ ~ ~ 3 9 ~ ~;:tm ~
N;~3~~)...!3 :l ~ ~ ~ ~ :d .:l -~ n ,g~ 2
: --1----4:__ I--_~~l·~ 1---!---,~~8~--I
: I I;' ~ ~I~
f--l--f--+-+--I-_;_:f---+-1-+t:-+---+ICY+- u ~
f---+-+---L-....L..-..J'--4-- +-1-,1---1--1 1--1--
- - - --[1- --1-1--- - I - - §
-~ I~: ~l+-------+---t--I----l §
-: -1/'--1---I--l---I---- ~
r--r--t- ~I-J"-r-t--~--+-~
- I P ____+_--1-_1 § ui
I---+--t-~I+--+JI ~al--I---+-4--1 ....illOJ
§
;a
I
t
/
I
Jr6 -1--+--+--I--1 g~ '":::l0
";::
c:I
f--+-+--f--+-+--,H,f--+--t--HI i,. /-+-JI--+--l---I----I-----l~
po
I ....0
1
1-------+r;:1--+--+-----t--__-t_-___--1_ § rn
bJJ
:::l
;Q
<0
0::: 0:1
::> p.
0 "Jj
J:
;§
I.J.J ..r.
0::
a.. 03 p::
Q k
CJ
....0
~
'"OJ
W i:;
!;:( :::l
<0
0::: ll)
X '-'
:::l
::> 0:1
~
-l
lL.
I.J.J
0:: ......
k
OJ
P.
OJ
.~
~
E--i
r..:
bil
i:i:
:il 2 ..
.
:aun~:iu" 0IlfJl....OfU.V
I.J"IX!~ 159l ,uox.IjOI,.(oIAttI~·.UDld.H-u 141 Aq X"uoy 11IIOJ. IDpeu!wJoleOI
SS3N3All0lli3 9NIlVNOu.aI1M;l S31111d
IV. LIQUE1<IF.D GA~l"S AND LmV-BOILI~G LIqUIDS 397
fect has been observed in ordinary distillation. o The maximum throughput

of the packing is stated by Podbielniak (private eomnl1lnieatioll) to be in
the neighborhood of 150 ec. gas/mill. at standard temperat.ure and pressure.
Heli-Grid Packing. Two different types of Hcli-Grid paeking(A and B)
have been described (and patented) by Podbielniak, til as discussed in
Chapter II, and shown in Figlll'c 15 there. The capilbl'Y spaces between
wires cause thp liquid tospl'eaditself evenly over the packing, so that channel-
ing and flooding a~'e minimized. Even if such packing is tilted nearly hori-
zontally, the liquid will not channel but will run spirally lip and down fol-
lowing the ea,pilIary wire coil path. Type A paeking is used only in small-
diameter eoluIllns (up t.o 6 mm.) but Type B may be made in Itn'ger diam-
etors by winding a seeoud spiral around the outside of the inner spiral.
Pnckings of I-in. diameter an~ reported. Both types of packing are Pl'C-
tensioned, to twist during; fabrieation and fit very closely to the eolumn
tubing after t1flsembly. Thifl :mug fit avoids channeling between the pack-
ing and the column wall, due to slight irregularities.
It is cw,tomary to lise a corrosion-resistant wire such as Ineonel or
Niehrome, so that eorrosion is avoided and active cleaning agentR may be
used. The wire sizp is approximately 0.008 ill. in diameter with clearance
betwet~n adjaeent turns of about 0.002 in.
The H.E.T.P. of Heli-Grid pae,king of ll-mm. diameter is reported lO
as 0.2 in. when tested in a l4-in. section withmethylcyelohexane-n-heptane
mixtures at total reflux. The efficiency v!Lries cOllsiderably with the di-
ameter of the paeldng and with the boil~up rate lO as is shown in Figure 7.
Even at high boil-up rates the H.E.T.P. is not over an inch. Figure 8
shows the H.E.T.P. of Heli-Grid packing compared with other varieties
of packillg. 10 The comparison is, however, at about IOOoe. with methyl-
cyelohexane-n-heptane, and not at low t.emperatures.
The range of throughput or boil-up rate of thc Heli-Grid paeking is
given by Figure" 7 and 8. The holdup of various paclcings is summarized
in Figure 8. Podhiellliak has llsed an al'bitmry factor ("Fn,ct.or A") ill
order to express the utility of a paeking. This factor is the hoil-up rate,
in cubie eentimeters per hour, divided by the holdup per theoretical plate,
in cubic centimeters. This involves the major faetors that [lffect the sharp-
ness of separation and the time required for a partieular distillation. Fig-
ure 8 compares various pae-kings with rasped to valum; of "Faetor A."
Simple Wire-Coil Packing. This was the packing used in the early designs of
the Podbielniak low-temperature fractional-distillatioll apparatus. Its nature ll
9 Houston, J\'1.8. Thesis, Pennsylvanitt State College, HJ47.
10 Podbielniak, Ind. Eng. Chem., Anal. Ed., 13, (139 (1941).
11 Podbielniak, Ind. Eng. Chem., Anal. Ed., 3,180 (1931); u. S. Pat. 2,009,81-1;.
398
a.;
r-:
LLi
I
rr:
o
t;
it
>-
<.>
z
!!:!
<.>
E
i*~-!-
Ll.l
'0
ii
OJ
.....
2.0 ----.--I---+_
1.0
'j. -I~J
---J---f---..,\ 5
..;
"
cL
::l
o
...J
o
I
o 100 300 400 500 600 700 800

BOIL UP RATE, cc.lhr.
Fig. 8. Comparison of Heli-Gril] with othor high-eIliuicrtey lahoratory column packings.

(1) Heli-Griel, 11 mm. I. D., 35 em. (14 ill.) (.n Spinuill)l; hllnd, 6.7 mm. 1. D., 5,10 em.
Hcctioll. (f2n) Fenske, single-tum met!ll (21G in.) section.
Iwlkes, UJ8 em. (O.7\) in.) 1. D., 277.5 (5) Bruun lOO-phLte bubble cap, 105 em.
em. (Ill in.) scotion. (i!Jb) Fenske sin- (78 ill.) sectioll.
gle-turll hplicas, 1.55 em. (0.02 in.) 1. D.,
[)f No. 34 IltllillloBS stQel wire, 25.4 mtn.
un BUl'k-Splker, conccntrie tubes, 14 mm.
1. D., IIiO cm. (f\O in.) seetion.
1. D. pnckillg, 100 llllll. length of pal.~ked (7) Stl~dmall, conical ",h'e eloth, 0 ..5 mIn.
section. . I. D., no em. (24 in.) SCl!tiOll.
(8) Punlue, spira1 screen, 1.875 em. (0.75 (8) Urey, rotating cone, 10.5 meter (35 ft..)
ill.) 1. D., '~2.5 em. (17 ill.) seetioll. eolumn.
IV. LIQUEFllm G.\SES AND LOW-BOILIXG I,IQI'IDfi
is depicted in Figun~ O. The (mt~id[J diallleter" reportell runge fWIll 2 ..~ to 7.0 mill.
In sOllie ease~ the elli! <[iamekI' W:l:'i made s(H1lcwhut Ie"" th:llI Uw illside diameter
of the column tube. Thc diameter !lIld pitch 0:1' til(.' coil were adjusted to secure the
greatest cnpillary film effect of the reflux iiqui,l between tlll' cllil turns and between
the packing and tlw tube wall. Thi~ proyiricrl maximum eontact with the ascend-
ing vapors. Precise and optimum dimcl1;iiollB and supporting; test data hitYIi not
heen de~crih()(l in the litcrntllrr'.
Oppositely wound
coil
Outer coil
: ._ Fractionating
lube
(A) (C)
Fil!;. I). Simple wire-('oil lIaekiug: (A) ellitlrgC'd f>L'etioll llh(]\\"ing packing of fi- 01' 7-
turn-pel'-ilwh #20 wire with (·.lose tit iuto 3.H-lllIlJ. I. ]). tnt""; (ll) with eoil diameter
lCHs than imide llianwtt!l' of tulle; (C) with (lppusitd~' wound inuof spiral.
The RE.T.P. on O1le type of this p!wking was repnrterilts approximately 2 inY
but the work of Rosel:! indicates that the H.B-T.P. actually might vary ovcr a wide
range according to the vapor velocity used. The diallwtCl'S of the tuhe, wire, and
coil cerhtillly also exert a major effect on the seplLratiolls achievecl. NIcMilIull 14
reported that almost identical rc;\\llt~ were I)btained in the llhltillation of an iso-
butane-butane mixture regardless of whether the wire roil was present or absent.
The packillg used in his experiments \Va.'; R-turn-per-illeh spirals of 1}22 B & S gage
wire ill [1, 3.8-mm. diameter COIUlllll. He describes an experiment in which a gas
Clmtaining suspended fog partielcs was introducer[ !Lt the bottom while liquid was
flowing down the column. There were two streams of ascending vapor, one moving
along the tUl'nfl of the :;piral, and a larger stream riHing directly up the ccnter.
This showed that there was channeling. McMillan found that addition of an inner
spiral resultm[ in but little impl'Overnent. A~ far as is known, simple wire eoil
packing has been entirely superseder! by more emcient type~, lLllll it is mentioned
here only hecause of it~ llistorical importance in the development of preeisc low--
temperature frtlctional distillation.
l~ PodbielIliuk, "Apl)amtus l1ml Methods for l'wc:ise Fractional Distillation Analys[)s,

VI," Am. Chem. Soc., AtJUlltir. City Meetirlg, lHH, p. 7. t\vl\ibbJe from author.
13 Rose, Ind. Eng. Chem., 28, 1210 (H)311).
14 Mel\!iJlan, J. Illst. Petro/euYn Tee/moL, 22, (iHi (H)36).
400 A. AND E. ROSE
McMillan Wire-Coil Packing. As the result of his exhaustive studies

of tIle distillation of mixtures of exactly known compositioll,\· McMillan
finally suggested the llse of a coil packing consisting of 6-turn-per-inch
#15 gage wire spiral with a #23 gage stmight wire insert to eliminate en-
tirely the center free space. MeMillan stated that the coil should fit the
column tube snugly and the insert should also make a force fit into the
center space. Materials used were brass, aluminum, or Nichrome, ac-
cording to the type of corrosion resistance desired. These paekings were
tOHtecl in t\ 3.8-mm. diameter column 36 In. long. Figure 10, taken from
2385mm
o
Iii
~ -1.0
./ VV ./
o
>
.J
_J
If
~ -2.0
ANALYSIS
w
a:: ~ _Q[L ~
~ -30 Propane 17.40 IS.77 17.19
i:t
Isabulane 41.93 34.17 41.93
eJ
Il. Bulon. 40.67 49.06 40.88
Old/ New
~ -4.0
1-. _,)
-5.0
o 20 40 60 80 100 120 140 160 180
MILLIMETERS DISTILLED
Fig. 10. Com]J!Lrisoll of simple wire-coil p:wking (old) with M(JMillf1ll wire-spirul-und-
insert, packing (new).
McMillan, compares simple 8-turn-per-inch 1122 gage packing with the im-
proved type having the straight wire insert to eliminate the central free
space.
The MeMillan packing is stated to function efficiently at all rates up to
200 ce. gas per minute at standard temperature and pressure. However, no
actual H.E.T.P. data have been published on mixtures boiling either above
or belo'w room temperature. ~o comparison of this packing has been
made with other more efficient packings such as glass or metal single-turn
helices, or the Heli-Grid packings.
Dufton Spirals. Booth and Bozarth 16 have reported the use of a
II Tho procedure for prep!Ll'ing gaseous mixtures of known oomposition has been de-
scribed ill detail by Busey, Barthauer, and Metler, Anal. Chern., IB, 407 (1946), and by
~JaveJ1i, Seyfried, and Filbert, Ind. Eng. Chem., Anal. Ed., 13, 868 (1941).
16l\ooth and Boza!th, Ind. Eng. Chern., 29,470 (1\)37).
IV. LlQUEFIIm GASES .\ND LOW-BOILING LIQCIDS 401
DuHon or Widmer spiraP7 (see Chap. II, Part 1, Fig. 4), which is nearly
the same in construction as the nile Millan packing. A modified form of
Dufton spiral is described by Booth and .l\lcNahncy.18 The column con-
sists of a straight glass tube of 8 mm. inside diameter and an inner glass
tube. of 5 mm. outside diameter and 90 cm. long. The inller tube is sealed
off at both ends, and separnted from the outer by meau" of it wire spiral
consisting of two #26 Nichrome wires, one of which i::; wound ~pirally about
the other. No low-temperature H.E.T.P., throughput, l)()ldup, {)r com-
parative distillatioll data have been pUbliHhed.
Helix Packing. The lise of glass or metal heliee8l~ or "mall glass
triangles as packing for low-temperature columns has been fmggested by
Rose. ~o The glass triangles used bad an H.E.T.P. of approximately 1 in.
and were fragil~~ and difficult to make. Small-size single heli('cs of fine
wire or glass are preferable, hut, no reeord of tests of these at low tempera-
tmes has been published. The very small portable {!olurnns described by
Simom;21 nse. glass ring paeking of this type. Benoliel 22 used 14.5 in. of
3/ 3~-in. single-turll heli(',CB made from fl30 B & S stainless steel wire, and
14.5 in. of a/IG-in. helices of #26 stainless stepl wire in a 1. 38-ill. illside di-
ameter pressure column that tested 100 theoretical plates at toUt! reflux
with n-heptl1ne - methyleyelohcxanc at normal atmospheric: pressure and
temperature. A glass eolumn (8 mm. LD.) with 29 in. of "/61-in. diameter
single-turn helices of #40 B &: S stainless steel wire tested 70 to 80 theorcti.,
cal plates at a boil-up rate of 150 mI. liquid per houl', again when tested
with n-heptane-methylcyclohexane at normal atmospheric pressure and
temperature.
Unpacked Spiral Tubes. This type of contacting device WUH proposed
by Davjs2~ and much more recently by Shepherd 2-1 (see also Chap. II,
Part I, Sect II, 1A). Shepherd duims that, because the length of the
paths of ascending vapor and descending liquid nre equal, an equilibrium
is achieved that is not possible when the vapor path iH shortened.
MisceUaneous Types of Packing. The classified bibliography lists several other

types of packing which have been reported. In none of these cases are there suf-
17Dufton, J. Soc. Chern. Ind. London, 38, 45T (1919).

18Booth and MoNt1bncy, Ind. Eng. Cham., Anal. Ed., 16,131 (Hl44).
19 WilSOll, Parker, and Laughlin, J. Am. C'hem. Soc., 55, 27% (1933). Fenslw, Tong-
berg, and Quiggle, Ind. Eng. Chem., 26, U(\{l (1\134).
20 Rose,lnd. Eng. Chem., A.nal. Ed., 8, 478 (UI36).
21 Simons, Ind. Eng. Chem., Anal. Ed., 10, U-18 (l!~38). R!\Ul,Jcr !wd Simons, ibid., 14,
430 (1942;.
22 Benoliel, M.S. Thesis, Pennsylvania Stnte College, HI41
'3 Davis, Ind. Eng. Chern., Anal. Ed., 1,61 (1\l29).
24 Shepherd, J. Research Natl. Bur. Standa1"lis, 26 1 227 (1941).
402 A. AND l!:. ROSE
ficiellt tldvantag,cs 01' test data for detllilerl r\iscm;~ion. A eOllsidcmble variety of
packing:; have heen systematically tested ill recent investigatioll8 on liquid-air dis-
tillation. However, the:;e data are of limited interest for laboratory rlistillation
columns, because the packings were all of relatively large size, and intenrled for
columns of diameters of 2 in. or more.
B. METHODS OF INSULA'l'ION OF COLUMN PHOl'ER
In ordAr to achieve proper adiahatic operatioll, the eolumn proper is

usually insulated by URe of a vacuum jac:ket of Home kind. In many of
the early types of upparatwl the VaCtlllIll ,iaeket waH mEtde integral with
the fractionating tube,2u but all later' types URe n separate vacuum jacket
through which 01' into which the column tube is inserted,26 tlR in Fig. 11.
The outer dimensions of the column tube must of course closely match the
imler diameter of the innor tube of tho jacket. The Ulmal precautions are
necessary l'ogrl>rding evaetmtion a.ncl hltking out, of finch vaeuum jacketfi.
In some designs the jackets arc unsilvcl'ecl, but it is more wmal to silver the
entire jacket except for vertical st.rips permitting observation of the column
itself. Murti11 27 encountered superheating iLncl Hooding with a st,rip-sil-
vered jaeket, but this has not been reported by others. Poclbielniak 28 has
substituted thin c:yr:lindrieal metal reflcc:tors fo), the silver in order to reduce
heat transfer ac:ross the evacuated jackeL. These l'efieetol's are perforated
in order to permit some visibility of the column. Comparative data on
heat transfer throngh sueh jackets are shown in Figure 12. In early
models the inner jacket was mt1.cle of quartz to avoid breakage when there
was a marked temperature differential between the inner and outer jack-
etH. Recent practice iH to make the entire jacket of Pyrex glass, and to
use a bellows or spiral to allow for differential expansion.
Simom;29 and Rose 20 have secured insulation by placing the entire dis-
tillation apparatus in t1 suitable DewHI' tube. In the Rose apparatus
(Fig. 13) the cold nil' c:oming from the evaporation of liquid air supplied
to the condenset' was cireulated through the Dewar tube: This maintained
a temperature somewhat below that in the column itself. An air jacket
prevented direct contaet between the column and the circulating c:old air.
The more recent improved models of the Porlbielniak column liRe a similar
system in that cold ail' from the evaporation of the liquid air supplied to
the condenser is circ:ulated through the annular spac:e between the column
propel' and the iuner wall of' the vacuum jacket. This tends to compensate
21 Podbielniak, U. S. Pat. 2,OO(),814.
26 Podbielniak, Oil Gas J., 30, No. 46, 69 (1932).
" Martin, Australia Council Sci. Ind. Research, Bull. No. 197, p. 11.
28 Podbielniak, Ind. Eng. Chem., Anal. Ed., 13,640 (1941).
29 Simons, Ind. Ena. Chern., Anal. Ed., 10, 048 (1938).
IV. LIQUHl"mn GASES AND I,OW-llOILHiG 1,lt~Flf)S
for the residual heat IHuk of the VHeullJl1 jackeL, hut may alw hlt ve more
complex effects. PodbidlJiak ao Rtatcs that the ~quin!lon t numher of
plates is markedly inere[L~ed by the introduction of the air circulation
o 20 40 60 fJC KIO IZO 140 140 180 ZOO 220 240 Z&Q 280
g
AVIiRAc;.E OIFI"E'REIICE. IN TIiMP&.A.ATURE e. SIf.TWIlIiI"l
IflSIDIL op. VACUUM JAtKIi'T J> SUP,ROUWOIWc.s
Fig. 11. Podhiplniltk Fig. 12. Heat kansfer through various jackets. Onli-
]ow-tempemture (Jol- nate reprCHentR lellktlge of he lit through walls of VltcUUm
umn with still pot, jlwket, ('xIll'csserl ill 1I'atts (I Iratt = 14.34 I:tll./mill.).
fused to column.
feature, particularly at high reflux ratios. He ahio stateR that be(!tLUSe

the heat lJapacity of the packing and column tube is rebtively high com-
pared with the dist.illing capacity of small diameter Heli-Grid paekings,
30 Handbook Butan8-1'rnpllllc GaRI"~. 3l'il [ILl., Jenkini'! .Publie!ltinI\:'\, Los Angeles,
1947, p. 65.
A. AND E. ROSN .
there is a considerable saving in time of distillation if dry air of room tem-

perature is circulated through the annular spnce during the period when the
last of one component is being taken off. As shown in Figure 14, auto-
matic vnlves substitute the circulation of dry room-temperature air for
FLEXI8LE LEAD
TUBING
CONDENSER-
/
DETAIl. OF UDPER
PAIlT Of CONDENSER
LARGE DEWAR
TUBE WITH
FRACTIONAT-
ING COLUMN
INSIDE
BAFFLE
Fig. 13. Low-temperature eolumll with circulation of cold air

from condenser through Dewar tube eontailling column ..
cold..air when liquid-air feed is cut off and column temperature is rising. 31
BenolieP2 secured a controlled low-temperature environment by en-
closing the column in a vacuum jacket, and winding this with resistance
wire. A hollow copper cooling jacket was then placed around the outside,
31 Podbielniak, Analytical Distillation and Its Aplllicat'ion to the Petl'0le1!m Industry

(available from the author), Chiel1go.
12 Benoliel, M.S. Thesis, Pennsylvania State College, 1941.
IV. LIQUEFIED GASES AND LOW-IlOIUNG LIQUIDS 405
and cold methanol circulated through this at It 1:emperahu'c below that de-
sired in the column. Enough heat was then added hy the resistance wire
VAPOR DISTILLATE
_...... -_ ---------- ---.,
V[.NT \
DELI VERY TUBE I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
SEAL
INLET TUBE
CARTRIDGE"
Fig. 14. Diagram of SUpcl'c()ollow-tempcl'aturo coluIllIl showing cireula-

tion of air through jlll1kct.
so that the temperature on the outside surftlee of the vacuum jacket was
the same as that inside.
40() A. AND E. ROSE
C. CONS'l'HnC'l'IQN OF S'!'ILL PO'!'
M~\llY of the ea,rly designs for ltpparatus had a Btill pot which was at-
tached to t,lw base of the column by means of ~1 stancitml-tltper ground-
glatls joint. 'rhis has the advantage that still pots of several sizes can be
used with ~L single column a,(\cording to the Hample necessary or available.
Fig. 15. I'odbi()lniuk jackoted and ulljackct(l(l still pots and various
sizes of fractioIllLting colllmn~.
Both vu.euum-jacketed and single-wall still pots have been successfully

used. Typicnl arrallgements 28 are shown in Figure 15. In either case, if
the I:ltill pot and columu are fiepamtc pieces, it; is desirable to use a vac-
unm-jacketed standarcl-taper joint at the base of the column, since this
lV. UQU}<JFUm GASES AND LOW-BOILING LIQUIDS 40i
gives [et-;s trouble with leaks at this point. A luhl'imwt with high graphite
(Jontent is recommended. A mercury-seal joint gives further assurance.
'While the use of a joint between column amI still pot it4 practical, it is
preferable that it be eliminated and a fURed glass Heal be u~ed instead. This
has been done as illustrated in }i'iglj.re 11 !lnd more recently a~ ill Figure 14.
The latter arrangement is highly prefemble from the point of view of proper
insulation of the upper portion of the still and its conneetion to the column
proper. Sueh insulation iH absolntoly essential for the smooth operation
and high effieiency nece~sary to separate the various C'l hydrocarbons.
1vlelvlil1an3~ noted that inadequate insulution of UU'l upper portion of the
still pot dcstroyed the effediveness of the lower 15 ern. of the eolumn, since
it was functioning merely to remove superheat from the rising vapor. He
also found that use of a De\vul' tube was not sufficient to correct this dif-
ficulty but that much improvement resulted from a double wrapping of
ashestos cord about the upper still pot phIl:< the use of a tluffieielltiy tall
Dewar flask closed at its upper end with glass wool or cotton. Sueh pre-
cautions arc not n{,CCRS[U'Y when the entire appamtus if:, plll(·.ed ill a lu,l'ge
Dewnr tube, nor when the main vneuum jaeket extends clown to alld below
the bottom of the HtiIl pot..
All the more recent models of eolumlls have an auxiliary opening into the
still pot for entering the Hample, and/or fa!' drainage of residues and for
cleaning (see Figs. 11 and 14). The cleMl spare in the eonneetions for
sueh all auxiliary opening may he eliminated by filling them with mercury
nearly to the still itself. It is possible to pitSS a sample into the fltill pot
through the column, and thus eliminate additional openings. However,
this scheme is successful only when noncondensable gas is nearly or en-
tirely abHent from the sample. Pre-evit(~u!ttion is of course required in this
case.
Early designs of low-temperature fractionating apparatus controlled the
boil-up rate hy adjustillg the level of the Dewar tulle around the still pot.
All recent apparatus nses an electric resistanee coil to supply the necessary
heat input. This is highly preferable, because irregularities in boiling dis-
turb the balance of heat lLnd material How in the column. MeMiIhn 83
reported the marked clifferenee in re;;ults shown in Figure 16, which eom-
pares boiling by means of an electric heater and by atmospheric heat.
Internal heaters have the advantage of eliminating bumping, which
sometimes occurs with external heaters. H(}\vever, the former introduce
two additional metal-glass seals into the apparatus, so that external heat-
ers are to be preferred. The latter allow filling of the still pot with mercury,
which is sometimes desiru,ble. Bumping can be avoided by fusing glass
fibers or coarse crushed glass to the inner wall of the stJIl pot.
a3 McMillan, J.ln.~t. Petroleum Technol., 22,616 (1\)36).
LI08 A. AND E. ROSE
232 mm
0
-1.0 2r:::..--
en 7
~ -2.0
0
;> -3.0
1/ ANALYSIS
Wealh- 8011-
:J
_.J
~
-4.0
~
)
I Ethane
--
True erlng ~
19.40 18.53 19.18
w -5.0 PropoM 27.16 29.09 27.37
a:: II Isobulane 22.41 25.01 21.98
~
-6.0 Bulane
~ 31.03 27.37 31.47
a::
LIJ -7.0
a.
Ii
::;;
w -8.0
I-
~
-9.0
-10.0
20 40 60 80 100 \2.0 140 \60 \80 200
MILLIMETERS DISTILLED
Fig. 16. Oomparison of the effect of boiling by el()etric heater (curve 1) and of "weather-
ing" ((lurve 2).
Most recent low-temperature apparatus uses a cylindrical-shaped still

pot rathel' than spherical, in order to give a better fit into tL vacuum jacket
01' Dewar tube. It ii:l deHirnble to have an extension or finger from the bot-
tom of the still pot, and to apply the heat at this point. Such an arrange-
ment keeps the entire heating section below liquid level and gives uniform
heat input until all but a small fmction of the sample h[1,s boiled away. It
also reduces the possibility of overheating the charge. Podbielniak col-
umns use a l'e-entmnt tube for the location of the heater coil. Typical
arrangements are indicated in the various figures given in the preceding dis-
cussion. It is customary to use It re-entl'ant tube as a thermocouple well,
and to place a drop counter at the bHse of the column or upper entrance to
the still pot.
D. CONDENSEE DESIGN AND REFRIGERATION
The condenser section of every column described in the literature con-

sists merely of a straight tube extension of the column itself. In compara-
tive experiments, McMillan 33 obtained sharper separations when the con-
denser volume was kept to a minimum and some indentations were used
ill the lower part. The indentations have not been reported in other col-
umns but it is standard pnwtice to keep the volume at a minimum.
The more important part of condenser design relates to the device for
distributing the refrigerant. With few exceptions this device consists of a
meta~ cylinder, closed at both ends except for a central longitudinal tube
IV. LIQur':FmD GASES ,\'~D LOW-BOILE\G LIQUIDS 409
just large enough for the glass rondenser of the column proper. Typical
condensers 31 are shown in Figl1l'es 17 and 18. Othel"~ are shown in Figures
11, 13, and 14. It has been found e~::;ential to usc II :,;ystern of baffles to
Clamping Yoke
IA:JJii'l,',J(':"'':T
-.'\_.-.~::.
_1t:',,·r()~I.·.'5~'\~
CE!1.'tt.'r
Nichrome Spiral
CA) (B)
Fig. 17. Typical cOIldl~1l8er designs: (Al still head of Booth and McN[,blleYi (B)
Podbielniak cooling; vel3seL
keep the upper end of the condenser at a lower tempemture t,han the lower
end. It is also desirable to use metal packing such as copper shpt to reduce
34 Booth and McNabllcy, Ind. Eng. C'hem., Anal. Ed., 16, 132 (1944). McMillan,
J. Inst. Petro/emit Tech nol. , 22, 638 (1936). Podbielniak, Oil Gas J., 30, No. 4G, 6n
(1932).
410 A. AND E. ROSE
random fiuetuat-iom; in t.he temperature of the condenser. The alternative

is [t more complex system for supplying refrigerant at a very even rate.
Mal't,in:15 preferred the latter because it improved automatic control of the
(:olUllln prcssure. In the designs of Rose and Podbielniak the eold gas from
the r0irigel'l1nt ill the condenser is passed along the packed section of the
column in order to maintain [I, suitable temperatnre gradient. In other
designs this cold gas is merely vented to the atmosphere. Rose 36 suggested
n, scminutornatie arrangement fo)' supplying refrigerant to the condenRer at
Fig. 18. Typieal eondensel' design: McMilllln liquid-ldl' distributor.
just the right rate to maintain steady temperatures. The liquid refriger-
ant was forced from a Dewar flask by air pressure. The lat.ter was COll-
trolled by escape of an air strelLm through a definite head of water in a
leveHng tube. Podbielniak columnR are fitted wit.h entirely automatic:
devices for c:outrolling the air pressure, which in turn controls the refriger-
ant feed and the condenser temperature. Douslin;11 has described an auto~
matic refrigerant supply and pressure-control device. He has als0 38 de-
signed a condenser that avoids erratic supply of refrigerant and resultant
pressure changes. A similar al'1'angement was used by Martin. Booth
and McNabney 8U used a pressUTe-control apparatus for a purifieatioll col-
umn which intercepts injeetion of the refrigerant at a column pressure
higher than that at whieh injeetion is automatically Rtartcd. The result
is that excess cooling is reduced or eliminated.
35 Martin, Australia Council Sci. Ind. Research, Bull. No. 1£)7, 11.
3! Rose, Ind. Eng. Chem., Anal. Ed., 8, 478 (1.936).
37 Douslin, U. S. Pat. 2,388,312.
38 Douslin, U. S. Pat. 2,379,953.
39 Booth and McNabney, Ind. Eng. Chem., Anal. Ed., 16,131 (1944).
IV. LIQUEFITm ClASW,; AND LOW-BOILING LIQUIDS 4:11
The be~t refrigerallt is liquid Jlitrogen. Liquid ail' alway:> presents some
danger due to the pOSHibility of aceidental mixtnre with f'omhustible ma-
terial and suhsequent igllitioll or explosion. Other refrigerants have been
llsed, hut are less satisfaetol'Y Hille;.;:,; it is certuin that lc)\\'-boiling gases are
not present or lleed not be separated. Podhielninlcl11 and Booth and
Bozarth 41 have described apparatus and procedure for using carbon dioxide
as n refrigerant. TheAe are to he used only for gases with normal boiling
points above about. -80°C. Lower-boiling substances must he treated Its
nOll('olHlellsahle gases. Other articles have deseribcd the 11"e of (:irculating
flllid~ as refrig(·~mlli~, theRe being in turn eooled by an ammonia or similar
rei'rigemtioll coil. Thu;.; Lucas and Dillon·'2 used enId brine; I(istiakowsky
ct al. .1".\ useel ethanol, (tncl Benoliel H methanol as condenser refrigerants.
The WiC of such cooling ;;ystclIl:-i iH limited to HampleH of not too low boiling
point.
B. 1mMmHEMl~N'l' 01,' l'ItODUC'l' BOILIN({ POINT
The boiling puillt of the product of a low-tempemture fractional distilla-

tion is determined by ,t thermo('onple placed Itt about the highest point at
which liquid is present.. In i:iome cases a thermocouple well is uRed. A
liquid-in-glass thermometer has entirely too much hulk, heat capacity,
and sen~iti\'ity, lwd is too limited in mnge to yield anyth.ing like aeCUl'I\,te
retmlts on the ll:-;ually qnite small Ramples distilled.
The readings obtained from product thermoeoupleR such as the above are
used to control the column and are also plotted against. the volume of gase-
ous product or its equivalent. The resulting curve (Fig. 3) forms the
basis for decisions as to t.he composition of the sample being distilled. This
method is entirely sueceHsful when the boiling-point differences of the com-
ponents present are large, as in the separation of methane from ethane,
ethane ft'Om propane, or propane from isobutane.
When the boiling points are relatively dose, or when the nature of the
components is not ecrtain, flpe(~ial preeulltions are necessary in order to
ma.ke the boiling point;; pl'eeise and aeenrate enough to avoid misinterpre-
tation. The rliffieulties in determining boiling points in ordinal'Y batch
fraetionntion nbove room temperature are well known. It is surprising
indeed t.hat. so mudl use haR been made of boiling point.s in low-temperature
40 Podbielllink, "Low-Tempcrat,ufe FmGtioll!Ltion Analysis of Natural Gus and
Similar High-Methane Gnses without Use of Liquid Air," Am. Chem. Soc., Chicago
Meeting, 1933.
41 Boot.h and Bozarth, Ind. EnfJ. Chcm., 29, 470 (1937),
4a Lucas I1nd Dillon, .T. A III. ClwlIl. Soe., 50, 1-160 (1928).
4' I\:istinkowsky et (fl., .T. A m. elII'm. Soc., 57, 87[) (lH35).
H Beuolie!, M.S. Thesis, P(~Jlnsylvl111ia Stal!;' College, 19J L
A. AND Ie. ROSB
distilllltion ~Lpp~Lratus. The N. G. A. A. procedure 4fi directs that the ther-

mocouple junction should be two-thirds of the condenser length down from
the top of the condenser, and the junction should not touch the walls of
the tube or the packing. If a thermocouple well is llsed it must be similarly
... l2'"
:;:~
.....'"
flO
~~
lti~
:IIi •
g •
'" .
......
.....
TOTAL ANALYSIS
TIME:- :3 HOURS
~~~L_~J_-L~_ _~-L~~~__~__L--~ •
..
-140 ·lOO -60 -20 0 20 60 so o.
STD. HYDROBOT TEMP. SCALE ·C.
Fig_ 19. Distillation curves for C 4 hy<il'oCI1l'\lOu mixt,urc.
located t1nd the junction must touch the end of the well. Podbielniak has
tested various devices, including a multiple-junction thermocouple, a spe-
cial design of the upper 8 in. of the column, and a commutating multiple-
thermocouple arrangement. One important source of erratic boiling point
valucs 46 is the instability of the uppermost vapor condensation ring of the
46Natmal Gasoline Association of Amel'iel1, N. G. A. A. Recommended Procedure fol'
Analysis of Satumted Hydrocarbon Ga8e.~ by L01ll Temperal1t.re F1'aclimwl Distillation,
Tulsa, 1946, Pub. 1146.
46 Podbielniak, Analytical Di.~tillation and Its Application to the Petroleum Indu8try
(available from the author), Chicago, pp. 9-11.
IV. LIQUEFn~D GASES AND I,OW-BOILING LIQUIDS 413
reflux. This wanders above and below the junetitm of it single thermo-
couple. Figure 19 (Hclitronic record) illustrates th" results ,dth t.he special
arrangement con"isting of tieveral independent therllluwuple junctions
placed l~t about 3/~-in. intel'vn.ls. A commutation dcvi{'c rapidly couneets
each of these in turn to un explHHlcd nmge temperat1ll'e reeorder, and the
lowest temperature is taken as correct. The Hycl.-Robot record is from a
single thermocouple. It is surprising that there ha~ been so 1ittle chwelop-
ment of other propert,ieto than boiling point to measure the composition of
product at the head of a low-temperature eoillmn.
F. m~l\IOVAL Op PRODUCT
It is nearly univen;ul pmcti(~e to remove product from a low-temperature

column from tlw top of the C011del1tier. When the gaseous product iH col-
lect.ed in an eV::teuatecll'eceiver (see Sect. G below), the rnte of take-off is
controlled by a stopc~oek or valve, which may be manually or nutomatieally
c:olltrolled. In either ease thr. \'a.lve adjustments must avoid sudden
chauges in ('olumn preSimre. For those caseB in which gaseous produc:t is
eollected in a buret over some c~onfini!lg liquid, the rate of take-off is best
controlled by the mtc at \"hich the ('onnning liquid leaves the buret. Thc
pressure of the gas is then the Harne in the buret as in the eolmnn.
Benoliel H describes prmlsure apparatus in whieh distillate is removed as
liquid, after all vapor is eOl)clensed in a total condenser. Simons,11 also
use" a t.otal eOlldenser in a pl'CHSUre eolumn, and the liquid product passes
out through a collection eup and U-tube (Fig. 20). This column has 110
stopcocks and the rate of talce-off is controlled by a heat or on the side of
the U outside the column. This controls the rate of evuporation to an
auxiliary vessel. Rose 4R designed a column with side arm for removing
gaseous produc:t either just below the eondenser, or from the top of the
condenser. Welshl111s·lo compared the two designs and found that take-off
through the top gave somewlmt sharper separation, paxtieularly at higher
reflux rat.ios. He also reported that smooth operation was mueh more
easily attained with product take-off below the conclenser, whercas con-
siderable experiellee and constant attention were required to get good re-
sults with take-off through the top of the condenser. Automatie control
equipment such as is commonly used is eertainly the equivalent of experi-
ence and eliminates the need for constant attention. However, for cir-
cumstances in which such controls are not available and experience has not
been obtained, the pro duet. take-off below the condenser is worthy of serious
consideration.
41 SimoIls, Ind. Eng. Chern., Anal. Ed., 10, 30 (1038).
48 Rose, Ind. Eng. Chern., Aned. Ed., 8, 478 (HJ3fi).
'9 WelshaIls, B.S. Thesis, PennsylvlInia Stltte College, 1939,
4U: A •.\ND .E. HOSB
RegtLrdless of the type of product take-off device, it is important that the

volume of the eonclemml' and the connect,ing tubing to the mttllometel':.l
Hnd prodnet collection bottle be 11S small as pmlsihle. lYklvliUau60 showed
that dC!Ld HpHces of thi~ type caused serious inaeenraeies in analyses for the
Fig. 20. tlimons pressure column.
lightest and heaviest components. In Podbielniak's columns the Heli-

Grid packing extends almost to the upper end of the condenser, even though
the product thermocouples are approximately opposite the lower end of the
condenser. MeMillan recommends the use of Y-type manometers so that
the dead space in the closed end of the manometer can be almost entirely
eliminated. It is adviHable to place a check valve or float valve to prevent
mercury from accidentally entering the column manifold.
~a McMillan, J. Inst. Petroleum Techno!., 22, 621 (1936).
!Y. LIQmmnm CtA[';ES AND LOW-HOlLING LIQUIDS 415
G. COLLIDC1'ION AND MEASurlElIIENT OF PHOD[JCT
Prodllet if) most; frequently eollectecl by pasRillg it in a guscott:-; form into

an cvaeuated reecivel' llottle which is maintained at con~tant temperature
(:1:0.2°0.) by a thermostat. The quantity of produet can be measured
direetly by noting the pressnre in the system, provided the volume has been
determined. In some in:;;tances a seeond, smaller bottle is used to secure
z 30
o
F
:::>
...J
o
III
26
/
~
;;\
o
w
!;i
tr
:::>
!;i
22
7
7
Ill",
trJ:
~ E 18
/
oE
ffi
!i:;:
"-
o
W
tr
14
v
:::>
III
[l ,0 /
V
a::
Q_
a::
oQ_
;; 6
10 18 26 34
TEMPERATURE. ·c.
Fig. 21. Vapor pl"e~S\lreof water over slItul"ntrd sal to
soluti<m.
greater sensitivity wheu small gas samples are distilled. Colleetioll bottles
used range in size from 1 to 5 liters. They should be carefully ealibrated
by det,crmil1ing tho water-weight eapacity at a Imom} temperature and by
mea~Ul'emel1t of the pressure riRe when a known volume of gas is admitted
to the system from a calibmted buret.
It is also possible to collect. the gaseous product in {metions by displacing
a confining liquid in a buret of 100 to 500 ru!. capacity, using a water jacket
held to eonstant temperature. In this cml('. the rate of withdrawnl of prod-
uct is best controlled by the rate of flow of the confining liquid from the
bUl'et. It is eouvenient to have at least two burets so that when one is
filled with product the flow mtty be diverted to the second. Fl'l,1ctions may
415 A. AND E. IWSm
be transferred to bottles or other ;;torage containers) or to auxiliary analysis

apparatus. This proeeciUl'c requires ehoiee of a suitable ('ol1fining liquid
for 11;;(1 in the buret, and herein lies the chief disadvantagr, of this method.
Mercury is eminently suitable because gases do Ilot dissolve in it.. How-
ever) it is too exp(lDsive and heavy to be eDnvenient for use in the rela-
tively large volumes required. Nearly sat.urated Ralt Rollltion eontaining
1% sulflll'ie Heid, or 50% glycerol solution is satisfaetory for somB purposes,
but nse of "ueh confining liquids reqnireH preliminary experiments to deter-
mine solubility corrections or t.o establish th:1t they are negligible. Correc-
tions for the presence of water vapor (Fig. 21) are alsu required in many
eases.
When uistillation is for the purpose of purification l'l1ther than analysis,
product iH advantageously collected as liquid or solid in refrigerated bulbs
[lS described by Booth and Bozarth&l and by Shepherd. 52
H. AUXILIARY APPARA'l'UR FOR ANALYSIS
The flimplest type of low-temperature distillation apparatus depends

entirely on observation of the temperature at the head of the column for
intel'pretation of the results. There arc, however, many instances in which
this is inadequate. In such cafle:-; it is necessary to transfer a portion or all
of each fraetion of gaseous product from the collection bottle or buret to un
auxiliary u,ppnratus. This may be done in several ways. If the fraction is
in [t buret over a confining liquiu it is merely necessary to reverHe the posi-
tion of thc three-way stopcoc.k at the top of the buret to connect with the
auxiliary apparatus, and raise the confining liquid to the desired extent.
If the fraction to be further analyzed is in an evacuated bottle receiver, it is
usually neeessary to u:-;e a Toepler pump of some kind for the transfer. In
other cases t.he Hample may be transferred by distillation to a refrigerated
bulb attached to the auxiliary apparatus. In such instances it is import.ant
that the entire sample be transferred without. partial evaporation or con-
densation that will modify the composition.
Perhaps the most frequently encountered auxiliary is ordinary Orsat
absorption apparatus for the determination of oxygen, carbon monoxide) or
unsaturated sub:-;tanees. Closely related is the special apparatus for de-
termining isobutylene by eombination with hydrogen ehloride63 and buta-
diene by l'eaetion with maleic anhydride. 54
61 Booth and Bozarth, Ind. Eng. Chem., 29,470 (1937).
62 Shepherd,.T. Research Natl. Bur. Standards, 26, 227 (1941).
53 Universal Oil Products Co., U. O. P. Laboratory Test Methods for Petroleum and Its
Products, Chicago, 1947. McMillan, Ind. Eng. Chern., Anal. Ed., 9, 511 (1937).
541Iandboolc Butane-Propane Gases, 3rd ed., ,Jenkins Publications, Los Angeles, 1947,
pp.56-57. See also Tl'Opsch and Mattox, Ind. Eng. Chern., Anal. Ed., 6, 104 (1934).
IV. UQUEFIED GASES AND LOW-BOILING LIQUIDS 417
A gaH-density balunceo[l is a valuable aceessory to any low-temperature

distillation unit. A means of measuring vapor pressure is equally useful
(see Booth and Bozarth51 ). Freezing-point determinations 51 require more
complex apparatus. Benoliel56 describes a mociifiell Spencer Abbe refrac-
tometer for measuring refractive indices down to -60°C. and also special
dilatometers for density measurements to the Slune temperature. There
has heen very limited use of physical properties such as thermal conduc-
t.ivity for auxiliary analysis in low-temperature fl'adionation. Spectro-
graphic methods ~tre well known.
1. APPAltA'rUS Fon SAMPLING AND FOR PRE'l'HEA'rJ\lENT OF SA-MFL!!,
The details just discussed are meaningless unless the original mixture
has been properly slLmpled and tmnsferred into the distillation apparatus.
For instance, stmtification must alwu,ys be kept in mind as [L possible cause
of samples that are not representative. The sampling of completely gase-
ous mixtures is [t relatively simple matter, requiring only the usual precau-
tiolls to avoid contamination from air, or from materials previously present
in the sample container or distillation apparatns. The most; sntisfaetory
arrangement is to connect the duct or gas holder 01' other source of sample
directly to the inlet end of the distillation apparatus, evacuate or purge the
connections, and enter the sample. It is more often necessary to transfer
the sample to a container of SOIlle sort, transport this to the Rite of the ap-
paratus, and then make a second transfer, 'When the transfel' is made
from a duct or gas holder, it is best to use :1 sample container with valves
at each end, so that purging may be thorough. The valves are finally
closed with the sample under slight pressure. Transfer from such a con-
tainer usually requires the use of a Toepler pump, The transfer can also
be achieved by complete liquefaction of the sample in the still pot of the
distillation apparatus, if the sample contains only easily liquefiable com-
ponents, so that the final vapor pressure is negligible, It, is of critical im-
portance in any ease that all stopcocks and connections be vacuum tight.
Transfer from a sample container snch as the above may also be carried
out by introducing a confining liquid into the bottom of the sample tube,
The N, G, A. A.67 suggests 15% sodium chloride in water. Care must be
taken to avoid introducing air which might be trapped neal' the stopcock.
All sampling operations with gafleous mixtures should be made through
precision-ground glass stopcocks or diaphragm valves, and with full recog-
nition of the possibility of leaks when rubber cOIlnections are used.
s, Edwards, Natl. Bur. Standards Tech. Paper 89 (1917).
6G BenoUel, ]\I[,S. Thesis, Pennsylva.nia State College, 1941.
51 N, G. A. A. Publication 1146.
418 A. AND E. ROSE
SampleR of liquefied gas are considerably more difficult to deal with than
completely gl1RCOUS Ramples, because of the changes in composition that
ocellI' by partial evaporation or condensation during handling. It is im-
possible to give specific directions for such cases because of the many cir-
eumstances under which transfer of samples may occur. The essential
point is thut. a liquid Rumple in a tank or pipeline must be t.ransferred into
the still pot without appreciable evaporation. Sample containers should
be of metal with diaphmgm valves at each end, and should meet all safety
regulations for use with the pressures to be encountered. Connecting
lines Hhould be of metal up to the point in tho distillation apparatus where
the refrigerant cools the sample and l'eduees pressure. Pressure gages
should be connected to the cylinder unless pressllres ure definitely known
from the previous history of the sample.
The best way to fill a sample containel' iH first to evacuate the container
and then COllnect it directly to the source of the sample. In general the
fil'Ht portion should be diseal'ded as not being representative. The above
evacuation procedure CaURGS the sample container to be completely filled
with the liquid Rample. It is therefore best, after disconnecting the con-
tainer from the connecting tubing, to allow a few milliliters of liquid sample
to escape from the lower valve. Thii'! will avoid leakage or bursting of the
eontainel' due t.o expansion of the liquid from a rise in temperature subse-
quent to the completion of the sampling operation.
A sample container may also be filled by allowing the sample material to
flow in one valvealld out the other to purge the container and connecting
lines. Safe and Imitable disposition of the effluent must of course be pro-
vided. Direet filling; without evac.uation or purging is sometimes used,
but always results in some cont!lmination due to air.
The preferred method for transfer of a liquid sample into the still pot
of a column is to UHe a Hmall-diameter metal tube passing through a (looling
bath to a suitable valve. Before refrigerant is placed in the eooling bath,
the line is pUl'ged and filled with liquid sample. With valve dosed, the
line is then attached to the column, and the connecting tubing and the
column evacuated. Refrigerant is then applied to the (~ooling bath and to
the column still pot. When t.he connecting valve is opened the sample is
cooled sufficiently in the cooling coil so that it flows to the still pot as liquid
without appreciable evaporat.ion. It is of (Jourse necessary to avoid use of
excess refrigerant, which would freeze the sample, restrict the flow, and
eause composition changes. Simpler methods of transfer are often useJ
without serious errors, but these require good judgment as to the magni-
tude of the IosHes or dl!lllges in composition.
Any low-temperature fractionating apparatus should have the inlet
sample line equipped with a pressure gage and an arrangement for removing
IV. LIQUEFmD G,\Sl~S AND LOW-DOU.•I1\(l LIQUIDS ,UO
moisture, ('arbon dioxide, hydrogen sulfide) or other ga"l'" that will solidify.
The preSClwe of water can abo canse erratic hehavior and poor Reparation
of the unsaturated C:I and G1 hydrom11'hons due per]:Utps to tho fOl'IUlttion of
hydrates. Podhieilliak58 hus stated thut snch "hydmte~" M'C not neces-
sarily decomposed hy any drying agents.
Before choosing a drying agent for an unkllown sample, preliminary ex-
periments may be needed to make sure there is not a eomplicating reaction
or preferential adsorption. Thu:.; olefin gases arc polymerized by phos-
phorus pentoxide. Absolute ethanol can sometimes be uHed for the dual
purpose of chaser and dehydrating agent. However, it dom; form azco-
tropes with pentaIles, and interferes with their Reparation from one another
and from hexancs. Other alcohols arc free from this disadvantage and will
also remove water. The problem of hydrate removal is complex and still
not clearly understood (Podbielniak, private communication). Solid ab-
sorbents are preferable to liquids, although solut.ions of potassium or
sodium hydroxide have been used to absorb carbon dioxide. Ascttrite is
sa tisfactory for this plll'pose. \Va t.er vnpor may be removed with eaieillm
chloride, sodium sulfate, calcium sulfate, 01' phosphorus pel1toxirie. The
last cannot be used with gases containing ole fins , aromatic hydrocarbons,
or naphtheneH. Podbielniak reports that ealcillm ehlol'ide absorbs ole£ns
and that barium oxide is tlH~ best absorhent. Magnesium pCl'chlomte hns
also been used as a dehydrating agent.. The N. G. A. A.57 suggests use of
calcium chloride, Ascarite, and Drierite in thiR order for purification of
saturated hydrocarbon gases prior to liquefadion and distillation. The use
of indicating varieties of drying agents to give wurning of exhaw,tion is to be
recommended.
Burrell and Guilcl5H have described the use of a Halt-activated ehareoal to
lLdHorb higher-boiling hydrocarbons from a sample prior to distillation.
These higher hydroeal'hollS are then desorbed, eondellsed, [Lnd measured.
2. Procedure
It cannot he too strongly emphasized that smoothness of operation,

meaning absence of sudden changes ill boil-up rates, take-off rates, or rate
of heat input or refl'igemtion, is of m!1jor importance in suceessful operation
of ttny distillation apparatus involving close separations. This is eonsid-
erably more difficult to achieve with low-temperature apparatus than when
operating at and above room temperature. Although use of automat.ic
and semiautomatic controls has greatly reduced the possibilities of sudden
6S Podbielnink, Technical B'ull. No.1 12. Frost [mel Deaton, Oil Ga,~ J., 45, No. 12, 17('
(l94li).
,9 Burrell and Guild, U. S. Pat. 2,399,095.
420 A. AND E. ROSE
changes in operating conditions, it is still very true that experience and

understanding nre essential in low-temperature fractionation. GO
A. GENEHAL OPERATING PROCEDURE
Since specific operating procedures arc a vailablc for use with commercially
supplicd apparatus, and for special types of mixtures, the following dis-
cussion is confined to details of a general procedure. These will be de-
scribed in terms of the hydrocarbons present in natural and refinery gases,
since they are the most frequently encountered and are also illustrative of
the less common g!tseOlls mixtures. Unless otherwise stated, the use of a
partial condenser is to be assumed. (}peration is usually at or near atmos-
pheric pressure, but it is frequently necessary to distil at, reduced pressure
when eomponents are eneountered which boil above or slightly below 01'-
Jimtry room temperature.
Preliminary Operations. Preliminary operations include the cleaning
and greasing of fltopeoeks, the introduction of clean mercury into man-
ometers, and repbcemont of reagents for removing water and carbon di-
oxide from gas samples. Other steps include checking the thermostat
bath temperat.ure, the thermocouple eold junction, and various electrical
equipment. Any containers [or storing product fractions should also be
c:lcaned and prepared for use. It may occasionally be necessary to clean
the column paeking itself, either by disassembly or by a suit.able solvent
or reagent. It should be noted that if alcohol has been present it is danger-
ous to dean with nitric acid, since a fulminate may be formed.
The preliminary operation of gl'e!Ltest importance is a careful check for
leaks over the entire apparatus. This is best done by evacuating the
apparatus and letting it stand overnight or for at least an hour before use.
Thore should not be a readable increase in pre$sure during the test period.
The use of a low-ratio McLeod, Pit'ani, or other similar gage greatly reduc~s
the time required for the leak test. .
Entering the Sample. The various· operations involved in ente~ing , .
the sample have alre:v\ been discussed in connection with the descl'iption
of the apparatus. O:df:l' types of apparatus often required introdli~tion of
sample through the top of the c:olumh and this was difficult when there was
a considerable proportion of low-boiling gas present. Modern apparatus
introduc:es the sample directly into the still pot, so that 110ncondensable
gases pass up the column and are collected in the produc:t receiver as al-
ready described. If there is any possibility that a gaseous sample has
undergone partial condensation whil0.c6nfined to a sample container, the
80 Savelli, Seyfried, and Filbert, Ind. Eng, Chem., Anal. Ed., 13, 868 (1941).
IV. J.IQUEFIlDD G:\SIlS AND LOW-BOILING IJIQUIDS 421
entire container should be warmed to 15 DC. above the temperature of

collection of the sample before allY part is removed to the columll.
In general the size of it sample should be as small us is possible, since this
will reduce the time required for the distillation. However, it is doubtful
if less than 100 ml. (gaseous) of 11 component can be detected with assur-
ance, and certainly such small amounts eallIlot be estimated quantita-
tively.
There should be a separate pressure-inclieating device on the inlet sample
line, and this should be checked during and particularly at the end of the
sampling procedure to make sure its pressure reading is substan tially the
same as the manometer conneeted to the head of the column. .There is
always a possibility thnt a plug of frozen water, carbon dioxide, or even
butanes (freezing point about -12.5°C.) may for111 in the upper part of the
coluJllll. Such a plug is especially likely to oecnr with samples eontaining
air and 0 4 hydrocarbons without lower-boiling hydrocarbons. A frozen
plug nan of cmURe cause trouble due to riRing pressures as the still-pot
contents warm up. The plug not only bloeks the exit from the lower por-
tion of the column, but prevents any indication of the rising pressure on the
column manometer. Difficulty from this score ean be avoided by allow-
ing the stopeocks and valves connecting the still pot to the sample-line
manometer to remain open for a short time after the sample is entered,
until the absence of a plug is assured.
Establishing Initial Reflux. Venting of N oncondensable Gases.
The procedure for entering the sample involves cooling the still pot suf-
ficiently so that pressure in the system remains less than atmospheric.
If methane or any lower-boiling gases are present, it will be difficult or
impossible to keep the pressure down. In any case refrigerant should be
applied to the condenser simultaneously with its application to the still
pot. If the condenser is cooled to liquid-nitrogen temperature, the methane
will condense. A continued rise in pressure as ~noJ'e sample enters, with
the condenser at or near liquicl-nitrogen temperature) indicates the presence
of appreciable nitrogen, oxygen, hydrogen, or othm' kt,w;-boiling gases.
The~proper procedure is to allow these noncond.i.;r!~~l?le gases to escape
slmvly l!,lto a product receiver. However, this ml.l"t;'bt) clone only while the
condenser is kept at ~onstant low temperature (b(,~io\v -:-.-170°C.) by con-
tinual application of refrigerant. It is also essential to allow the still pot
to W1;Lrm up somewhat, so that some reflux liquid is for~ed in the condenser
and flows down the column. .This reflux should be visible for at least half
the length of the column and pressure variations should not exceed 50 mm.
from atmospheric. Under these ,conditions none of the higher-boiling
hydrocarbons (ethylene, ethane, or other component:-; of similar or higher
boiling IJoint) will escape wi.th the noncondensable gases and only a mini-
A. AND E. ROHE
mum of methane will be carried with them. Starr et al_6ll'eported that as

much as 250 cc. per minute could be entered into a Podbielniak Hyd Robot
column without contaminating the noncondcnsable gases with ethane or
ethylene. Their samples contained 50 to 75% noncondensable gases.
Product should not be removed unless the pressure is above atmospheric.
The completion of the removfl,l of this first fraction will be indicated by a
tendency for the pressure to drop in spite of continued boiling in the still
pot. When it is apparent that the pressure will no longer rise above at-
mospheric, the rate of refrigerant supply should be reduced slightly so that
the temperature of the condenser rises to the normal boiling point of
methane. This marks the end of the Iloncondensable fraction. If fur-
ther information on this fraction is desired, it must be transferred to an
auxiliary apparatus for density determinntion, Orsat analysis, or similar
operations. If no further information is desired it is merely necessary to
note the receiver pressure or the volume of product collected up to this
point in the analysis.
Separation of Methane Fraction. If appreciable methane is present,
the pressure will rise as the condenser temperature rises after the nOIl-
condensable fraction is removed. If the pressure does not reach at-
mospheric when the condenser is at the normal boiling point of methane,
the latter is presumably absent. In this case the condenser is allowed to
warm up until the pressure cloes rise to atmospheric.
If methane is present, condenser temperature must be closely controlled
to check the rising pressure. After the column is operating smoothly with
reflux visible throughout its length, product should be withdrawn into the
receiver. During this operation the pressure in the column should remain
near atmospheric and.the temperature at the normal boiling point of meth~
ane. For hydrocarbon-gas analysis the rate of withdrawal may be as high
as 150 co. pel' minute since the separation from the next higher boiling
component (ethylene or ethane) is relatively easy. The N. G. A. A.57
procedure specifies that, if the temperature rises as much as 2°0., take-off
rate must be decreased. Its procedure allows pressure variations of 25
mm. but it is far better if these are eliminated. When most of the methane
is removed, the pressure will have a tendency to drop and the rlLte of take-
off should be decreased to 5 co. per minute.
The still-pot contents should be kept boiling at all times. The column
will probably flood at some time during the removal of noncondensable
gases or methane. If it does not, or if nOllcondensable gases and methane
are absent, the column should be purposely flooded at some time prior to
commencing the more difficult separations. Flooding can be achieved by
81 Stan', Anderson, and Davidson, Anal. ahem., 19,409 (1947).

IV. LIQUEFIED GASES AND LOW-BOILING LIQUIDS 423
increasing the heat input to the still and stopped by the l'Cverse operation.
It is essential that no product be removed during flooding, and that smooth
operation be established for at least 15 minutes after flooding, before any
separation except that of methane from ethane is attempted. In this last
case the interval may be reduced to 5 minutes.
The temperature of the condenser, pressure in the column, and pressure
or volume in the receiver should be recorded periodically during the re-
moval of the methane fraction, and these readings should be taken more
frequently as the end of the fraction is reached.
Separation of Intermediate-Boiling Fractions (C 2 from Ca, Ca from C4).
The general procedure follows that already described for noncondensable
gases and methane. The heat input to the still pot should be measured
with an ammeter, and will vary with the diu,meLer, height, and packing in
the column. For columns in which the still pot is vacuum jacketed, the
heat input can usuully be maintained constant for most of a distillation. In
this conneetion Podbiclniak has shown G2 that constant heat input to the
column causes the formation of hydrocarbon vapor at practically the same
rate for methane to butane, inclusively. Variations in heat input should
be as small, as gradual, and as infrequent us pOHsible since they always
disturb smooth operation and result in poorer separation and waste of
time.
A separation such as that of methane from ethane or ethylene is naturally
very sharp, but even in this c[LIse it is highly desirable to decrease the rate
of take-off to nearly zero (not over 5 ce. pel' minute) as the last of the meth-
ane is removed. This higher reflux ratio will aid sharp separation and also
will allow time for the lower-boiling component to work its way to the top
of the column. As soon as it is certain ;ilmt the pressure in the column will
continue to decrease and remain below atmospheric ifproductis l'cmoved, the
condenser temperature is allowed to riHe, and when column pressure reaches
atmospheric, product l'enlOv[11 is again commenced. Readings are taken
frequently during this transition from one component to the next higher
boiling subHt[tlH:e. Particular (:a1'e is taken to obtain readings a.t and neal'
the midpoint in temperature. In some cas-os it is desirable to change re-
ceivers at this point. This must he done without upsetting column opera-
tion in any way. For more difficult separations it is prefern,ble to change
receivers along a pl[1teau. It is extl'emely important that there be no
sudden changes as the end of [1 plateau is reached. Thus addition of excess
refrigerant will result in decrease in pl'essnre, collection of excess liquid in
tho condenser, and subsequent poor separation.
The sopumtion of (~thylene (b.p. -104°0.) and ethane (b.p. -89°0.)
62 Handbook B1.tane-Propane Gnse.~. 3rd od., .Junkins Publications, Los Angeles, 1947,
Ch:lptor 4.
424 A. AND E. ROSE
is not exceedingly difficult but it is seldom attempted because time can be

saved by collecting the two components in one fraction. and then deter-
mining the proportion of each by Oreat analysis of the entire fmction.
The separation of ethane from propylene or propane is sharp because of
t.he large boiling-point ditTerence, and the same is true of the propalle-
isobutane separation. The propane-propylene sepan\,tion is avoided be-
cause of the nearly identical volatilities, and bccn,use Orsat analysis of fl,
fraction containing these two hydrocarbons is simple. The analysis of
mixtures containing the several saturated and unsaturated C 4 hydrocarbons
requires both a high degree of fractionation and the use of auxiliary methods
of analysis on fractions containing two or three components. Sharp separa-
tion of these C4 hydrocarbons and the preparation of pure samples by dis-
tillation is of even greater difficulty.
The time required for a naturally sharp separation Huch as methane-
ethane may be as little as 5 minutes, but for closer-boiling components an
hour or more may be needed, depending upon the efficiency of the column
and the boiling-point difference. For the easier separations the controlling
factor seems to be the rate of temperature rise of the column itself.
. The N. G. A. A. method63 for saturated hydro(Jarbon gases specifies the
limits shown in Table III in its operating procedure and these are indicative
of those to be used for other caHes.
TABLE III
N. G. A.. A. SPIDCIl'ICATIONS FOlt RATES OF TAKE-OFF IN Drs'rn,r,ATION OF SA'I'URA'l'ED
II YDIIOCARBONS
Max.
I1UownbJo Max. rate
tNnIl. of take-off if
inercase CLhovo temp. Max. allowablo
Component jJl~tea\l reaches limit Pl'e..':iSUro chn.nge
Ethane ....... 2°C. 5 ce./min. ""lOmm. Hg

Propane .......... 1°C. 3 ce./min. *'5 mm. fIg
Is~b~t!1ne} ........
n- u ane
1°C. 3 ce./min. *2 mm. Hg
Pentanes ......... 1°C. 3 en./min. *2 mm.Hg
As soon as all of a lower-boiling component has been removed and the

condenser temperature has reached the boiling point of the next oomponent,
product removal will be possible without decrease in COhUUIl: pressure or
further increase in condenser temperature. The rate of take.~ffof product
may then be increased until column pressure again indieatefj;that' the com~
pOlleut being removed is near exhaustion. In cases in whicli thoro are only
small or negligible amounts of lower-boiling hydrocarbons present in a
53 N. G. A. A. Publication 1146.
IV. LIQUE1!'IED GAS1~S AND I~O\V-nOILING LIQUIDS 425
sample, it is necessary to allow the column to operate at total reflux for

30 to 50 minutes before taking off prouuct. This assUl'cs tempcmture and
component equilibrium before difficult separations are commenced.
There has been some confusion between the terms reflux and reflux
ratio. Strictly speaking, the reflux is the liquid flow down the column,
while the reflux ratio is the relation of the reflux flow to rate of take-off
of product, both being in the same units, either of volume or weight. In
general a high reflux ratio is necessary for sharp separation, particularly
for close-boiling mixtures. The reflux ratio may be allowed to decrease
on plateaus but should be high at breaks in the curve. McMilIan 64 found
reflux ratios of 10 to 1 to 20 to 1 were necessul'Y for most hydrocarbon sep-
arations. Large reflux alone cannot ensure good separation, since it is the
reflux mtio (together with the number of theoretical plates) that determines
the sharpness of separation. In general, the flow of reflux should be kept
at the point at which H.E.T.P. is at a minimum, so that plates are at n,
maximum and reflux mtio can be reduced (and take-off rate increased)
to the greatest extent possible without reducing distillate composition.
Receiver pressure must of eourse always be maintained appreciably less
than column pressure, so that use of a second reeeiver is sometimes neces-
sary. If the pro duet is not needed for further tests, the originall'eceiver
may be re~evacuated during a period when no product is being removed
from the column. Another procedure is to re~evacuate the original receiver
with a Toepier pump, and thus transfer the product to a storuge vessel of
some sort.
A further and impOl'tHnt reason for maintaining low pressures in the 1'8-
eeiver is that at ordinary pressure there are apprceiable deviutions from the
perfect g~tS laws for the higher hydrocarbons.. Corrections for these arc
avoided by using pressure below 100 mm. for butanes and below 50 mm. for
pentanes and higher.
Separation of Fractions Boiling near Room Temperature. When deal-
ing with components boiling much above O°C. it is desirable to conduct
the diRt.illation below atmospheric pressure. Thus in the distillation of a
mixture of normal hydroearbolls, ufter n-butane has been removed in the
manner deflel'ibed in the preceding sectioll, the pressure will deCl'Cafle to a
eomparatively small value if condenser temperature is held at about O°C.
In this case. the condenser temperature should be ~tll()Wed to rise only
enough to 'h~ing column pressure to ahout 350 mm., and the rem(}val of
pentanes cart'led out lLt this pressure. (The N. G. A. A. method directs
that pressnrc~:be redueed well before the n-butane is exhausted, rather than
wilen it is completely removed.) It is of course necessary that reeeiver.
a4 MIlMUl!~n, J. Inst Pct·roleum 'l'echnol., 22, 632 (193().

426 A. AND E. ROSE
pressure be kept still lower. It should not be above 100 mm. for the bu-
tanes and higher-boiling compounds. This is low enough to permit. a rapid
rate of take-off and also to avoid the necessity Jor corrections due to devia-
tions from the perfect gas laws, except when highest accuraey is desired.
Table IV gives values for column pressure and receiver pressure as sug-
gested by Podbielniak6z l1nd the N. G. A. A.63
TABLE IV
COLUMN PRESSURES AND RECBlVER PRESSURES
Column pl'essnre l\.fax. recoiveI' pre~Buro
Hydrocarhon Podhielniak N. G.A.A. Podbielniltk N.G. A. A.
Methane Atmospheric Atmospheric Atmospheric Atmospheric:

Ethane Atmospheric Atmospheric Atmospheric Atmospheri(l
Propane Atmospheric Atmospheric 300 mrn. fIg Atmospheric
Butanes Atmospheric Atmospheric 100 mrn. fIg 400 mrn. IIg
~:~~~~:s}. .......
fIexanes, hexenes
300mm. Hg
100 mm. Hg
300 lllm. Hg
200 mlU. Hg
(i0 mm. Hg
20 mm. Hg
200 mm. fIg
75mm. Hg
Heptanes 20 mm. fIg 5 Hun. IIg 25 mm. Hg
Withdrawal and Measurement of Residue. It frequently happens that.

not all of a sample can be distilled eonveni.ently in a low-t.emperature ap-
parat.us because some high-boiling components are present which cannot
. be distilled even with a good vacuum. Podbielniak6'2 suggests 20 mID.
pressure as the lower limit of successful operation in the usual equipment
for low-temperature dist.illation. In other cases it mi1y not be desired to
take the time necessary to distil components boiling up to 100 0 e. even
though this can be done.
When distillation of these higher-boiling components is not necessary,
the most satisfactory procedure is to use a column having a still pot with a
bottom connection through which the still residue can be drained into a
graduated vessel. It is; of course, necessary to htwe the top of the column
open to the atmosphere when this is done. The sample ean thus be meas-
ured, weighed, and otherwise investigated by such methods as are suitable.
The above procedure involves some drainage and holdup errol', which is
usually neglected, although it may be corrected by suitable preliminary
calibrations. An alternative procedure is to wash the mt1terial from the
column using a solvent that can be easily sepamted from the desired re:-:;idue.
Still another procedure is to use a noncondensablc, inert gl1S to v!tporize the
residue and carry it to a cold trap for collection. The N. G. A. A. pro-
cedure63 directs the evaporation of the residues (eomponents boiling higher
than the pentanes) into an evacuated collection bottle. The still pot must
IV. LIQUEFIIDn GASES AND LOW-DOlLING LIQUIDS 427
be warm and tho final collection bot.tle pressure il:l leR:'; than 25 mm. Ham-
blen and Thorstenberg 6'" suggest the use of a graduated conical tube as a
still pot, so that the volume of tho residue can be measured directly.
Podbiclniak (private communieation) suggests still another excellent
method for withdrawing this residue without appreciable loss of liquid or
vapor, or inaccuracy of measurement, that is, to connect the drain stopcock
of the column (at end of distillation proper) to an evacuated, tapered,
graduated centrifuge-type test tube, with suitable cOllnections, chilled in
dry ice or liquid nitrogen, then to open the stopoock to let residue drain into
this test tube. The top of the column should not be vented to the ail', and
it is essential with this method that there be no appreciable ail' 01' n011-
condensable gases in the column. It wjll be found that the bulk of the
liquid residue will drain promptly into the tube, and that, within a rela-
tively few minutes, the liquid films in the distilling bulb, within the packing,
and in the connections will h:.we evaporated and will hnve recondensed in
the chilled test tube, HS will all vapor propel' in the column. The test tube
is then permitted to warm up to the desired tempemture at which the
volume of the residue can be read accurately.
It is highly desirable to test the apparatus for lcaks after the distillation
as well as before, and, when possible, during the distillation. No noticeable
leak should be evident after 30 minutes at 2 mIll. pressure.
Automatic Operation and Standardized Procedures. It has become
common to use an almost completely automatic procedure for distillation
of hydrocarbon mixtures. The advantages and details of s11ch separation
have been worked out and described by Podbielniak6; and confirmed by
Savelli ct al. G6 An automatic pressure- and temperature-recording appara-
tus has also been described by McMillan. 67 Standardized methods have
been developed for saturated hydrocarbon gases and for the butadiene-
containing gases. The former are given in detail in N. G. A. A. publica~
ti011 1146. The U. O. P. manual~B describes procedures for the gasoline
content of gaseR (Method G-49-40), for analysh'l of complex gaseous hydro-
carbons (G-59-40), and for analysis of 0 4 and C6 hydrocarbons by fractional
distillation (G-162-40).
Errors. Important c[tuses of errors n,re: inadequate reflux ratio or
too rapid rate of take-off, erratic application of refrigerant or any other
disturbance causing fluctuations in the vaporization rate or the pressure
0;" IIumblcn nnd Thorstenberg, Ind. Eng. Chem., Anal. Ed., 18,1.53 (11)46).
au Podbielnink, U. S. PittS. 2,275,648 and 2,3 112,366,
M Savelli, Seyfried, !tud ll'ilbert, Incl. Eng. Chem., A.nal. Ed., 13,868 (1941).
a1 McMillan, U. S. Pat. 2,342,20(i.
68 Universal Oil ProdlH~ts 00., U. O. P. Ln/)()rato1"Y Test ]I.[ethods fo,' Petroleum and
Its PI'oducts, Ohiet.go, 1947.
428 A. AND E. ROSE
in the eolumn, presenee of unrccog,nized 01' unsuspected components,

lc~akagc 01' contamination, and dead-space errors.
Probably the most important cause of errors and difficultics is applying
too much heat to the distilling bulb, thereby either flooding, pl1rtially
flooding, or overloading the packing with reflux. This results in poor
separation inside the column, violent pressure fluctuations, and poor con-
trol of position of uppermost eondensl1tion ring in the distilling tube, hence
poor temperat.ure IIlel1SUl'ement and large errors in int.erpreting the distilla-
tion curve (Podbieiniak, private communica,tion).
Errors due to dead space in the column may be quite important for small
amounts of componcnts unless design and operation eliminate these diffi-
culties. If the amount (as vapor) of the lowcst-boiling component is less
than the volume required to fill the condenser, the column manometer,
nnd t.he connecting tubing, this component will not be detected in the analy-
sis. Such errol' is best avoided by keeping the volume of the critical parts
as small as possible, preferably a total of only 1 or 2 ce. When necessary,
the first portion of product may be subjected to further tests to establish its
true composition, 01' the apparatus may be ealibrated and suitable cor-
rections then made.
A seeolld dead-space errol' arises from the vapor remaining in the column
at the end of a distillation. This is best avoided by using memury to fill
the still pot toward the end of the distillation, thereby redueing the error
to the volume of the free space of the packing. An alternative method is
the use of a "chase)'," i.e., a high~boiling component such as toluene or
xylene. This drives all the lighter components ahead of it until finally the
still pot and column free space arc filled "lith tho vapor of the chaser, and
the entire sample is accounted for. A chaser cannot be used if there is a
possibility that some relat.ively high boiling components are present in the
sample.
Reduction in the pressure in the column reduces the final dead~space
enol'. In some cases the distillation is stopped, and the contents of the
column are completely condensed and tranRferred to a smaller~size column.
B. INTERPRETATION OF DATA
Calculation of Results. The data resulting from a low-temperature

distillation are most frequently recorded as a graph of pressure in the
receiver bottle versus the temperat.ure of the product at the head of the
column. A typical example is shown in Figure 19. When product is
eoUected in a buret the total volume collected is plotted instead of the
pressure in the receiver bottle. In either case the length of the plateau for
a particular component divided by the total length of the curve is the vol~
ume or mole fraction of that particular component in the mixture. When
IV, LIQUEFIED GASgt; AN'D LOW-BOILING LIQUIDS 429
the break is very sharp, as between methane and ethane in Figure 3, the
decision as to the length of a plateau is easy and the resulting precision is
high. When the break is gradual 11 more or less arbitrary method is used
to decide upon the point at which one eompOllellt ends and the next higher-
boiling component begins. Most freqnently used for this purpose is the
mid-point or midte~perature method, by which the cut point is chosen
as that pressure in the receiver bottle at which the condenser temperature
is midway between the boiling points of the two components being sep-
m'ated.
Martin69 has used the boiling points of ethHne--ethylene mixtures of vari-
ous compositions, and the actual di!>tillatiol1 ellrve (boiling point versus
volume of product) to outHin a theoretical value for the cut point. This
was done by using the equation:
7GO - p" XC P,
;t: c = ----- and YB = 760
pc - pa
in which X, is the mole fraction of ethylene in the liquid and Yi! its mole
fraction in the va.por. The vapor pressures of pure ethane und ethylene
~
W
0::
::I
I-
«
a:
w
a.
::;
w
I- 1.0 1.25
ETHYLENE, mole per cent VOLUME, liters
(A) (B)
Fig. 22. (A) Boiling-point curve:,; of ethylene-etlu\.llo mixtul'('~. (13) Dist.ilitttion emVH
Hhowing othyl()Illl-ethl111e bl'Ouk.
are Pa and Pe, respectively, at the tempemtnre in question. The equatiolls

are based on Raoult's law, and give a typical vapor-liquid equilibrium curve
as in Figure 22A. In order to find the proper cut point in the actual dis-
tillation curve (Fig. 22B) , the volume of the break has been subdivided
into segments ilS shown in Table V. The percentage of ethylene is then
read from Figure 22A and the ar:t.Uf11 volume of ethylene computed. The
tot.al volume of ethylEme in the sample ir; thou the portion prior to the break
(1.25 liter!» pluH the 0.14::3 uuleulated itS present in the mixture distilling
during the break. The total volume of ethylene (1.3U3liters) correspondt;;
in Figure 22B to -97.5°C., which is the COl'l'ect cut point.
nn Martin, AustraLia COlLncit Sci. Ind. Research, Bull. No. 197, pp. 20-21.
4.30 A. AN"D E. HOSE
TABLE V
CALCULATION o~' CUT POINT BY METHOD m' MARTIN
C,H,in
Volume, segment.
liters Mean telllp.) °C. "a[o. %
1. 250-1. 275 -103.7 nO.3

1. 275-1. 300 -103.5 n8.3
1. 300-1. 325 -103.1 n6.8
1. 325-1. 350 -102.3 03.0
1.350-1.375 -100.3 83.3
1. 375-1. 400 96.4 59.2
1 . ,100-1 . ,125 92.0 27.0
1.425-1.450 !l0.3 11. ,t
1.450-1.475 89.0 0.0
'J'otal fat aU segments. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . a .14,"l
Total volumo ethyleno ........ , ........ 1.250 + 0.143 = 1.393 liters
Volume C2 fraction .. , .......................... . 2.199 liters
1.393
Percentage ethylene ........ ' ........ 2. 199 X 100 63.4%
True cut point ..... ' ... , . , ............ , ....... . -97.5°C.

---------------------------------
Podbielniak70 has recommended the equal-area method, illustrated in
Figure 23. McMillan/\ however, has stated that the equal-area procedure
gives less accurate results than the midpoint method. He recommended a
procedme72 in which the cut point is ehasen as that temperature itt which a
50-50 mole per cent mixture is in equilibrium in the vapor state. This has
apparently not bee11 generally adopted. The N. G. A. A. method7a
allows alternative use of either a micltemperature cut point or an equal-area
cut point, using the temperatures in Table VI.
The midtemperature and the equal-area cut points usually give the same
result if the fractionation has been properly carried out so that the break in
the distillation curve is symmetrical. The equal-area method probably
is preferable when the break is not symmetrical. The equilibrium cut
point of McMillan usually gives a slightly greater quantity of the lower-
boiling component than the other methods. No thorough discussion
of the subject, with substantiating data, has been published.
Deviations from Boyle's law amount to 1% for propane to 5% for iso-
pentane, at atmospheric pressure. However, if pressures in receiver bottles
are maintained below 400 rom. for propane, 300 mm. f01' butanes, and 100
70 Podbielniak, Refiner Natural Gasoline Mfr., 8, No.3, 55 (1929).
71 McMillan, J.ln~t. Petroleum Technol., 22,616 (1936).
72 Fitch, Natl. Petroleum News, 23, No. 25, 66, 70 (1931).
73 N. G. A. A. Publication 114G.
mm. for pentanes, as generally l'ecommende<l, the deviations are gren,tly

reclnc~ecl anclllo corrections need be mnde. vVhen eorrcctiollS are necessary
they may be made by use of the cOl'l'ection fl1etors 7.1 in Figure 24.
X
=:!
~ 110
ILl
a:
z
::!E 100
=:!
_J
o
u AREA "A"· AREA "s"
l-
e:(
I-
Z 80 _ _ _ _ _ _ _ _ +~
oQ.
DIFFERENCE
~ 70
_J
o
CD
DISTILLATE QUANTITY COLLECTED

Fig. 23. Illustration of temperature miupoint [tnd equn,l-:tl'Cl1 cut-point methods.
TABLE VI. CUT-POIN'l'TEMPERATURES

----_-----------_----_- ------
Column
prBEII'Illre,
Mixture Temp., "C, mm. Hg
---- --------_---_--------_
Methane-ethane ........ _.... . -115 760
Ethanc - propane ........ _ 05 760
Propane-isobut!tne ....... . 28 760
Isobutane-n-butallc ... . (j 760
n-Butane-isopcntane ... . 10 300
n-Bntane-isopenl,ane ... . 1\\ 200
ISOP{lutl1fi!l-n-)lcutauo .. + 7 aoo
- 2 200
n-Pontm!e-heXltne .......... , . +11 200
Precision and Limitations of the Method. Podbielniak76 and Me-

Millan 71 both give figures of 0.1'10 for the precision oflow-temperature dis-
Courtesy Universal Oil Prouucts Co.
H
Podbiolniak, Analytical Distillation and Its Application 10 the Pel1'oICll11! Indl18try
75
(ttvailable from the author), Chicago.
432 A. AND E. ROSE
<.0
'<:l
'Cl
'Cl
......
~o
f3<ll
S
00 '&, :~
., 00 '0
...... :r:<l
:."
o 1... • L'-
'0
:N
, .
'<:l
o
N
IV. LIQUEFIED GASES AND I"OW-BOILING LIQUIDS 433
tillation when properly carded out. In any [Lctual case, the pl'ccision and
accuracy vary widely with the apparatus, procedure, and mixture. Pre-
cisions ranging from a few hundredths of one pel' cent up to several per cent
have been mentioned. Savelli, Seyfried, and Filbert76 state that samples
I HZ
2. N2,CQ
3 02
.., CH..,
~ C02
6 'z H4,HC.1
7 C 2 HZ
e CZH&
9 C 3 HS
10 C 3 HS
II 50Z
12 C" He. (I) C4 He ISO
13 ISO '4 HIO
14 C4 He
I~ C4HB'(Z)
16 HC.., HIO
17 ISO c~ HI2
IS N C5HIZ
1.021--~1-----"--"';----+---".,L..--j,"'::::"-=~~9e:._.--+--:;:=-f--;;:;;;;--i
100 200 300 400 SOO 600 700 1100
MILLIMETERS PRESSURE
Fig. 24. Compressibility (Jol'l'eetion factors for deviations from Boyle's IIIII' (at 25°C.).
containing no olefins can be analyzed in the automatically controlled

Podbielniak Model L coluIllll with a maximum errol' of about :1:0.3%
on any cOIllponent based on the entire sample. McMillan reported a some-
what larger error for a manually operated Podbielniak Precision column.
The U. O. P. test method 77 G-162-40 is stated to have a precision of 0.5%
if there is a minimum of 100 ml. (gas) of the component present in smallest
76Savelli, Seyfried, and Filbert, Ind. Eng. Chem., Anal. Ed., 13,868 (1941).
77Universal Oil Pl'oduets Co., U. O. P. Labomtol'Y '['e8! Method8 Jor Petroleum and Its
Prod1tcis. Chieago, 1947.
434 A. AND E. nOSE
amount. No suitable method has been developed as yet for objectively

evaluating a particular apparatus or procedure.
The limitations of low-temperature fractional distillation have been
listed as follows by Podbielniak :75
(1) The method is cumbersome and lengthy for the determination of a single
component.
(2) The method is inapplicahle or inacclU'ato for sepamtions in which relative
volatility is small.
(3) The method is of limited applicability for components present in relatively
small proportion.
(4) The method is not suitable fo], components Bueh as acetylene and carbon
dioxide which do not exist as liquids at normnJ pressures.
(5) The method cannot deal with substances boiling below liquid-nitrogeu
temperature, nor with substances that decompose during distillation.
Table VII (p. 482) from Podbielniak's work illustrates the reproducibility
and accuracy of analyses of hydrocarbon samples.
C. PRINCIPLES OF LOW~TEMPHHA'l'Um] DIS'l'ILTJA'l'ION
Eose 78 (see Table VIII) has pointed out the importance of relative
volatility and number of theoretical plntes in determining the purity of
distillate for a particular still-pot composition inlow-temperature distilla~
tion. Podbielnin,k70 has extended this reasoning to obtain graphs such as
Figure 25 of the minimum plates at totalrefiux which are required to pro-
duce various distillate compositions, assuming a 50-50 mole per cent
still-pot composition, and various values of the relative volatility. Pod-
bielniak has also used the equution;
ROmln
.
= [ a -
aJ[
xp - - - 1 1
-1--
- XI
J[xvJ[
--
a - 1 XI
( 1
1- XI
)] ,
in which XI represents mole fraction of the lower-boiling component in the

still pot j Xv, the mole fraction of the lower-boiling component in the distil-
late; while R oml•. is the minimum value of the reflux ratio (moles reflux
retul'lled down the column divided by moles product, each per unit of time)
at infinite plates. This relation was used (see Figure 26) to compute
the minimum reflux ratio, at infinite plates, required to obtain product
purity of 99.5 mole pel' cent for various still compositions and relative
volatilities. This showed that minimum reflux is relatively insensitive to
small changes in product composition even near 100%.79
78Rose, Ind. Eng. Chern., Anal. Ed., 8,478 (1936).
7&Podbielniak, "Apparatus and Methods for Precise li'ractional Distillation An(l,lyses,
VI." Am. Chern. Soc., Atlantic City Meeting, 1941. Available from author.
IV. LIQUEFWD GASES AND LOW-DOlLING LIqUIDS 435
To compute the appl'oximate change in minimum reflux for other product purities
than 99.5 mole per cent multiply minimum reflux, as obtained from chart, by the
ratio of desired product purity to 99.5 mole pel' cent, times a correction factor
varying as follows: (a) 0.88 for 50% in kettle; (b) 0.95 for 30% in kettle; (c)
0.965 for 5% in kettle; and (d) 0.998 for 0.1% in kettle.
Table IX and Figure 27 indicate the application of this information to

the separations encountered in hydrocarbon-gas analysis. Figure 27 shows
TABLE VIII78
H.Er,A'l'ION OF RELNl'IVE VOLA'rILlTY AND MINH1UM NU:\!DER OF TIIEORETICAL PLA'l'ES
Vapor Plates
COm!lll. of pressnre in Enrichmont
liquid in still ratio CUhUI111 factol" ,Compll, of produut, %
------------ ------------- -------~----------~~
Butene-l n-Butnne
50% butene-It 1 1.2 54.5 45,5

1.2 59
50% n-butane] 2 1.4<1 41
5 2.5 71 29
10 6.2 86 14
20 39 97.6 2.4
30 240 99.4 0.6
Isoblltanc n-Butl1ne
---- .. _---------._------_._------------------.-----
50% isobutuue} 1 1.6 01.5 38.5
1.6
50% n-butane 2 2.56 72 28
5 10.5 91 9
10 101 99 1
20 12,000 100 o
30 1,300,000 100 o
Prop.no IBobutanc
---- .-.~----.----
,50% pl'opane } 1 3.5 78 22

3.5
50% isobutane 2 12.2 92.3 8
5 520 99.7 0.3
10 5350 99.95 0.05
20 400,000 100 o
30
EthanQ Prop ann
50% ethune } 1 10 90 10
10
50% pl'opane 2 100 99 1
5 10,000 99.99 0.01
10 10,000,000
20
30
Methane Ethane
50% ethane } 1 50 98 2
50
50% methane 2 2500 99.96 0.04
436 A. AND E. ROSE
rate of change of composition of the still-pot contents as distillation pro-

gresses, [LS well as dii:ltillate boiling points, maximum distillate l'l1tes, mini-
mum plates required if totall'eflux i8 assumed, nncl the rate of increase of
total distillate with time. This chart ie; COllHtl'w:ted OIl the aSHumption of
99.5 mole per cent distillate purity except at the actual intermediatefractibilS
between plateaus. Binary mixtures are also assumed in order to simplify
the reasoning and calculations, and holdup is assumed to be zero. Insteaq
of plotting minimum reflux ratios required to maintain the desired distillate
IV. LIQUEFIED GASES AND LOW-BOILING J,lQUIDS 437
,0 RELATION BETWEEN
MINIMUM REFLUX,
'0 ALPHA and COMPOSIlION I-I--I-++~ ..+~~+--I--I--I-+---~-I--H--PI
OF KETTLE fOil. BINARV
MIXTURES WITH CONSTAKT
OV~~HEAD PURITY of <J9.511Ul1.%
Q.J 10- b2 I .' I I I

L'- _ __:. - -2 3
I-- -I- -I-- -I- +--I - - I _..
4 " '" 7 & 9 10 t2.
--+- --!- --!-
!~ 20 ,25
-l-
30
+--
40
f---I -+- I -+-t-
~ 60 10 &0 90 K>O
ALPHA· RATIO' ~~R mnu: W","i1 ~Itmg €§~~§n:.w
Fig. 26. Relation between minimum reflux, relative volatility, and still composition for
binary mixtures with constant distillate purity of 99.5 mole per cent.
,138 A. AND E. ROSE
...,
:z
d
~I
~i
E'
.
E
0 l ~
~ .
~ ~
.. --- .
~ ~
0
~
"''"
~
iii
I
~ ~
~
..
\
w
1\
z
0:
::>
I-
O!S
::l!
I-
...'
oI[l.J
1/1 0
z ~I ~ ~ ~ I>l
.
tI .. ~"aI
1IlII')l/Iw'I
fII ~ N N
)
~
~\!)
<> ..
S
~
0
~
~c
0
co
0:
<!
u
0
a:: ;;
g~ ~~t~
~ ~ ~~~f
~
N
1\ ~
~
is
~
•
\
\
1\
0 E
>- ~I
~
~j
J:
s. 6 .,
~ ... 0
, 0; \ 'iN'IOd'
S ,..0 li! ~ ~ ~
l)N'ntO~
o'~
~I
3lV'l1IUUJ 'lIIWn) 3.lV~
"': "'! I!l Ifl -
N
~
"
E
E
Ii ..
0
~i w
W
0: '" tI"""''' .!;" '"
~
~;~~~~ ~ ~
".
t-
1 5~ -:3
:::J
l-
X
.,.. z '
"
t:) a lillo!
i u !: ~~ -:- ~ ~'" w
-I-
I *2 ;
u Ul ,.. I- 1.11 ~
z
0 g~gia~ ~
0: ~s~@~;
CO
:\"
<! ~
U
0
a:
N
iii •
E
0
>-
J:
."'
~
0
~
o ~ N 0 ~ 'f ~ ~ , , ,
ON'
J. lNIOd t)N11101J 3J,W"ILSIO
1--.1...-,r-- 3
~6
~,. 10" ...... ~. ..lNICd ~N1110Q 3!"111S.U~

( - - I 13<.', Jl'.'''' AY IllnNlII • l"'l
lHW/,uUVM l""",.&SHI,"""")MI ONI11/J.!i1O "JOJ.
IV. LIQUEFIED GASES AND LOW-BOnING LIQ1JIDS 439
purity as the still charge becomes exhausted of a particular component,

the graph plots the corresponding actual distillate rn,te, as~uming the usual
values for rates of boiling corresponding to the heat iuputs indicated on the
graph.
TABLE IX
MINIMUM PLATES AND REFLUX SPECIFIED SEPAR!l!rION OF VARIOUS BINARY
RA'l'IO FOR
HYDROCARBON rAmS"
Binary Vapor
hydrocarbon preS9ure
mixture ratio A B C D
b
Methane-ethylene 61.2 1.32 1.82 0.03 0.1H4
Methane-ethane" 107.0 1.14 1.60 0.018 0.093
Ethylene-ethane 2.3 6.5 9.2 1.60 7.H5
Ethane-propene 8.1 2.52 3.57 0.27 1.4
Ethane-propane 9.3 2.38 3.3(j 0.23 1.2
Propene-propane 1.3 20.2 28.6 6.65 33.0
Propane-isobutane 3.3 4.G 6.5 0.89 4.5
Isobutane-n-butane 1.5 13.0 18.4 3.99 20.0
Isobutane-isobutene 1. 20 28.8 40.8 10.0 50.(1
Isobutene-butene-1 1.03 170.0 249.0 05.5 328.0
Butene-1-n-butane 1. 23 25.5 36.0 8.98 43.5
n-Butane-isopentane 2.8 5.3 7.5 l.lG 5.8
Isopentane -n-pcn ~ane 1. 33 18.G 20.3 6.25 30.1
n-Pentane-n-hexane 3.0 4.9 0.05 1.04 5.25
n-Pentane-diisopropyl 2.15 7.0 g.O l. 75 8.7
a A = plates required at total reflux for 99.5 mole per cent overhead from 50-50%
kettle composition. B = plat.es required at total reflux for Hg.5 mole per cent overhead
from 10% low-boiling component in kettle. C = minimum reflux l'a~io for 100 mole pel'
cent overhead from 50-50% kettle composition. D = minimum reflux ratio for 100
mole per cent overhead from 10% low-boiling component in kettle. Minimum reflux
ratio (at infinite plates) is relatively insoIlsitive to small ohanges in rlesired overhead
purity especially for low values of CI and smu,11 cOIl(Jentl'l1tions of low-boiling components
in kettle.
~ '" = 185 at boiling point of methane.
• '" = 642 at boiling point of methane.
The most interesting portion of this graph is the third section from the
top, showing maximum distillate rates and minimum plates required at
total reflux. The rapid decrease in the permissible rate of take-off as a cut
point is approached is not surprising, since this had been well established
by practical experience, and is in accord with general distillation theory.
The same may be said of the increase in plate requirements as the cut point
is approached, and as closer-boiling mixtures are encountered.
These graphs are the first attempt to establish approximate numerical
values for the proper take-off r!1tes and plate requirements for various sep-
440 A. AND E. ROSE
arations. The values given are not intended to be thosc actually used in
practice and must not be used for this purpose, because of the various
simplifying assumptions made. These are: the assumption of infinite
plates for estimating reflux and take~off rate, the assumption of total reflux
for estimating plates, and the assumption of negligible holdup. The actual
take-off rates recommended by Podbielniak are half those shown on the
graphs. These are based on practical experience rather than thcOl'Y.
2.0
/l;
mZ I- .0
j(_ /
/ ft5
.
I,
'7'
/ ,/
'/ A~
WW
>Z r/
0 0
CDc..
<{:::!:
~O
!flU
.~
r /
1/
Ih V
@5 ~c;;
jif/ # ~/'
"(31 W
~
Ii'
.r,lZ; ~
.t •• LEGEND ~
(!)
V
h If
I.M.'hCl~-llh,rlna
~Z
o::i
_ 0.1
f0 2.JMth."t_ ,,"'M'
3.E'I\:rlfMIO" e'hane
4. EthaM" Pfopy'lnD.
wO 5, Elhal'lD" pfapanl
o:DD V ~. 6.Profl)'I ..... ~an.
::l 1/ '7. Prapan,· 1'0 DUIClnl
I- 0:
.05 a buione- 110
<{WD4 / / 1£ 110 bulln.
0:3: lL_ v: //~ ~

9.110 ~tOf'l'"'' bulOnl
InSulIn, I .. 1'1 butane
~ ~
~V
.03
V
II. N butane .. 110 IMnlon..
L v:;1'
/) ~
Ii!. 1.0 "'",G"'. n Ptrllotll ...
il5 LI.. .02
"'- 13, N PlInian'" driI<lP(ClPJ' ..
h~
/ '*
v.~
1-0 300 (TIm. ab •. prilL
.0 I
.01 .02 .03 _04 .06 0.1 .2 ,3 ,4.5 1.0 3 • 3 10 20 30 JIll
MOL % HIGHER BOILING COMPONENT AS IMPURITY IN EQUILIBRIUM VAPOR
Fig. 28. Theoretical relation between temperature rise above boiling point of lower-
boiling componen~ and mole per cent of higher-boiling component present as impurity.
Similar calculations could be made 80 to obtain reflux, ratio and take-off

rate using actual number of plates, but such calculations have not yet been
published. Theory has not yet developed to the point that holdup can be
taken into account in such calculations.
Podbielniak81 has also calculated curvcs (Fig. 28) showing the percentage
of higher-boiling component corresponding to various boiling points at the
head of a low-temperature column. 'These give an idea of the temperature
variation that oaube tolerated along lJ. plateau during a distillation.
8Q Rose alld Long,lnd. En(J. Chem., 33, 681(1941).

Handbook Butane-Propane -Gases.
SI 3rd ed:, Jenkins Publications, Los Angeles,
1947. ' . ..
IV. LIQUEFITGD GASES AND LOW-BOILING LIQUIDS 441
3. Special Apparatus and Procedure
A. MICRO COLUMNS
Simons82 developed the very small and convenient low-temperature

fractionating column shown in Figure 29. The entire apparatus can be
placed in a one-quart Dewar tube, and is easily moved about the labora-
tory. No elaborate controls are required.
The capacity is about 5 cc. liquefied gas.
The column is not intended for difficult
separations, &ince the height of packing is ~
only a few inches.
A column of the Podbielniak type has also
been built with very small dimensions by
Fitch. 83 It is designed to be used in con-
junction with a larger column in order to
reduce dead space and holdup when dealing
with small samples. Construction and
operation are similar to larger columns.
B. TOTAL CONDENSER COLUMNS
The Simons column differs in one impor-

tant respect from those described in the pre-
ceding sections, in that it has a total con-
denser rather than a partial one and gaseous
product is withdrawn from the bottom rather
than the top of the condenser. Rose78 de-
scribes a flexible total-condensing low-tem-
perature column. A Dewar tube longer than
that of Simons is used, so that better separa-
tion can be achieved by suitable lengthening
of the packed section of the apparatus (see
Fig. 13). This column is operated with the
top of the condenser open to the air, and
product is withdrawn as gas through a side . Fig. 211. Simons mi(Jl'o column.
arm ,iust below the condenser. This arrange-
ment eliminates the need for the very close control of condenser tempera-
ture that is characteristic of partial-condensing columns. Welshans 84
compared the two types of operation and concluded that the total-con-
82 Simons, Ind. Eng. Chem., Anal. Ed., 10, 618 (1938). Ramler and SimonB, ibid.,
14, 430 (1942).
83 Fitch, Natl. Petrolcum News, 23, No. 25, 66,70 (1931).
M Welshans, B.S. Thesis, Pennsylvania State College, 1939,
442 A. AND E. ROSE
densing type was particularly useful when only iuexperienced operators

were available, as in a laboratory where only oecasional low-temperature
distillations are performed. The Rose column was the first to emphasize
smooth operation as a requisite for suceel::ls in difficult separations. This
was achieved by using a completely insulated still pot and n high-heat-
capacity metal-packed condenser, and by ensuring constant pressure by
having the top of the condenser open to the air.
C. OTHER SPECIAL COLUMNS
A variety of other special low-temperature columns have been described.

Kistiakowsky et al. su and Lucas and Dillon86 have devised apparatus
particularly suitable for separation and purification of considerable quan-
tities of materials boiling near O°C. Booth and co-workers87 have devel-
oped apparatus for purification with particular emphasis on accessories
such as vapor-density and vapor-pressure measurements. Bosschart88
has recommended inverted distillation for the determination of hydrocar-
bons in natural-gas samples. The range of his column is from +200°
to -170°C. Components are colleeted as liquid in the still pot, the high-
est-boiling component first. Askevolcl and Agruss 8Q have used an auxiliary
column for separu,ting small proportions of gaseous substances from
petroleum. The column is packed with pellets of sodium hydroxide.
These remove water and hydrogen sulfide when the volatile portion of the
sample is distilled through the auxiliary column to the still pot of an ordinary
low-temperature column. Podbielnill,k90 has described apparatus and
procedure for bringing a sample to equilibrium in a column at total reflux,
using such a small amount that there is practically no material remaining
in the still pot, and basing the analysis on temperature measurements made
throughout the length of the column.
D. PRESSURE COLUMNS
The petroleum industry regularly carries on the fractionation of hydro-

carbon gases which are maintnined in the liquid state by pressure alone.
This has the advantage that low-temperature refrigeration and insulation
problems are eliminated, and ordinary water can be used for condenser
Hi Kistiakowsky et aZ.,.T. Am. Chem. Soc., 57, 65, 876 (1935).
86 Lucas and Dillon, .T. Am. Chem. Soc., 50, 1460 (1928).
87 Booth and McNabney, Ind. Eng. Chem., Anal. Ed., 16, 131 (IBM). Booth and
Bozarth, Ind. Eng. Chem., 29,470 (1937). Booth and Stillwell, .T. Am. Chem. Soc .. 56,
1529 (1934).
88 Bosscbart, Ind. Eng. Chem., Anal. Ed., 6,29 (1934).
sa Askevold and Agruss, Ind. Eng. Chem., Anal. Ed., 17,241 (1945).
90 Podbielniak, U. S. Pat. 2,377,900.
cooling. The equipment must of course be constructed to withstand the

considerable pressures that are encountered. The lower-boiling hydro-
carbon gases cannot be handled because of the excessive pressures neces-
sary. Industrial liquid-air distillation appamtus also operates under some
pressure, although refrigeration is used in this ease to avoid excessive pres-
sures.
Benoliel91 describes a stainless-steel laboratory pressure column with
still-pot capacity or 16 liters for distilling C 4 hydrocarbons without refriger-
ation and using ordinary tap water for cooling the condenser. This can be
operated up to 200 p.s.i. and 200°C. The column itself is a 1.38-in. I.D.
stainless-steel tube packed with stainless-steel helices.
SimollsU2 describes an all-glass column made without stopcocks, stated
to be useful up to preSi.me of 5 atmospheres. Pressure control is automatic,
and product is taken off as liquid, which is revaporized at a controlled
rate.
n. SIMPLE DISTILLATION
1. Repeated Simple Distillation
Simple distillation of liquefied gases is similar to ordinary simple distilla-

tion, except for the use of refrigerants and the necessary avoidance of leaks
and contamination. These special features inevitably make the apparatus
and procedure more complex. "When Ramsay and Travers 98 made their
study of the argon group and other gases about 1900, their equipment was
essentially two bulbs refrigerated to different temperatures. But in general
the early history of the distillation of liquefied gases consisted in the deve1-
opment and use of elaborate apparatus and processes for repeated simple
distillation as a method of analysis and purification. This means the
initial separation of fractions by simple distillation, and the subsequent
recombination and redistillation until reasonably pure fractions are ob-
tained.
Beginning about 1915 there was developed apparatus04 with a multiplicity
of distilling and condensing bulbs, eonnecting tubes, and accessories to
permit storage, combination of samples, repeated distillation, meaSUl'ement
Ql Benoliel, M.S. Thesis, Pennsylvania. State College, 1941.
92 Simolls, Ind. Eng. Chem., Anal. Ed., 10,30 (1938).
93 1'ra.vers, Experimental Study of Gases. Macmillan, New York, 1901, pp. 213':"223.
D4 Burrell, Seibert, and Robertson, U. S. Bur. Mines Tech. Paper 104 (1915). Shep-
herd and Porter, Ind. Eng. Chem., 15, 1143 (1923). Shepherd, Bur. Standards J. Re-
search, 2, 1145 (192\). Shepherd, J. Research Natl. Bur. Standards, 26, 227 (1941).
Tl'opsch and Dittrich, BrennstojJ-Chem., 6, 169 (1925). Tropsch and Mattox, Ind.
Enll. Chem., Anal. Ed., 6, 235 (H)34).
444 A. AND E. HOHE
of volume, pre~sure, and vapor pressure, and other propcrties. Relatively

low pressures are used bemtuse relative volatilities arc greater under such
conditions. Excellent results can be obtained and only small samples are
required. It is essential that leaks and contamination be rigorously pre-
vented and a considerable number of redistillations are required even for
easily separable mixtures. Separation of closfl-boiling components is
generally recognized to require a prohibitive amount of time. As a result,
there has been little reference in recent literature to the use of repeated sim-
ple distillation. It is recommended only for very eaRity sepamble mixtures
when it is not worth while to use the more complicated fmetionnting column.
TABJ~E XU.
SEH'ARA'l'ION TEMPERATURES AND CUT'rING PRESSURES EMPLOYJilD
WITH WARD HIGH-VACUUM ApPARATUS
Condens~r temporature °0.

j
l':r;cflsHl'e,
SCPll.ration mm.Hg: No.5 No.4 No.3 No.2 No.1
Methane-ethallll ..... 0.25 Liquid-air -175 -160 -142 -128

temp.
Ethane--prl)pane ..... 0.1 " -HiO - 143 -128 -113
Propane-butane ..... 0.1 " -142 -128 -113 - LOO
Buta.ne-pen tane ..... 0.1 " -117 -103 - 90 - 77
Ward's a ppara tus95 is typical of those operating at very low pressure to

take advantage of high relative volatility. His is a modification of the
method described by Shepherd and Porter. He uses a series of five dis-
tilling and condensing bulbs, No.1 being mnintained at a relatively high
temperature, while No.5 is n,t the lowest temperature. Only the lowest-
boiling component passes through the train, and others are condensed out at
intermediate points. Typical temperature:,; and pressures employed are
shown in Table X. The lightest component is gradually removed and
transferred by diffusion and Toepler pumps to a buret, measured, and then
subjected to Orsat analysis. Higher-boiling components are treated simi-
larly. The Ward method is especially suitable when only small samples are
available. A minimum of 150 ce. gas is required, and 110 attompt is made to
separate relatively close boiling mixtures such as butane and isobutane.
Savelli, Seyfried, and Filbert96 have described further details of such a
procedure and conclud('l that it is comparable in aeeuracy and somewhat
faster and more amenQ.hle to standul'tlization than low-temperature frac-
tionation with a COlUm!l.
9. Ward, Ind, Eng. Chem;, Anal. Ed., 10,109 (1938).

~6 SavelIi, Seyfried, and Filbert, Ind. Eng. Chem., Anal. Ed., 13,868 (1941).
IV. LIQUEFIED GASES AND LOW-BOILING LIQUIDS '145
2. Single Simple Distillation

The use of a single shnple distillation has been developed for routine
malysis of samplei'J whose eharader docs not vary widely. Considerable
-42 o % ~UTA~E
1%
a"/p_
-40
... 3"1•
---
Q)
~ -38 4%
~ ~%
~ -36 r--~
.
<..i
«
:I:
tJ -34 I - t : i
6%
7% -, -I--
-I-
~
8% -,
0: ~ ~
=f:::::-~ =-
::l Q
~ -32 !!1 I- ~ ~ -;,t
~ l-
"' 0
,..,<n
- -
0: N
L1J
~ -30
I--
9·'" '"
L1J I--
~
I-
-28
r-- t-,,_
I-
-45 -47 -49 -51 -53 -55 -57 -59 -61 -63 -65
TEMPERATURE, ·C., 10% over
Fig. 30. Empirinal curves for analysis by Rimple distillation. u7
A 8
-31 -34
<..) -37 -40 _"..,. /./'

• V
W
0:
:::> -43 -46
)---: ~ ~
~
!;t
0:
l:e -49 -52 .,/ 'i
/ Y
::l!i
w
I-
-55 -58
~
/
-61 -64
v
o 80 160 240 320 400 480 560 640
AMOUNT OVER,mm. Hg
Fig. 31. E;.,.'perimental simple distillation curves. 111
preliminary work with more elaborate apparatus, or with known mixtures,

is required to standardize such methods properly. Rosen and Robertson 9'1
n1 Rosen and Hobertson, Ind. Eng. Chem., Anal. Ed., 3, 284 (1931); ibid" 6,12 (1934).
Happel and Hobertsol1, ibid" 6,323 (1934).
446 A. AND Ie. nOSE
have described the simple distillation of three~eomponent hydrocarbon

mixtures. The proeedure if> very Rimilal' to an Engler distillation, and the
resulting distillation curve is intcrpreted in terms of cmpirical curves based
on prior experiments with known mixtures. The method may be illus-
trated by reference to Figuros 30 and 31. In Figure 31, which al80 indicates
the reproducibility of the method, the temperaturc for 90% over is close
to - 38°C., while for 10% over it is near - 58.5 DC. The corresponding
point in Figure 30 reads about 4.5% butane, 25% ethane, and, by difference,
70.5% propane. The method is stated to be relinNe to 0.5% of any com-
ponent. Other graphs can be constructed for the composition ranges
encountered in routine analysis. Corrections can be mnde for the presence
of olefins or methane.
Echols and Geluso 8 describe the semimicro analysis of multicomponent
hydrocarbon-gas mixtures by simple isothermal distillation. Samples of
2 to 50 mI. are used. The sample is vaporized nnd a series of observations
made on the residue of one or more colligative properties such as vapor
pressure, dew point, refractive index, or density. The Rayleigh equation
is used to relate these, as for instance still vapor pressure to still composi-
tion. A curve of vapor pressure versus pel' cent vaporized is then obtained.
Determination of the slope of the curve serves to fix the composition of any
three-component system, while additional slopes at other points permit
determination of composition of mixtur~s with more than three compo-
nents. Experimental results are not close enough to determine more than
four components. If five or more components are present, the pl'oeedul'e
is repeated for two or more temperatures. Zook, Oakwood, and Whit-
more0 9 have deseribed a similar procedure, applicable to simpl~r mixtures.
08Echols and Gelus, Anal. Chern., 19, 668 (1047).
99Zook, Oakwood, and vVhitmortl, "AnalysiS of HydroclLrbon Gas Mixtures by Iso-
thermal Distillittioll," Division of Analytical Chemistry, Am. Chern. Soc. Meeting,
Sept., 1944.
·~.10Ut '{ISU'I ~
'ff,TO 'oslW ~}
'pda tI ~
'p{IO S ~
'[ldo a~
'pda T8 ~
N '.HJOUI ~
x x x x
I.t~IV !:; I'
3UGlUll.dUN ~
'019 '~HClO ~
uUIJIOICI ~
'Al~~rot~ ~ x
'Ol[ 'p~tlS;:::: IX X
'l)llo I JO JU t";;
'!ldo 1;,)" X
'v<l,) l1::ncM
.10 Ol;).ll ~
'pllD e)fIJI
.10 D~I c:o
'ptlJ .!ll~ x xxx XXXX
(ON) l~SO.l'-HlI
JO lAuoqJuQ "'1'
01'1"0" X xx x
OP!lpaH x
O-}
x
~ualllom ~ X X X X X X X x X X
c.i0 0:
00
<0 10
00
"..;
0: cq tq
10 lQ .-< I ' '0
....
I~
.-< "..; .-<
.... .... '"...
l' 0 ." '" It, ~ "" "', ,...
'-<l 00 t~
<:',
cq
.-<
~
~
1"- 00
~ 0,... 00.....
~
.f.> ,'-'
",i oi r.O I,
<0 ~
ci.
Pi
G'l
I ""I
'"I'
G'l
OJ 00 00 00 <0 10
'".....
.-< .-< .-< .-< .-< .-<
I I I I I I
,.... .-< 0
..... .-<
I
Ol ll')
Ol 01
I I I I "" 00 00 00
I I I
..... ,,-, ,...
0 ,... .... a..... '"' ,'-''"
<0 l'- DO Ol 0 "'G'l1 ~
"" '"' ::.!
.-< Cf,) if) C/O <0 00 0 <0
~
;-; .--< .--< .....
'"' .-< .-< <>1 "'I <>1 <>1
448
':1..0 '~)BtW ~~
'plio il 0; x
'[lIlIlSal
'pdo a ~ X
'pcla 18 ~
X
N '3JO~ x
N 'fiJOUI ~ XI
l'~)nv ~ I X
! )
'Ol0',ll'::> ~ X
lIuolola ~
'AT.lOp JO
uUolO ::: X X
'OIl 'PlUS ::;
I
'plIo I JO .19: '" X , I
'pclo to '" X XI X X
'Plio H;)I:M XI
JOmD,r'"
·Pd".I~M~~ \0 x IX ,X x X X
'plIo ,I '0 X X X XXX ,x X XX X x
(ON) I~SOJnll
.10 It'i:UCHIJ'UD"";' X
x
x
....
~
.0 '2::J
.§ .!!!-
~ 1001'-- IN q e-, "'! INC"! o;a<:q
tt~&iii5 (fJg~ ~?2 c-i c-i ..... 0

",j
<i 00 ~ t-o t- ~ t...
ill I I I I I I I I I I I I I I J
00
~
IX XI ] 449
.~
~
I X i
I X I X IX XI XI
::; X I IX
XI
IX X I
XX >< x X IX
x X XI
X
X x
IX
X X X x
XX X X X I X X X xxxx X
XX X
I X I
X X
x
~OOOlQcqC(l~C-l ~
~ ~f3~~~m~~~~g~fJ~8 .....
•r.>
I I I I I I I I I I I I I I I
450
'l'l.IO '0f1jJ-\I gl 'X
'pI{n (I~
'ptIiJ S g I X I
'IllfO IT ~
'l)(Ii)IS~ I X I
N·.fI·10~ XX
N 'II.lonI ~ IX
lA:3[1V ~ X I X
aumnIldUN::! x
·[)1Q'Zf.[I{) ~ x x
nyGIOT(I ~ X X'
't\I.WP JO
lIU010 ::: x
',HI'Pl1J8;::
'P<Io I .\0 Jfl 0)
X
x X
'vela IlDICl./I
JO OlD . f( t--
X X
~PdJJ[?g~ (0 x
'pd. II '" 'X X XX X X
(DIS!) I~HOJlIU X
JO 18uorpuD "-1-
OPIXO"
apI1l)t{H 01
o
"'I"
", 0
l~ t;..... l~ ""'0 ..... 'OJ< 00
CI':i e-i cq
~~~ NOOC;;
1 I 1 I ""'''''''''
I 1 "'"
1'1'" "'"1 "1
1
o 0
Z t--
'X
I X I
., IX X
X
X
XI X
XI X
X
XX I X I X X X X 'X
XI
xx X XX
X
X x
XX x
X 'X X
" X X X x X X
IX
X X
X I X
"0
~
M
I m
•~
'"
".)
.....
I
~ CI'l l"""I .,....I 1t:) t:.OO~ C!
~ ~~~;?j~ '<!'".)e<!
.-< ......... .... ,..,
00
00
I I I I I I I I I I I
4.52 'fj.lOUI 'OSI1\[ ~
'fJ.lO 'u9n~ ~
°Dt[;) S ~
"lHIJ H 0:.
x I X I
'PlI;) IB ~ x
N 'B,lO ~ X
N "ifJOlII ~ x
l'~J:lIV I~
~ x x x X XX X X
amnl'lt{(l1JN ~ X X
'Ola '!:T:lli[J ~ x
UUlJIQICI ~
'A).lOp .1O
ulIOIO ;:: x X
'~Iq '{1'ltJ8 ~
X
'p(l,-, I .10 .Ia ~
xx X X X
x X X X X
"[lila H;)[O.!I
JO OI;)~I ~ X X
·I1dn.l¥H~ ~ x
·lHla.iI I~ X x X X X
(O~I Ij;\~:::~ll:~ ,~ X X
"PIXQ '" ,I X
OpppA:n ~,
~
Po 00 (0 co '<'J
ai C'i .N (!.l <>l <>l 00",,,, '<I!"'lI:> *
COI:-;OOll:>"''''
I I I I I I I 0,", <>l <>l C'i M MCr.lM ... .,.;...<'"
~~Ol~LOOO
~('I'JO'J "':'fIlr';)~~t...t_a.tOt-..
..Q ~ ~ t-- t": t': 00 00 a::i r,i.. 00 ~ 0; :::~;j~~~
454 '1J'Jonr '019U\[ ~
. :I,IO ';)SIl'J: ~} x
'!Hl,l d ......
"pdo S
""
0,
'pdo H ~ x I X I
"PU" IS ~
N '!i1JO ~
X X I X I x
N '~JOUI ~
Ih'lIV ::; X XX I X I
tUWlnQIlUN ,;- X
~
'~1-a 'jSlItO ~ X
UHolola ~ x
'ATJap JO
uUOIO ::: IX x IX
'ot! 'PlUS ~
'plln I JO .1U O'j

x
"pdO ID '" X X
"llll\)HDI,),1
.10 Dl~).!t c.. X
'IHJa ;HI.!!
JO O.!l r,o
'I J\I;).!I l~ X X X X
(ON) 1"{"OJlIU
.10 tJtuocpua "'''1-
°PlxO '"
°PPP'!I ., X
lnmtHlt8:
I I I I I
I I I X I I xx
XXI I I I I I I
x I I I I I I
I X I I I I I I I
I X J I I I I I
X, x I I I I I
I I I I
,< I I I 'XX
I I I I I
I X I I X I I I I
IX I I I I I I
::: X xxx , I I
X, I I I I I
I I I I I
I I IX
I I I I
I I
I I X
I I I I
x I I I I
I 01
x I I I I I
I I I I I
s
~
,_,_a
<q
,..,
~ ~
..; O'l
0 b
,; ,_
Q!
..., '0
:::;r, ....
""<ci ,_ 1<
,,~
;ri 00 0>
.Q <ci ""
q 0() ""0
'"
(,'1....-1~ "''_O> q .<,>
~ ~ ~~~
.Q 00 00 ~.....:,...;
'<'> '0 '0
C'1 ""'" C'l "i
'" <:'1
'" C'l C'l
'" '" '" '" '" NO'lo"') «>0",)1>';) 0",)
""
,.., <a ,_
z
0
<!)
t-
I,
... ....
[',
00 0
"" ....
,..,
I ' 00 ao '" ..... .....'" ,.., .... .... ao....
00
....
I>';)
00 00
It:> <!) I '
00 0() ao 00
...
0>
ao
Ql""""iC'l
cr>o>O>
CT:I oo:tt u';)
a>r1>O>
!::; ..... H ........ .... ......... ""..... ""....
456
x x
x x
x x x I I
1::-
x x x x
x
IV. LIQUEFIED GAl:lE8 "I.ND LOW-BOILIl\'O LIQt:IDS .f57
Chronological Bibliography of Low-Temperature Distillation Apparatus
(Patents not included)
1915. Burrell, Seibert, awl RohertHon, U. S. Bur. Mines Tech. Paper 104 (1915)
(simple distillation into fractions).
1923. Shepherd and Porter, Ind. Eng. Chem., 15, 1143-0 (1923) (improvement on
Burrell, Seibert, and Robertson).
Leslie, Motor Fuels, Chemical Catalog Co., New York, 1923, p. 555 (fore- .
runner of Podbielniak-type columns; not a low-temperature column).
1927. Frey and Yant, Ind: Eng. Chem., 19, 492, 1358 (1927) (low-temperature
fractionating column based un Leslie column; 5-mill. brass rings for
packing, liquid ail' in condenser and hydrogen in space between condenser
and column).
1928. Lucas and Dillon, J. Am. Chern. Soc., 50, 1460 (1928) (cut-glass I'ing pack-
ing, jacket filled with cold brine; for distillation at about -woe.).
Podbielniak, Ph.D. Thesis, Univ. Michigan, 1928.
Oberfell and Alden, Oil Gas J., 27, No. 22, 142 (1928) (first published de-
scription of Podbiclniak columll, from Phillips Petroleum Co., which had
supported Podbielniak's work at University of Michigan).
1929. Podbielniak, Oil Gas J., 27, No. 35, 38 (1929); ibid., No. 52, 30 (1929);
ibid., 28, No. 21, 161 (1929); Rejiner Natural Gasoline Mfr., 8, No.3, 55
(1929).
Podbielniak and Committee of California Natural Gasoline Association,
Petroleum World, 14, No.1, 102 (1929) (comprehensive description of low-
temperature distillation apparatus and procedure; apparatus consisted
of a glass fractionating column, packed with a single wire coil, surrounded
by a silvered evacuated glass jacket and using liquhl ail' in a copper vessel
in a gasoline bath fur condensing).
Davis, Ind. Eng. Chem., Anal. Ed., 1,61 (1929) (column of spiral glass tubing
surrounded by silvereu evacuated glass jacket; condenser cooled with
acetone-carbon dioxide mixture).
Shepherd, Bur. Stnnclards J. Research, 2, 1145-1199 (1929) (isothermal dis-
tillation) .
1930. Podbielniak, Oil Gas .T., 29, No. 20,235 (1930).
Schaufelberger, Oil Gas J., 29, No. 16, 46 (1930) (packing of uuuble spiral
Nichrome wire; liquid air in liquid butane for cundenser).
l\)3L. Podbielniak, Ind. Eng. Chern., Anal. Ed., 3, 177-88 (1931) (complete de-
scription and history).
Ebl'yand Engelder, Ind. Eng. Chern., 23, 1023 (1931) (Midgley wire spiral
coated with carborundum).
Fitch, Natl. Pet1'oleum News, 23, No. 25, 66 (lOal) (small auxiliary column
!tdded to Oberfell and Alden apparatus).
Rosen and Robertson, Ind. Eng. Chern., Anal. Ed., 3, 284-289 (1931)
(simple distillation analysis of three-component mixtures).
458 A. AND E. ROSE
1932. Davis anrl Daugherty, Ind. Eng. Chem., Anal. Ed., 4, 193 (1932).
McGillivray, J. Chem. Soc., 1932, 9·'l1 (modified Podhielniak column for puri-
fication of ethane and methane; packing of nickel-silver wire wound
rounrl glass rod, forcing vapors through a spiral annular space).
Podbiclniak, Oil Gas J., 30, No. 46, 68-()9, 1 UJ-120 (1932).
1933. Frey and Hepp, Ind. Eng. Chem., 25, 4.41 (1933) (ice water in condenser;
for liquids distilling at -10 0 to + 10 0C.).
Podbielniak, Ind. Eng. Chem., Anal. Ed., 5, 119, 135, 172 (1933).
Podbielniak, "Low-Temperature Fractionation Analysis of Natural Gas
... without Use of Liquid Air," Am. Chem. Soc., Chicago Meeting, 1933.
1934. Rosen and Robertson, Ind. Eng. Ch ellL , Anal. Ed., 6, 12--18 (1934) (exten-
sion of 1931 work).
Tropsch and Mattox, Ind. Eng. Chern., Anal. Ed., 6, 235-24,1 (1934) (im-
provement of fractional-condensation method of Tropseh and Dittrich,
1925).
Booth and Stillwell, J. Am. Chem. Soc., 56,1529-1530 (1934) (first of Booth
columns for purification of low-boiling compounds: dichlorosilane).
Bosschart, Ind. Eng. Chem., Anal. Ed., 6, 29-33 (1934) (inverteei column;
highest boiling components recovered first, as liquid in bottom of still;
Range +200 to -170°C.).
1935. Kistiakowsky et al., J. Am. Chem. Soc., 57, 55-75 (193.5); ibid., 876-883
(1935) (special column;; for purification of ethylene, propylene, butenes).
Lang, Ind. Eng. Chem., Anal. Ed., 7, 150-152 (1935) (combination Poclbiel-
niak column and Shepherd apparatus for complex gaseous mixtures).
193ft McMillan, J.Inst. Petroleum Technol., 22, 616-545 (1935) (review of previous
work and comprehensive comparison of Podbielniak and McMillan (Cali-
fornia Natural Gasoline ASRoc.) columns; packing is brass spiral wound
round vedical wire; condenser cooled by liquid air flowing over metal
beads).
Rose, Ind. Eng. Chmn., Anal. Ed., 8, 478-483 (1936) (glass triangle packing;
total condenser and side arm take-off j Dewar tube around still, column,
and condenser).
1937. Booth and Bozarth, Ind. Eng. Chem., 29, 470-475 (1937) (condenser cooled
by liquid air or solid carbon dioxide; apparatus autoillatically controlled j
auxiliaries for determining properties of gases purified).
Fischer, Z. Vel'. deut. Ing., Beihejt Verfahrenstech., 1937,118-124 (separation
of ethane-ethylene mixture).
Gooderham, J. Soc. Chem. Ind. London, 56, 25-3t1T (1937) (Dufton spiral;
buffer vessel at top of column, inside condenser, filled with gaseous hydro-
gen, like Frey and Yant column).
Laird, Oil Gas J., 36, No. 24, 58-59, 62, 65 (1937) (metal column of Podbiel-
niak type; solid carbon dioxide used for condenser; quick and accurate
enough for rough plant work).
Wustrow, Z. anal. Chem., 108,30.5-309 (1937) (used high vacuum; liquid
nitrogen in condenser j column itself is spiral tube; reflux produced in
annular space in condenser; 1000 cc. gas required for analysis of satur-
ated paraffins methane to pentane or methane, ethylene, and ethane).
1938. Schultze and Weller, Ol u. Xohle Erdiil u. Teer, 14, 998-1011 (1938) (ap-
paratus for analysis of hydrocarbon mixtures, especially butane and iso-
butane).
Simons, Ind. Eng. Chem., Anal. Ed., 10, 30-31 (1938) (column for -5°C.
to room temperature; vacuum jacket nonsilvered; condenser cooled by
ice, ice and salt, etc.; in same article glass pressure column is described
for pressures up to 5 atm.).
Simons, Ind. Eng. Chern., Anal. Ed., 10, 648 (1938) (capacity 5 cc.; column
fits into 1 qt. vacuum flask; glass ring packing; total condenser).
Ward, E. C., Ind. Eng. Chern., Anal. Ed., 10, 169-171 (1938) (uses pressure
of 1 mm. Hg; l'Cpeated simple distillation).
1940. Groll, Reichsamt Wirtschaftsausbau, Prilj.-Nr. -43 (PB52003), 79-88 (1940);
Chem. Abstracts, 41,5757 (1947) (packing of Cr-Ni spiral; column 7 mm.
I.D.; analyses on Cr C6 hydrocarbons).
Koch, Reichsamt Wirtschaftsausba1t, Pruf.-Nr. -43,93-102; Chern. Abstracts,
41, 6492 (1947) (6-meter spiral of glass tubing 5 mm. LD.; chart of separa-
tion propane, n- and isobutane, and isopentane).
Koch and Hilberath, Brennstoff-Chem., 21, 197-203 (1940) (review; same
as Koch column above; total condenser; separation described of pro-
pane-propene, isobutane-butane).
Schultze, Reichsamt Wirtschaftsau,,~ba1t Pl·uf.-Nr. 43 (PB52003), 7-14
(1940); Chem. Abstm-cts, 41, 6492 (1947) (Podbielniak-type column with
2-turn brass spiral; liquefied buffer gas in head; 100 ce. gas mixture
analyzed with accuracy of 1%) .
• 1941. Podbielniak, Ind. Eng. Chem., Anal. Ed., 13, 639-645 (1941) (Heli-Grid
packing; other important improvements).
Savelli, Seyfried, and Filbert, Ind. Eng. Chem., Anal. Ed., 13, 868-879
(1941) (comparison Podbielniak, Ward, and other types of analysis).
Shepherd, J. Research Natl. Bw·. Standa1'ds, 26, 227-244 (1941) (most recent
account of Shepherd apparatus).
1942. Hamler and Simons, Ind. Eng. Chem., Anal. Ed., 14, 430 (1942) (modified
Simons small coluIlln for -30to -5°C.).
1943. Hilberath, Ol u. Kahle, 39,875-880 (1943) (review of apparatus with figures
and comparative tables).
1944. Booth and McNabney, Ind. Eng. Chern., Anal. Ed., 16, 131-133 (1944)
(autoIllatic controls).
Douslin and Walls, Ind. Eng. Chern., Anal. Ed., 16, 40-42 (1944) (pressure
control for low-pressure apparatus.)
Zook, Oakwood, and Whitmore, "Analysis of Hydrocarbon Gas Mixtures
by Isothermal Distillation," Am. Chcm. Soc. Meeting, Sept., 1944.
1945. Askevold and Agruss, Ind. Eng. Chern., Anal. Ed., 17,241 (1945) (auxiliary
still containing sodium hydroxide pellets to remove water and hydrogen
sulfide).
Thomas, W. H. et al., J. Inst. Petroleum, 31,16-22 (1945).
1946. Hamblen and Thorstenberg, .Ind. Eng. Chem., Anal. Ed., 18, 153 (1940)
(graduated centrifuge tube for kettle; heating wire wound around capil.
lary tube of measured capacity at bottom).
460 A. AND E. ROSE
Martin, Austmlia C071nciI8ci. Ind. Research, Bull. No. 197 (1946) (McMil-
lan-type packing of aluminum wire WOllnd l'Oaud hard metal insert; elab-
orate reflux condenser; detailed cliseu8sion apparatas, procedure, and in-'
terpretatioll of results).
1947. Echols and Gelus, Anal. Chem., 19, 668-675 (1947) (multiisothermal dis-
tillation) .
Shepherd, J. Research N atl. Bur. Standards, 38, 19-51 (1947) (comparison
of analyses of standard sample by various lahoratories; agreement not
very good).
Starr, Anderson, amI Davidson, Anal. Chem., 19,409-412 (19-17) (study of
rates of charging samples).
Classified Bibliography
Recent Articles on Industrial Distillation of Liquefied Gase.~
L[Q"UID-AIR DISTILI,A.TION
Aston, Lobo, and Williams, "Liquid Air Fractionation," Ind. Eng. Chem., 39,
718-731 (1947).
Conway, "Manufacture of Oxygen in I,arge Qnantities for IIldm:trial Uses,"
Iron Steel Engr., 24, No.3, 53-58 (1947).
Dennis, "Still for Liquefied GaRes," IT. 8. Pftt. 2,386,297 (Oct. 9, 1945) (to Air
Reduction Co.).
Linde Meeting OIl Oxygen, Chem. Eng. Progress, 43,7-16 (1947).
Lobo, "Low Pressure Oxygen Process," Petrolewn Engr., 18, 120 (1947).
McKoon and Eddy, "Determination of TracGs of Acetylene in I.liquid Oxygen
in Rectifying Columns," Ind. Eng. Chem., Anal. Ed., 18, 133-136 (1946).
McMahon, "An Efficient Packing for Rectifying Columns," Ind. Eng. Chern., 39,
712-714 (1947).
Rice, "Separation of Gaseous Mixtures Such as Air," U. S. Pat. 2,327,4.59 (Aug.
24, 1943).
Rushton and Stevenson, "Developments in Oxygen Production," Trans. Am.
Insl. Chem. EngTs' 43, 6H39 (1947).
j
Van Nuys, "Liquid Air Distillation," U. S. Pats. 2,423,273-4 (July 1, 1943) and
2,424,201 (July 15, 1943).
Weedman aml Dodge, "Rectification of Liquid Air in a Packed Colnmn,1I Ind.
Eng. Chem., 39, 732-74<1 (1947).
DISTILLATION OF ACETYLENE
Babcock et al., IISeparation ane! Purification of Acetylene by Distillation and by

Extraction," U. S. Pats. 2,236,953-5 (April 1, 1941).
Balthis and Taylor, "Separation of Acetylene from Gas Mixtures," U. S. Pat.
2,236,966 (April 1, 1941).
IV. LIQUEFIED GASES AND LOW-BOILING LH~UIDS 4(H
.'izeotroll'ic and Extractive Distillation of C4 Hydrocarbons

Buell and Boatright, "Furfural Extractive Distillation for Seplll'ution and Puri-
fication of C4 Hydrocarbons," Ind. Eng. Chern., 39, 695-705 (1947).
Engs, Wile, and Roberts, "Separation of Monoolefins from Diolefins by Azeo-
tropic Distillation with Methylamine," U. S. Pat. 2,414,639 (Jan. 21, 1947).
Gerster, Mertes, and Colburn, "Vapor-Liquid Equilibrium Data for Ternary
Systems," (n-butane-1-butene-furfuralj isobutane-1-butene-furfural), Ind. Eng.
Chern., 39, 797-804 (1947).
Matuszak and Frey, "Separating Butenes from Butanes," Ind. Eng. Chern., Anal.
Ed., 9, 111-115 (1937).
Mertes and Colburn, "Pl'Operties of Binary Mixtures of n-Buttme, Isobutane,
and I-Butene with Furfural," Ind. Eng. Chern., 39, 787-796 (Hl47).
Nutting and Horsley, "Separating Hydrocarbons Having Different Degrees
of Saturation," U. S. Pat. 2,414,761 (Jan. 21, 1947).
Miscellaneous Recent Articles

Klliel and Slager, "Etl1ylene Purification," Trans. j1m. Inst. Chem. Engrs., 43,
335 (1947).
Kravets and Belyaev, "Separation of Butane-Propane Fraction from Shale Dis-
tillation Gas," U. S. S. H. Pat. 66,314 (May 31, 1946) j Chern. Abstracts, 41, 2233
(1947).
Schultze and Hillyer, "Recovery of a Butadiene-Isobutene Concentrate," U. S.
Pat. 2,416,647 (Feb. 25, 1947).
General References
I~llis, Chemistrll of Petroleum Derivatives. Reinhold, New York; Vol. I, 1934,
pp. 1104-1115; Vol. II, 1937, pp. 1138-1141.
Stanley, "Analysis of Petroleum Gases," in Science of Petroleum. Vol. II, Oxford
Univ. Press, London, 1938, pp. 1524-1533.
See also review articles listed in the Chronological Bibliogmphy. l11nong the be,lt are:
Podbielniak (1931), McMillan (1.936), Hose (193H), Gooderham (1987), Koch
and Hilberath (1940), Podbielniak (19'11), Hilberath (1943), and Martin (1946).
CHAPTER V
DISTILLATION UNDER MODERATE VACUUM
JOHN R. BOWMAN AND R. STUART TIPSON, Mellon Institule of Industrial Research
I. INTRODUCTION
Distillation under moderate vacuum will be defined as that in which the
pressure of the vapor in the system is less than atmospheric but sufficiently
high that the mean free path of the vapor molecmles is small relative to the
gap between the distilland surface and the condensing surface. In practice,
this includes all vacuum distillations performed in equipment not specifically
designed for molecular distillation (see Chapter VI). In the moderate-
pressure range, certain additions to and modifications of the equipment and
procedures employed for distillation at atmospheric pressure are necessary.
The principal advantage of distilling under diminished pressure instead
of at atmospheric pressure is that the boiling point is usually lowered con-
siderably (Fig. 1). It will be noted that this diminution is much greater
at lower pressures; this is because the pressure of a vapor varies logarith-
mically as the inverse of its absolute temperature. In some cases, this
permits distillation of substances subject to chemica'! alteration at the
normal boiling point. For example, mannitol or duleitoF can be distilled
at 276 0 at a pressure of 1 mm. of mercury. Thermally induced changes
of the distilland thus avoided, or greatly diminished, include pyrolysis,
rearrangement, condensation, and polymerization. Oxidation is usually
obviated. Vacuum distillation is also useful for heat-stable compounds
having low vapor pressure which would boil at inconveniently high temper-
atures at atmospheric pressure. Examples of the two main types of sub-
stance commercially distilled under diminished pressure are certain lubri-
cating oils and some essential oils; the first group comprises difficultly
volatilized but highly stable compounds, and the second contains reactive,
moderately volatile materials. In the second group are many of the
monomers, e.g., styrene,! used ill plastics manufacture.
The principal objectives of vaCUum distillation may be either of the two
which are basic to all disti1lal~iolls, viz., sepamtion or resolution for analysis
1Krafft and Dyes, B~r;; 28, 2583 (1895).
2Marks, U. S. Pat. 2,336,493 (19·13); C. A" 38, 3IGn. Gadwa, U. S. Pat. 2,370,948
(1945); C. A., 39,5122. Cermnk, Insll'1~mentation,2, No.3. 7 C1(46).
463
464 J. fl.. nOWMAN ANn H. S. TIPSON
of a mixture. The firflt is by far the more important. Although the sharp-
ness of separation depends on the same factors, regardless of the pressure,
the transitions between component plateaus during a vacuum distillation
are usually not so sharp as those of an atmospheric-presslire distillation
and are therefore more difficult to follow. SometimeR, however, transitions
are sharper under moderate vacuum.
A much less frequent liRe for vacuum in preference to atmospheric dis-
tillation of a two-component liquid depends on the inerease in relative
volatility of the components which may occur as the pressure is diminished.
This application may be found convenient for systems in which the relative
volatility is sman at atrnosphel'ie pressure but large under reduced pressure.
Homologous series of compounds often display this property, flome strongly.
200~----------------------,
..;
•
......
z
o
n. 100
(!)
z
::;
oCD
0~-I~O~O----~3~OO~---5~O~O~----~7~60
PRESSU RE, MM. OF HO
Fig. 1. Effect of pressure on boiling point of benzaldehyde.
However, when the substances to be separated are very different chemically,

there are exceptions; specifically, if the heat of vaporizat,ion of the more
volatile component is much less than that of 1;he other, the more volatile
will have a steeper vapor pressure - temperature dependence, and the
relative volatility may decrease with decreasing pressure. A special
application of this principle is the separation, by distillation under reduced
preSSl1re, of compoul\ds which forIn azeotropes 3 at atmospheric pressure.
In nearly all snch systems, reduction of the pressure causes the azeotrope
composition to become richer in the more volatile component. 4 Further
reduction of the pressure may then debar formation of any azeotrope (see
Swietoslawski and Andersono). The ethanol-water system, which gives no
I Horsley, A.nal. Chem., 19,508 (1947). See Chap. I, Sect. IIl, and Chap. III, Sects.
I and III, this volume.
I Roscoe, J. Chem. Soc., 13, 14B (18Bl); 15,270 (1862). Britton, Nutting, and Hor-
sley, Anal. Chem., 19,601 (1947), Nutting and Horsley, ibid., 19,602 (1947). Horsley,
ibid., 19, 603 (1947).
• Swietoslawski and Anderson, "Determination of Boiling and Condensation Tem-
peratures," in Weissberger, ed., Physical Methods of Organic Chemistry. 2nd ed., Inter-
scienoe, New York, Part I, 1949, Chap. IV, Scct. II.
V. DISTILLATION UNDER MODERATE VACUUM 465
azeotrop(" below 70 mm. of mercury, is an example. 6 Two other, less im-

portant, features of vaellum distillation may be mentioned. One is the use
of vaenum to attain eeonomy of heat in production operations involving
several stages; the other is to permit heat transfer from a low-temperature
source, e.g., low-pressure steam.
To summarize, vacuum distHlation is principally used in the laboratory
for separation of compounds which would change chemically or distil too
slowly at the tempern,ture needed for n,tmospheric-prcssure operation.
Furthermore, in some cases, the yield of a purified product is higher than
is obtained by recrystallization or other separation methods. In addition,
ecrtain mixtures inseparable by equilibrium recrystallization, e.g., a eutec-
tic composition, may often be readily separated by fractional vacuum
distilla ti on.
Although the history of ordinary distillation goes back to ancient times, the in-
troduction of vaCllum distillation is relatively recent. In 1662, Sir Robert Boyle7
reported the cvaporation of water and of turpentine under diminished pressure, but
did not effect the condensation necessary to complete the distillation cycle. Thii:! is
rather surprising, since he was greatly interosted in distlllation at atmospheric
pressure, and was probably the first to employ it as a general method for separating
volatile compounds and the first to recognize its usefulness as an analytical method.
Lavoisier found that the rate of evaporation of diethyl ether is increased under di-
minished pressure, and, during the French Revolution, Labon8 proposed the use of
vacuum in distillation. The prossure was to be diminished by letting the water
run out of a long, water-filled pipe attached to the receiver, but it is not known if he
tested the procedure. Howarcl,9 in 1813, appears to have been the next to study
the phenomenon. He presented a tahle showing the lowering in boiling tcmpera~
ture caused by diminii:lhing the pressure with a pump, and described the evapora-
tion of aqueous sucrose solutions in the first steam-heated vacuurn pan. He em-.
phasized the desirability of maintaining the vacuum throughout the course of the
evaporation by keeping in operation a vacuum pump connected to the apparatus.
Tennant lO designed a still for VILCuum distillation of sea water aboard ship, and
shortly thereafter Tritton ll described a complete vacuum distillation operation in
terms readily underAtandable today. He observed that, in the distillation of spirits,
heating of the still pot with a direct fire caused decomposition which led to inferior
8 Merl'iJmcn, J. Chem. Soc., 103,628 (1913). Lecllot, L'Az€otropisme. Lamertin, Brus-
eels, 1918.
7 Boyle, Sir Robert, New Experiments PhysicD-mechan£cnl TDuchl:ng the Air. 2nd
ed., Hall, Oxford, 1(\62, pp. 202-20,1.
8 Lebon, Philippc, DisWlation au moyen dn vide et dn/raid, French patent of 25 fructi-
dol' de ]'llon IV (i.e., Sept. 11, 1796).
9 Howard, Brit. Pllt. 3754 (1813); compare Romershauseu, Destillier- und Abdun-
atungsap1Jarat. Zel'bst hei Fuchsel, 1820, and Rillieux, D. S. PL\t. 3237 (Aug. 26, 1843).
IQ Tennltnt, Trans. Roy.·Soc. London, 102,587 (1811).
11 Tritton, Brit. Pat. 4140 (1817);· Annals of PMlosophy, 11,445 (1818). Compare
Barry, J. Chern. PhY8. (Schweigger's), 28, 250 (1820); Bdt. Pat. 4376 (May 24,1819).
466 J. R. BOWMAN AND H. S. 'rIPSON
taste) and coloI' of the distillate. He reasoned that, if the boiling tempel'a,tllre of the
c1istillanci could be reduced, the still pot could be heated by external application of
steam without any danger of overheating. To reduce the boiling temperature,
he applied vacuum to the still. In 1823, ChevreuP2 distilled stearic acid and oleic
acid under diminished pressure, but laboratory vacuum distillation in the modern
sense of 11 process for isolation and purification of chemica'! compounds was inde-
pendently introduced by Dittmar and by Anschlitz;t3 The apparatus they de-
scribed would not look unusual in the chemical laboratory today.
Industrially, vacuum distillation was discovered independently and by accident.
In 1867, .10shua Merrill was distilling a gOO-gal. batch of petroleum in Pennsylvania
when the condenser became plugged. The chargc was "too heavy for illumination
and too light for lubrication,"14 and the stoppagc was probably due to deposition
of wax in the condenser. The pressure became so great that the fire was removed
and the still allowed to cool, thereby producing [L vacuum. When it was opened,
a clear, neutral distillate was found in the condenser. Mel'l'lll noted later that a
similar distillate could be prepared by thc URe of superheated steam which also
acted as an entrainer. JJarge-scale application of vacuum distillation to petroleum
oils followed quickly, and by the 1870's the Vacuum Oil Company was thriving in
Rochester, New York. Extension ofIarge-suale vacuum distillation to other indus-
tries did not come until the present century. Early applications in this field were
the distillation of phenol and cresols 16 and steam·-vacuum distillation of glyceroJ.1G
The use of open steam in distillation is the invention of Count Rumford 17 in
1802. He not only described the process in detail but gave an excellent theoretical
treatment of the mechanism, which he described as "sweeping away vapors." The
process may be regarded as the forerunner of entrainer-vacuum distillation,U a
process which entaIls judicious admission of an inert gas to speed up distillation and
obviate "bumping." Distillation under diminished pressure with steam,19 super-
heated steam,20 or other suitable vapor can be accoIllpliHhed.
12 Chevrenl, Recherches Chimiqucs sur les CorlJS G1'US d'Origine Animale. IJevmult,
Paris, 1823, pp. 23, 77.
13 Dittmar, J. Chern. Soc., 22,4·16 (1860). Compare KekuIe and Fmnchimont, Ber., 5,
\)08(1872) and ThOmer, ibid., 9, 1868 (1876). Anschiitz and Reitter, Die Destillation
unieT vermindertem D1"1.~ck im Laboratorium, 2 Aufl., Cohen, Bonn, 1895.
H Peckham, Rep01't on the Production, Technolog1J, and U,ge,~ of Petroleurn and It.~
Products. Govt. Printing Office, Washington, D. C., 1885, pp. 162-163.

1& Raschig, Z. angew. Chern" 28, 409 (1915).
16 O'Farrell, Brit. Pat. 3284 (1881). Compare: W. Clark, ibid., 5348(1881); Van
Ruymbeke, U. S. Pitts. 522,132 and 522,135 (1804); Stbckmltn, Chem. & Met. Eng., 52,
No.4, 100 (1945).
11 Thompson, Benjamin, Count Rumford, "Of the Use of Steam as [J. V (lhicle fo!' Trans-
porting Heat," in Essa1Js, Political, Economical and Philosophical. Vol. III, Caddell
and Davies, London, 1802. '
16 Dittmar, J. Chern. Soc., 22, 446 (1869). Kekule and I<'mnchimont, Ber., 5, ~)08
(1872). Compare Pellogio, Z, anal. Chern., 6, 396 (1867).
19 Steinkopf, Chern. Ztg., 32, 517 (H)08), Monhaupt, ibid" 32,573 (1908). Anon.,
ibid., 32, 1083 (1908). Hoering and Baum, Ber., 42, 3076 (1909). WeIde, Biochem. Z.,
28,504 (1910). Edelstein and WeIde, Z. phy,~iol. Chern., 73, 152 (1911). Edelstein and
Csonka, Biochem. Z., 42, 372 (1912).
20 Hanies and Haal'mann, Ber., 51, 788 (Ell8).
V. DIS'l'ILLATION UNDER MODBRA'l'E VACUUM 467
II. THEORY
The effects of pressure reduction on distillation processes fall into three
general categories: vapor £low, rectification efficieney, and vaporization and
condensation phenomena.
1. Fluid Flow
The flow characteristics of vapors in general are of three types. There is

a substantial distinction bctween turbulent flow and laminar flow. 20a The
first is characterized by a more-or-less extensive, small-sca.le, random
motion, and the pressure properties of the flow are largely dependent on the
density of the fluid. In laminar flow, a steady nonreversing movement
prevails through the fluid, and the action may be described as a pure shear-
ing movement of sheets of fluid over each other. Under these conditions,
the pressure characteristics largely depend on the viscosity of the fluid.
In liq'uids, the change from turbulent to laminar flow operation is always
sharp and is associated with a critical value of the Reynolds number, R., a
dimensionless quantityZl given by an expression of the form:
Re = (mass rate of flow) (a length)/(viscosity)
In gases, on the other hand, the transition from Olle type of flow to the other
is gradual, especially itt low pressures.
At extremely low pressures, a third kind of flow, generally known as
Knudsen flow,22 prevails. 'rhis flow differs greatly from the two more
familiar ones, and appears only when the mean free path of the vapor
molecules is of the same order of magnitude as the diameter of the duct,
i.e., under the conditions of molecular distillation (see Chapter VI).
In nearly all distillation operations at 01' slightly below atmospheric pres-
sure, the vapor flow is turbulent. Under these conditions, the pressure
drop in the vapor lines is substantially independent of the absolute pressure
prevailing in the system. This is a consequence of the fact that, for a given
throughput, the linear :flow rate increases in inverse proportion to the re-
duction in pressure, and the density of the vapor decreases in direct pro-
portion. These two effects, therefore, counteract each other. This prin-
20a Egly, "Hcating and Cooling," in Wcissbcrger, cd., 'l'echniquo 0/ Organic Cherni.~try,
Vo]' III. Interseicnce, New York, 1950, Chap. I, Sect. ll3B.
21 Reynolds, 'l'7·ans. Roy. Soc. London, 174, 035 (1883). Compare: Ln.mb, Hvdro-
dynamic8, 6th ed. (1st Am. ed.), Dover, New York, 1045, p. 663 Bt seq.: Drew and
Genereaux, "Flow of Fluids," in Porry, ed., Chemical Engineers' Handbook, 2nd ed., Mc-
Graw-Hill, New York, 1941, p. 790; Duslumm, Scienl1'fic jl'o1t1ulalions oj Vacuum Tech-
nique, Wiley, Nl1w Yo!'k, H)4l), p. 89.
22 KnudHIlTl, Ann. Ph118iit, 28,75, noo (1900). See DUAitnuUl, op. cit., p. 90.
468 J. H. BOWMAN AND R. S. 'l'IPSON
ciplc call1lot, however, be extrapolated to extremely low pre:,;sures. Even

though the value of the Reynolds llumber at COllstant throughput for a
given I1pparatus is superficially independent of the pressure, n different type
of flow, probably laminar, appears at a pressure of about 10 mm. of mer-
cury. Below this pressure, the pressure drop at constant throughput is
almost inversely proportional to the total pressure. Hence, to keep the
pressure di·op constant, the throughput must be reduced proportionally with
the pressure. This is pl"Obably in consequence of the fac:t that, under Iam-
illlll"-flow conditions, the viscosity controls the pro8sure drop; to keep the
prest'mre drop constant, the lineal' flow rate, not the total mass throughput
rate, must be kept constant.
The appearance of htminar-flow churactel'ifltics in the tmnsport of vapor
at intermediate preSRures may be associated with shock-wave phenomena.
The rates of linear flow obtained by elementary enlenlatioll approl1eh the
velocity of sound, but so little is known about eluuacteristics of gas flow
at high velocities and low pressmes that very little speculation can be
made as to the fundamental effects concerned. Perhaps high-altitude,
high-velocity, wind-tunnel data will some day find applimd,ion to vaCUUm
dil:!tillatiOll. A numerical example illustrates the magnitudes of these phe-
nomena. Consider the distillation at atmospheric pressure of a material of
molecular weight 200 at a rate of 1000 g. per hour through a cylindrical
vapor line 1.0 cm. wide and 10 cm. long, n set of conditions easily rcalized.
The pressure drop will be of the ardor of 1 mm. of mercury and the vapor
velocity will be about 400 cm. per second. If the pressure is now reduced
to 75 mm. of mercury, the velocity will increu8e to 4000 cm. pel' second, a
high yet not unreasonable figure, but the pressure drop will alter only
slightly; hence, no change in the distillation rate or in the apparatus is
necessary. Now consider another tenfolclreduction of the prCi:lsurei even
neglecting expansion of the vapor as it flows along the tube, the pressure
drop becomes of the same order of magnitude as the total pressure desired
in the still pot, and the Vltpor velocity mU8t exceed the sonic value. Ob-
viously such operation is impossible. 23 Any attempt to attain it will result
either in a decrease in throughput, or an increase in still-pot pressure, or
both.
Hence, columns intended for distillations at atmospherie pressure may be
of lit.tIe use at low pressures; with high vapor velocit,y in a packed column
containing reflux, the pressure drop is so high thltt excessive temperatures
in the pot may re:mlt. Furthermore, equilibrium between reiiux and vapor
is not approachod, and the liquid reflux may be swept through some parts
of the packing, causing vapor channelling in other parts.
~3 Griffiths, Trans. Inst. Chern. Engrs. London, 23, 113 (1945).

V. DIS'l'lLI,A'l'lON UNDl1lt MOD.l<JltA'l'E VACUUM 469
2. Rectification
A. COUNTEIWURRlDN'l' CON'l'AC'l' nmC'rIFICA'l'ION
The usual method of fractional distillation involves gravitational reflux

to give contact rectification. A two-way transfer of material between liquid
and vapor streams flowiug couutercunelltly is effected by reason of the
spontaneous tendency of liquid and vapor in contact to approaeh equi-
librium. The vapor at !tlly point i.s always poorer in the "light" material
than the hypothetical vapor composition which would be in equilibrium with
the liquid at the same point. As a consequence, there is a spontaneous
transfer of "henvy" material to the liquid, and of "light" material to the
vapor. This is discw:iHed in detail in Chapter 1. Analysis of these phe-
nomena at low PI'CSSl/./'CS, however, brings out certain dependencies that are
unimportant at atmospheric pressure. In the latter case, the interphase
transfer rates are so small relative to (;he absolute rates that the liquid and
vapor can, for practical purposes, be said to be in equilibrium with each
other at the interface, and the factor controlling the process is diffusion of
material to and from the interface. On the other hand, the [~bsolute rates
of vaporization and condensation become small at low pressures (see Chap-
ter VI, Part I). At pressures of the order of a few millimeters of mereury,
they may become the slow step in controlling and limiting the enrichment
proeess.
A compensating, favorable effect of low pressure on enrichment is the
increase in diffusivity in the vapor phase. This is, approximately, in-
versely proportional to the pressure, but its advantage lies only in im-
provement of transfer in the vapor phase, because diffusivity in the liquid
is substantially unaltered by pressure changes. Combining the two effects,
it is obviom: that, at pressmcs below tttmospherie, diffusion in the liquid
phase controls the rat{l of the process a,nd a further increase in vapor dif-
fusivity docs not eaUse appreciable improvement in the transfer rate.
The incren:-oe in vapor Lliffusivity at low preSSUl'es 1ms [111 additional effect;
and this one is unfa.vorable. It illGl'eaSes longitudinal mixing, so tha.t the
vapor stream becomes unable to support large coneentrution gradients.
This is not. serious until pressures of a few millimcters of mercury are
reached, but then a. profound effect on the enrichment is exerted. In the
limiting case, when the vapor diffllsivity CfLn be considered infinite, any
column, however long, would have fI, uniform vapor composit.ion along its
length, aud the ()vel'~all effieiOllCY lJl1n be ~h()Wll to be exactly two plates.
Another unfavorable effect of pressure reduction is the great increase ill
linear velocity of the vapor. This velocity becomes large relative to the
interphase transfer rates, and further contributes toward pOOl' efficiency.
Combining these effeets, nnd taking into account the difficulties COll-
470 J. n. BOWMAN AND U, S. 'l'IPSON
cerning pressure drop discussed in Section III, it becomes evident that

gravitational rectification is not usually effective at extremely low pres-
sures. No criticallowel' limit of pressure can be given, because the relative
effects of the phenomclllt coneerned depend greatly on t,he design and
dimensions of the apparatus used. From pr::wtiCLtl experience, ordinary
atmospheric-pressure stills (see Chapter II, Part 1) will generally function
satisfactorily down to a press'Ure of a few centirneters of rnG1'C'U1·Y. Below
this value, specially designed columns should be used down to a few
millimeters of mercury.
B. BA'l'CHWISE, COUN'l'mWURRl~NT RBDIS'l'ILIJA'l'ION. 'l'REHMAI, HlDC'l'IFICA-

TION
Finally, in the 1'!tugo between abont 1 mill. and molenulal'-di8tillation

conditions, contaet rectification cannot be employed to any useful extent.
Instead, recourse must be had to rn'ultiple Tedistillat'ion to achieve enrich-
ment (see Chapter I, Sect. I). In such processes, simple distillations are
used repeatedly; each achieves very nearly one-plate I)epal'ntions, and many
stages are cascaded with suitable intermedifLte blending of fractions, as in
cascade recrystallization. 2,1 The schematic principles involved resemble
those of countereurrent liquid-liquid extraction. 25 The mechanism by
which a simple dist,illation aehieves separation il) somewhat different from
that of enrichment by contact rectificatioll. In a simple distillation, the
separation is achieved by the controllable net change of phase. As a liquid
is vaporized, the composition of the vapor is fixed either by equilibrium
relations or by absolute vaporization rates; it in no way depends on spon-
taneous approach toward equilibrium. In the range of a few millimeters
of mercUl'Y, the spontaneous meChalli8fi1 of contact rectification often be-
comes so slow as to be almost useless, and the phase-ehange mechanism is
the ol1ly one that remains usefully avnilable. It, has been termed "thermal
rectifieation"26 to distinguish it from the cont.lwt rectification obtainable in
conventional columns.
In thermal rectification, independent control of the two interphase trans-
fer rates is possible. Enrichment may be effected either by a partial vapor-
ization or by a partial cOndCllS!Ltion, and the l'l1tes of both can be controlled
by appropriate addition of heat to, 01' removuHl'om, the system. Of course,
both processes occur in contact rectification, but since they are simultaneous
and spontaneous, no control of them is possible. The finlt well authenti-
See Tipson, "Crystallization and Recrystallization," in Weissberger, ed., 'l'echnique
M
ojOr(fanic Chemistry, Vol. III. Interscience, New York, 1950, Chap. VI, Sect. IIr3A.
25 See L. C. and D. Craig, "Extraction and Difltl'ibution," in Technique oj Organic
Chemi,l!j'y, Vol. III, Chap. IV, pp. 171-311.
26 Byron, Bowml1u, [mel Coull, Ind. Eng, Chent" in J)rps,~.
v. DISTII,LATION UNDER MODEItATE VACUUM 471
eated use of thermal rectification was simple multiple redistillation 27 with

appropriate blending of intermediate fractions between distillations. This
method of enhancing the separation obtained by a single simple distillation
probably goes bade to the Middle Ages, and is, therefore, much older than
contaet rectification, which was not made precise nntil early in the nine-
teenth century.
Rccently, however, several deviecs, horizontal and vel'tica,l (see also
Chapter VII), have been described which, in principle, cffect in a single
stage the equivalcnt of Ruceessive mUltiple redistillations with appropriate
blending. The first of these was proposed for semimicro operation. 28
A small sample was sealed in a long, evacuated glass tube which was then
slowly pushed horizontally through a temperature gradient. As this
I
2~1
=!~1
I
I
I
i
! ~~[,'--------------------4 r
'
=~~~,----------------~~ i
I : i
I VACULIM JAC KET
GLASS
~~~~~::::C~O~O~L~ER~:l~~~~~~::::~H~E~AT~E~R~::~~~~~
T~BE "
1~COPPER TUBE '/11TH OBSERVATIO~ SLOT

I I
!BEFORE SEPARATION
i- I
: : 1
! AFTER SEPARATION kI !
I
·.;. . n:·;·,-l
I
3
Fig. 2. Horizont.al still for thermal rectification. jR
operation proceeded, the material was repeatedly and continuously l'edis-

tilled from i1 region near the hot. end nncl condensed in a cooler region
(Fig. 2). As the process eontinued, a separation was achieved ill which the
more volatile components appeared in the cool end and the "heavier"
maLerial nearer the hot end, For eompounds of low vapor pressmc, the
method is superior to gravitational reflux procedures in whieh there is
difficulty in maintaining 11 rel1,~onably high reflux rate. It is excellent for
rectification of a small volume of distilland; because of loss due to holdup
this is not feasible by contact rect.ifieation.
Another ·method of thermal rectification employs a vertical, unpacked
tube, 5 ft. high, in which alternate sections are eleetrically heated and air
27 Wolcott, U. S. Pat. (Oct. 6, 1835).
28 Von Elbc and Scott, U. S. Pat. 2,108,8-18 (1040); Ind. JiJng. Chem., Anal. Ed., 10,
284 (1938).
472 J. R. BOWMAN AND H. S. TIPSON
coole(1,29 This device (lI'ig. 3) resembles n conventional, vOl'tieal column

nncl is used with an ordinary still pot I1nd I1ccessoriet;. The reHux flows
J!'ig. 3. Vertical still for thel'HlId

recti fleet t.ion. 29
down as a film on the inner surface of the tube, and, at eneh heated zone,
a fraction of it is revapol'ized. This vapor is generated in contact with the
main vapor stream, mixes with it, and begins to flow up. At each cooled
19 Schaffner, DOWID,l11, IIlld Coull, Trans. Am. lnst. Ghem. En{JTs., 39, 77 (1943).
v. DISTILLA'l'ION UNDER MOmGRATE VACUUM 473
zone, a fraction of the mixed main vapor stream is conclr))1sed directly into
the reHux stream, mixes with it, lLnd starts to flow clown. The heated and
cooled zones may be regarded as a cascade of simple vertical stiIlR and par-
tial condensers. The net effect is to achieve separately, and in a con-
trolled manner, the two effects which any rectification process must pro-
vide: transfer of "heavy" material from the vapor into the reflux, and
transfer of "light" material from the reflux stream into the vapor. The still
was used for separating a mixture of 1n- and p-tricresyl phosphates. The
column was equivalent to about 15 theoretical plates.
Experience has shown that this type of column is only effeC1tive in a
rather narrow and critical p:cessure range, which, for most systems, lies
between about 0.5 mm. and 5 mm. of mercury. The failure of the process
at higher pressures is probably attributable to inadequate mixing in the
vapor stream. The flow of the vapor is apparently laminar, and, unless
the pressure is low enougl1 to give high diffusivity in tlmt phase, the
vapor formed at the hot zones does not mix adequately with the main
stream but remains outside it and flows upward to the next cooled zone,
where it is totally condensed. This gives a recycling effect which does not
contribute to enrichment because the main V!1;por stream merely travels
through the apparatus as a streamlined core. The low-pressure limit is
fixed by longitudinal diffusion. As mentioned before, if the diffusivity
in the vapor becomes too large, the stream can no longer support a concen-
tration gradient, and no more than a two-plate separation can be realized.
This lower limit of pressure is probably nearly universal for columnar
separations.
Another type of thermal rectifying column, whieh obviates the difficulty
of the high-pressure limit, employs coneentrie tubes 2G (Fig. 4). The reflux
stream fiows as a film on the inner surface of the outer tube, and external
heat is uniformly applied to this tube. The inner tube is rotated on its
vertical axis and is cooled by circulation of a coolant through it. In this
device, the outer tube serves as a continuum of simple stills and the inner
as a continuum of condensers. The main How of the vapor is radially
inward to the inner tube, where it is condensed, and then the condensate
is centrifugally thrown outward as a spray into the flowing film of reflux.
Fresh vapor is continuously and uniformly generated ovor the entire surface
of the reflux film. Superposed upon the inward flow of vapor is an upward
component of fiow which is equal to the throughput. Columns of this
type have been operated efficiently from atmospheric pressure down to a
few tenths of a millimeter of mercury, at which pressures the longitudinal
diffusion effects seriously impair the results.
The theory of the performance of thel'mall'eciijication devices is far more
complicated than that for contact rectification stills, and as a rule each de-
474 J. H. BOWMAN AND R. S. TIPS ON
vice must be subjected to individual mathematical analysis. For detaih;

on specific cases, the reader is referred to the original literature cited. In
general, the equations of performance for thermal rectifiers are quite differ-
ent from those for contact devices. For example, in thermal rectification,
there is no exact equivalent of a theoreticf11 plate as n usefully invariant
concept. Although, for any given run, an equivalent number of plates
could be computed for any given column using the conventional eontact-
enrichment formulas, the number so derived will vary for different opera-
tions of the same column.
The rational unit for invariant evaluation of thermal reetification col-
umns has been termed the "heat ratio",26 l1lld defined as the ratio of heat
applied to the column, useful in effecting enrichment, to the heat applied to
the still pot, useful only in effecting distillation. This quantity is invariant
and is quite the equivalent of the number of' theoretical plates in contact
rectification. Surprisingly, for many types of separation the "heat ratio"
is numerically nearly equal to the equivalent number of theoretical plates.
An intuitive explanation of this coincidence followi:l from recognition of the
"heat ratio" as vaguely equivalent to a simple distillation, which, in turn
is, in a poorly defined way, equivalent to a theoretical plate. These
equivalences are by no means quantitative, nor can they be made so.
A few generalizations can, however, he made. The broadest is that, as the
difficulty of the separl1tion increases, the required value of the "heat ratio"
becomes numerically slightly less than the required number of theoretical
plates. This divergence increases with increasing number of plates and
with decreasing relative volatility, charge concentration with respect to the
light component, and reflux ratio. Exceptiolls are, however, readily found.
3. Vaporization and Condensation

Vaporization and condensation phenomena are both aJfeeted by reduc-
t,ion of pressure. The most important and best-known effect is the eon-
eurrent reduction in temperature at which the two processes will occur.
rn fact, this is the principal phenomenon which makes vacuum distillation
useful. According to Hickman 30 many compounds which boil with decom-
position at about 350°0. at 760 mm. will distil unchanged at 160 to 210°
at 10 mm., at 100 to 130 0 at 0.01 mm., and at 40 to 60° at 0.0001 mm. The
theoretical aspects are discussed in Chapter 1.
One of the annoying effeetR of pl'flSRUl'fl reduction on the vaporization
process is the great increase in the tendency of liquids to "bump" when
boiled under vacuum. No rigorous theoretieal treatment of bumping can
be given, but a qualitative explanation is easy. The effect is probably
30 Hickman, J. Franlclinln.st., 213, 119 (1932).
V. DIS'l'JIJI,ATION UNDEH MODEHA'l'E VACUUM 475
largely due to surface tension. The surface tension tends to give high
pressure in the bubble seed relative to the pressure at which the distillation
is being performed. Since the pressure, generated by surface tension, in·
side a bubble increases rapidly as the size of the bubble is decreased, it may
easily approach and even exceed the pressure at which the distillation is to
be performed. Bubbles can then form only by extreme superheating, and,
as soon as growth commences, they may enlarge at an explosive rate. Once
started, the reduction in internal pressure as the bubble grows permits rapid
vaporization and relieves the superheat. The effect of surfaee tension is
illustrated by the general experienee that liquids which foam do not usually
bump, and that the foaming property is usually associt1ted with low surface
tension.
Another eause of bumping is the hydrostatic head in the pot. In pot-
type distillations, heat is almost always applied to the bottom Df the pot,
i.e., to a portion of the liquid under a hydrostatic head whlch may be many
times greater than the operating pressure desired. When this prevails,
boiling is obviously impossible without great superheating.
In distillation in the lowC1' region of the pressure range under consideration
here, true boiling in the ordinary sense usually does not oceur at all.
There is, instead, a rapid evaporation from the exposed surface of the
liquid, and no bubbles form. In this event, there is always substantial
superheating throughout the liquid, and sometimes a large amount in the
layers in contact with the heat-input surfaee. In the range of moderately
high vacuum this general superheat is necessal'y. The very term "boiling
point" is an unfortunate misnomer. The boiling point is conventionally
taken to mean the temperature at whieh this vapor pressure of the material
has a given value. 31 It is thus an equilibrium concept, not a kinetic one,
and we have the curious paradox that no liquid boils at its boiling point.
The boiling point may be redefined as that temperature at which, at a given
pressure, the liquid just does not boil. To effect a change of phase from
liquid to vapor we must have superheat. At atmospheric pressure, the
extent of i:luperheat required to effect useful vaporization rates is generally
small, but in vacuum pl'OCeSSeH it may be very large.
Application of V~W\1um has little effect on the tot~\l condensation process,
as employed in dist.illation, ot.her thfLll t.o reduce the temperature at which
it OCellI'S. The effeet on part.ial condensation is, however, profound. In
partial c:ondensation at pressures of the order of atmospheric and higher,
the "heavy" components are selectively condensed only at the start of the
operation. This leads to an accumulation of "light" mn,terial in the vapor
phase neal' the interface, and to the building up of concentration gradients
n Sce Phy,q'ical Methods of 01'oanic Chemistry (Vol. I, this series), 2nd ed., Part I,
HIM), Chap. IV (Socii. I), Itnd C!1!\p. V (Scct.. III2).
476 J. n. BOWMAN AND H. S. 'J'll'SON
extending lXLCkward into the bulk of the vapor. As a result, the interface
film in the vapor phase reaches a concentration which is in equilibrium with
a hypothetical liquid having the composition of the vapor feed, and the
condensate has nea,rly the same composition.
As pointed out before, reduction of pressure increases the diffusivity in
the vapor phase. A sufficient reduction has a strong, favorable effect on
the selectivity of a pl1rtial-condensation process. The increased diffusivity
permits diffusive flow of the "light" component away from the interface
against the mass How of the vapor stream toward the interface, allowing the
heavy mat,erial to be selectively condensed in acconlance with the equi-
librium relations for the system. This favorable effect is observed in a
pronounced way in thermal reetificatioll, espr~cially in rotary-type columns.
At pressures of the order of atmospherie, the enrichment probably arises
almost solely from the partial vaporization notion. The pal'tial condensa-
tion undOl' these conditions is nearly nonselective and is required only to
maintain relatively uniform flows along the two streams. III the moderate-
vaeuum region, however, the apparent efficiency of rotary thermal colnmns,
expressed ill heat ratio units, nearly doubles. This must be ascribed to the
contribution of selective partial condensation to the over-all separntive
action.
III. AI)P'ARATUS
Reduction of pressure from atmospheric to 11 moderate vacuum, c,g., a few
centimeters of mercury, requires additions to the equipment conventionally
used for !Ltmospheric-pl'essllre operation (seo Chapter II, Part 1); these
inc:lude a pump, gage, trap, and device for ehanging receivers. In this
pressure range, theRe are used in mueh the same way as for distillation under
a few millimeters of mercury, but the latter range requires more profound
alterations in equipment design,
1. Stills
A still eonsists essentially of a dil:i1Jilling i1a:-;k or "pot," a head, a con-
denser, and a receiver. It may include a column. Stills fLre of two general
kinds: in one (rctort), the condensing surface is situated to one side of the
vaporizing surface-dcstillatio ad lat'us; in the other (alembic), it is above
the vaporizing surface-dlJstillatio pm' ascens. Combinations of the two
are often employed. All glassware, preferahly Pyrex, must be strong
enough to withstand the external preSl-lUre; nevertheless, the worker should
weal' goggles and plane a safety-glass shield in front of the still. All stop-
pers must be tight.; rubber stoppers IUay be lightly smeared with castor oil.
For high' temperatures, Silicone rubber stoppers or ground-glass joints
lubricated with Silicone oil are preferi1ble. The interchangeable joints
V. DISTILLATION UNDER MODERATE VACUUM: 477
ml1,y be used even for low pressures. For det[1iJs of joints, conneetiolls, and
sealing compounds, see Chapters II, Part 1, and VI, Part 2. Conven-
tional methods.of heating the pot may be employed; these include water,
glycerol, oil, or metal baths, ancl electric ffirwtles (Hee Chapter II, Part 1,
and EgIy, this series, Vol. III, ell. I, Sect. IV).
A. HORIZON'rAL STILLS
Horizontal stills are used for separation of certain mixtures of liquids of

greatly different volatility, for concentration of solutions, and for batch-
wise, fractional redistillation. A typical aRsembly is depicted in Figure 5.
The ordinary distilling flask shown is satisfantory for distillations at pres-
sures down to 15 mm. of mercury. The distilland is placed in this flask,
and boiling stones or glass wool are added (see Sect. IV). A thermometer'
is inserted in the neck of the flask, with its bulb opposite or slightly below
the side arm. To the side arm is attached a Liebig condenser and cooled
receiver, and the whole evacuated by means of 11 wl1ter pump, COllnected by
stout rubber tubing to a trap and mercury manometer as shown. The
distilling flask is surrounded by H bl1th whoso upper level is nbove the
distillandlevel, to lessen the tendency to bump. In somo' cases, it may be
desirable to immerse the flask as eompletely as possible, or to lag the neek
with asbestos cord. The bath, which may be covered with an asbcstos
plate, is gradually heated until diRtillation commences, and is then kept
Fig. 5. Horizontal still for distilhltion under diminished pressure.

478 J. n. BOWMAN AND R. S. TIPSON
Fig. G. Simple horizon tl11 stills for Fig. 7. Modified Cbisell flasks.
entruincr-VtlCUUlll distilbtioll.
at a constant temperature, adequate for steady, not too rapid, distillation,

un til disimn.tion slows. 1'he vItCuum iB now gradually released, the re-
ceiver changed, the system l'e-evaeuated, :md the bl1th tcmpemture raised
until distillation recommences. In this wa.y, a number of fraet:iolls of
distillate may be obtained.
III enlrainer-vacuurn distillat'ion, bumping is obviated by admission of a
fine stream of bubbles of inert gas, e.g., hydrogell,32 at the bottom of the
distilland, through a fine, hairlike capillary which should be drawn out in at
least three stages. The lower the opemting pressure, the finer the capil-
lary needed. A plastic balloon filled with inert gas may be employed
for introducing entraine1' into the top of the capillH.l'Y when distilling
readily oxidized compounds. The same device may be used in conjunctioll
with flutter-valve manostats. For use with a capillary, a, two-necked
Claisen Hask 33 as depieted in Figure 6a, is convenient. 34 The Rimplificd
32 Richter, Bel'., 49, 2339 (1916).
38 Claisen, Ann.,2'77,162(1893).
31 Barnitz, J. Am. Oil Chemists' Soc., 26, No.3, 10,1 (1949).
assembly35 shown in this figure is useful if a quantitative yield is not desired.

The water-delivery tube may be swung to give various degrees of cooling.
The apparatus may also be employed for evaporating solutions of solids.
The still head shown in Figure fib may be used for continuous addition of
solution to the distilling flask. An improved Claisen flask (Fig. 7a) is
provided with a pear-shaped bottom so that there may be less still residue
and the capillary may reach the bottom even near the end of the distillation.
In addition, the upper ends of the necks are constricted and the capillary
and thermometer are held in place with gum rubber tubing lightly Hmeared
inside with castor oil and wrapped with adhesive tape. This device ob-
viates contact of hot vapor with the rubber. The distilland is introduced
into the flask through a funnel having a long, narrow stem. Small models
of this apparatus may be employed for semimicrodistillations; the receiver
may be cooled in dry icc - chloroform.
A further modification (Fig. 7b) consist,s in elongation of one neck and
introduction of packi ng :l:l,36 or provision of bulbs, or of a series of indenta-
tions almost meeting in the center, to give iL fractionating column. The
still then employs both vl'lrtical and horizontal featmes of condensation.
The end of the elongated neck may be sealed off and provided with a sealed-
in mercury well into which a thermometer miLy be inserted. Columns may
0
be used with compounds which boil below about 170 under vacuum and
which do not undergo chemical change when considerably superheated.
The column may be made removable; it is then somotimes useful in the
separation of a low- from a high-boiling compound as follows: After all the
low-hailing component has been distilled off and the high-boiling material
has started to distil, the distillation is intel'l'Upted and the rest of the charge
distilled, without, a, column, into a single receivCl'.
If the distilhtte erystallizes readily in the side a-rm, this tube should be
made wide and be heatecl,37 e.g., electrically,3S to a temperature slightly
above the melting point; it may be necessary t() use a fresh delivery tube
for each fraction. 39 At pressures below about 15 mm. of mercury, the width,
35 Emil Fisehel': see Wahl, Chem. Ind. Berlin, 42, 270 (1919); I{ummer, Chem.
Fail1"U" 1,7 (H128); Burger, J. Lall. Clin. 111"ed., 25, 1221 (1940). Compare: Davis,
Ind. Eng. CIUlIlI., 11 nal. Ed., 14, 548 (1942); Hii.ussJer, Chem. Tech. Berlin, 15, 240
(1942); Barthel, Ind. Eng. Chem., Anal. Ed., 16,374 (1944); Barthel and Huller, iliid.,
1'7,529 (1945); Grant, ibid" 18,729 (1946).
3a Compnre: Michl1lJI, J. pralct. Chem., 4'7, 197 (1893); I,ederer, Chem. Zig., 19, 751
(1895); Michael, Ber., 34,4028 (1()01); Willstatter, Mayel', and Hiini, Ann., 3'78, 73
(1911); Hillyer and Deutsehman, AnaL Cham., 20,1146 (1948).
37 Haehll, Z. angew. Chern., 19, H)(l!] (H10G).
38 Bredt and van del' M!1al'en-Jausen, Ann., 36'7, 354 (1909).
3U Emory, Ber., 24, 596 (1891).

480 J. H. BOWMAN AND It. S. 'l'lI'SON
length, and position of the side arm become inereasingly important. 40

First, if this tube is too narrow, perhaps by only a few millimeters, t.he
velocity of the vapors is greatly diminished, even though the pressure at the
receiver iH satisfactorily low. Second, the side arm should bo shOl't. Third,
if it is too high,41 the temperature needed to give a certain dist.illation rate
at a given pressure is mueh higher than if t.he tube is nearer to the surfaee
of the liquid. Pressure drop through the side !Lrm may be obviated 42 by
condensillg the vapor in an alembic head in the neck of the flask, as in a
total-condensation, variable take-off,
still head (see Sect. In5).
A water-cooled condenser is usually
needed only for compounds boiling below
about 130 0 under vacuum. For high-
boiling substances, an air-cooled con-
denser or a cooled receiver generally
~ ,{ suffices. If the distillate is sensitive to
~ ail' or moisture, the receiver may be so
constl'ueted that it may be scaled off,
without ingress of air, at the end of the
distillation. If the hot distilbnd and
Fig. 8. "Udder" types of receiver
vapor are readily oxidized, or if very
chango!'. precise fractional distillation is neoessary,
a device to permit interchange of receivers
during the distillation, without interruption of the vacuum and of ebullition,
must be incol'pomted. Such changers are of three kinds. In one type,
the "pig" 01' "udder," provisioll is made for altering the direction of the
delivery tube relative to a receiver, 43 01' vice versa, so that successive frac-
tions of distillate flow into different receivers in turn (Fig. 8). In addition
to those shown, a modified vacuum desiccator, having two side tubes,
e.g., in the walls, and a central hole in the lid, lUay be employed. 44 A
~ Hicknmn and Sanford, J. Phys. Chem., 34, 637 (1930); Rev. Sci. Instrwnents, 1, 140
(1930). Hickml,n and Weyerts, J. Am. Chem. Soc., 52, 4714 (1930). Hickman, J.
Franklin h!.~t., 213, 119 (1932).
4] Krafft, Ber., 36, 4339 (Hl03); J. pralct. Cltc/n., 80,242 (HIOU).
42 Hickllum and Weyerts, J. A/n. Chem. Soc., 52, 4714 (1930).

U Gautim', Bull. soc. chim., 2, 675 (188\)). Bl'f~dt and van del' Maal'en-Jansen, Ann.,
36'7,354 (H10!l). Hnlm, Ber., 43,1725 (1910). Glaser, Chern. Zig., 36, 437 (1912). K.
Meyer and Schoeller, Bel'., 53, 1410 (1!J20). Holmes, Ind. Eng. Chem., Anal. Ed., 13,
61 (1941).
~4 BevlLli, Chem. News, 37, 183 (1878); Z. anal. Chem., 19, 188 (1880). Konownlow,
Ber., 17, 1531 (1884). Gorboti and Kessler, ibid., 18,1363 (1885).· Bruhl, ilJid., 21, 3339
(1888); 26,2508 (1803); Raikow, Chem. Ztg., 12, 6U4 (1888). Wislieenus, Ber., 23, 3292
(1890). Sehulz, Be1·., 23, 3568 (18\)0). Anderlini, Gazz. chim. ital., 24, 190 (1894);
Footnote continuod ~n next page
V. DISTILLATION UNDER l\WDERA'I'E VACUUM 481
set of test tubes is arranged circularly in a rack which may be turned by

meant:: of a handle extending thI'Ough a stoppel' in the center of the lid.
The end of the condenser enters one side arm and di8tillate is caused to fall
into each test tube in turn. The other side arm serves fot' connection to
ig. n. Stopcock tYllu~ of f(!ceivcr clllLllgUI'. Sue p:1ge 482 et .~elJ.
the pump. Thel-le "udder" typel:J are ullslLtisfaetol'Y for low-boiling liquids
since all the receivers intercommunicate, and there may be interchange of
vapors between the fractions.
Hull. 80C. chim., 12, 1057 (1894). Ktthlbllum, Bel'., 28, 3\)2 (1895). Biltz, Chern. ZtO., 19,
304 (1805). BUl'styn, 08t. ClulIn.-Ztg., 4, 5G3 (lIl01). Haulin, Z. angew. Chem., 19, l6GO
(1\)OG). Ubbclohde, ibid., 19, 757 (WOG), Krafft, 13m", 40, 4778 (Hl07). Dcl6pillu,
lhtll. soc. chim., 3,411 (1908). Pl1ttefl-lOll and Vall Dobh, Inri. Eng. Chem., Anal. Ed.,
14, 511 (1942).
482 J. n. BOWMAN AND R. S. '.rIPSON
The second kind of changer makes lise of stopcodcS,45 three of which are,
in principle, needed: one to close off the still from the receiver, one to
From Vacuum
Head System
ll
Vacuum System
or Vent
-~ Vent to Atmosphere
Fig. 10. H.eneiver ehangcl' with meLlt! va]ve. 1G
45 Thome, J. Chem. Soc., 43, 301 (1883); Ber-., 16, 1327 (1883). L. Meyer, ibid., 20,
1833 (1887). Lederer, Chern. Ztg., 19,751 (1895). Fogetti, ibid., 24,374 (1000). Skraup
Monatsh., 23, 11G2 (1902). Bertrand, B1l11. soc. chim., 29, 778 (1903). Alber, Chem.
Zig., 28, 819 (HJ04). De Florin, ibid., 32, 50 (1908). Roseanu, ibid., 32 r 13\) (1908).
Kolbe, ibid., 32, 487 (1908). li'rouncllich, ibid., 32, 820 (1908). Bogert, Ind. Eng,
Chem., 7, 785 (l9I5). I{ohen, Chern. Ztg., 45, G38 (1\)21). LOl'l1ng aJld Kluizenaal',
Chern. Fabrilc, 7, OG (1934). Willingham nnd ltossini, J. Research Nail. Bur. Standard8,
3'7,15 (1946).
4ij Williams and G!asebrook, priv[\te communication.
V. DISTIf,LATION UNDEH MODERA'l'E VACUUM 483
break the vacuum on the receiver after the o;till has been shut off, and one t.o
permit evacuation of the fresh receiver 11fter it has been put in place.
Several arrangements are possible, some involving multiple-bore stopcocks
(Fig. 9). Such "triangle" changers are more convenient than the "udder"
type for collect.ion of a large llumber of fractions of distillate, but have the
disadvantage that the pressure in the system builds up while the receiver
is being changed. In the device shown in Figure 9 (upper left) a first
fraction is collected in D, which may be gl'aduat,ed, with A and B open and
C closed. Then A is closed, C is opened, and the distillate flows into E.
C is now closed, A is opened, and the three-way cock B is turned to let air
into E. E is removed, replaced with an empty receiver, and the operatioll
repeated. The receiver shown in Figure 10 is satisfactory 46 for operatioll
in the range of 1 to 760 mm. pressure. A metal valve,47 which eliminates
leakage at the most critical place, is conneeted between the two sections
of the receiver by means of Kovar-to-glass seals. 4B First, the valve is
soft-soldered to a brass block, 1 X 3 inehes, which is thon bolted to a Tran-
site sheet, 29 X 6 X 0.25 inchcs. The valve is set at a slight angle to the
plane of the Transite to ensure complcte drainuge from the valve. The
short sections of Koval' tUbing are soft-soldered to the valve to make a
vacuum-tight connection. Stopcock A may be closed as required to clean
ILnd lubricate three-way stopcock B; C may be lubricated when the lower
receiver is at atmospheric pressure. After removal of a fraction from the
lower receiver, the latter may be re-evacuated to column pressure either by
opening B very slowly or by using an auxiliary pump to lower the pressure
to that of the column and then turning B to eonnect the lower receiver to
the column. Distillates corrosive to Koval' or the metal valve should not
be collected in this receiver.
The third kind of fraction cutter"·1 consists of a ground-glass, ball-and-
socket valve opened and dosed manually by changing the tension in an
attached stainless~steel spring.
B. VERTICAL S'l'ILLS
For the operation of vertical stills, Chapter II (Part 2) should be con-

sulted. Down to the pressures required for molecular distillation, no
modifications of the pot are required. Two side tubes may be sealed on at an
angle, for introduction of a capillary tube and thermometer. If the pot is
sealed to the column, a side tube for introduction of distillancl into the pot is
necessary. The pot may be lagged with asbestos cord (see Chapter II,
Part 1).
47 Hoke, Incol'pOl'}l,ted, Englowood, New Jersey.
46 Stupakoff Ceramic nnu Manufllcturing Company, Latrobe, Pellnsylvania.
484 .T. R. BOWMAN AND R. S. 'l'IPSON
The principles of col~Lmn design for reetificatioll in the rangc of a few

centimeters of mercury have been disetlRKed in Seetiolls III and 2. Here it
may be re-emphnsized that the mail! consideration ill designing columns for
operation under diminiNlwcl pressure is (L low 1J1'c8swrc drop, and this can only
be achieved by employing a wide, relatively short tuhe with little, if any,
packing. N early all the eon ventional types of column paeking (see Chapter
II, Part 1, and Chapter IV, Sect. IlA) are useful'18n down to pressures of a
few centimeters of mel'eury. Below t.his, none give efficient performance,
and recourse must then be had to thermul rrct,ifieation (see Sect. II2B).
Columns particUlarly suitable for distillations nnder diminiHlwd pressure in-
clude thoRe containing; special glass or metal heliees, ,wand the Widmer
column (see Cha.pter II, Part 1). They may be lagged with asbestos cord,
or be vacuum-jacketed (see Chapter II, Part 1). Another excellent column,
useful for micro-,flil semimicro-,51 and ma(~rodiHtil1atiolls,r'2 eOllsists of a long
tube placed cOIwcntriL:ully in the eolumn tuhe (soe Chapter II, Part 1).
A column of this kind is depicted in Figure 11. It has boon 1160<16:1 suecess-
fully for the separation of mn.tArinls boiling anywhere between -30 and
250°C. at pl'eHSnreR from 5 to 7(iO mm. of mercury, for measurement of
vapor pressllres of pure liquids, for the annlysis of mixtures of close-hailing
componcntR, and in the preparation of pure eompOllllds for synthetic pur-
poses. Sinee the holdup is only abo1lt 0.25 mI., slLmples n.s small as 2 ml.
may be employed. Rtills ha,ving fnwtiouating efficicneies of 50 to 85
theoretical plates luwo been made. The temperature of the outside of the
vacuum jacket neal' the middle of the (Jolumn and the distillate tempera-
ture at the top of the column are moasured by means of copper-constantan
thermocouples. The upper thermo(JonpJe is inserted through the opening
at the bqse of the centntl tuhe and has the junction located at its top.
The design of the take-off regulator is shown in the detail drawing on p. 486.
Dro~~ fro;tJ; the eccentric tipof the refhix condenser are split on the partitioning de-
Vice."1'·:.
0 that a readily varied fraction of each drop returns to the column and the
rest'eJ s through the V-shaped opening, V, into the receiver. The reflux return
.impinges on the center of t1 grooved glafls ball which is sealed to the top of the inner
. ~8. Oompare: Feldman, Myles, Wender, and Orchin, Ind. Enf/. Chern., 41,1032 (1949);
Struck Ilnd Kinney, ibid., 42, 77 (1950).
49 Wilson, Parker, and Laughlin, J. A,rn. Chem. Soc., 55,2795 (1933). Longenecker,
J. Soc. Ohern. Ind. London, 56, 1H9T (11)37); Oil and Soa]), 17, 53 (11)'10). Norris,
Rusoff, MilicI', and Burr, J. Bioi. Chem., 139,199 (1941). Compare: Todd, Ind. Eng.
ChBIn., Anal. Ed., 17,175 (1945);u. S. Pat. 2,387,479.
50 L. C. Craig, Ind. Enf/' Chem., Anal. Ed., 9,441 (1937).
11 Naragon, Burk, and Lankelma, Ind. Eng. Chern., 34, 355 (1942).
U Selker, Bnrk, and Lankelm!t, Ind. Eng. Chem., Anal. Ed., 12, 352 (1940). Hall and
Palkin, ibid., 14, 807 (1942). Nal'agon and Lewis, i1>id., 18,448 (1946).
63 W. F. Sager, private communieation.
V. DISTIU,ATION UNDER MODERATE VACUUM 485
tube and ~epar:1tcd from the shoulder of the outer tube hy means of two gro(Jv(J([
glas,; straps Rmtled here across the anllular space. This ensures uniform distribution
of the reflux return and obvitLteH chaullolling. V provides a liquid seal to prevent
escape of vapor into the take-off side arm. The distillate leaving the latter is
cooled by jacket 13. At D if; cOllnected a leak-free system comprised of a dry-ice
trap, manometer, 1-liter ballast bulb, and a three-way stopcock which can be vented
to the atmosphere or connected to the pump. Stopcock C, in the pressum-equalizer
tube on the head, facilitates removal of water from V if a trace of water is present
in the distilland. By closing C and quickly varying the pressure, a clt·op of water
sealing the take-off is blown out. The distillate drops from the side arm into a
graduated centrifuge tube, T. Hole, H, in ground joint, J, is large enough to let
the drop fall directly into T. The receiver changer may be rotated on J for inser-
tion of a clean tube T. For operation at low temperatures, dry ice or liquid nitro-
gen may be introduced as coolant. The pot is insulated from drafts, and the
column, mounted exactly vertically, is vacuum-jacketed. The jacket temperature
is kept within 3 or 4 0 of the head temperature. The initial boiling rate for liquids
boiling at about 100° at 760 mDl. should be such that at least 30 minutes elapses
before reflux I1ppcars at the head.
A somewhat similar column has a stainless-steel rotor.M It operated at
maximum fnwtionating power at 300 r.p.m. At a boil-up raLe of 300 cc. per
hour, the pressure drop was 0.03 mm. of mercury. The pressure range
employed was 0.01 to 2.5 mm. of mercury for still-head temperatures of 80
to 200°C. A mixture of di-n-butyl phthalate plus di-n-butyl azelate has
been used66 for evaluating low-pressure distilling columns. Provision of
reflux offers no problems different from those encountered in ordinary dis-
tillation (see Chapter II, Part 1).
A typical verticllJ stil1 56 ,67 for fractionation under diminished pressure is
shown in Figure 12. Part of thc column is vacuum-jacketed and provides a
slight fractionating effect. As was pointed out in Section III, high-velocity
vapors exert. strong tangential drag 011 any liquid film past which they flow.
In the range of a few millimeters of mercury, this frequently becomes trou-
hlesome, and special designs must be provided to prevent the vapor from
drawing a liquid film lLlong with it. A brge bulb with a ring-seal insert or
annulus is a convenient device for accomplishing this (Fig. 12), Entrain-
ment of crude distilland presents a similar problem. If rough boiling
occurs, distilland spray is apt to be thrown into the vapor stream and
carried along with it into the distillate. Of the several devices that have
M Birch, Gripp, and Nathan, J. Soc. Ghem. Ind. London, 65, 33 (1947).
~6 Williams, Ind. Eng. Chem., 39, 779 (1!)47). Compare: Feldnlan, Myles, Wender,
and O]"eitin, ibid., 41, 1032 (1949).
66 Barr and Anhorn, Scientific and Industrial Glass Blowing and Laboratory Tech~
niqlle,l. Instruments Publishing Co., Pittsburgh, Pa., 1949, p. 364.

61 Compfire: Cowan, Falkenberg, and Teeter, Ind. Eng. Ghem., Anal. Ed" 16,90 (1944).
486 .J. R. BOWMAN AND R. S. TIPSON
been proposed for redueing entrainment, n, Himple off Ret. or bend in tho
column line (Fig. 12) iii probttbly t.he most cff()etive itS n mist trap. It hail
the advantage, which several others lnc:k, of only slightly increasing the
pressure drop.
Fig. 11. Coneeutric-tube cohunn.53 Fig. 12. Typical vertical

vacuum still. 60
Even for distillations performed at pressures of a few millimeters of

mercury, specially designed condcn,sers al'C unnecessary; those eonveIl-
tionally employed at atmospheric pressure (see Chapter II, Part 1) are
nearly always satisfactory. Ail' cooling, is, however, more frequently used;
in vacuum distillation of high-boiling compounds, condensation at the
temperatures employed in conventional atmospheric-pressure distillat.ion
of lower-boiling compounds may be unnecessary. In other respects,
vertical stills resemble horizontal stills. Similar receiver changer8 may often
be employed,58 but speeiu-l designs are sometimes necessary. Two examples
are shown in Figure 13. In the first,69 the drops of diHtillate fall from the
pointed tip of a "finger" condenser into a cup on the end of a capillary
tube, through which they pass to the collector and down a bent rod into the
~8
Wenk and Schuwirth, Z. ph1/siol. Chern., 267,260 (1941).
Ellis, J. Soc. Chern. Ind. London, 53,77 (1934). Compare: Bailey, Ind. Eng. Chem.,
69
Anal. Ed., 14, 71 (1942).
v. DISTILLATION UNDEH MODERA'l'E VACUUM 487
receptacle. If necessary, a trace of inert gas is ndmittcd occasionally

through the stopcock, to force the liquid through the capillary. The
other design 60 is self-explanatory.
Fig. 13. Receiver changers for vertical stills.""·GO
Vertical stills for thermal rectification (see Sect. II2B) are depicted in
Figures 3 and 4. The literatureO l should be consulted for operating in-
structions.
2. Pumps
The pressure given by t.he pump on a closed system should be tested
before starting iL diHtillation. There has been much difference of opinion
as to the type of pump most. advantageous in vacuum distillation. For
distillations down to about 10 mm. of mercury, a water-aspil'ator pump is
satisfactory and dependable. Such pumps give a pressure of about 10 mm.
of mercury in winter (water temperature, 10°) and about 25 mm. in the
summer (water temperature, 25°).
To obtain pressures below these, any of the small, mechanical, oil-
immersed rotary pumps offered by laboratory supply houses are satis-
factory (see Chapter VI, Part 2). If in good condition, these produce pres-
sures far lower than necessary for pot-ancI-column distillation inasmuch as
they should all give about 0.01 mm. of mercury, below which further
reduction of pressure causes no change in performance of this kind of
CDFlosdorf and Palmer, J. Rheology, 3,205 (Hl32).
!ISchaffnel', Bowman, amI Coull, 'buns. Am. lnst. Chern. Engr8., 39, 77 (1943).
Byron, Bowman, and Coull lnd. Eng. Chern., in pres,.
488 J. R. BOWMAN AND R. S. TIPSON
apparatus. There is no need to employ a diffusion pump in oreler to reduce

the pressure further, except when molecular distillation is required, and then
special apparatus mm,t be used (see Chapter VI). Sinco the pressure given
by an oil pump mny be too low, it may be necessary to incorporate !L mano-
stat (see Sect. III3) in tho system. There is also mueh misunderstanding
as to the nature of the oil required for such pumps. Its properties are not
nearly so critical as certain manufacturers would have their customers be-
lievo, alld no advantage is to be gainell by the use of speeial, supposedly
highly refined, pump oils. A good, SAE 30 or 40, automotive crankcase oil
is just as effective. Tho principal limitation of the vacuum that can be
obtained by mechanical pumps is the air diHsolved in the oil. On con-
tinued operation over a period of several hours the amount of air dis-
solved in the oil beeomes reduced, and the pump will give a higher vacuum.
It is therefore advisable to run the pump on a elosed system for several
hours before performing IL distillation, jf the maximum a,ttftinaLle vacuum is
desired.
The oil in vaeuum pumps tends to ahsorb volatile organic compounds
strongly, and after an appreeiable amount has been so dissolved the pump
will not give sueh a high vacuum (see Sect. III4). The condition of the oil
in this respect is far more important than the quality of oil with which the
pump was ol'iginally charged. Hence, it iH good practiee to usc a cheap
oil and change it frequently. Volatile impUl'itieR may sometimes be removed
by running the pump open to the air for soveml hours, or by heating the oil.
For operating rangeH of vacuum pumps, see Chnpter VI, Part 2.
3. Manostats
In attempting to reproduce conditiolls described by a previous investiga~
tor or in compal'ing boiling point!'> of a number of eornpounds, it is neeessary
to be able to produce any desired pressure. FUl'Lhel'more, it iH desirable
to maintain a constant pl'eSSUl'e or to produce a series of successively lower
eonstant pres~mres in the eoursc of a single distillation. Pressure regulators
may be manually operated, or automatic, i.e., manostats (see Chapter II,
Part 1, Sect. VI2, and Thomson 62 ).
4. Traps
For use with a water pump, the only kind of trap needed is that shown
in Figure 5, which preventfl baekflow of water to the still. In distillations
employing an oil ]Jump, it may be nenessary 1;0 prevent ingress of water
G2 Thomson, "Dnt,(lrmiul1t,ion or Vt~pOl' Pressure," in W()issi.lGl·!,!;er, ed., Physicltl l>feth-
ods of Organic Chcmistl·Y. 2nd ed., Intnl'seionco, New York, Part I, 1!J4{), Chap. V, Sent.
m.
V. DISTILLATION lINDER MODERA'J'E VACUUM 489
vapor, which condenses and causes the pump oil to foam; this may be
accomplished with a drying to\yor connected between the receiver and the
pump. In certain cases, use of n chemical trap, e.g., of soda lime, may be
advisable for removal of corrosive vapor". Absorption of vu.pors by the oil
can be largely obviated by eonnecting a cold trap in the vacuum line. 03
Suitable traps are shown in Figure 33 of Chapter VI, Part 2. For moderate
vacuum the connecting tubes may be of Hmallel' diameter than shown in
this figure. Cooling with dry ice in a Dewar flask is generally adequate.
A contact liquid should be provided. Acetone is convenient for this pur-
pose, but the liquid eutectic mixture of carbon tetrachlO1'ide and chloro-
form, which has about 1: 1 composition, is better because the dry ice floats
on it and gives uniform cooling throughout the depth of the bath by con-
vection. Flll'thel'mol'e, the fire hazard is eliminated. For operation down
to the 0.01 mm. of mercury range there is no need for liquid-nitrogen cool-
ing. For safet.y, Dewar flasks should be wrapped with adhesive tape .
.5. Gages
Indication of the operating pressure requires some kind of vacuum gage
(see Chapter VI and Thomson M ). From 760 to 10 mm. of mercury, the
~A
~
c __,--'
A B G
Fig. 14. Vacuum gnges.
simple gage shown in Figure 5 may be employed. An inverted U-tube

over 760 mill. high, one end of which dips into a contitiller of mercnry,
S3 Hickmnn, J. Franklin Inst., 213, 110 (1932).
490 J. R. BOWMAN AND H. S. TIPSON
may also be used. For operation fronl 10 to 1 mm. of mercury, ~t dosed-end

mercury manometer (Fig. 14A) is adequately l1ccumtc. Bulow 1 mm. of
mercury, 1L small-bulb McLeod gage 65 (Fig. 14B and C) is generally pre-
ferred. Rotary McLeod gages 66 (Fig. HB fmel Chapter VI, Part 2, :Fig.
41) are compact and particularly cOllvenient; they may be obti1ined for
the ranges 0 to 1, 0 to 5, and 0 to 10 mm. of mercury. For luwer pressures
there nre many satisfaetol'Y types of gage, discllssed in Chapter VI, Part
2, and by Thomson. 67 Fol' operating
mnges of vncuum gagcs, see Chapter
VI, Part 2.
All (jollventional gages nre sui table
only for measuring pressure8 in portions
of the system not containing Vf1por. To
avoid proHsure-drop errors, the gage
should be connccted into the system as
dose to the still pot as possible. This
is ordinarily interpreted to mean ne::.r
the eoid end of the condenser. At pres-
snres down to the range of millimeters
of mercury this is satisf:.tctory, because
the pressure drops due to v::.por flow
are small relative to the total pressure
in the system. In and below the range
of millimeters of mercury, however, the
pressure drop:,; are relatively large, and
Fig. 15. AutomI1ll0meter.@ determination of the true pressure in tho
pot is more difficult. Some efforts have
been mnde to achieve the determination by use of null-indicating, dia-
phragm-type gages, but a more ingeniOUf-l dovice G8 termed an "aut,oma-
nometer" gives the result directly. 60 In it, the vapor leaving the pot
passes upward through H small jet which iH part of 11 miniature diffusion
pump (Fig. 15). During operutioll, this ereates n relatively perfect
vacuum in a referencc vacuum spucc which communicates with th,e still pot
through a U-tube manometer. The U-tube becomes filled with distillate
65 McLeod, Phil ..M'ag., 48, 110 (1874). Gnelle, Ann. PhY8ik, 41,289 (1013). Raiff,
Z. InstrumenUc., 34, 97 (1914). Pfund, PhY8. Rev., 18, 78 (1921). Cox, J. Optical Soc.
Am., 9, 56!) (1924). R. ,J, Chtrk, ,J. Sci. Instruments, 5, 126 (1928). Hickml1u, J. Opti-
cal Soc. Am., 18,305 (192(l). Booth, Ind. En(J. Ohmn., Anal. Ed., 4, 380 (1932).
66 Hamlin, J. Am. Chern. Soc., 47, 70!) (Hl25). Brl)IlIllll', Ilelv. Chim. Acta, 13, (l15
(1930). VOll Meycren, Z. phllSilc. Chell~., AI60, 272 (lfJ32). MOHer, Physik. Z., 36, 1
(1935). Flosclorf, Ind. Eng. Chern., Anal. Ed" 10, 53,t (1038); 1'7,198 (194.5).
67 TholIlSon, loco cit., Seets. JIll-6.
68 Hickman and Weyerts, J. Am. Chern. Soc., 52, 4714 (1930),
61 TholIlSon, loco cit., Sect. III2.
v. DISTILLATION UNDER MODERA'rE VACUUM 4Hl
during opel'ation and indicates dircctly tho cli[f()l·cnce hetween the pressure
in the still pot and the refer{)me vacuum.
IV. OPERATION
Reduction of pressure in distillation processes necessarily introduces
several changes in the methods employed for atmospheric-pl"P.ssure opera-
tion; these include degassing, control of vaporization, following the course
of the distillation, and control of pressure.
1. Degassing
The charge for a batch vacuum distillation is nearly always saturated
with air, large amounts of which are dissolved by most organic liquids.
Air must be removed before the temperature reaches a point at which
oxidation occurs to a deleterious extcnt; this is ordinarily effected very
simply by reduction of pressure in the apparatus after putting the batch in
the still pot but before heating. With highly viscous oils, a moderate
amount of heat may be required to make the charge sufficiently fluid for
adequate convection. In any case, application of heat is to be avoided
before the vacuum is applied. A particularly striking example is in the
distillation of paraffin wax. If the wax is warmed just to the melting
point and then held undor vacuum until hubbling ceases, water-white
distillates are readily obtained. If, on the other hand, the wax is rapidly
heated to the distillation temperature under vacuum, all distillates will have
a dark color due to oxidation products.
Many liquids foam strongly, and caution must be observed in applying
vacuum during the degassing, to avoid carrying distillancl through the
column and iri.to the receiver. In such cases, the pressure must be reduced
very slowly. As a check on complete degassing, the stopcock in the vac-
uum line sllOuld be momentarily closed. and the rise in pressure noted. If
appreciable over a period of several minutes, low-temperature degassing
should be continued until a subsequent similar check shows substantially
constant pressure on standing. (Jnly then is it safe to raise the tempera-
ture of the still pot to the distillation value.
It is desirable to remove traces of volatile solvent before distilling a
high-boiling distill and at the oil pump. This is often readily accomplished
by warming in a hot-water bath, together with entrainer-vacuum tre&.tment
at the water pump.
2. Prevention of Bumping
Numerous devices have been suggested for preventing Ilbumping."
69a
These may he Bl)proximately evaluated. in advance on the basis of the

OOn Compare: Egly, "Heating and Cooling," ill Weissberger, ed., T'echniqu~ of 0,·-
ganic Chemi.~try, Vol. III. Illterscience, New York, 1950, Cbap. I, Sect. II3D. .
492 J. R. BOWMAN AND R. S. TIPSON
theorcti(m] considerations given in Section II3. Thn chRsicaJ expedient

is to introduce a few smnlllumpH of pnmice "Lono, asbes(;os, or porous pot.
This is generally effective [1t pressures clown to [L few millinwtfH'H of mercury,
as long as boiling is maintained. The action of the sLone is obviously to
supply many minute gas pockets from which bubbles can grow. If boiling
is interrupted, or if an attempt is made to rense the lumps, the results are
lIsually unsatisfactory because, once used, tho permanent gas entmpped
in the stone quickly becomes displaced with condensable vapor, and this
in turn is replaced by liquid as soon as vaporization stops. Consequently,
fresh granules should be employed for eaeh run, and steady boiling should
be maintained. Ebullator tubes 70 eonsisting; of fine capillaries scaled at the
upper end can similnrly be used but once. A better expedient is to pack
ghLss wool into the distilling fiask to a level just a,bove the liquid surface. 71
Such compounds a,s sodium bicarbonate,72 whieh deeompose with gas evolu-
tion on heating, are sometimes ndded to the distilll:ind if there is no dele-
terious chemical effect. Of course, in this and .other kinds of entrainer-
vacuum distillation, the entrainer Cl!tUses reduction of the boiling point. 73
A number of [1ntibumping devices depending on creation of local super-
heating have been proposed. Metal nnll glass bends fall into this class.
They [l,et, by confining a thin, stngmnt film of distilla,nd at small areas of
the bottom surfaee of the still pot, cmlsing that film to become strongly
superheated. SlIeh devices are, in general, unsatisfactory for vacuum
dist,illa,tion. Other methods of causing local superheating include use of
electric heating near the surfaee,74 an inner heating coil, or a wire 76 sealed
through the bottom of the pot.
A third type involves the use of lumps of material not wetted by the
distilland. Probably the most llsefll1'6 is Teflon (polytetmfiuol'oethylene).
This high polymer is chemically highly unreactive and withstands tem-
peratures up to at least 200°0. It is not wetted by most liquids and con-
sequently maintains a very thin film of gas over its entire surface when im-
mersed in a liquid. This film can serve as a nucleus for bubble formation
and growth.
Last and probably best of such antibumping devices are sintered-glass
particles fused to the inner, bottom surface of the boiler. 71 Such a surface
is readily prepared by putting a small amount of ground glass, preferably
10 Markownikow, J. Rus8. Phy.~. Chem. Soc., 19,520 (1887).
71 Angeli, Gazt. chim. ital., 23, II, 101 (1893). AndClrlini, ibid., 24, I, 190 (18lJ4).
12 Gettler,.Niederl, and Benedetti-Pichler, Milc"Qchernie, ll, 165 (HI32).
1a Schierholtz and Staples, J. Am. Chem. Soc., 57,2709 (1935).
H Lcschewski, Z. anal. Chern., 89, 50 (1932).
7& J. H. O. Smith a.nd Milner, Milcrochemie, 9, 117 (lB3!).
7G Bennett, Bcience, 106, 646 (1947).
77 Morton, Ind. En.(/. Chern., Anal. Ed., 6, 384 (1934).
about .50-mesh, into the empty still pot and heating its bottom until· the
glass softens sufficiently to cause the powder to adhere. The resulting sur-
face 'will nearly always give smooth boiling if smoot.h boiling is possible at
all. The action probably depends both on minute entrapped air bubbles
and on creation of minute spots which become superheated. An excellent
way of eliminating bumping is to stir the distilland mechanically. 78 The
::;tirrer may be motor-driven by a shaft through a vacuum seal 01' may be
magnetically operated.
Finally, entrainer-vacuum distillation may be employed. As mentioned
in Section II3, smooth boiling cannot be obtained at pressures below ahout
1 mm. of mercury except with great superheating. Under these conditiomi,
satisfactory operation is often obtained by quiescent vaporiz[ttion from the
surface of the liquid, no buhbles being formed at all. A superior method is
entrainer-vacuum distillation, in which an inert gas is ponducted in minute
bubbles to the bottom of the distilland through a suitably fine capillary
tube (see Section HIlA).
3. Determination of Fractions
In distillation at atmospheric pl'e::;8Ul'e, the course of the fractionation is

almost invariably followed by observation of the temperature of the vapor
just before it cntcrs the condenser. This type of control is no!; usually
applicable with precision to distillation at pressures below a few centimeters
of mercury, because the relative fluctuations in pressure, even with a good
vacuum-regulating system, may be large relat.ive to the total pressure in the
system, and the corresponding variations in temperature render that in(li-
cation inadequate for characterizing the course of the distillation. This is
especially true in the range of millimeter::; of mercury, where, even with the
best efforts to achieve reproducibility, temperatme differences in eheck rum;
are frequently of the o1'clel' of several degrees centigrade. N evel'theless,
valuable subsidiary evidence is sometimes acquired by recording the bath
and vapor temperatures. In precise work, the method of comparative
measurements 711 may be employed.
For following the course of a vacuum distill[Ltion, an adequa.te equivalent
of temperature has not been found. The method of comparative measure-
ments, employing differential thermocouples for a stl1ncll1,l'd distillancl and
the unknown mixture, respectively, is valuable. No other continuously
indicative property has beon developed for use during l1 dist.illat.ion, except
in those cases,ill which fractions are of different color, fluorescence, optical
Birch, Gripp, and Nathan, J. Soc. Ohern. 'Ind. London, 66, 33 (1947).
78
See Swietoslawsld and Anderson, "Determination of Boiling and Condensation
79
Temperatures," in Physical Methods oj Organic Chemistry. 2nd ed., Part, 1,19'19.
494 J. n. nOWMAN AND n. S. '['IPSON
activity, etc. The only means of control normally available involves

measurement of physical properties of the individual iradions tdter the
distUlation has been completed. Consequently, when sharp sepan),tion
is required and the compClsition of the charge is not accurately known,
recourse must be had to taking many small fractions, followed by deter~
millation of one or more physical properties of each which will serve for
characterization. Refractive index is most frequently used for this pur-
pose, but viscosity, density, surface tension, optical uc1;ivity, and other
properties are convenient for special cases.
4, Control of Pressure
Oontrol of the Vacuum h[LS been discussed in Section IlI3. In addition,
the following procedure is frequently useful: When a distillation is under
way, the minimum useful pressure at the cold end of t.he condenser cn,n be
determined by keeping the throughput constanti und reducing the pressure.
The temperature of the vapor wHl be observed to decrease asymptotically
to a rather definite limit beyond which n further reduction of pressure has no
observable effect. The pressure [Lt which this limit is reached is n char-
acteristic of the appftmi;us. For conventional pot stills, its value is usually
above 0.1 mm. of mercury.
General References
Barr and Anhorn, &ientific and Industrial, Glass Blowing and Laboratory Tech-
niques. Instruments Publishing Co., Pittsburgh, l)n., 1949.
Carney, Laboratoy Fractional Distillation. Macmillan, New York, 1949.
Dushman, Scientific Foundations oj Vacuum Technique. Wiley, New York,
1949.
Farkas and Melville, Ej;pcrimental Methods in Gas Reactions. Macmillan, Lon-
don, 1939.
FOl'bes, SlWl·t History of the Art of Distillation. Brill, Leidell, Holland, 1948.
Griffiths, "Some Problems of Vacuum TechniquG from a Chemical Engineering
Standpoint," Trans. Inst. Chern. Engrs. London, 23, 113 (1945).
Guthrie and Wakerling, Vacuum Equipment and Techniques. MQGraw~Hill,
New York, 1949.
nose, A. and E., Distillation Litc1'Uture, Index, and Abstracts, 1941-1945. Rose
and ROBO, State College, Pa., 1948.
Sanderson, Vacuum Manipulation oj Volatile Compmmds. Wiley, New York,
1948.
Stage and Schultze, 'l'heol'ie, Apparatc sowie Verfahl'en del' Destillation und Rek-
tifikation. Hobart, Washington, D. C., 1947.
Vilbrandt, Bryan, Couper, and Dorsey, "Distillation Bibliography," Bttll.
Virginia Poly tech. Inst., Sar. No. 62, XXXIX, No.9 (1946).
CHAPTER VI
DISTILLATION UNDER HIGH VACUUM
Part 1. Distillation. E. S. PrmnY. . . . . . . . . . . . . . . . . . . . . .. page 495

Distillalion Prod/Lets Industries
Part 2. The Vacuum System. J. C. HECKER. . . . . . . . . . . . .. page 540
Distillalion P1'Od'llcts Inriu,tl"ics
SYMBOL INDEX
Symbol Definition Page
n . ............ . Efficiency factor in ruLe equ!Ltion ....................... . 504

01, Cl2 . . . . • . . . . . Factol's in Knudsen equation .......................... . 545
i1. " ........ " Area .......... , ..................................... . 546
/) ............. . Elimination rate factor. , ......... , . , ... , ............. . 5HJ
C" Ce ...•..•.•. Factors in Knudsen equation .......... , .... , ........ , .. 545
c2 ••••••••• •.••• Root. mean squaro velocity ............... , . , .. , , ...... . 502
d ............. . Density .. , ............ , ............................. . 502
tl ...... ...... . Diameter ........................................... . 546
d . ............ . Distance octween successive molecnlar collisions ......... . Fig. 1
D ............ . Hazard index, ............... ' ..... , . , ., .. , .. , ..... ' .. 507
Dh ... ........ . Log of hazard index.... .... . ....................... . 507
Ii ............ . . Molecular diameter .................................. . 544
7] . • . . . • . • • • • • • • Viscosit.y ............................................ . 544
P ...... " " .... Flow through tubing; or system ........................ . 1142
P o.•.......•..• The factor A/v'2'1J'pt ................................. . 546
h ............. . Height ............................................. . 555
II . ........... . Perimeter of tube 01' duct ... , ......................... . 551
k ............. . Erg per degree Kelvin, Boltzmall constnnt .............. . 544
j ( " " " ..... " Factor in flow formub fOl' n!\,l'l'OW slits .................. . 558
[ ............. . Length ............................................. . 1546
A"." ...... ". M can. free path ...................................... . 544
In . . • . • . • . . . . . . l\1ass ............................................... . 502
1/1. . . . • • • • . . • . • Elimination mte factor (units of time) .................. . 510
.AI ............ . Moleculr1l' woight .................................... . .503
J1. • •••••.••••••• Micron (O.OOlmm.) .................................. . 541
n . .. , ......... . Number of molecules ............................... , .. 502
NA " .. · . . . . ,,· Avogadro Number ........................... ,., ...... . 544
P"" ........ .. Pressure ............................... : ............ . 501
P/L'"'''' " .. .. Pressure in microns .................................. . 507
P/LI,"""""" Pressure in microbars ................................. . 541
495
496 E. S, P1TIRRY
RYlllbol Definition
-~------.----------------------
I'WRBIll'(l in millinwt(,],H ............................... . . 555

p ............. . Part.ial Jll'l~RHUJ'(~. . • . •• • . . . . . . . . . . . . . . . . . . . . . . . . . • . • • . • 508
(2 ............ , J';vaporatioll rate .................................... . 501
(2 ........ . Hate of gas flow ..................................... . 542
R ............. . Univ<JrRlll gas eOIlHkLIIL ......•.......................... 1101
R ............. . H.!'YlloldH Nmnher ................................... . 562
r .. ... , . , ..... . j':limiwtj.ion rate fnr,tor ......................... , . , ... . IiHJ
1' . . . • . . . . . • • . . . Hadius" .... " ... ,' ".,., ..... ,.,.,', .............. . 546
p ...•. . . . . . . . . . DOllsiL.v of gaR 0]' VLL]>Ol· ............................... . 54.4
,.'), ............ . Pumpillg s]JUuli ............................. , ........ . 542
'IT, ...... , ..... . DegmeR rL)Jso]uie (Kelvin) ............. " .............. . 501
t ............. . Time ............................................... . 507
'n • • . . • . . . • . . . . . Vcloeity ...................... , ..................... . 502
" ............. . Unit volume ......... , .............................. . 502
V............ . Gram molcclilar volume ........................... , .. . 502
1' ............ . VolUlne of spane heillg (-)vaclIaLn(l ...................... . 573
w ......... .... . MaRS Ht.l'ildng unit. ll],(llL in unit. t,imo .................... . 502
Wi' ........ , .. l~lilllinllt,joll rat.n ...................•................•. 51\)
~V ............ . .Ii\wt,m in mo\('(\u\ar !low C1lutdioll ...................... . 551
X·, ••••....•.•.. FILetor in fiow formula fill' ('Ollcl!utl'ill tuhe .............. . 558
Z ............ .. Fado)' in gaR flow eqlllLtion .. , ..................... , .. . 548
Zi············· J';liminatioll l'llte fact.or ...... , ........................ . 5UJ
Part l. J) 1 S 'r I L L4. T ION
I. INTRODUCTION
In the chemist's desire to purify organic substances of ever-increasing

molecular weights by distillation, he has had tiO resort to subatmospheric
pressures to avoid thermftl destruction of his material. Vacuum distilla-
tion has long been practiced, but its progress has had to paee the develop-
ments in the technology for creating vacuums. In spite of innovations
introduced by Langmuir, Gaede, ILnd others in the methods of producing
vacuity, the vacuum still reached a limit beyond which lower pressures
yielded no further lowering of the boiling point. This condition was fi-
nally recognized as being due to the impedance offered by the still to the flow
of vapor at low pressures. The tortuous paths interposed between boiler
and condenser in the conventional distilling appamtus no longer benefited
the process but became a harmful barrier and obstruction in the natural
path of the vapor molecules, thus causing their destruction.
Subsequent developments in high-vacuum distillation were aimed at
altering the shape of the apparatus. The necessity of simplifying the stiIJ
VI. DIS'l'ILLA'I'ION UNDER HIGH VACUUM 497
gradually became u.pparent and the early batch-type stills showing marked
departure in design from the conventional vacuum still were similar in
principle to the apparatus reported by Bronsted and Hevesy, I which they
used for the concentration of the isotopes of mercury. In the immediate
years following the first-published applications of moleculu,l' and unob-
structed-path distillation to organic substances, the process was intensively
examined by several independent groups: one under Waterman in Holland,
a second under Burch in England, and a third under Hickman in America.
The results of this pioneering work have brought to the chemist a useful
laboratory tool. The molecular still permits distillation at temperatures
50-I50 a C. lower than those required by other vacuum stills. It provides
a means of distilling organic substances having molecular weights up to
1200. Heat-sensitive materials having molecular weights as low as 150
may even be successfully handled by the method. .
The term "high-vacuum distillation" will be used throughout this sec-
tion as a generic term to include all the specialized kinds of high-vacuum
distillation processes which have been variously known as "molecular"2
distillation, "unobstructed path"2 distillation, and "evaporative"8 distilla-
tion. "High~vacuum distillation" will be useful in the general discussio~l
when specific reference to anyone of the above-named processes is not
necessary. The following pages are devoted to a description of the tech-
niques of this type of high-vacuum distillation which are useful to the or-
ganic chemist. No attempt is made to review the past developments in
the field since this has recently been done proficiently by Hickman 2 and
earlier by Burch and van Dijck,4 and Waterman and van Vlodrop.6 A
comprehensive bibliography including scientific papers and patents has
been compiled by Detwiler and Markley 6 and extended py Todd7 and by
Rose and Rose. s Pertinent patents have been noted in connection with
several of the stills discussed but no attempt has been made to include all
patents which bear on the processes and apparatus described, Part 1 of
this chapter is devoted to the general theory of the method and a descrip-
1 BriiIlRted and Revesy, Phil. Mag., 43, 31 (1922).

2 Riekman, Chem. Revs., 34, 51 (Hl44).
3 Bureh, Proc. Roy. Soc. London, A123, 271 (1929).
4 Bureh aud VI1U Dijck, J. Soc. Chem. Ind. London, 58, 39 (1939).
'Waterm!lu and van Vlodrop, Rev. chilli. indo Pm'is, 48, 314 (1939).
• Detwilel' and Ml1rkley, Oll & Soap, 16,2 (1()39). See n,lso: Detwiler, ibid., 17,
241 (1940); AI,slracts of Articles ami Patents on Molecular or Short-Path Distillation,
U. S. Dept. Agl'., Bur. Agr. Chem. Eng. ACE-1l5, U. S. Regional Soybean Industrial
Products Laboratory, Urbanl1, Illinois, 1941.
1 Todd, Oil & Soap, 20, 205 (1943).
8 A. a.nd E. Rose, Distillation T,itel'atw'e, Index and Abstracts, 1941-1945 .. E, and A.
Rose, State Coll~ge, Pennsylvania, 1948.
E. S. l'}QRHY
tion of the main types of laboratory stills. Complete illstnwtions are

illchJ.ded for performing high-vLtCllUm diRtillatiollH. In Par!; 2 are de-
scribed vacuum upparLttus and techniques n,s npplied to high-vacuum
distillation.
n. GENERAL ASPECTS OF HIGH-VACUU1\1 DISTILLATION

1. Basic Differences in Distillation Processes
In cOilventionuJ atmospheric distillation procesl'JCS, distillation begins
at a well defined temperature, there is a definite boiling point of the ma-
terial being distilled, and the distillation process is accompanied by ebulli-
tion of 1;he distilland. A dynamic equilibrium exists between the liquid
and v&por phases thus causing a large proportion of the molecules evaporat-
ing from the liquid surface to return to thc sarno liquid phase.
In contrast, in the high-vacuum distillation process there is no well de-
fi,ned temperuture at which dist.illation begins. Distillation occurs 11t any
~emperature as long as a thermal gradient exii':lts between eonclenser and
evapoI'[Ltol'. Due to the high vaCuum maintained during 1',he process,
there is effectively no superincumbent ail' pressure on the material being
distilled and consequently there is no well defined boiling point or attending
ebullition of clistilland. The diRtilling vapor molecules pass direetly from
the vi:Lpori~ing surface to the condensing surface without having to pas::;
through a barrier of air molecules and the number of distilling molecules
returning to the liquid phase is therefore insignificant. As a eonsequence
a. dynamic equilibrhlm between vapor and liquid phases docs not exist
dudng high~vacuum distillation. Ideal distillation conditions art'. attained
when the number of molecules being condensed is eqUfLI to the number of
molecules of distilland evaporating from the distilland surface ..
2. High-Vacuum Stills-General
Before turning to a consideration of the high-vtLeuum distillation process

itself, it is desirable to consider the general eharaei;eristies of high-vaeuum
stills which are such an integral part of the proeess. In general, high-
vacuum stills consist of an evacuable chamber in whieh a vaporizing sur-
face directly faces a condensing surface, the two placed only a short dis-
tance apart. The space between the vaporizer and condenser surfaees
has no physic""l barrier which can obstruct the pasflnge of vapor molecmles
while in flight.
In the high~vaouum still, it is necessary that a major proportion of the
molecules distilling from the liquid phase pass directly to the condenser
without returning to the liquid phase. ]'01' this reason, it is desirable for
efficiency of operation that the vaporizer-condenser sp~tcing be not sub-
VI. DIS'l'ILLA'l'WN UNDEH nIGH V,cCUUM 499
stantially greater than the "undefleded path" of the vapor molecules as

defined by Jacobs and Kapf'f.9 The term "undeflected path" may be said
to denote the distance that a major proportion of the distilling molecules
will travel under the high-vacuum distillation conditions without serious
deflection from a rectilinear path. Ol'dinarily the spacing is kept small so
that a variety of materials can be distilled without the neeessity of experi.
mentally determining the maximum desirable distance in each case.
The stills used in high-vacuum distillation are of two general types.
These are: (a) pot-type stills; and (b) flolving-film stills. As the name
implies, pot.type stills are those in which
the distilland is contained as a pool of
liquid in all evacuablc vessel; distillation
takes place from the quiescent surface of
the liquid clistilland to an adjacent con-
denser. In these pot-type stills, the boiler
01' evaporator eontains the entire batch
being distilled and the depth of the liquid
usually varies from a fraction of a milli ..
meter to several centimeters. The still is
necessarily of limited capacity when em- DISTILLAND
ployed with thermally unstable material
since the entire pool of distilland must be
maintained at distilla.tion temperatures
throughout the process. Typical pot stills
are the concentric flask still of Bri:insted
and Hevesyl and the concentric test tube Fig. 1. Concentric tube-type still.
still shown in Figure 1. Various still
types are illustrated and described in greater detail later in the chapter.
In high-vacuum distillation, only the molecules at the evaporating sur-
face enter into the process and, at any given instant, all other molecules in
the bulle of the distilland layer need not be considered as part of the distilla-
tion process. Their function is to replace the molecules evaporating from
the surface. The problem of molecular diffusion in the distilland is thus
an important fttctor because of the absence of the ebullition which accom-
panies atmospheric distillation. The problem of molecular diffusion is par-
ticularly aggravated in the case of binary mi}..'tures of large molecules
wherein these large molecules move slowly and with difficulty through their
closely touching neighbors. Stirring of the ditltilland serves to renew the
distilling surface but, like boiling, stirring is usually accompanied by splash-
ing with a resulting contamination of the distillate.
o Jacobs and Kapff, Ind. Eng. Chem., 40, 842 (1948).

500 1<i. S. FERRY
The need for thin layers of distilland and for the renewal of the liquid
surface was envisaged by the earliest workers in the field. Thus, Burch's3
first experimentnJ evaporative still waS designed so that the evaporator
consisted of n, shallow tra,y Imving n, large rtttio of surface area to depth.
The first progress toward increasing turbulence in thin distilland layers was
shown by Burch 10 in his industrial "cascade-tray" still. In this still, the
distilland was made to follow a Ryst.ematic course through a series of
troughs stacked one above the other and the distillate was collected on cold
plates whieh were interleaved between the troughs.
True "flowing-film" stilhl followed the "cascade-tray" still of Burch.
Examples of such stills are the "falling-film" still of Hickman,ll the "col-
umn" stills of Watermltn and Oosterhof,12 and Carl' [mel Jewell,13 and the
"centrifugal" still of lIiekman.14 In t.he "flowing-film" st.ills, distilland is
ml,used to flow in a thin film over a vaporizing surface either by gravity or
by other mcam. Distillation is from the thin flowing film directly to a
condenser which is coextensive with the vltporizing surfaee. "Flowing-
film" st.ills enabled distillat.ion to be effected from distilland films approach-
ing 0.1 mm. thickness and extended t.he effective range of dist.illation to
thermally unstable materials having molecular weights of as much as 1200.
3. Distillation Process
In considering the difltilbtion process, several ftwt.Ol's are import,ant to
the chemil'lt. These include the rat(~ of di8tillation, the thermal hazard,
!md the separating or fractionating ability of t.he process. These factors
will be briefly considered in order.
A. RATE OF DIS'rn,lJATION
As has been pointed out, there is no well defined t.emperature at which

distillation begins under high-vacuum eonditions. Instead, dist.illation
oecurs at any temperature as long us there is a thermal gradient between
the evaporator and the eondensel'. The rate of distillation, however, in-
creases rapidly with increase in t.empemture and the process is performed
at elevated temperatures merely as an expedient to effeet distilla,tion of a
given quantity of material in a reasonable length of time.
The rate of distillation for a specific substanee ean be computed from
10 Burch, Brit. Pat. 303,078 (Den. 21, 1928); reisslled itS U. S. Pat. 1,\)55,321 (Apr.
17, 1934).
il Hickruan,Ind. Eng. Chem., 29, 968 (1937).
U Waterman and Oosterhof, Roc. trail. chim., 52, 895 (1933).
U Carl' and Jewell, Brit. Pat. 41.5,088 (Aug. 17, 1934).
11 Hickruan, Chmn. RClm.! 34,51 (1944); U. S. Pat. 2,210,928 (Aug. 13, 1(40).
VI. DIS'l'lLLA'J'IIJN UNDER RlGIt VACUUM 501
Langmuir's equation lO for calculating the vapor pressures of metals. The

evaporative rate, Q, in terms of moles per second per square centimeter of
surface is:
(1)
where P is the vapor pressure in dynes per square centimeter measured at
absolute temperature T, 111 is the molecular weight, and R is the universal
gas constant equal to 8.3 X 107 ergs per oK. per mole. Langmuir checked
the validity of the equation by measuring the rate of evaporation of hot
metal filaments under high vacuum. The logarithms of the vapor pressures
calculated by equation (1) using his experimental data plotted against the
reciprocal of the absolute temperatures gave a.pproxirnate straight lines as
predicted by the Clausius-Clapeyron equation. Also, the vapor pressures
of certain metals were found to be in agreement with those previously re-
ported as determined by othel' methods. The analogy between metal
evaporations and high-vacuum distillation can readily be appreciated and
the rate equation was found to apply to high-vacuum clistillation processes.
The derivation of the Langmuir equation will therefore l)e considered in
the following paragraphs.
The theoretical derivation of thc rate equation is based on simple princi-
ples of the kinetic theory for the behavior of molecules in the gaseous state.
The kinetic theory postulates that there is a continuous flight of molecules
from the surface of solid or liquid matter to the free space above it. Simul-
taneously, molecules of vapor rcturn to the surf::Lce at a rate depending
on the concentration of the vapor. When the rate of condensation even-
tually equals the rate of evapol'l\,tion, the two phases are said to coexist in
a state of dynamic equilibrium. The concentration of the vapor, or more
precisely, its pressure, is dependent upon the temperature.
The vapor at the interface of a rapidly boiling liquid can justifiably be
considered to be saturated because the rate at which the vapor is being re-
moved as distilbte is undoubtedly very small as compared to the rate at
which th~ liquid is vaporizing and again recondensing to liquid. As the
temperature of the system is lowered, these processes of evaporation and
condensation continue at equilibrium but their rates correspondingly de-
crease. At temperatures so low that the vapor pressure of a substance
does not exceed a millimeter, Langmuir considers that the actual rate of
evaporation of the substance is independent of the presence of the surround-
ing vapor. In other words, the rate of evaporation of a substance in a
high vacuum is the same as the rate of evaponttion in the presence of its
saturated vapor.
The process of evaporation is difficult to treat theoretically because it
16 Langmuir, Phys. Rev., 2, 329 (1913); 8, 14G (1916).
502 E. S.PEItRY
involves among other things a consideration of molecular diffusion in the

liquid phase. The eDndensation process, however, can readily be accounted
for by certain aspects of the kinetic theory. 11; involves a gaseous state
and because of the low vapor demdties anticipated in the temperature
ranges of evaporative distillation, the vapor can be assumed to obey the
laws of a perfect gas. If an expression can be derived to account for the
rate of condensation, it is evident that this expression should also give the
rate of evaporation because of the original assumption that at equilibrium
the I'lI,te of condensation equals the rate of evapomtion. In the derivation
that follows, therefore, the expression finally arrived at. will be that for the
rate of condensation. Consider a unit; volume of vapor in contact with its
liquid smoface and suppose that the uSlml nssnmpt;iolls of t,be kinetic theory
of gases apply, namely, that the VELpor molecules arc small as compared
with the distance between thom, that; there are no forces between molecules,
and that, collisions between molecules are perfectly elf;1stic. Equilibrium
between vnpor and liquid is reached whon hltlf the molecules in the unit
cube are moving toward tho liquid surface and the other half are moving
away from it. If n denotes the number of molecules of mass rn in the unit
volume v at saturation, the quantity of vapor approaching the liquid sur-
faoe will be e/2 mn)/v or 1h el, where d equals the mass per unit volume or
the density. The average (IOmponent of velocity either to or from the sur-
face is 1/ ~ ~L, where 11. is the mean veloeity of the vapor molecules. There-
fore, the mass 'W of vapor which strikes a unit surface of the liquid in unit
time is ·the product of the vapor concentration and its mean velocity, or:
w=1/2dx 1/2u=1/4du (2)
The density of the vapor can be expressed as a function of its pressure
and temperature by means of the gas law. Since d = lYI/V the gas law
j
PV:;=; RT may be written:

d = PlYI/RT (3)
where V is gram molecular volume, P is pressure, JliI is molecular weight,
T is absolute temperature, and R is the gas constant. By the elimination
of d between equations (2) and (3) and rearranging terms, the mean-velocity
component becomes:
'U = 4w(RT/PM) (4)
The fundamental equation of the kinetic theory;
PV = 1/amnc2 (5)
relates· the pressure-volume product of a gas with the "total kinetic energy
of its molecules. In equation (5) nand m have their usual meanings, and
c2 is the root-mean-square velocity 01' the velocity obtainod from the
VI. DISTILLATION UNDER HIGH VACUUM 503
square root of l/nth of the sum of the squares of the individual velocities. Hi
By substitution of RT for its equivalent, PV, and M for mn, and solving
for c, equation (5) becomes:
c = V3RT/M (6)
where JJ[ is the molecubr weight when V is defined as the gram moleeular
volume.
Since the mmm velocity, !~, is related to the root-mean-square velocity,
c2, as follows: 16
or:
(7)
the mean velocity can be evaluated by combining equations (6) and (7)
to give:
'U = V8RT/rrM (8)
Finally, by combining equations (4) and (8):
w = PVM/27rRT (9)
which is the rate in grams per second pel' square centimeter at which the
vapor condenses on the liquid surface. At equilibrium, then, equation (9)
also gives the rate of evaporation in accordance with the premise set forth
at the start of the derivation of the rate equation.
Langmuil· 17 used equation (9) to calculate the vapor pressures of pure
metallic elements by measuring the loss in weight of a filament which
was held at an elevated temperature for a definite length of time under high
vacuum. His experimental results, particularly with tungsten, platinum
and molybdenum, clearly show the validity of the equation uncleI' these
conditions.
The vaporization of atoms from metal filaments at high temperatures is
not a process of dynamic equilibrium in the sense that the rate of evapora-
tion from the hot filament is equnl to the rate of condensation on the fila-
ment. Instead, the evaporating molecules pn,ss from the hot filament
surface to the colder walls of the container, where they condense and are
thus removed from the sphere of activity. A condition is thus established
IG The square root of the l'Oot-Dlean-squnre veloeit,y, c 2, differs from the mean velocity,
u; the lat.ter is v'8/31!' times the former. For a more extensive explanation the reader
is referred to: Glasstone, Textbook of Physical Chemisil'Y, 2nd ed., Van Nostrand, New
York, 1946, p. 253; or Loe!:), The Kinetic Thc01'y of Gascs, 2nd erl., MoGraw-Hill, New
York, 1934, p. 103.
17 Langmuir, Phl/s. Rev., 2,329 (1913); 8,149 (1016).
504 E. S. PERRY
in whiel! the rate of condensation on the hot-metal surface is practically

zero and the rate of evaporation proceeds normally without regard to the
condensation process. In general, equation (9) is applicable to the evapo-
rative distillation process. Discrepancies between measured and calcu-
lated distillation rates are largely due to molecular collisions and the devia-
tions from the theoreticall'ates can be accounted for by an efficiency factor,
a, as follows:
(lO)
The validity of the rate equation has becn demonstrated by several work-
ers. Washburn's data l8 for the distillation of mercury in a pot-type molec-
ular still at 0° C. are given in Table I. The distillation rate attained was
84% of the theoretictLl raLe calculated by equat.ioIl (9). Also using mer-
cury, Knudsen 19 showed, by the use of an eqUrttioll simibr to that of Lang-
muir, that a approaches unity if the mercury surface is kept cle~m.
TABLE I
DISTILLATION OF Mlmcunv IN I'O'1'-TVl'l~ MOLE:CUI,AR SUI,L AT 0 De. IB
Time of distillation ........................... . 4.0 hr.

Area of distilling surface ...................... . 48 crn. 2
TemperatUl'e of distilling surface ............... . -0.2°e.
Vapor pressure at distilling surface ............. . 0.18,..
Weight of distillate obtained .................. . 5.2 g.
Weight of distillate calculated ................. . 6.2 g.
Efficielwy ................................... . 84%
The dew-point data of Kapff and Jacobs 20 for the vapor-pressure deter-
minations of high-boiling esters calculated by equation (9) gave results
in close agreement with values obtained by Perry and Weber21 using a
method of directly measuring force per unit area. The efficiency factors
under these conditions are unity or very nearly so. The results of Verhoek
and Marsha1l 22 using dynamic methods for vapor-pressure determinations
also show that coefficients of unity can be obtained. The fact that coef~
ficients approaching unity can be obtained indicates that intermolecular
collisions under certain conditions and to a limited extent are permissible
in high-vacuum distillation.
Equation (9) can be put into a simplified form which is useful to the
18 Washburn, et al., J. Research Nail. Bur. Standards, 2, 467 (1929).
19 Knudsen, Ann. Physik, 47,697 (1915).
;l() I{apff and Jacobs, Rev. Sci. In8truments, 18, 581 (1947).
21 Pcny and Weber, J. Am. Ghem. Soc., 71, 3720 (1049).
22 Verhoek and Marshall, J. Am. Ghem. Soc., 61, 2737 (1939).
VI. DISTILLATION UNDRR HIGH VACUUl\{ 505
chemist. R is equal to 8.3 X 10 7 dyncs-cm./g. mole-oK. and to convert P

to mm. it is multiplied by 1333. Thus:
tv = 0.0583PVMjT (11)
where w is the distillation rate in grams per second per square centimeter
and P is the vapor pressure in millimeters of mercury.
Although hypothetical, the calculated d!tta tabulated in Table II will
serve to show the general utility of equation (11), In column 1, the molec-
ular weights of several compounds are listed in increasing order. Column
2 shows the distillation rates in grams per second pel' square centimeter at
1 j.J. saturation pressure, while column 3 shows these values as mole per
second pel' square centimet.er.
TABLE II
HYPOTHETICAL DU,'l'lLfA'l'ION RA'l'ES
(1) (.~) (3) (4) (6) (6)
1V = 0.0583 Q = 0.0583 P
Mol. PyM/T P/VMT P, at393°K. VM/T y;';7
Substance wt. (g./scc./cm.') (mole/sec./cm.') (120°0.) at 1 I' at 3na°K.
Stearic acid 28'1 0.52 X 10- 4 0.0021 X 10-' 35.0 0.90 2.07
Cholesterol 387 0.56 X 10-4 0.0014 X lO-.j 0.5 0.97 0.025
Tricaproin 387 0.60 X 10- 1 0.0016 X 10- 4 20.0 1.10 l. 01
Tricaprylin 401 0.63 X 10-4 0.00l(} X 10-' 0.5 1,20 0.025
Tristl'ul'in 891 0.76 X 10-4 0.0009 X 10-4 0.0001 1.32
For any given temperature, it is apparent from equation (1) that the
rate of ev:aporation is controlled by the ratio P / M. The quantity distill-
ing therefore is a function of this ratio and the relative qU[1ntities of each
constituent dist.illing arc:
PI P2 1Jn
VM1' V1I1~" ':VM"
where p is now the partial pressure of the constituent. By contrast, in
equilibrant or conventional distillation, where the effect of the molecular
mnss is lost, the relative quantities of constituents distilling from a mixture
are proportional only to their partial pressures such as PI, P2 .. , pn.
B. 'I'Hli:RMAL HAZARD
One of the greatest virtues of high-vacuum distillation is the ability to

distil successfully substances of high molecular weight which are thermally
unstable and which will not survive conventional distillation processes.
'l'he~ thermal 'exposure becomes increasingly important as the size of the
mol;cllle.becomes larger and special precautions must be taken to minimize
506 E. S. PERHY
the thermal hazard. The quantity of distilland being heated and the time
of hOfLting ~ontribute to the thermal exposure.
As has been indicated, distillation under high vaeuum is preferably ef-
fected from a thin film; fiowing-film stills are particularly adapted fll!'
thin-film formation. A comparison of film thickness and times of exposure
compiled by Hickman 23 for the pot-type and flowing-film stills is given in
Table III. The thiekness of the distilling layers for the flowing-film stills
is approximated for ideal conditions. Under IO:;i; favorable condit.ionH
generally encountered, the t.hickness of the distilland layer may be as much
TABLE III
COMPARISON OJ" FILM TmcKNEss AND TnmR m' gXPosuRms
APPl'OX. fL Appl'ox.
distilland time of
Still thickneBH (IXpOAUro
------------------------.------------------
Laboratory pol; still ............... .
1-5 em. 1-5 hr.
Inuustl'it11 falling-film still .......... . 1-3 IIlm. 2-10 min.
Luburatory falling-film still ........ . 0.1-0.3 HUll. 10-50 see.
Inuustri(11 ccntrifuglLl still ......... . 0.03-0.0(\ mm. 0.1-1 sec.
Labomtol'Y centrifugal still ........ . 0.01-0.02 mIll. O. O'J-O. 08 sac.
a Assuming ~imilar throughput fot' smue unit area of all sLills.
as 2 to 5 times the values given in the table. As can be seen from the
table, the exposure timf) differs markedly between the pot-type still and
the flowing-film stills. In a cyclic distillation requiring as many as 20
passes, the distilland is maintn,ined at the highest temperature in the falling-
film still for a total time of only ,5 to 15 minutes; on the centrifugal evap-
orator the total time is reduced to less than 10 seconds. '1'he halibut-liver
oil distillation reported by I'Iickman 24 serves to illustrate the comparative
performance of the falling-film still and the simple pot still, the vitamin A
recovery in the pot distillation being less than 50% of the original charge
while the recovery in the falling-film distillation is 80% of the original
charge.
The thermal hazard encountered during a given distillatioll is a function
of the distillation time and tempel'atme. Based on the olcll'ule of thumb
that reaction rates double with ellch 10° rise in temperature and that ther-
mal decomposition is a chemical reaction, IIicloll[LIl 2o oomputed possible
relative thermal exposures from the boiling point of dibenzyl phthalate
for a number of stills. The results are given in 1'able IV. 'rhe centrifugal
23 Hickman, Chern. lie.V8., 34, 51 (1044).

24 Hickman, Ind. Eng. Chern., 29, 1)68 (1937).
2, Hickman in Baitsell, ed., Science in P1'o(Jr88s. 4th Series, Yale Univ. Press, New
Ha.ven, 1945, Chapt. IX.
VI. DIS'fIIJLATIQN UNDER HIGH VACUUM 507
TABLE IV
THEitMAIJ EXPOSURES OF VARIOUS STll,LIS
Relative Iielatjve
Pressure, Temp.,Il tllcrlnnl decompn.
Kind of atill 111m. Hg Time °c. coati. b hazard
Simple tia8k ............ 760 1 lu·. 360 223 1.5 X 10 12

ChLiscn flnsk ............ 10 1 hr. 270 21-1 3.0 X 109
Wide-Heck pot still. ..... 1 1 hr. 220 29 9.2 X 107
Petroleum flash st.ilI ..... 1 1 min. 220 2' 1.5 X 10'
Molecular pot still ....... 0.001 1 hr. 130 1 1.8 X 105
Molecular falling-film
still .................. 0.001 1 miu. 130 1 3.0 X 10 3
Molecula.r cent,rifug!tl
still ................. , 0.001 sec. 130 1 5.0 X 10 1
Molecular centrifugal
still .................. 0.001 0.02 sec. 130 1 1
" nailing points of dibcnzyJphthalate, for example.
; Assumillg that the hazard danLleH with each 10°C. rise.
evaporator with a contact time of 0.02 second per pass has a relative thermal
coefficient of 1 and a relative decomposition hazard of 1. The falling-film
still with exposures of the order of 1 minute has a relative decomposition
hazard of 1/0.02 X GO, or 3000.
More recently, Hickman and Embree 26 employed a "haza.rd index" to
assign numerical values to thermal decomposition in stills. According to
this concept, the decomposition hazard, D, is the product of time, t, in
seconds and pressure, P, in microns, expressed as:
D = t'ec. X P~ (12)
To give a compact range of decomposition hazard values, D is preferably
m..-pressecl as log D or Dh. Thus the centrifugaJ still operating for 1 second
at 1 J.L pressure will have D = 1, or Dh = O. The conventional Claisen-
flask still distilling for 1 hour at atmospheric pressure will have
D = (3600 X 76,000) =" 2.74 X 109
or Dh = 9.44
C. SI~PARATION OR FRA.CTIONATION
The degree of separation effected by these high~vacuum stills compares

very favorably with the unit separation of the conventional equilibrant
distillation process. The phase equilibria of conventional distillation are
absent in high~vacuum distillation since ebullition is prevented and the
residual gas pressure is so low that collisions between molecules return only
26 Hickman antI Embree, Inri. Eng. Chem., 40, 135 (1948).
508 E. S. PERRY
a negligible number of cHstilling molecules to the evaporating surface while

the majority pass directly to the condenser. The fract.ionating power of
high-vacuum stills is therefore restricted to the preferential evaporation of
the most volatile constituents from the surface layers. Replenishment of
the surface layers is maintained by diffusion of the desired constituents
from the body of the liquid. When the rn,te of removal of a certain consti-
tuent from the liquid surf:we exceeds the mte of diffusion of other similar
molecules to the surface, a state of depletion sets in. To obt.ain maximum
efficiency in the high-vacuum distillation process it thus becomes necessary
to induce a violent but i:iplashloss agitation in the liquid phase to hasten
the inherently slow process of molecular diffusion. This was a prime ob-
jective in the evolution of the modern flowing-film stills such as the falling-
film still in whieh distillancl flows downward by gravity over a vaporizing
surface and the centrifugal still, where distilland is spread by centrifugal
force over the surface of n diseliiee evaporator.
In the centrifugal still, the centrifugally induced turbulent How of dis-
tilland enables separations of about 1 theoretical plate per pass. Although
differing in pl'ineiplo in that the quantities of material distilling are propor-
tional to lJ/ VM for each cOll<ltituent, the results fwm the centrifugu.l
high-vacuum still per pass are comparable to the simple equilibriuIn still
whose separating power depends only on differences in the partial pressure,
p, of the components. Turbulent flow induced by the centrifugal evapora-
tor, therefore, contributes to high-vacuum distillatioIl that which is con-
tributed by ebullition in the equilibrant still. Helative perfOlmanees of
the high-vacuum stills have been assessed in terms of the "elimination
curve,"24 which is discussed in some detail in connection with analytic
high-vacuum diHtilllLtioIl.
Fractionation, or more precisely the aehievement in a single operation
of a greater degree of [;eparation than cnn be attained by It single simple
distillation, does not occur on the ordinary laboratory flowing-film stills.
Separations better tha.n those obtained by single distillations are usually
realized by repeated distillation. The fractions resulting from a cyclic
distillation are l'Ccombined according to the distribution of the desired
constituents and the composites individually redistilled. The operation
may be repeated as many times as necessary to effect the desired separation.
Schemes for connecting stills together to achieve fnwtionation have been
set out by Fltwcett27 nnd by Fl'aser. 28
\ Several devices have been proposeu to increase the fractiom1ting ability
of a single high-vacuum still. The use of a semipermeable barrier 29 inter-
27 Fawcett and McCowen, U. S. Pat. 2,073,202 (Mar. 0, 1037).
28 Fraser, U. S. Pat. 2,128,223 (Aug. 30, 1938).
29 Hickman, Ind. Eng. Chem., 39,686 (1947).
VI. DISTILLATION UNDER HIGH VACUUM: 509
poscd betwecn evaporator and condenser hilS been suggested to stop the
least volatile mulecules of the prim[Lry evaporation while nllowing the more
volatile molecules to puss to the condenser. A cone-type centrifugal still 3fJ
has 'been recommended for mllltiple redistilln,tion. In this still, a' conical
rotor is sectioned into three zones. Distilland is introduced onto the rotor
at the bottom of the upper zone nnd travels up to the gutter. Distillate
obtained during this short passage drips back onto the rotor at the bottom
of the middle zone amI is redistilled during travel up the rotor. The second
distillate drops to the bottom zone of the rotor and gives rise to a third
distillate as it travels up the rotor. The third distillate is withdrawn as
product.
Brewer and Madorsky 31 hrtve described a lO-cell countercurrent reflux
vacuum still which they used for the sepamtion of mercury isotopes. The
ten evaporators are arranged in a row, one slightly above the other in such
fashion that the condensates drain from the condenser to the evaporator
cell once removed upstream in the series while the distill and overflows one
cell downstream. A similar still was proposed by WoUner, Matchett, and
Levine 32 and used to purify acetylated marihuana "red oil."
HI. HIGH-VACUUM DISTILLATION TECHNIQUES
For the purposes of this section, distillation techniques will be illustrated

by reference to a few typical distillation methods. Suitable high-vacuum
stills will be described in connection with each method. Variations in
apparatus and techni'ques will be obvious in connection with the several
methods.
1. Simple High-Vacuum Distillation

The simplest form of high-vacuum distillation is the separation of a de-
sired material from a residue of higher molecular weight. The distillation
is particularly adapted for purification of a material which is difficult to dis-
til under ordinary atmospheric distilhLtion conditions. For such simple
distillations, particularly when a small quantity of material is being treated,
the extremely simple "cold-finger" pot still may be used. Two useful types
of the "cold-finger" stills whieh are easy to make and which arc also com-
mercially available 33 are shown in Figure 2. These stills consist of concen-
tric test tubes sealed by rubber-to-glaRs seals (Fig. 2A) or flange seals (Fig.
30 Hickman, U. S. Pat. 2,234,166 (Mar. 11, IH41).
31 Brewer and Madorsky, .T. Re.~earch Nail. Bu1'. Standards, 38, 129 (1947).
32 Wollner, Matchett, and Levine, Ind. Eng. Chern., Anal. Ed., 16, 529 (104'1).
3' Injormat'ion on High Vacuum Distillation. Distillatiol). Products In.dustries, Roehes-
ter, N. Y., April. 1947.
510 ~;. s. PERRY
2B). Heat is sllppEed by an oil bath whidl is heated elreLrienlly or by lL

Bunsen {-bIlle. The tI;.;o of the oil biJ,th permit;.; temperature measurements
which are, for practical purposeH, the tcmperaturcs of the distillanrl.
A typical simple high-vacuum distillation is illustmted by the separation
of tocopherol from lard, as reported by Quaife and Harris 3 <1,35 employing !1
cold-finger still as illustrated in Figure 2A.
1 g. of lard containing a small percentage of toeophel'ol is charged into the outer
tube of the still, preferably as a thin film in It met,al-foil cup. 'rhe sample is de-
gassed by closing the f)till with 11 rubber stoppel' and reducing thc pressure to less
T
C\.I
C\.I
----,.--,.
11
lb.~ ~48-l
1 Jf1~-75-~
(A) (8)
Fig. 2. cICold-finger" RWIR «(limcnsions in lllillirnotel's).aa
than 1 f.4 at room temperature. Thn pressure is returned to atmospheric pressure,

the inner tube is immediately placed in positioll, Hnd the pressure is again reduced
to 1 J.I,. The inner tube is then filled with dry icc and serves as a condenser. The
temperature of the oil bath is raised rapidly to 220°0. for 30 minutes whereupon dis-
tillation takes place. The oil bath is removed and the still is allowed to cool to
room temperature under vacuum. The condenser tube is then removed and the
distillate is washed off quantitatively by means of chloroform. In this way, more
than 90% of the tocopherol content of the lard is recovereci as distillate.
The method is readily applicable for separation and recovery of small
amounts of many organic materials which are admixed with a large amount
of a higher~boiling material. The smaller still shown in Figure 2A is sllit-
34 Quaife and Banis, Ind. En(J. Chem., Anal. Ed., 18, 707 (1946).
36 Quaife fl,n<! Harris, Anal. Chem., 20, 1221 (1948).
VI. DIS'l'ILLATION UND.~m HIGH VACUUM 511
ably employed for samples of leRs than 5 g.; samples up to 25 g. ca.n be ac-
c:ommodated in the larger apparatus.
2. Cyclic Distillation
The previously described methocls have been concerned for the most part
with relatively simple materials. It. is often desirable, however, to sepa-
mte complex mixtures into component fradians. The component fractions
may vary widely in volatility, molecular weight, and thermal stability.
Since the nature of the mixt.ure is often unknown, it is difficult to decide
what type of high-vacuum still to use. In such a case it is desirable to run
n trinl distillation in a boiling-point still to gain nn idea of the distillability
of the material with regard to the stability and the viscosity of the distil-
land. When the boiling point of the material is high and the thermal sta-
bility is low, the use of a flowing-film high-vaeuum still is indicated. If the
thermal stability is good, the boiling-point still or a conventional vacuum
fractionating colunm can be used. A limited but useful working guide in
choosing distillation apparatus is that those substances having molecular
weights up to 300 can be processed in boiling-point sWls, while those in the
range of 300-1000 require flowing-film high-vacuum sWls.
In many cases it is not possible to complete a distillation in one kind of
still becaul:\e nearly all crude materials contain substances which boil at
lower and higher temperatures than the main body of material being proc-
essed. A typical example of the distillation of a natural substance is
furnished by the eydic distillation of a diethyl ether extract of oranJ};e peel.
The extract, eonta,ins the ether, terpenes, heavy essential oils, and relatively
involatile waxlike substances. The cyclic distillation of orange-peel extract
to sepani.te it into a number of fractions will be described in detail to illus-
trate the general technique employed.
In the first step of the process, the ether is removed from the extract by
means of a water aspirator leaving the orange-peel extract which is to be
separated into fractions. Since thc extract contains relatively volatile
terpeneH, it is tlrst treated in t~ modifiecl boiling-point sti1l36 wl1ich is useful
for distillation of substances which are too volatile for convenient handling
in the flowing-film high-vacuum stills and too thermally unstable for dis-
tillation by conventional methods. Several models of these stills are shown
in Figure 3 and Chapter V, Figure 13. They consist of boiler, A fitted
with a. eolumn, B, whic:h terminates in an enlarged bulb, 0, and a collecting
alembic, E. Vapors of distilland from boiler A are passed up column B
where they are freed of entrained material and expand into bulb C and are
condensed ["wl colleeted in n1embic E. By attaching; a side manometer, F,
8C Hiokman, J. Jl'ranklin Inst., 213, IHJ (1\)32).
512 E. S. pmnRY
as shown, which will collect a small amount of di::;tillate, the exact pressure
corresponding to the vapm' tempera,ture in C can be obtained. This pres-
sure is the sum of the pressure indicated by the manometer and the prcs-
-8
Fig. :3. Boiling-point Etills. ~3
Fig. 4. Boiling-point still assembly. '3
sure indic[tted by t.he indepcnclont gltge which measures the prcssure of the
system.
In the orange-peel extract distillation, the extract, freed of ether, is
charged into flask A of the boiling-point still assembly shown in Figure 4.
When the stopper B carrying thermometer C is inserted as shown the system

is ready for evacuation. Cold trap D and beaker E were previously charged
wit.h dry ice and acetone. The system is evacuated to a pressure of 20 mm.
mercury by mechanical pump F and maintained at that pressure by con-
trolled leak H, the pressure being indicated by manometer G. During the
early stages of evacuat,ion, the charge may froth and foam considerably due
to the violent escape of solvent, dissolved gases, and possibly moisture.
The ether passes directly to trap D, condenses, and is collected in receiver
K. Moisture remains in D as a frost.
Oil bath I is then slowly heated by means of the electrical immersion
element and the oil bath temperature is measured on thermometer J.
The applied heat increases the ebullition in the flask for a short time while
the distill and is freed of the last traces of volatile matter. As the tempera-
ture of the oil bath increases, the pressure in the system is kept constant
by adjusting leak H. Finally, the rising vapor in the lagged column reaches
thermometer C causing the temperature to rise rapidly and distillation be-
gins. The condensate g[\,thers in droplets which collect in the alembic
and pass through vacuum plugcock M. The distillate is refluxed through
M back to the column until the desired distillation rate is established by
final adjustment of temperature and pressure. After steady distillation
has proceeded for 10-15 minutes, stopcock M is opened and the terpene
distillates are collected. Thermometer C indicates a constant value for
the greater part of the distillation because the terpene fraction comprises
77.0% of the original weight of the extract. As the last of the terpenes dis-
til, the temperature begins to increase and the distillation is terminated.
The flask is cooled and the residue is transferred quantitatively by the aid
of solvents to the flowing-film high-vacuum still.
For purposes of this example, a small laboratory centrifugal still of the
type shown diagrammatically in Figure 5 is uscd, the method being the same
whether a falling-film still or a centrifugal still is used. The centrifugal
still assembly proper consists of an inverted bell-dome enclosUl'e containing
an aluminum rotor with embedded electrical heaters and an aluminum cup
supported just below the rotor to serve as a residue collector. The bell
dome is connected through a cold trap to an oil diffusion pump which in
turn communicates with a mechanical fore pump. A Pirani gage is used
to indicate the pressure within the dome and a variable transformer regu-
lates the power to the electrical heaters. The waxy residue from the
boiling-point distillation is drawn through tube 13 into dome A, which serves'
as a, reservoir, while the still is under vaeuum to effed preliminary degas-
sing. The circuhtting pump, C, is immediately put into operation and the
distill and is circula,ted over rotor D. A little heat applied to the rotor as-
sists the final degassing. Heating of the rotor must be done continuously
514 E. S. PEHRY
or distillate mn,y be evolved and lor;t to tntp E, which is l'ci'l'igcrated with

dry ice and acetone. The tempemturo of the rotor is kept below no°e.
The distilland is circulated through ono 01' two cyelos to effect complete
degassing, and during the final stlLges of degassing, the oil diffusion pump is
st:1l'ted. When the diffusion pump is in full operation, the rotor tempera-
ture is slowly increased until distillation begins. The first distillate is 1'0-
tiurned to the still and the temperatlll'e is adjusted to 8()OC. which is the
nearest 10 0 0. mark. When the temperature is stabilized at 80°C., the first
FORE PUMP
Fig. 5. Small labomtory COil tl'il'lIglLI still RSi:lBmbly.
cycle is begun. The ball valvo is closed a,nd the first fraction of distillate is
collected in receiver I by opening vacuum valve H. As the last dregs of
distill and are pumped out of the bottom of dome A, the bull valve is opened,
allowing the residue from the first cycle, which has collecLed in the alu~
minum cup G below the rotor, to flow into the lower reservoir where it. then
becomes the distilland for the next cycle. The first cycle is continued for
an additional minute to flush the pump and the lines to the rotor. The dis-
tillate stopcock, H, is then closed and the temperature of the rotor is in-
creased to 90°0. While making the temperature adjustment, a new dis-
tillate receiver is put in place. This is done without interl'Upting the prog~
ress of the distillation by using two stopcocks. When the temperature
reaches 90°0., the ball valve is closed and stopcock H is opened to collect
the second fraction.
VI. DISTILI,ATION UNDER HIGH VACUUM 515
These operations are repeated ai; each 100 interval up to 2~10°C. to give
a series of 16 yellow~to-orange fractions. At the end of the distillation, the
diffusion pump and the rotor are allowed to cool before air is admitted to
the system. The residue is allowed to circulate over the rotor until the
temperature drops well below 100 0 e. The mechanical pump is then
stopped and the vacuum is released. The residue is drained from tube B
into a sample bottle. The still is cleaned by circulating a solvent through
the system by means of the feed pump while the rotor is still hot. The
rapid evaporation of the solvent floods the whole interior of the still and
washes the viscous residue into the lower reservoir, from which the washings
are then drained.
The course of a typical orange-oil distillation is shown in Table V. The
data are presented on a typical laboratory record forIll which gives all the
TABLE V
ORGANIC RESEARCH LABORATORY
Distillation Data
M!tteriaJ: EXTRACTED ORANGE PEEL OIL Number: 0100
Distilled for: THE SQUEEZE-IT CORP. Date: 6-1-49
'remp'j GroS~J '11nrc, Net, %
No. ac. Pl'eHBUre g. g. g. cut :!:% State Color
Charoe* 5000
A 68- 20 3850 77.0 77.0 L Ww
SO mnl.
Residlle 1150 23.0
It 80 250 " 95.4 00.1 35.3 0.7 77.7 L Ww
2 \)0 200 175.5 89.1 86.4 1.7 79.4 L Ww
3 100 100 147.8 89.7 58.1 1.2 80.6 L Y
4 llO 60 118.7 88.8 29.9 O. () 81.2 L Y
5 120 '10 119.5 90.1 29.4 O.H 81.8 L Y
() 130 25 115.7 89.5 2B.2 0.5 82.3 L Y,G
7 140 15 173.5 90.2 83.3 1.7 84.0 . L Y,C
8 150 8 139.9 89.6 50.3 1.0 85.0 Ss y
\) 1BO B 140.0 88.9 51.1 1.0 86.0 S Y
10 170 (I 147.1 90.1 57·,0 1.1 87.1 S y
11 180 4 134.4 89.9 44.5 O.!l 8S.0 S y-o
12 190 4 121.3 90.1 31.2 0.6 88.6 S y-o
13 200 4 135.3 88.8 46.5 0.9 89.5 S Y-o
14 210 4 149.7 89.9 59.8 1.2 90.7 S 0
15 220 6 198.7 89.5 109.2 2.2 92.9 S R-Br
16 230 10 266.5 143.2 123.3 2.5 95.4 S R-Br
Residue 272.4 147.8 124.6 2.5 97,9 S BI
* Boiling point. t CClntrifugoJ.
L, liquid. Ss, semisolid. S, solid. Ww, w!\ter-white. Y, yellow. 0, orange.
R, red. Bl', brown. Bl, blade. C, c:rystalline.
Still Used: MOI,ECUI,AR. Operator: J. Doe
516 E. S. FIDUUY
pel'Lincnt informtttioll regarding; the distillation. In this case, the resultH

of the boiIin~-point distillation are shown in the first lines. The remaining
data for the centrifugal still show the temperature::; of' the residues as they
left the rotor at each cyele, and the pressures as rond on a Pirani gage. The
results are calculated on thc basis of a 5000-g. chargc of the O1'iginal oil to
the boiling-point still. Thul'I, the terpene fraction removed at 20 mm.
pressure amounted to 77% of the charge. The residue was a yellow-to-
brown semiviscous liquid having the fragrant orangelike odor of the orig-
inal material. Fractions 1-16 were obtnined fmm the centrifugal still.
....
Z 60
w
(.)
0:
~ 50
w
...J
o
::;:
w- 40
~
::i
~ 30
Ci
3 20
c[
Z
iii
~ 10
10 20 30 40 50 60
ROSIN ACID IN FEEO, MOLE PER CENT
Fig. 6. Results of tall-oil distillations.
Fractions 1 and 2 were watcr-white and appeared to consist mostly of ter-

pene material not removed in the boiling-point distillation, which was ter-
minated before completion so as not to harm the heavier constituents.
Fractions 3-7 were yellow oils with some erystals appearing in the last ones.
The remaining fractions were solids ranging from a bright yellow through
orange to brown. These fractions progressively increased in hardness.
The residue was a hard and brittle black solid.
This general procedure of· cyelic distillation is very useful with many sub-
stances. Sometimes the distillation amounts to little more than removing
small quantities of impurities and color from a single. substance, as, for
example, a high~rllolecular plasticizer. .
VI. DISTILIJATION UNDER HIGH VACUUM 517
The cyclic distillation of tall oil is summarized in Figure 6, The curves

represent the relation between feed und distillate compositions at distilla-
tion rates of 40% of the feed 1':.1te,
3. Analytical Distillation
Analytical distillation is a systematic procedure of distillation which was
devi::;ed by Hickman 37 for the determination of a relative boiling point--
like temperature and an estimate of the purity of materials allegedly com-
posed of a single substall(~e, This method involves the determination of
z 8
o
~
z 7
:::;;
:J6
w
lJ..
05
OJ
~
0:: 4
w
>
~ :I
..J
~ 2
To h T4 To Ta Ta TI~ TI4 TIG TIB

TEMPERATURE
Fig, 7, Theoretical elimination Cllrves,
the temperature at which t.he rate of removal or elimination of a constitu-

ent from a specially prepared substrate reac:hes i1 maximum, The tempera-
ture value at the maximum is a specific property of the distilling species
and is designated the "elimination temperature," It is reproducible to a
high degree of accuracy under controlled conditions, and substances can
thereby be charncterized by a temperature measurement in much the same
manner us by the boiling point in equilibrant distillation. The procedure
consists of subjecting the substance under investigation, c:ontuined in a
special solvent medium, to repeated dist.illations in a cyclic batch still,
such as a centrifugal stHl or a falling-film still, with eaeh sllec:essive distilla-
tion carried out at a progressively higher temperature. In this way the
~11iickman, Ind, En(l, Chern" 29,968 (1\)37).
518 :E. S. I'[i]]{H.Y
rate of elimination of the desired constituent; incl'PHses steadily to a maxi-

mum value and then elro])S off. As the diKi,iIlnnrl becomes impoverished
of the distilling; constituent, the rate of l'emovul neem;sl1rily diminishes to
zero when the (~onstitucnL is completely dist-,illed out. The variation in
coneentmtion of tlw constituent, from fmction to fraction, when plotted
as n function of the tempemture at whieh the fmotion distils, resembles a
pl'Ohability diRlxihution eurve fUld was etLUed the "elimination curve" hy
Hickmun}7 The theoretie[ll Rhape of the eiiminai",ion eUl've for it single
Buhst1tltee was determined mathemtLtieally hy E;mbrce:18 j theoretieal
elimination enrves are shown in Figure 7.
The climination method depends upon thc intrinsie distillability of the
substance, that is, the ratio of the number of tnoleeules of no given species
leaving the distilling surface to the number of the HILIlle kind of molecules
remaining in the surface layer of distillancl in unit time, and upon the rate
of elimination of the dBsired eonstituent as shown by the rate of accrual of
TABLE VI
COUHSN 011' ELUllN"A rl'ION
~-~----~.-'~"------~--'''.~.-------------~--.--'----~ -._-------- ... _
---- -------- __..Arithnwti(~~7
_- , , ·_ _ _ _ _ _ _ _ _ _ _ _ h_
Cnleullls:l 8
TI)tnl '1'ota\
per ceuL pel' cent
Pnl' (Hmt Distilhthility ulfLteriul mo.tcriul
'l'ill$! lun.tm'in.l m' il\trin~ic A'llltOX.. otimin. clhnin.
intervul loft, in di.Htn. l'ftto of from Per cent, Pm- eont, from
No. diHtill:Uld !"fl,te
--------.-----~---,-----------------~_-------~-
nlilliinatiull diHtilianrI
....• -.-.-
Zi Wi -- _._._--_
dis tillaml.._
0 100 l J 1 100 1
1 fl9 1.26 1.25 2.25 99 1.19 2.19
2 07,75 1 . fill 1.51) 3.S0 97.S1 1.5G 3.75
a 90.2 2.0 l. 02 G.72 9fl.25 l. 92 5.fl7
4 IH.28 2.52 2.37 S.O$ \)4•. aa 2.26 7.\)3
Ii 9Ull 3.17:') 2.1J1. 11.00 D2.07 2.S0 10.70
() 80.0 4.00 a.5G 14.5(i 80.21 3.48 1<1.27
7 85.44 ii.23 4.31 IS.87 85.n 4.20 18.47
8 Sl.13 G.30 5.17 2J..(H 81.1)3 5.0() 23.53
9 75.9(i 8.00 fL07 30.11 70.47 5.88 29.'11
III 70.00 10.0n 7.04 37.15 70.59 (U8 36.19
11 02.81) 12.67 7.05 ,ill. 10 63.81 7.!l0 43.79
12 53.9 HLO 8.G3 54.73 56.21 8.32 52.11
13 45.27 20. ]5 n.la (i3.86 47.S9 8.77 60.88
14 3(i.l'1 25.4. 9.17 73.03 30.12 8.80 69.68
]5 26.97 32.00 8.04 81.m 30.32 8.30 77.98
Ifl 18.33 40.7 7.'19 89.16 22.02 7.29 85.27
17 ]0.84 50.4 5.46 95.02 14.73 5.87 91.14
18 4.38 04.0 2.80 98.'12 8.80 4.20 95.M
19 1.58 80.7 1.27 4.(iO 2.58 97.92
20 101.61 2.08 1.29 99.21
as Embree. Ind. Eng. Chem., 29, ~175 (1937).

VI. DIS'!'ILIJA'!'roN l1NDI1R HIGH VACUUM 519
condensate. At. constant temperature, the distillnhility is constant while

the rate of elimination gradually decreases to zero as the distill and becomes
depleted in the distilling cOnHtit.ucnt. By raising the temperature of each
successive cyele in discrete steps, the distillability is increased at each step
and the mte of elimination rises to n maximum nnd then falls off to zero.
Table VI shows hypothetical elimination dntH as calculated by Hickman:17
arithmetically and hy Emhree:l S using calculus. Curve A in Figure 7 i:;
based on these data, and the shape of the eurve has been verified experi~
mentally. An increase in cycle time causes a dispbeement of the maximum
to lower temperatures (see curve C, Fig. 7).
The experimental equation derived hy Embree:mfor the compntation of
elimination rates from a nonvolatile solvent in a series of progressive inter-
vals has the following form:
\\"ho1'e Wi il:l per eent of original amount of snhsbmcn I~liminatod in nny ith
intelYal; Zi is pel' rent of original amount of substHnce remaining in
distillancl at the start of thl' ith interval; Ii if-) per cent of Rubsta,nee distilled
out at end of first, interml; /' i::; factor by which Ili::;tillability at the ::;tart
Ii increa::;eH at each of slleucssive temperatureH; and In iH the units of time
nt eaeh temperature.
Curve B of Figure 7 was based OIl rates of elirnin11tion euleulated by
eqlluJ.ion (13) llsing the conditions of clistill:tbility assumed by Hiekman in
the arithmetically calculated hypothetical distilbtioll. The eurves of
Figure 8 were ealeulated by Embree using In vnlues of 1, 2, and 4 unit timc
intervals.
In currying out th(' elimination distillation pI'Oee~H, the distillate frac-
tions are usually small and difficult to remove from the condensc!'. To
facilitate colleetion of fmetions and minimize dminage errors, a substanee
is preferably addecl to the distilland which co-diHtils wit,h the desired COll-
stituent and whieh has fL etlllstant distillation yield ov(,1' the range of tem-
peratures employed during the proeeSH. The ehoiee of thi:,; so-called "eon-
stant-yield oil" will clepenclllpon the cliHtillation ('ouditions. Typical eUJl-
stant-yield oils are mixtures of synthetie glycerides and of cliHtilleJ hydro-
carbon lubricating oils of SAE viscosity 20 to 40. The synthesis of !t tri~
glyceride oil has been set out in tlcta,i[ ill the liten),t,ure,:19 the l\ompouncied
oil being a blend of mixed trigly(~erides of (teetic !toid, butyric ueid, the
mixed myristio ancllaurie add:,; from saponified coconut oil, and the highly
unsaturated C l8 acids from saponified pel'ilIa oil. The constant-yield oil
should distil over the range of 100 t.o 220°C_ yielding nearly equal w(~ights
30 Baxter, Gmy, :tnd Tisd\l'l', [,ul. En!!. (;/WIII., 29, l112 (1037).
520 1~. S, PERRY
of distillate at (,ttl',h tmnpemture. The oil should yield liquid fractions over
the entire j.('mpemtul'c range, be thermally stahle lInder the eonclitions of
15~----------------------------------------~
~
" 10
:z
o
~
<!
0::
LL
!:
o
.J 5
UJ
>=
L-J-...l.-..I..._J.'_l-..J--'--'--'--'-......--"=-<,__;........_..:.::.-&--'---'----'
T5 Tlo TIS
TEMPERATURES Or DISTILLATION
Pig. "'. Eff :rd; of cycle time on eliminatioll CUl'VPH. 3B
distillation, nnd be in",:t tuward the constitueut under examination. A

typical pilot distilLtio)l of i'L constant-yield oil is shown in Table VII.
TABLE VII
-_ _--- ..
"ILO'" DrSTIr,r,A'l'ION OF CONS'rANT-YIJ.iJI,Il
Tc (': -eru.tU\,l!,
, ,---~-"'-"-,-,------------
llcr eeut
OrL
Tempel'attIre, Pel' eent
Ii'raction O. Cut °C. Cut
-- ..,,--'------------"'-----
'
1 100 4.0 10 190 (i. 6

2 llO 4.8 11 200 6.6
3 l20 6.1 12 210 7,4
4 130 6.0 13 220 6.4
5 140 5.0 14 230 7.2
(j 150 7.0 15 240 6.8
7 l6ti 0.2 16 250 5.9
8 171) fLEi 17 260 4.8
!J lRO 7.7
-~- ,--,-----.~-. ,,---~-,
Equati« ,[1 (13) cun be altered to twcomIUodate the effect of the co-distilling
c:onstant-lield oil, ;,~ Thus if 2% of the constant-yield oil distils into each
fraction, the equa i:ion becomes:
nr, == 100 Zi (1 _ e- brim/(lOO - 2i)) (14)
• 1 - Q,Q2i
521
TABLE VIII
BLIlIlINATION TEMI'ERA.'l'UR1~S OF ANTlIRAQUINONE DYE:,;"
Ito!.
eHm.
NOrlno Formula max., ClC.
CeJanthrenl' 127
Red 3B
Dimethy ldiamino-
unthraquinone
1'11
Diethyldiamillo-
ant.hraquinone
car
153
Dimethylmonoace-
tvldiaminoan- Hi8
thrnquinone
Dipropyldit~mino
CO NCR,
I ]62
anthraquInone COCH,
Dibutylrliarnino- 171
an tliraquinone
NHC.HII
Diamyldii1Inino-
anthraquinone
Dixylyldiamino- ~HCH20CH' :Q,."

~H'OOH', NHO"'"
anthraquinone
Quinizal'in
Green
Anthraquinone
NE,
/Qo», 217
D'
183
Blue Sky
Nfl,
522 E. S. I'EHRY
Since the elimination rate will depend upon the particular conditions of the
distillation, inelwling the apparatus employed, it, is desirable to employ a
st!Lndard material and correlate the elimination maximum of the material
being investigated to the elimination maximum of the standard material.
Fig. 9. Cyclic falling-film ll101eculm' still. 33
This will give a relative figure of, for example, lODe. above the standard
material, rather than an absolute figure dependent upon the distillation
conditions. These standard materials, or pilots, are suitably volatile
dyes because of the ease of measuring concentmtlons by colorimetric
means. A group of pilots useful in the elimination distilh~tion process is
the homologous series of anthraquinone dyes examined by lIickman 3r and
listed in Table VIII. Tho oliminntion temperatures shovin in the table
were determined using a lO°C. temperature interval and the procedure de-
scribed below.
VI. DIS1'ILLA'1'ION UNDEn HIGH Vc\()UlTM 523
Before describing a typical elimiw1tioll distillation in detail, the use of a

llondistilling cltlTier oil should be mentioned. 8inee the volume of distil-
land diminishes with each cycle, it is neeessary to adjust the feed rate to
maintain a constant time interval. In order to minimize the necessary
decrease in feecl rates, the constant-yield oil should be dilutcd with an equal
volume of a nondistilling oil thcreby l'edueing the change ill feed rate by
half. Suitable nondistilling carrier oils are ohtained as residues from strip-
ping petroleum lubrieating oils, soybean oil, east or oil, 01' cottonseed oil
at 230°0.
10
9
8
~ 7
uj
>- 6
o
l:; 5
o
cd 4
:;:
3
2.
~O~O~~1~20~~1~4~0~~16~O~~IB~O~~200
TEMPERATURE, ·C.
l?ig. 10. Typieal expmiment.al elimill:1\'ion curve [or di-
ethyjcliamil1oanthraquillOlltl.:l7 See page 525.
To illustrate the elimination distillation process, let it be assumed that

the elimination curve for cliethyldimninoanthraquil1oItc is desired. For the
purposes of the example, a falling-film still as illustrated in Figure 9 will be
used. The method in [\,ceord1mee with the procedure detailed by Hie1c-
man a7 employs a lOoC. temperature interval over the ternpemture range
from 100 to 200°C. A solution is prepared contuining 2 mg. of the dye
and 100 g. each of constant-yield oil of the propel' temperature range and
nondistilling carrier oil. The solution is charged into reservoir A of the
still of Figure 9 through funnel I while the still is under VHCUUln of the fo1'e-
pump alone. The distillll,nd is recyded at leust twice over evaporator C
from reservoir B and back to reservoir A by means of magnetic pump D.
The cycles arc l:1cparatecl by use of ball valve E. Finaltmd com.plete. degas-
sing is hastened by the application of preheat at F and heat onthe evaporator,
C. During the last; Htnge of the degassing period, the vacuum is reduced to
Iil. S. PEUHY
1 j.£ by means of the [Luxili:uy oil diffusion pumps. The feed rate over the
evttpol'atol'is then adjusted so that the magnetic pump empties the entire
bulk of t.he distilland now contained in reservoir B in exactly 9 minutes.
The heat to the evaporator is increased until thermometer G registers
100 0 e. and maintains that temperature. The ball valve, E, is closed and
6
ESTERS
~
o
_J
w 4 J~ ~~
):
)fj \"
2 FREE FI "
---
fLC~HOT
. - -'Q •.
/V x COD
'IHAILlB~T
1\ "-
::::..- /'
"r-.
o
80 100 120 140 160 180 200
TEMPERATURE, ·C.
220 240
'" 260 280
Fig. 11. Distillation eurVl1S of vitamin A-bearing oils.41 800 pnge 526.
Cl
_J
w
):
80 100 120 140 160 180 200 220 240 260

TEMPERATURE, ·C.
Fig. 12. Distillation of pltrtilLlly saponified cod-liver oil. 11
stopcock H to the distillate receiver is simultaneously opened. 'rhe first

fraction is now being collected. In 9 minutes, the oil has all passed from
reservoir B over evaporator C and into reservoir A. As the last bit of dis-
tilland is being pumped out of reservoir B, ball valve E is opened and the
distilland residue flows into reservoir B again. Distillate is collected for 1
minute longer during which time the distilland in the lines is being dis-
placed. At the end of the tenth minute, collection of distillate in the re-
ceiver is stopped by closing stopcoek H, and subsequent distillate is re-
VI. DISTILI,A'I'ION UNDER HIGH VACUUM 525
turned to the stilI. The voltage to the evaporator heater is increased to

full line value for a specific number of seconds and then returned to a new
value required to maintain the evaporator at 110°C. These conditions of
adjusting voltage are desirably predetermined with t,he settings for each
step tabulated on a chart for ready l'eference. At the end of the eleventh
minute, the temperature should have become stable at 110°C. In the in-
terim of the last minute, the sample bottle is changed and the magnetic
....... 1'\
, /\ /(\/\ / \
W ::! ~~ ILJ W
_J
~ a <t<t I- 2:
a a u.:
-0::
UI-
<t
0:
W
I LU a:: z I-
a 0::
D- o: Ww <t a
U lJJ 0:
_J «
u
:2u
a.: :2z I- <t
1- uj 00
(/) u,
« .'t :::> uu .'t QJ.
(f) (T" . <>- • 'q
2:
a I ..··.
.p \
i=
u J:'
« \
0:
LJ..
LJ..
/ :' .--
\. ./ .:-.~
a
// \.~., i '\
u>-
Z
lJJ
I-
,
'.
"'0'.--
.,.
61
.
'.~
a(L
-IJ-.. p.--.a..
..0- "0- \
.tY" 9..
0
80 100 120 140 160 180 200 220 240
TEMPERATURE, ·C.
Fig. 13. Fntf! of vita.min A compounds during ingestion by the rlLt.42
pump is adjusted to the !lew feed rate necessary to eirelliate the remaining
bulk of distilland in 9 minutes. The distillation proeedure detailed abovc
is repeated and the second distillate fraction is collected. This procedure is
repeated for each 100 interval up to 200°C. and 11 fractions are obtained.
At the end of the distillation the fractions are assayed for their dye concen-
tration and the elimin:1tion curve is plotted. A typical experimental
elimination curve for diethyldiaminoanthraquinone is shown on p. 523.
The elimim1tion distillation process is useful for identifying a substance
by determining its elimination maximum and comparing it tn the known
value. For example, early observations'lO indicated that vitamin A exist.ed
in marine oils as an ester. Because both the free vitamin A alcohol and t.he
~o Bacharach and Smith, Quart. J. Pharm., 1,539 (1928).
521) I':. K. l'I~JWY
vitamin A ("8tC.~l'H responded to Lhe same identifying teKt, t\,nalysis wus dif-
ficult. By Ltnulytie:tl diHt:iIlatioIl, Hiekman 4 ! waR able to show that vita-
min A. oer~(ll'red prineipa.lIy in Lhe ester form in both eod- and halibut-liver
oils. The emves resulting from distillation of the original oils are shown
in Figure 11. The eurve obt,~tine<l by distilling another sample of the
original oil to whieh had been added a qnantity of vitamin A aleohol is
ShOWll in Figure 12. An analytical-distillation study of the metabolism of'
vitamin A in experimelltnJ rats was mnde by Gmy et al. 12 Their curves
me shown in Figure lao
A modification of the elimination-cmve determination was reported by
Fletcher et (tl. 1 :l using the temperature at which 50(lv of the experimental
subRtanee is eIimim1ted (elimination temperature 50). By thifl method,
the elimination curves cOllfliHt of It plot of enIDultttive pel' Gent agninst dis-
tilbtion temperatures. Distillation data employing this modiii.ecl method
for t,he distillation of dlOlestel'Ol eaproate nrc shown in Table IX.
'fABLE IX
CnOLgS'l'lmoJ, CIcP1WA'l'li: III EUMINA'l'ION MAXIMUM DA'l'A'"
dyo, 5.0 m!. ()f aDo millif,!;l'nm pur (lent Holtltioll. Totnl st.erol, 300 mg.
Totll!
_._---"-- . ""---_.
"'-'~"--'----''''-'---~-'''''''----'-~
tln,lllp)C Total OIlUl.

Temp., wf~igllt" 8h~I'()I. Dl'(~, ~tm(Jl. flterol, St(lrol,
°C mg. Ull-!:. Ill!. % iJ{, probi t~
-----.--~ ----.------.-.-"-.--~-- ... '_'-'_--"-'._...- " -..... ~---.----.--,-- ...-.-- .-..
13() ·l.fia G.1l O.OW o.a 0.8 2.2
HO 4.25 o.a O.O;3r; 1.1 1.4 2.8
150 1.45 7.il O.llO 2. Ii 3.1l 3.2
100 3.32 15.3 0.3·12 5.0 8.9 3.6
170 3.29 30.0 0.'115 9.8 18.7 4.1
180 3.35 48.1 0.ij,5U 15.7 34,04, 4.6
190 3.39 (iO.S 0.822 19.9 5<1.3 5.1
20() :~.37 (H.G O.87G 20.1 74.4 5.6
210 3.27 ,1A.5 0.763 14.5 88.9 6.2
220 3.2U 24.7 O.51() 8.1 07.0 6.9
230 3.11 9.5 0.240 3.1 100.1 8.7
A proeedure almost iclentieul with the elimination-eurve technique and

employed to enlmnce the separation of two adja(~ently boiling constituents
is i,he method of "amplified distillation"H in which a cttrl'ier is provided to
minimize holdup of distillate ana the resultant mixing of utljaeent fraetions.
The results obtained by Weitkamp 14 in tlH~ amplified distillation of methyl
41 Hickman, Ind. Eng. Chelll., 29, 1107 (Hl37).

,2 Gray, Hi(!kman, and Bl'OWII, .T. Nutrition, 19, 3D (H)40).
,3 Fletcher, Insu.iltllO, Cobkl", allli Hodgo, An(ll. Chem., 20, 043 (1\J48).
41 Weitkamp, .T. A.m. Oil ClwmisiH' Soc., 34, 23G (1947 J.
VI. DIS'l'II,LA'l'ION UNDJDR HIGH VACUUM 527
esters of eottonseed fatty acids arc shown in Figure 14. The broken curve
represents the eourse of distillation of the carrier oil; the solid line that
of the carrier oil-ester mixture.
B. P. OF OIL ______ .- ,.
""
"-
:r:
a
:.::
ci>
E
:;!i
:!i ~
«_j
'"
I-
_j
i=
<t (f)
I-
Z C1
a
Q_
LL
a
t!l
Il:
Z
w
ro
::i ~
a
ro
::>
z
i§
i=
60 «
u
li:
10 z
0
Q_
<'!
if'
o 20 40 60
WEIGHT DISTILLED, g.
]rig. 14. Amplified distillation of cottollHeed methyl esters. ,.,
IV. STILL DESIGNS FOR LABORATORY USE

A brief discussion of still designs will serve to acquaint 1ihe chemist with
the various apparatus employed in high-vacuum distillation. This section
is devoted to the still itself. The characteristies of the pumping systems
and pressure gages are described in Seetioll 2.
1. High-Vacuum Pot Stills
A. CONCIDN'l'IUC-TUBl!J 'l'YI'ES
These arc the simplest of all pot stills and many variations have been
proposed. Two useful types whieh are eommercially Lwailable 46 have al-
ready been discussed in some detail and are shown in Figure 2.
,6 Information on H1:gh llawmn DisWlatio1t. Distillation Pl'oduds Industries, Roches-
ter, N. Y., April, 1947.
528 lG. S. PERRY
B. HICKl\iIAN 'l'YPE
Improvements in the pot stills have been directed to facilitating con-

'tinuouR withdrawal of distillate. Several variations similar to those pro-
posed by Hickman and Sanford JG are shown in Figure 15. These one-
pieee all-glass ~tills have withdrawal tubes which emerge Rideways (Fig.
15A) or dOWllwnrd (Fig. 15B) from the still. The design in FigurA 1513
-r-
o
'"
L.'--_ _ _ ___'
~mi~~ASBESTOS
1----
60 -~-_j ~ \""-MICA AND
WIRE
(A) + INSULATION
has it SUPPol'Ling strut from bottom to top to prevent implosion of the

concave condo<ser surface, a necessary precaution as the diameter of the
still benornes ];I.l'ge. The distillate collects on the top and runs down the
strut (:0 tIle on tfiow tube. A variation in design consisting of separate
boile!' ;)!ld condenser iA shmvn in Figure 15C. The distillate can either be
fr(':~I"n 01' c!)ll('d~ed in .sample bottles by means of the withdrawal tube and
stl'I.l('fJek arl'fu~gement ,hown. The vlteuum stopcock shown in Figure
Ii> ; lid,ruM illid Sanford,.l. Plt1/8. Chem., 34" 637 (!930).
VI. DISTILLA1'ION UNDER HiGH VACUUM 529
15C permits the removal of ilny number of fractions. Its construction

permits the passage of distillate through the plug and into the receiver
when in one position, and, when turned 180°, It vertical groove in the plug
returns the distillate to the boiler. At intermediate positions, the distillate
collects in the withdrawal tube.
C. MA'I'CH]]TT AND LEVINE TYPE
A small still suitahle for collection of two fractions has been described
by :Matchett and Levine:17 This still, shown in Figure 16, is suitable for
samples ranging from 0.25 to 5 g. of distilhmd. For operation, the still,
5
I
A
'FrtfQFG
____~-Ta____~~~~~~-~~'~~~~---,B~r--
JQ + -,
, SECTION
THROUGH S-S'
s·
Fig. 16. MaLehett and Levine molecular pot still. 47
held in level position, is charged through tube H. The still pot C is formed
by two transverse creases, D, ju the outer tube, A, with the inner tube, B,
acting as eondenser. With the cap replaced, vacuum is gently applied
and the pot, C, is heated slightly to effect degassing. When the distillfLtion
vacuum is established, the still is tilted towl1l'd one ell (1 and heating is in-
creased until distillation begins. The distillate colled: on the (!onder·ser
tube, B, which is air or water cooled, and drains to the prod, G, whenee it.
drops off into receiver F. After one fraetion has been ,'ollected, tIll' "dm
is tipped towl1rd the other end and n, see and fraction :" collected ill j he
same manner. The distillates are thus obtained quantitaTively since tmns-
fer losses are avoided by colleeting the fractions in separaie condenser".
D. RIEGEL, BEISWAN(}ER, AND LA.NZL 'rYI'E
Modifications of the eold-finger still have been described by Riegel oJ. aZ. 48
and are shown in Figme 17.
,7 lVItttchd,t, 1111d Levine, Ind. Eng. Chern., Anal. J~d., 15, 2~1(i (1\143",1.
IB Rip-gel, Beiswanger, Iwd LlliIzl, Ind. Eny. Clwm., ,{nal. Ed., 15, 4]7 (. W4:-1!"
530
F
+-
ng. 18. Carothers !\!ld Hill pot 8tH!. 40 Fig. H). Strain ItJld Allen pot stilJ.62
VI. DIH'fILLA'fION UNDIDlt nIGH VACUUM 531
re. CAItO'rHEHS 'fYPE
Another geneml type of pot still which has e\Tolved in several modifica-
tions is designed for nonvenience in neeessibility of both distillate and distil~
land. One such still type ereditE'cl to Carothers and Hill·1iI is shown in Fig-
ure 18. It consists of an onter Rhell, A, made from t,,·o domes of Pyrex
distillation apparatus, a condenser, B, and a distilling pan, E, and heater,
D, Rupported by glass support C. The heater and pan are contained in an
outer eopper pan, as shown; the heater is connected to leads F brought; up
through support C. Carothers employed thiR still for studies of the poly-
merization of the trimethylene estel' of hDxadccnmethylenedicarboxylic
acid 50 and mallY other similar reactions!'! The still can he used for distilla-
tions of any type involving separations, eoneentmtions, 01' stripping of
volatile constituents.
F. S'l'H,\IN SND ALI,IDN MODIF1CNl'lON
A modification of Carothers' still was designed by Strain [tIllI Allenf'2; in

this still the tempemtnre of diHtillation is readily eontl'olled. This still,
HllOwn in Figure 10, was lIsed by Strain lLnd Allen for the pnrificn,tioll of the
hormone progestin.
G. GOULD, HOLZMAN, AND NIEMANN DESIGN
To effect effieient transfer of distillate from condenseI' to receiver in

centigram 01' decigram quantity distillation, Gould et Ill.50 devised a micro
pot still utilizing it unique receiver. The still, shown diagrammatically in
Figure 20, consists of a boiler, A, and it cold-finger nonrlenser, B. A Hange,
C, on condenser Baets as a collecting tip in incipient contact with a eapil-
lal'Y tube, E, eommunitmtillg with receiver D. The details of the COll-
denser-receiver systemfi<l arc show11 in Ii'igurc 2Gb. Using this still, Gould
et al. distilled butyl phthalate, under varuum and at temperatures below
50°C., ut a rate of 20 tLl. pel' minute.
49 n,rot.herH LInd Hill, .T. Am. Chem. Son., 54, 1557 (1032).
60 Carothers and Hill, J . .11m. Chem. Soc., 54, 155H (11)32).
61 Mark awl \Vhitby, edH., ('oll{)ctivc Pape/'s oj lYaliace Hurne ('w'ollwl's on High Poly-
meric Bubstl!nce.~. Intcl'scienee, New York, 1\140.
62 Strain !Lnd Allen, Ind. Eng. ('hem., A netl. Ed., 'T, 443 (1935).
53 Gould, HolzIntlIl, tHld NielllmUl, ,tnal. Ghem., 20,361 (H)48).
G·l The Fabrication of the c:olldcnS(iI'-rcl~Biver sYHtem illuHtl'!lted is diffieult, though Olle
maJlufaetul'(]]' (Ace Glass, Illeol'pomteil, Villoland, New Jel'HlJy) has made 11 !lum]Jer of
these st,ills.
532 Ii" S, PlmnY
VACUUM
__..
WATER IN
(0 ) (b)
"'ig, 20, Gould, Holzumn, lUttI Niemlwll mim'o pot sti]l,53,""
H. AI,MQUIS'l' DIFFUSION 'l'YPE
A diffusion-type high-vacuum still as shown in Fignre 21 was reported

by Almquist5» nnd llsed for the pUl'ifieation of vitamin K. The still is de-
signed to effeet fraetional condens!l.tion of distillin!~ Vapors and consistR
~
~~~#~~~~~------~~
A B c
Fig. 21. Almquist tuhullll' st.ill. 65
essentially of a tube divided into three (:ompartments, each compartment

being separately heated whereby a temperature gradient exists between ad-
jacent compartmont".
2. Falling-Film Stills
The falling-film still consists of a heated vertical tubular evaporator,
over which distilland is allowed to How by gravity. In so doing, the dis-
tillancl spreads over the entire surface in a thin film which is usually 0.1,
6& Almquist, J. Biol. Chern., 120,635 (1937).
VI. DIS'rILLA'l'ION UNDER HIGH VACUUM 5:3;3
WIRE GAUZE
GROOVE
EVAPORATOR
COLUMN A COLUMN B
Fig. 22. Falling-film still distributors. See page 034.
COLO TRAP - - - I
VACUUM - I---!~-- CONDENSER
!&"x"i(",\lf...!---...f+--- EVAPO RATOR
PREHEAT TUBE
RESERVOIRS
Fig. 23. Falling-film moleculal' still. Seo page 536.

534 lD. S. PEnRY
to 1.0 Hun. UJiek, depending upon the feed ratio and viscosity of the distil-
land. The evaporator is SUl'l'olUlCled by a (~()ndensel' tube (',oneentric with
the flvaporaLo)' and separnted from the evaporator by generally unob-
strneted spaee which is cOlltinually exhuusted to ::t pressure of about 1 fJ.
Fig. 2,1, Falling-Jilm moleeulal' Htill. "U,60 SOC page 53(1.
of mereury. Dj~tilland is stored ill a, reservoir from whieh it is pumped by

means of [1, small displaeement pump to [~ tl'oughfiG or groove which sur-
rounds the evaporator column (Fig. 22). The trough (Fig. 22A) contains
evenly spaced serrations in contact with the column which divide the dis-
tilland into n ~umbel' of streams around the cireUtnference of the evapora-
tor. These st,reams nrc then intercepted by a fine wire gauze whieh i;ntns-
66 IIiclmlllll, Ind. Eng. Chcm., 29, nOB (ln37).
VI. DISTILLA'l'ION UNDIDR HIGH VACUUM 535
forms them into a continuous film. In a modified design (Fig. 22B) the
incoming distilland travels around a groove and, on overflowing the groove,
it passes through the fine gauze to forIn the film.
A. HICKMAN DESlGN
An early cyclie bateh falling-film still was designed by Hickman"7,~g

The somewhat modified form of this still now being used is shown diagram-
DISTILLATE
RECEIYER--
RI:::SERVOIR
Fig. 25. Tllylol' falling-film molecular still. 65 Suo pagu 538
.7 Hickmtm, Ind. I£ng. Chern., .titwZ. Ell., 29, 968 (1937).

5~ Hickman, U. S. Pat. 2,117,802 (May 17, 1938).
536 E. S. r.ErtnY
matic!LlIy in Figure 23. A photograph of a typical still which IS com-

mercially uvailuble Go • 60 is shown in Figure 24.
B. QUACKENBUSH AND S1'1~.ENDOCK D.ESIGN
A modified form of the falling-film still waH described by Quackenbush

and Steenhoelc 61 It, has the cvapor:Ltor surrounding the eon denser to pro-
• --3-WAY
4-rnrn. BORE
+---THERMOCOUPLE
DISTILLATE
RECEIVER
l?ig. 20. BJ'(~f.!;'ll' ~(lmimicro falling-film molecular sUII. 66 See page 53\) .
•0 See HIgh Vcwuu.;'! jiJqIdplnlmt, Dist.iJlltf,ioll Products Indust.ries catalogue.
60 Information on H iflir Vacuum Distillation. Distillation Products Industries, Roches-
ter, N. y" April, 19 17
61 Qurlcjwllbush al,,1 St,e(mboek, Ind. /iJn(l. Chem., Anal. Ed., 15, 4(18 (19·13).
VI. DIsrrrLLA'l'ION UNDER HIGH VACUUM 537
vide a large evaporator surface and a relatively smaller surface from which
condensate must drain. Distillalld is spread over the evaporator by meaIlS
of glass rods which are made to scrape over the evaporuting surface. Other
modifications of this method of distributing distilland are the rotary blades
proposed by Hiekman and Pcrry fi2 and the rolling rods suggested by Se-
mon. 53
C. DgTWILER AND MARKLEY DESIGN
To effect satisfactory distrihution and flow of distilhtnd OVCt the evapora-

tor, Detwiler and Markley 61 used a tapered-sleeve guide to conduct the
Fig. 27. Ditlgramnu.t,ic elevation of slllall centrifugalilloiecullll' ~r.i11.67 See page 540 .
•2 Hickman anti Peny, U. S. l'!lt. 2,403,H78 (Jllly 16, 1\)4(\).
63 Semon, U. S. Pitt.. 2,460,602 (Feb. 1, 1\)40).
64 Detwilor and l\hrkley, Ind. Eny. Chern., Anal. Ed., 12,348 U!:iO).
538
diiltilland over the mushroom-shaped !lend of n. rliHtilling mlumn. The ('ll-

tire Blll'faec of the eolumn ht18 n roughened cOll,ting of frl1.gmented glass
fused to ii, to provide illllumem.ble tortuous r:lutnnels for the How of dii5-
tilland.
D. 'l'A YLOH D1GSIGN
Taylor SB utilized the pl'ineiple of (listillaml diKtribuLion of Detwiler and

Markley in constructing the still shown on ]mge 5:35. In this still, distil-
land is pumped into an inverted enp whieh Slll'J'OllIHIs the upper portion of
the (WllporatOI' but does not touch it. Tl10 annulus hei;wPfm the nip and
Fig. 28. Centrifugal molecullLl' sl;ill.G~
65 Taylor, .1. RcsearchNatl. Bur. Standal'ds, 37, 173 (19'16).

68 Breger, Anal. Chern., 20,980 (1948).
V[. DIS'rILI,Nl'ION UNDER HIGH VACUUM
the evaporator is narrow elloug;h to retain oilnntil sufficient head is built

up to induce flow. Once stmtt'd, the clistilhnd continues to flow uni~
formly down the surface of the evaporator.
E. BHEGER DI~SIGN
The conventional falling~film stills are limited to mucro-size samples bc-

eause of the quantity of distillaud necessmy to \Vet the surface areas of the·
reservoirs, feed tubes, fLnd pump. Bregm· 1ifi described [1, semimiel"o falling-
film still which is suitable for h<1lHlling samples of as little as 4 g. This
still is shown on page 5an. Hecycling is aecompli:.;hecl by rotlLtion of the
Rtillnhollt the exhausting tulmlatioll. Both (mds of the still u,re identical.
The material to lw distilled is clmrgecl in at olle end find allowed to flow
over the evapomto[" to the other eud. After withdrawal of the distillate,
the still is turned around and the pl'oeess is repeated.
Fig. 20. A]l-glll~H (J:lsrmcie i"r:wLionatilip; molecubr still (measurements in

millim(]tel"s).70 SCI) pltge MO,
61 Biehler, Hiekmnn, nnc1 Perry, Anal. Chern., 21, fl38 (Hl4fl).

OR Hiekman, U. S. PILtS. 2,1l7,H02 (lVIlW 17, 1\)38), [Iud 2,210,\)28 (Aug. 13, 1940).
60 Illformation on FIiyh VOI'nuII! Distilllttion. DistilhLtion Pro<incls Industries, Roches-
ter, N. Y., April, 1\)47.
;',) Wollner, l\btnhett, nnci LI~virll1, II/d. gnr/. CIWIII., Anal. Ed., 16, 52!) (Ill-H).
540 ,To C. HECKlm
3. Centrifugal Stills
A smull centrifugal still has been described previously in connection with
the discussion of the cyclic distillation of orange oil. This unit is shown
diagranunatically in Pigm'e 27 (p. 537); for a dctniled description, refer-
ence can be made to the literature. 07 ,OB A somewhat different design hav-
ing greater capaeity has been dcscribed08 ,GD and a completc unit of this kind
assembled for cyclic bateh operation is shown in Figure 28 (p. 538),
4. Cascade Fractionating Stills

The cl1scade-type stills have been previously referred to itlld consist of a
number of evaporator cells arranged in cascade with a number of reversely
sloped condenser stages nrrnnged above the eVII,poratOl' eells to move dis-
WhLte countercurrent to the flow of distilland down the cascade. The
first cascade still, credited to Wollner, lVbtehett, and Levine,70 is shown in
Figure 29. A more elaborate cascade [1'aotionl11;01' has been designed and
reported by Brewer and Madorsky.71,72
l'al't2. THE VACUUM SYSTEM
I. INTROlJUCTION
tlo far ill this chapter, atteniiion Ims been directed to the design and appli-
cation of high-vaeumn stills. A neeessary and important pt1rt of II,ny vac-
uum still is t.he vacuum system. Under this heading are included such items
118: (1) high-vacuum vapor pumps, and mechanical fore-pumps; (2)
gagesj (3) vacuum plumbing, whieh includes interconnecting tubes and
ducts, valveR, gaskets, stationary and rot!tiiing seals, etc.; (4) cold traps
and baffles; and (5) accessories, which include leak detectors, sealants,
ete"
To understand the working of a vacuum system, it is desirable to know
the specific. function of ·the various components, to have a knowledge of
their basic design features, and to appreciate the general principles of high-
vaCUUlll production and practice. It sometimes seems anomalous that, as
the operating pressure gets lower and lower, lal'gel'-capadty pumps and
lal'gl'J' interconnecting piping must be used. A knowledge of the deriva-
tion, application, and limitations of gas-How formulas nt reduced pressures
is helpful in understanding this problem.
Tn assembling a piece of distillation equipment, the still and auxiliaries
II Brewer and Madol'sky, J. Research Natl. Bur. Standards, 38, 129 (1947).
;' Brewer and Madorsky, U. S. Pat. 2,446,997 (Aug. 17,1948).
VI. DIS'l'ILLA'rION UNDER HIGH VACUUM 541
should be considered as a vacuum-distillation as:;embly and the component

parts so designed and so sized as to perform adequately the job in hand. It
frequently happens that a skillfully designed vacuum still cannot function
properly because an inadequately designed vacuum system limits its per-
formance. It is not the purpose of this section to present all the rules for
design of laboratory vacuum systems, but to present basic fundamentals
and simplified guides which may be helpful to the chemist who wishes to
apply molecular or high-vacuum distillation to his problems. In a good
many cases the vacuum system is 1l1rger than the still itself, that is, larger
than the evaporator and condenser portion of the assembly. This is ex-
emplified by the photographic reproduetions in Figures 24 and. 28 (Part 1),
and schematically in Figure 50 (Sect. IX). It is evident that the distilla-
tion portion is overshadowed by the auxiliary vacuum equipment. The
reasons for this seemingly unbalanced relationship will become apparent as
the chapter progresses.
n. DEFINITIONS Ole TEHlVIS

To facilitate the discussion which follows in this section, a brief definition
of terms, sevemlrecnrring constants, and a few basic formulas commonly
used in high-vacuum distilltttion practice are given.
Siah-vac'1lwn distillai£ons, in generl1,l, are conducted in the 0.1 to 1000 I"
region, while molecular distillations, in particular, are usually conducted in
the 0.1 (0.0001 mm.) to 10 I" (0.01 mm.) of mercury range. The rm;cl'on, jJ.,
of mercury is the unit of pressure (not vacuum) which corresponds to a mer-
cury column 0.001 mIll. high and this unit has been widely adqpted in high-
vacuum discussions. Pressures in microns will be referred to as PI' and
in microbars as P Ph' Remembering that in the e.g.s. system the unit of
pressure is 1 micro bar (1 dyne pel' square centimeter) and that 1 atmosphere
is the pressure exerted by a column of mercury 76 em. high a.t O°C., the
relationships shown in 'Table X are derived. It is interesting to note that
in German literature 1 Tor, or Torr, is Ul'led to designate 1 mm. mercury
pressure. Also, engineel'l'l frequently use as their standard 1 atmosphere =
29.921 in. mercury at 32°F. = 14.G9(:i lb. per square inch.
TABLE X
PRESSlffiE UNITS
Atm. Bat· Pm.1Jl. Ppb
l.013 7(iO 760,000 1.013 X lOG

1 750.0() I X 10'
1 1,000 1333
0.001 1 1.333
0.715 1
J. C. HlJ;(lKEm
In dcali ng; wi Lli gatl flow at l'cchwed PI'(:BSUI'PS, Lltt: ml.e of flow itl usually
expreHRed ill one 01" two ways:
(1) Vollt"HW liV/" 'uwa time, Jllea~ured at. the prf',~HUre at whir:h the volume is de-
termine!!. Common units arc iitll)",q per sr,cmul, or culn:(: 11(:1I.l'I:me(crs per second.
The letter F will be llsed to ileRigllate flnw through tubing, am) 8 for plllllping speeds.
(2) !'I'e,~sttrc-1Johl1lU: pel" nn'it time, red ueerl to a sek:eted standard unit pressure.
Common units are micl"on Uien pel" second; that; i~, Ute volume of gas fiowillg if the
pressure wer~ 1 p.. The lottm Q will he used to express the quantity of gas flow.
The liter mieron flow, q, i:; eOIlHtunt t1mlllghout; allY ehosen Iwstem while
the liters pel' second, S, vu,ry at:co]"(liug to tho PI'f1KSlll'O. ThiR may be
madr. clearer by thinking ()f mic:roll liters ILK lL maRR whieh. is independf1nl; of
pressure, whereas voiliitletl'ie How in liters (per H(1C'.ond) is a function of pres-
sure. The volumetrie rate of How in liters PCl' so('ollli iH obviously rc1ntcd
to micron liters per K(K',OIl(L HR follows:
8 = (J/P 1 £
l,arge commercial vamnnn insLnllaLillIls Jn~qlwntly llRtl pmmdR per hour

t.o reprcsent gas How. The large vollllnes that must be pumped t.o handle
n relatively tlrnaU mass are evident, from Lhe relation for ail' at 25°C., where
11b. per hom is approximately 80,000 mi(1)"()1l liters per second.
Two useful exprH~RionR in dmding wid] gas flows nt, red\lced l)l"eSsures arc
conductance and I'Csi8la.nce. Conduotml('.(\ is the l'n.tn of flow 1)01' ltnit (hi-
Jej'cnce oj p1"e,~,~tti"(! whi<:h wn shall dnsigllate n:-;:
(1)
and llF LIte l'GHistanee. Tlw analogy followH front Ute oIe(:trieul pattern of
Ohm '8 hw when P is compnt'orl to pol,ontia! E, and (J to the current I:
R = lIt P = Q(l/P)
As in cleetrio1Ll eircnits, tubes ill ~nJ"i()s have tL total resistum:e, P, calculated
as follolVR:
H = Hj + 1l~ + '"
1
F
:_-1 + -_.+
= FJ 1
p~ '"
(2)
while tubes in parallel luwc: the 1'oeipJ'Ocal:-; of the msisl;urweH, 01' the eon-
ductances added:
1 1 1
-- = --- + -_- +
Il Rr Rz
I? = PI + F2 + '" (:1)
VI. DIS'l'ILLATlON UNDEH HIClH V"CUUM
The application of this concept will become lUnl'C evident in the discussion
of complete :;;ystems and in speed caleulations in the molecular-flow region.
The actual 8peed of e:t:ha1l8t of l1 vaeuum ehrtlIlhel' connected to It pump
d(~pends on the resistance to flow of gases through connecting tubes as well
as on tllf~ Hpeed of the pump. It. can he sholVn that, if the speed of the
pump is Sp, the term 1/8p can he regarded as [L resistance; and for any
system having a resistunee of I/F up to the pump, the eomhined effect, is:
(4)
8
where ,,) is the speed of the system. This ean be rewritten as:
The "rule of thumb" to llUtke high-vacuum eonne(~ting tubes as wide and

short as possible can be dearly grm;ped hy extllllining the above relation-
ship. The effect of the l'cHistance of conllceting tubing in throttling the
Hpeed of exhaust is shown in Table XI. Values of F' as muHiples of Sp are
given and values of 8, the effective exhaust speed, as a fraction of the pump
speed 8[,.
TABLE XI
EFFEC'l' 01<' TllBlNIl H.IGSIRTANcm ON SI'IGElJ (ll' EXIIAURT
._------._---- _ - -
PIS11
---.--~----------.-.----.---.---.----~-.-.
1 ........ 1/2 s .. ..
2.. 2/:, Hi .. ..
4. . "f,
When the eOIlnec:ting tuhing 11m; a speed equal to that of the pump the
resultant speed of exha\1Ht. of the vacuum chamber iK half the pump speed.
Even when the eonneeting tubing has [\. tlpE'erl 9 times the pump speed, a
reduction of 10% of t.he effcetivc vapor pump c:apaeity results. And as a
corollary to the above, very little is gained in oxhu.ust speed by increasing
the Ilize of the vacuum pump if P, the c:ollneeting tuhing conductance, is not
large compared to 8[" the speed of the pump. These consiclm.·ittions are of
great importance in arriving aL an efficient pumping system.
The bnsic: constantHia used throughout this section are listed l)elow for
l'efel'enee:
73 DnshnltlIl, Scientific ji'owulatilJns q( Vacltwn 'l'cc/miql1l:. ''''iley, New York, 1949,

Chap. 1.
544 ,J. c. H1~CKlDn
pTf',~mll'e
in miel'0I1H
pl'e~Hlll'n
in miel'oban;
viseosit,y in poises
1. 8'11; X 10'" poise (e.g.ll. uuits) for [til' :l,t 25 DC. (5)
11, = the numher of llIolecules per cubi(~ ('(mtimetiel'
P/kt = f).(iM X 10'[' (PIL/']') (0)
2, G87 X 10 19 at. (J°C. n.n<l 760 nun. --- Lmwhmidt; numb('!'
2.43U X lU I3 at 25 DC. and 1 mierobul'
3.240 X ]OJ., at, 25 DC. and 1 IL
2.462 X H)ltl nt 25 DC. lIud 760 mill.
7' dcgl'Ct'H nbsoltlte or de!!;l'ees Kelvin
273 X 1(; +
t "'here I = degrees eent,igl'wle
N.4 1i.023 X 1023 mok, -" Avogndro !lumber (7)
Ro 8. :31,,) X 10'/ ergs per dogree Kdvin pCI' gram mole (8)
k Ro/NA = 1.381 X 1010 cl'gs pel' degree Kelviu-- Dolt.zlIlILun (!onstaut (H)
ill molo(lulnl' weight
Mfol'nir 28.98 (clIIIJulntnd from It d()I1~ity of 1.293 X 10- 3 g. pel' cubit! ef!utimeter
at ODC. 11m] 760 lllm.)
m MiN,! = I.BOO X 10- 2 ,' M grmns-the mass pr,!' molecule (10)
I' mil = M/TTo = MPI'I,jll,,'l', the dem;ity atP!'b (11)
1,004 X 10-" MPILI'J', g. P[)l' oubin ('.(mtilllctel'
L 5fiO X ]0-11 g. pel' ouhil! e(,lJt.illwter fO!' nil' ILl 1 iJe nut! 25 "C.
1.170 X 10 --9 g. pel' (mbie (1()Ilt.imet.el' for ail' at 1 ",/) lLnd 25 DC.
8.580 X 10-' {l2)
The quan tit.y of gus 01' va.por striking unit at'ea in unit t.ime:
= mass = fj,833 X 10-(' PIL'VMj'l' g. per sq\ll~r(\ (lontimotel' pel' second
1//, (13)
u = VOlllIlW = 3G38V'l'jlli ('.[11. 3 pCI' HC[UfLI'C <',ent,ilrwt.el' per second (14)
V'T/M = 3.207 fol' !Lir lit 25°C.
Moleeulal' dirLmetel'S:
li llloleculitl' dillllll'tel' in (,(mtimetel's
,j2 Z.714 X 10-'21 VllrT/~ nm. 2, 1f heing tho gas (Il' vltpor viseosity Itt t(~lI1-
pemtlll'C 7' (15)
OJ'; Ii 1.329 X lO- H (M/,»I/ a, in whieh p is the dellsity in t.he (l(]ndell~ed ph!1se
(liquid) (10)
Moleeular dat.a for the common gases H,ro given in T!Lble XII.
TABLE XII
_ .. _.•.. _....... .. _----
...
,-
-
_----,
(bfl
MOJ,ECur,AU DA'l'A (25 "C.
H, He O.
AND
N,
1 }J Pm~ssuRE)"
Air CO, n,o Ifg
9.31 14.72 5.40 4.96 5.09 3,30 3.37 2.6G

'I
"X 10'1 0.892 1.986 2.059 1.85 1.845 1.496 0.964 2.54
Ii X 108 2.73 2.17 3.59 3.74 3.70 4.56 5.55 5.15
mX 10 23 0.335 0.665 5.31 4.65 4.81 7.31 2.99 33.3
('1# X 10
to 0,813 1.615 12.90 11.30 11,70 17.75 7.27
, ). is mean froe path in centimeters, 1'/ is viscosity ill pois[)s (c.g.s. units),.u is molecullH'
di,,:;nder, rn is muss pel' mulecule, P II' density in grams per cmbic centimeter at 1 )1..
vr. D1ST1LIJA'1'10N lINDEn HIGH VAcuuM 54ii
HI. GAS FLOW AT nm:mCm) PRESSURES

1. General Discussion
The physieist is usually better a.eql1ainted wit.h the beluLVior of gases at

low pressures than if; the chemist.. The fundamental theory and matheroa-
tieal analysis have been adequat.ely covered in tl1f~ litemture. 74 However,
a practical interpreta.tion of the flow formulas is usefnl t.o the laboratory
worker who is primarily concerned in assembling apparatus and putting it
t.o use on fl particular distillat.ion problem. Therefore, cert.ain liberties
have been taken with the basic facts in redueing them to approximat.e
working formulas which are sufficiently aeellrate for vaeuum-dist.iIlation
applications.
Gas fiow through channels at reduced pressure call be divided int.o three
regions: viscous, rnoleC1.~lal', and 'intermediate. This follows from the fact
that, at preSSl1l'es in which the frequency of collision between rnoleeules is
predominant, the coefficient of viscosity as introduced in the Poiscuille
law75 is most important, whereas, at lower pressures in which the frequency
of collision between molecules is less frequent than that of molecules against
the walls, molecular flow as defined by KuuclHcn76 resnlts. Since the type
of flow changes gradually from one pressure region to fLllother, t.here is the
third, intermediate transitory region ill whieh the flow is partly both. In
some eases \vhere only a rongh approximation is d(}sil'erl, this intE'l'lnediate
region can be neglected.
2. Flow at Intermediate Pressures. A General Equation

Knudsen 71 endeavored to arrive at an equation which would enver the
three fiow ranges-viscous, intermediate, and molecular, This probletn
was approached empirieally llnd from a series of mCnSlll'Chlents Knudsen
deduced the relation:
p 1 + CIP~b
ao - . ' - - - eubie centimeterH pel'
·1 + C2P"I' Heeonci (17)
74 Dushrnan, loco cit., Chnp. 2. KellluLl'<l, IGneti() 1'heorll (if Uases, MIlGnu\'.lIill,
New York, 1938, Chaps. 3 and 8. Loeb, Kinetic 7'heorll oj Oases, 2nd ecl., !H1.emw-
Hill, New York, H)34, Chap. 7. Taylor, it 7'l'eatise on Phl/sical ChemistI'll, arc! ed" ·'.'IUI
Nostrnnd, New York, H)42, p. 174.
76 POiseuille, Societe PhilornathiqlLc de Paris Bullet'in, 28, 77 (1038); Compt. re;/.,,'., j 'I
961, 1041 (1840); 12,112 (1811); 15, 1161 (1892), Loeb, loco cit., p. 230.
76 Knudsen, Ann. Physik, 28, 75, \I()9 (1009), and Hllbseque.\t paplll'S. Loeb, lUi.:. i (/.. ,
p.290. Taylor, loco cit., p. 108.
77 Knudsen, Ann. Physik, 28, 75 (lHOg).
MG
in which PJl./, the avnrngc pt'e~Hure in mierolml's

11" ,/,.,
~1 (18)
n viscouR-How fae-Lor
8r A 4rl A 4 rl
_. - PI (19)
Xt' V'~~ ~ 3 l V211"Pl :1 I I
= n molecular-flow fador
in which mdiu:-;, rl = cliametnl', [mel I length,
2TVp; r
(2m
C\
nnd
'l)
= 2.507
"I
rl
1.547 -- (21)
where Al iH the mmm fret) path at. 1 mierohal'.

"I
The mean fme p:tt,h78 is defined a~ the (t/!(J/"(I·ue <iiKtallce t,ravel'~(',(l between
sueeestiive eollisions hy :Lll moleeuleH [l]'e~c.m(;. Hinee the froe path is baser!
on the 0011(·Opt. of eolliHiolts, lmel Ki!l(:o the l'n:qucuey of eollisiol1s depends
on vel()(:ity, it is n)lp:mmt. that (.he frpp pld.h iH lIot a i:ipoc·ifie value bllt
a statistical tW(\I'!LgC of lLll the distmtens tmvm'sml fur allY given Het of eou-
ditions and typml of moleellles preHent. When one RtatC)S tlmt the mean
free pat.h of an air Illolec:llle at 1 J.L and. 2FjOC. if; ahnat 5.1 em., it is meant.
that, on the i1vcmgc, ail' molc(,llies ill the Pl'OSCll(\(\ of othol' ltil' molecules
will travel that distnneo l)(\Lwccn ('olliHiollS. (Thel'o is of eOlIl'HO no tmeh
thing ~tH nn air molecule, hut it is eOllvenien[; (;0 assume HlWh exist.H. The
free path is ealeulnt,<xl [rom a vi::leosity valllo, HS will be shown later.)
However, somo may tmvd only tt fr[l.(:t;iol1 of a. millimeter, whilcI a few may
escape collision fol' nmny mctol's. In Fip;un: :{O the relationship is shown
between tho mean free path and the di~tllnee truvelml between sHceessivo
collisiollK. .For lL glU., in which the molnGnleR ttm all of t.he samc) I'lpeeics t.he
mean free path/9 A, is mInt.ed (;0 t.he moloC'ulm' clinmeter by:
(22)
where n number of molceul(:)s pel' cubin eentimet.er = P /Ic'l'

= D,65(i X 101[] P 1'/'1' = 2.4(\2 X 10 19 at 2fioc. lWei 7no mm.
and G = molecular diameter in <:nntime[;cl's.
18 .Kenn1Lrd, lo~.
rEt., pp. 10(-]02. Loeb, lor. !'it., CIllLp. 8.
tD Ch!1jlll1l1I1 nml Cowling, The l1Ialhmnnlir.(11 'l'heor,l} oj Non-uniform (}nMiS. C,III]-
hridge Univ. Pross, Lon(I(Jn, luau, Cluljl. 12, p. 218.
VI. DIS'l'ILLATION UNDBll HIG-U V,\CUUM 547
Substituting the above value for n givcs:
t.. = 2.:3:3 X 10- T

17
e e •• _ _ _ _ _ _ e - -
(23)
52 PI'
The molecular diameter is l'(~la.tecl to the viscosity, ?), by:

52 2.714 X 10- v1l1Th tlquHl'e centimeter
21
in which 11[ molecular weight, of the gns or vapor

?) viKcosity in c.g.s. units (poises)
T ahsolute temperature
From the viscosity the values of 02 can be culeulated and hence the values
of AJ as given in Table XII. Assuming air contains all like molecules, the
100
90
80
\
70
\
60 \
50
\
40 \
30 1\
20
\
\
0 1""-
~
o
I--
234 5 ~ 7
Fig. 80. HelttLionship between free Pflth (}..) l\1ld distnnee (d)
Il1lHlll
trnV()I'RC(! bet.ween sueel1ssiVl1 (iollisiolls; x = rl/'/..; 1/ = ex.
mean free p!tth is 5.09 centimeters. Since the mean free path is inversely
proportional to the pressure, for air at 25°C. and at ally pressure, P" the
menn free path:
548 J. C. lIMJKI~lt
5.09 .
A = - PI' - eent.1meters (24)
.Formulas fo1' froe P1Lths of molecules in the presence of unlike molecules

h:we been derivecl Ro and serve their purpose but are usually not needed for
t.he da.y-to-day practical ealculations of vacuum-distillation systems. In
dealing with large IlollspherinnI molecules, mel:Ln free paths calculated from
equat.ions (22) or (23) tHe likely to he in error. cl'here is some indieation
that large nouHpherieal oil moleeules, sueh as amyl phthalate, amyl seba-
eate, and 2-ethylhexyl phthalate, IHwe "undofleeted meau free paths" sev-
eral times longer than those calculated by elassieal theory.81 Also, in dis-
tillation opemtions the (~ffeet of tempemtul'c on the mean free path may be-
come significant. A few values ealellluted for ail' from equation (23) are
shown in T~Lble XIII and inclieu1;e Lhe nu\gnitucle of the liempem1iure effect
TABLE XIII
A vs. t D
t, DC .. _............ 0 25 50 100 150 200 250 300

A...... _........... 4.7 5.1 5.5 (i.4 7.2 8.1 8.9 9.8
---~ --------~ .. -- .. ------.--------------.--- ------------------
at a constant pressuro of 1}.l. Tlw moleeuiar diameter has been assumed
constant; however, for large temperature nmges S2 lweount must be taken
of the increttse in thn value of 02 with 'iT. The conversion from viscosity,
'f1, to mean free path at 1 mierobar, Al, resultfl from the relationship 'YJ
O.5P1VaAl, nud !la = 4!-V21fPlI the average molecular velocity. Then:
2P1AI 2-VPIAI
1] = -: /-=" = -~/-c::.-='-
v 21fPl 'V 211'
Since P micl'obl1rs = .l/aP mierons, and Al at 1 miel'ohar = 4/a ]l. at 1 It:

P,LI/t Pl'd Pel
---' = -- = .-,-
Al'b AI' t..
in which P without; a suffix signifieR pressure in mic:rolls and t.. the mean free
path at l}.l. Therefore the fa.ei.ol':
r.' 1 +
CIP~b
Z = -------.----- = ~~.25?_~~d/A) (25)
+
1 C12P"b 1 + 1.547(Pd/A)
8' Loeb, }{_'inetic Theory oj Gases, 2nd cd., Mc\Graw-HilI, New York,
DU8hmall, Scientific F'owwati(JIIs oj Vacwum 1.'echrliique, Wlley, New York, 1
81 Jacobs !tnd Kapff, Ind. Rna. Chern., 40, 842 (1948).
82.Chapmau and Cowling, 'l'he Mathematical 'l'heol'lj of Non-uniform C!I.m-

bri(ige Univ, Press, I,ondon, 1939, p. 24l,
VI. DISTILLATION UNDER HIGH V.'\CUUM 549
in which rl is in centimeters. It is to be notcd that Z is the same whether

P and A are expresseu in units of microns or microhQrs a;,; long [LS consistent
units are employed. For ail' at 25°C.:
1 + O.24GIPd (26)
Z = 1 + O.3040Pcl
Converting the first term in equation (20) from microbm's to microns,
and substituting the value of Z, as defined above:
F = 4/3alP + Cl.2Z cubic centimeters tier second (27)

where P is in microns. For air at 25°C., land cl in centimeters, and p:'es-
sure P in microns:
a
F = 0.177 -l
d4
P + 12.2 rl-l Z (28)
cl a
= 12.2 - (O.D145Pd
l
+- Z) liters pel' 6'eeond (29)
It is readily seen that for small values of P the fin:;t term vanishes, Zap-
proaches unity, and F beeomes equal to molecular !:low for long cylindrical
tubes (equation 3G). Also, as P becomes large, Z approaehes 0.81 as the
limit, and the flow reduces to viscous flow and bc(mmes practically equal
to equation (55). A discussion of molecular flow and viscous flow will
follow in this section.
In Table XIV are given the values of Z and of 0.0145Pd for various
values of Pel. For all values of Pd above 1000 a constant value for Z of
0.81 can be used, and for all values of Pd below 0.01 a value of 1 can be
used. The chart in Figure 31 relates Pd, Z, and O.0145Pd. For any given
value of Pd, and at right angles to the vertical Pcl ~calc, both Z and 0.0145-
Pd can he read on a horizontal line and ean be SUbstituted in the appro-
priate formula at the right of the chart to arrive at the conductance of the
TABLE XIV
VALVES OF Z AS A FUNCTION OF F>d
Pd Z Pd Z
10' ..... '",.... 0.81 0.2 .......... 0.988
10 3 •••••••••• 0.81 0.1 .......... 0.993
10 2 •••••••••• 0.815 0,05 ......... 0.997
10 ...... : ..• 0.856 0.02 ......... 0.998
1 ....... '... 0.956 0.01 ......... 0.999
9i 5........ 0.975. 0.005 ........ 1.
~' 0.001 ........ l.
550 J. C. HECKErt
tube in question. H is apparent both from equation (20) and from Fig-
ure 31 that for values of Pd grentor i,]U111 1000 the flow is a,bout \-)5% vis-
COllS, and for values of Pel Jess than 2 the How is OVOI' 95% molecular.
These two points Imvc been selected l1S al'bitml'Y values for separating the
zones of gas flow, remembering that the numOl'ieal uocfflCients arc for ail'
at 25°C. Knudsen suggested that dj'A equal to or less than 0.4 be used to
0.0145Pd Pd Z
80
'5000
60 4000 4
50~ d
VISCOUS 40 - 3000 F'O.I77T P
2000
20
1000 0.810
800
600
- 500
400
_- 300
3 N
+
--.:_ 100 0.815 "a..
({)
w 80
1.0 :!
~
0
60
50 0.820
0
d
w ': 40 0.825
:::E
0::
0.5
30
'%1-
w 0.4 0.930 C\J
I- 0.3 _ ~
z 20 -0.935
-0.940
0.2 0,95
lJ._
10 0.96
0,1 -. S
0.97
- 6 0.98
5 0.89
4
0.05 - 0.90
0.04 3 -0.91
0.03 0.92
-0.93
0.02 - 0.94 3
MOLECULAR 095 F = 12.21
-- 1.0 0,96
0.01- - 0.8 -0.965
Fig. 31. Chart relating Pd, Z, Q,0145Pd.
define moleeuln,r flow. This gives the valuo of Prl equal to or less than 2
whieh has been adopted in this section.
As an illustration of the application of equation (29) and Figure 31, as-
sume a tube 2 om. in diameter, 50 em. long, with nil' flowing through it at
25°0. at 10 ;..t pressure. What is the conductance of the tube? Pd = 20 is
selected on the Pd line and Z and 0.0145 Pd are read on the Pd = 20 hori-
zontal line.
VI. DIS'l'ILLATION UNDER HIGH VACUUM 551
Pel 0.0145PcZ (O.0145Pd + Z)

20 0.84 0.20 1.1:1
The flow is in the intermediate zone, and using the formula indicated:
el3
F :::: 12.2 -i' (1.13) = 2.2 liters per second
Equation (29) and relations derived therefrom give the conductance of
the cylindrical tube itself and take no account of end efIects. As a strict
interpretation, an end correction should be made when l is less than 50d.
Such corrections are m.ore significant in the molecular-flow region where
ordinarily the tubes are made as wiele and short as possible.
3. Free Molecular Flow

A. THl~ FAC'l'OH }V
The expression "molecular Bow" was suggested by Knudsen. 88 As the

pressure is progressively reduced, a point is reached at which the average
distance a molecule travels is comparable to the dimensions of the vessel.
The rate of flow is then governed primarily by the effect of the collision
against the walls and not by intermolecular collisions which determine the
viscosity. An analysis of the httter problem has been made by several
workers. 83 - R6 Applying the Maxwell-Boltzmann distribution law,87 Knud-
sen derived the relation:
Pz - P l
Q
wV;;;
Q 1
or: F= (30)
p z - PI w-J~
where P is in microbarH, Pl equals density at 1microbar, and W iA a constant
depending on the form and dimensions of the tube only. I'V Vp~ can be
considered ns a l'esistanee since it is equal to 1/F. The value of W can be
calculated for tubes 01' duct;; of most cross sections. For a long tube of
length I, of varying cross section A, and perimeter H, Knudsen l l deduced
the fundamental relation:
83 Knudsen, Ann. Physik, 28, 75, 999 (1009), and subsequent papers.
8, von SmoludlOwski, Bull. Acad, Krakau, 143 (1903). Hagenbltch, POrJ(jend01J',~
Ann., 109, 835 (18(iO). Brillouin, Ler;on.~ sur II! Viscosile, Paris, 1(l07.
8, Gaede, Ann. Ph,lls£k, 41, 280,337 (11J13).
so Clausing, Ann. Phy,~i1c, 12, 9fH (1032).
87 Maxwell, Phil. ll1ag., 19, 22 (1800). BoltzmllllJ1. SUziJCI'. A/mrl. lV-i.~s. W£en., 74,
503 (1870). "
552 J. C. HECKER
3 V_ 27r
TV = --
1()
F--l [)
l
II
---- dl
A2
(31)
For it cluet of uniform cross section, that i!>, one in which II and A are not
functions of the length, the value of W becomes:
W= 3~?i Hl (32)
w A2
16 A A If) A ,
1'herefore: jI'-= - --- - - - - - ----- F (33)
3 Hl Y27rp~ - 3 Hl 0
The term A/y2;-~ has been ffwtorized separately since it will be found
convenient to designate it as Fo.
Il. APERTURE
See Figure 32A for the special case of l = 0, th~tt is, an infinitely thin disc
with an aperture of any cross section, A, which is much smaller than the
area of approach:
TV = V2;/A
and: Fo = A/y27rPl = 11.7A 9.17d 2 (34)
A'
----[----- --1--
L > 50d ~ < J_ < 50

l 4 d
F = 12.2 Td F. 11.7 A
A' » A F =J.!:.lA__
1-1- 1...l.. 1+1.
Fa' 11.7 Ii 4 d d
(A) (8) (C)
'I-_~
t2
(Ol--------{:)
- - - _. __ ... _..J..""I
a
a » b a » 1
F = 11.7 Kab . - ( 1 -'I- ) Fo
F =8rz
31 '2
(0) (E)
Jiig; ;,2. Mo)o()u)ar now: (A) !LIlcrtUI'C; (n) long tubes; (C) short tubes; CD) reo-
tungu]u,l' slits; (B) long concentric tubes.
VI. DIS'rILLA'rION UNDER HIGH VACUUM 553
If the aperture area, A, is not I:lmaU relative to the cross-sectional area,

A', of the space above itJ mass motion will result and the conductance will
be increased by the large aperture correction factor:
---
A' (35)
A' - A
which becomes unity when A I iH much gren,ter than A and approaches in-
finity as A' approaehe::l A. The same consideration applies when an en-
trance resistance is to be added to a short tube length.
C. LONG CYLINDlUCAL 'rUBES
t:iee Figure 3213. DHing equation (33) and SUbstituting the values of A =
1Td 2/4, Ii = 1Td, and Fo = 9.17d 2 :
16 A d3
F=--F
3 In 0
= 12.2 T liters per second (36)
The formulas apply only if the pipe is long. The rate at \Vh1Ch gas can be
removed from the pipe is given and no account is taken of the rate of ef-
fusion of gas into the higher-pressure end. Any difficulty experienced by
the molecules in finding the entrance to the pipe must be negligible com-
pared with the difficulty of traversing the length. This will be true when
l > 100 r > 50ri.
D. SHORT TUBES Ol!' UNIFORM CROSS SEC'I'ION
Sec Figure 32C. For tubes in which l < 50d, end effects are no longer
negligible in t:omptwison with the resistnnt:e of the tube walls. One can
consider the resistance as the sum of: (1) the resistance of the pipe inlet
considered as an aperture, and (2) that of the pipe length itself. If Wo is the
resistance of the opening and WI that of the tube wall:
TV = Wo + Wt
1 1 1
--F =-+-
Fa Ft
Substituting the value of Ft from equation (35) and F0 from equation (34):
1 1 1
F= 9.17d 2 + 12.2(d /l) 3
9.17d 2 l1.7A
F= = (37)
3 l 3l
1 + -- 1 + 4d
4d
554 J. C. HECKER
Since land d appeal' in the denominator as a ratio, any consistent units can
bo used. For example, both can be in centimeters, inches, feet, etc., as
long us they [u'e the sume for anyone ratio. However, for the numerical
constants given, d in the numerator must be in centimeters and A in square
centimeters. Equation (37) is normally sufficiently accurate for vacuum-
distillation calculations. For values of ljd > 3/4, the maximum errol' is
about 12%. It can be readily seen that when l = 0, the expression becomes
the aperture formula, and when l » el, it becomes the long-tube formula.
For shorter tubes, and agreeing with the more complicated expression
as derived by Clausing,88 Kennard 8U has shown thlLt, for very short tubes
the 3/4(ljrl) term should be (lid), leading to the expression:
n.7A 9.17d 2
• i
]I' = ,--~--- --------.-- 11te1'13 per seCOll( (38)
1 + (lid) 1 + (lid)
when l/d :::;_ 3/1

If the eonductance of it short noncylindrieal dud, of uniform cross sec-
tion is desired, the same approach as used above ean be made. If we sub-
stit.ute from equation (33):
in the equation:
1 1
-=--+ . 1~
F Fo Ji'd
we obtain:
p p. (39)
I = 1 + 3/16(Hl/A)
and for air Ht 2SoC.:
11. 7 A . d
F = -.~ - -- - -- = hters per secon (40)
1. + 3/ w(IIl/A)
where A is m'en in squH,l'e centimeters, and II and l are perimeter and length
of the duct, respectively, expressed in centimeters. Here again the de-
nominator represents the effect of the channel walls and· the numerator
that of the entrance.
SS Clausing, Ann. PhysClc, 12, HOl (1\)32).

80 KenIl!1l'd, KineUc Theory of Gases. McGraw-Hill, Now York, 1\)38, pp. 306-308.
VI. DIS'rILLA'rION UNDJDR HIGH VACUUM 555
E. DUCTS OF NONUNIFORM OR IRREGULAR CROSS SEC'rrON
If the vn,lue of W cannot be obtained from the integral, ,f,,t (H / A 2) dl,

of equation (31), an estimated value of the conductl1nce can be determined
by dividing the channel into sections which will permit an approximation of
the integrated value. The resistance of the sections are then added to get
the total effect. An abrupt decrease in area from one section to another
will introduce an entrance resistance, which, corrected by the large aperture
factor, becomes:
(41)
Fa Ao - A
However, the transition from one exit section to an entrance section of the
same shape and an equal (or larger) area will not require an additional en-
trance resistance.
F. BAFFLES, VAI.VES, ETC.
Only in the simplest cases is it possible to calculate with [my degree of

certainty the conductance of such irregular-shaped channels. At best,
one can calculate upper and lower limits by visualizing the channel as a
duct or tube and arbitrarily selecting a reasonable vn,lI1l'l. If the conduct-
ance is critical, an experimental determination of the value is necessary.
G. COLD TRAPS
See Figure 33A and B. For complicated designR, the same general re-
marks which pertain to baffles, valves, etc. apply. Hmvever, some of the
simpler designs are amenable to mathematical analysis. As an example,
an approximate calculation of the conductance of a trap design shown in
Figure 33A will be made. It is assumed that the gas travcls equally by
two parallel paths around each side of the cold thimble. For an irregularly
shaped channel, from equation (32) :
W = 3-{?~ !!!: (32)

16 A2
The axial length around each side is calcul::Lted by using the average radius:
1'a
1'2
= ---
+ 1'1
2
giving l = 1I1'a' The perimeter H can be approximated by assuming the
channel to be an inverted trough of sides h high and base (1'2 - 7'1) = 6..1'.
55() J. C. HI~CICFJR
Or J1 = 311. + At' amI the area can be assumed to be the arOfL of section
h(/'2 - 1't) (It. A = 11 D.1'. From lDqnationH (aD) and (aa):
1 16 A.2
F= W v'~ = 3v'2;~ lIz
Evaluating the factor:
Az h2 ,1.1'2
III (2h + M) 711'"

We obtain for air at 25° c.;
VACUUM REFRIGERANT
CHAMBER
! VACUUM
j I
CHAMBER
_,..VACUUM VACUUM
PUMP PUMP
SILVERED
VACUUM
JACKET
(Al (B) (Cl

I<'ig. 33. Cold tmpA.
in cubic centimers pel' second. Since this is the flow around one side only,
the total flow by the two parallel paths will be twice this amount. Sub-
stituting the numerical values indicated and converting to liters per second,
we obtain:
in liters per second. If h > 5,1.1', an errol' of less than 10% results by neglect-
ing 1::.r in the (2h+ /lr) term. In such cases an approximate value of:
VI. DIS'.rILLATION UNDER HIGH VACUUM 557
h A1'2] (42)
Ii' = 40 [ - -
21'"
is () btained.
Consider it trap, a::; shown in Figure 33A, having the dimensions h = 8
em.,1'2 = 3.5 cm., and}'! = 2 em. Then A.1' = 1.5 cm., andTa = 2.75 cm.
Using the simplified formula:
0' Ar2 = 8 X [52 = 3.27
21'a 2 X 2.75
the trap itself has a conductance of 130 liters per second.
A gas molecule entering the trap from the upstream side arm of radius
1'0, will "see" an opening of area approximately A = 21'0 t.r. To determine
whether an entrance resistance, Fo, need be added to F!, the value of A
shollid he compared with the upstream area, A' = 7rJ'02. As long [LS A ;;:::
A " no allowance for Fo need be made, or, in this particular case, as long as
211'1'0 A.T ;;::: 11'1'0 2
01':
2Ar ;;::: 1'0
l,'or the example being considered, 2 AI' = 3, whieh is greater than 1'0 = 2.5;
therefore no entrance resistance, 1/F OJ need be added. In those caseG
in 'which there is a reduction in area on leaving the downstream side of the
trap an entranee resistance should be added in computing the resistance
from thc trap shell to the next point downstream.
The conductance of a trap of the general design shown in Figure 33B hae
been calculated by Dushman. 90 It should be observed that the conduct-
, ance for all eold traps will be lower at a lower temperature, T, by the factor
V'[,/273. The average temperature of the gas in passing through the trap
can be approximated from the temperature of the refrigerant being used.
H. ELBOW::; AND BENDS
KloseUl determined rates of flow of gases at low pressures and concluded

that F depends only on the total length of path and on the values of l' for
different seetions. The introduction of bends 01' elbows has no appreciablE'
effect. Therefore, in the high-vacuum flow region it is usually sufficient
to eonsider the effective length of an elbow or a bend as its axial length.
However, Sheriff92 indicates that each bend requires the addition of one
tube diameter to thc lcngth, l, in conductance formulas.
10 Dusbman, S(;-tentijic FoundaUO'IL8 of racnnm Technique. 'Wiley, New York, 194\J
pp. 105-110.
S\ Klose, Physik. Z., 31, 50::! (1930); Ann. Physik, 11, 73 (1931).
D2 Sheriff, J. Sci. Instruments, 26, 42 (1949).
558 J. C. HECKEa
I. SLITS OF mCC'I'ANGUl,AR SECTION
See Figure 32D. For tubes with rectangular section a » b and a :::.> l,
Clausing9•1 hus arrived nt the relation:
F = a.638J{etbVl'j]{ liters per Rccond
D
and for air at 25 C.:
F = 11.7Kab liters per second (44)
Values for J(_ as a function of lib ltppear in Table XV.
lib K 1/11 K
0.......... 1.0 1.6 ........ 0.B024

0.1. ....... 0.9525 2.0..... ... 0.5417
0.2........ 0.9090 3.0..... ... 0.4570
0.4 ........ 0.8302 4.0..... ... 0.3990
o .8. . . . . . .. 0 . 72mj 5.0........ 0.3582
1.0. . ... ... 0.68,18 10.0 ........ 0.2457
.T. CONCEN'l'RIC TUBES
See Figure 32E. For n long annular space we can again apply the gen-
eral equation (33) for conductance:
]I' = ~~~F
3 HZ 0
The area A is the open space between the two cylinders, or 71"(1'22 - )'1 2).
The perimeter}] is the combined circumferences of the inside diameter of
the outside shell and the outside diameter of the inside core, or 2'11"(1'2 r1)' +
Therefore:
(45)
1'2
= x - Po (46)
l
93 Clausing, Ann. Physik, 12, 961 (1932).
VI. DISTILLATION UNDEH HIGH VACUUM 559
For air at 25°C.:

Fo = 1l.7A (47)
Values of:
. rl
as a f unctIOn of - (48)
r2
are given in Table XVI.
TABLE XVI
CONCENTRIC TUBER - VALUES OF X .~s A FUNCTION OF r, IT.
---------
TIll''.!. X n/J''l
----------
------------
0 .... ,., ,.", 2,67 0,6.......... 1.07
0.1 .. , ..... ,. 2,40 0,7.......... 0,80
0,2.". 2.13 O,R., .. , ..... 0.53
0.3.......... 1.86 0,9.......... 0.266
0.4,.",.,.,. 1.60 1.0 .......... 0
0.5 .. , 1. 33
If the concentric tubes are short, l < 50Ar, the value of A/HZ as derived
above can be substituted in the short-tube formula equation (40). It is of
interest to note that, if the concentric tube is a falling-film still or similar
device in which gas is formed or liberated between the surfaces, no end
correction is necessary. Also, if the still is being pumped equally at both
ends, and the gas liberation is uniform along the length, it can be assumed
that half the length is being evacwLted equally in both directions.
It is to be remembered that all formulas given in this section apply only
to free molecular flow, i.e., conditions in whic:h the melLn free path is large
compared to the dimensions of the opening. As stated previously, for
cylindrical tubes as long as:
Pd ~ 2 Pr ::; 1. or i\ 2: 2, 5d
molecular formulas are aecurate to within 5% for air at 25°C. Also, when
numerical values are given they apply only to air at 25°C. unless otherwise
stated. Since the density is proportional to the molecular weight, the
value of the conductance for other gases at other temperatures can be
found by multiplying F or Po by:
v'Z9/M VT/273 (49)
where T is in degrees Kelvin. The temperature eorrection is to be con-
sidered in calculating the conductance of cold traps.
560 J. C. llECICER
4. Viscous Flow
A. LAMINAR RInGION
The flow of gas in the viscous laminar-flow (nonturbulent) region is

governed by Poiseuille's law. 04 Referring again to equation (20) it can
be seen that the second term becomes negligible at higher pressures.
When Pd is gret~ter than 1000, the flow is over 95% viscous. The formula
reduces to:
4
F = 'lfr4 P = 'lfd P (50)
8~l .a 128~l a
which is one way of expressing Poiseuille's law, where P a is the average

pressuI'e along the tube.
The more familial' c}"-pression for ViRCOllS laminar-flow is:
(51)
in which 111 is moles per second; 'Y}, coelfici8nt of ViilClosit.y in poises at tem-
perature T; r, radius of the tube in centimet.ers; l, lengt.h of the tube in
cent.imeters; P2, pressure at inlet in microbars; and P l , the pressure at
outlet in micl'obars. If Y is the rat.e of flow in pubic centimeters per second
at pressure P:
PV = mRT
and substitut.ing in equation (51):
'If ,.4
PT' = - -:-' (P 2 2 - P 12) (52)
16~ l
Since P2 2 - P1 2 = (P2 +
Pl)(P2 - P1), and alAo (P2 + [>1)/2 equals t.he
average pressure Pa, equation (52) becomes:
'If 1'4
PV = 8'Y} 1 P a (P 2 - P 1) = Q (.53)
Q ?f1"4
or: F=---=-p (54)
. P2 - P1 8l]l a
which is identical with the expression (equation 50) deduced from the
general formula of Knudsen.
As shown previously, when Pd exceeds 1000, equation (28) becomesF =
94 Poiseuille, Soci~te Philomathique de Paris Bulletin, 28, 77 (1838); Compt. rend., 11,
961, 1041 (1840); 12, 112 (1841); 15, 1161 (1892). Locb, Kinetic Theory of Gase8,
2nd eq., McGraw-Hill, New York, 1934, p. 230.
VI. DISTU.. LATION UNDER HIGH VACUUM 501
0.177 (d 4/1)P, in which F' is in liters per second per unit pressure drop. In
dealing with fore-pressure liJles, one mJl often tolerate pressure drops of 10
to 20%. Since F' = Q/(P 2 - PI), Q, the quantity flowing at a pressure
difference Pz - PI microns, becomes:
Q = 0.177(cl'1/I)P(Pz - PI) micron liters pel' second (55)
where cl is diameter in centimeters; I, length in centimeters; P, average

pressure in mierons, (P 2 + P 1 )/2; P 2, inlet (higher) pressure in microns;
and PI, outlet (lower) pressure in microns. It is obvious from the factor
d4(l that a much greater effect is obtuined from a change in diameter than
hy altering the length. Doubling the diameter is equivalent to shortening
the length 16 times.
As an illustration of the use of equation (55), let us consider a tube 1 em.
in diameter and 10 em. long which is to conduct air at 25°C. at an average
pressure of 100 f.L under a pressure drop of 10 jJ., or P z = 105 J.L, and PI =
95 f.L. The vallie of d4/l becomes /4/10 = OJ, and P(P2 - PI) = 100 X 10
= 1000. This leads to a value of 17.7 micron liters per second,
As pointed out previously, it is sometimes dcsirable to express flow rates
in terms of t.he volume of gas flowing per second at the prevailing average
pressure P. Equation (55) becomes:
,'{ = Q/P = 0.177 (rl'l/l)(Pz - PI) liters per second (50)
where P 2 and PI are in microns and d a.nd l in centimeters.
In the example given above t.he 17.7 micron liters pel' second is equiva.-
lent t.o 0.177 liter per second at 100 J.L. The two expressions micron liters
per second and liters per second should offer no difficulties if the conversion
from one to the other is considered as 11 gas compression or expansion using
1 micron as the standard base pressure. We have seleeted the micron as
our unit of pressure and the micron liter follows naturally.
Equation (56) can be rearranged t.o a eonvcnient expression for estiimat-
ing the size of a fore vacuum line whon meclmniclLl pump speed and length
of connecting tubing are known. Assnme a vessel is to he conneetcd by a
length of tubing 100 cm. long to 11 mcehanical pump with a speed, Sp, of 2
l./see. (a Cenco Hypervae-20 has a rated capacity of 1.9 1./sec. at 100
jJ.). The speed of 2 liters pel' seeond is at the pump inlet and with a finite
pressure drop does not equalS in equation (56). It can be shown that with
pressure drop~ of less than 20% between the mechanical pump and the
vessel, an error of less than 10% results if the pump speed, Bp , is substituted
for S. The maximum error of 10%, with the top arbitrary limit of a 20%
reduction, is usually well within the limits of other errors in estimating
pipe sizes.
562 J. C. HECKER
Rearranging equation (5G), and for the speeial case 'when b.P = P2 -
::; 0.21\:
b..P = 5.04 (lid) Sl' ::; 0.2 P l
or:
glvmg a general expression fo1' determining diameters of fore-presi

lines, of length l em., conducting ail' at 25°C. at tlw fore-pump intake I=
sure P l microns, into a fore-pump of capacity 8]' liters pCI' second, and
pressure drops of less than 20%. Rubstiliuting the selected values of
100 cm., Sp = 2 liters per second, and P = 100 /J., the diameter shoul<
no smaller than 2.74 cm. to keep the pressurc drop or reduction in 81
below 20%.
B. TURBULEN'l' REGION
All the equ!1t.ions derived so far from Poiseuille's law are applic
only for viscous laminu,7' flow. However, in normal V1:LCUUm practice
for !1 properly designed distilbtion vacunm system, turbulent flow raj
if ever, occurs. This can be shown by ~Ln examination of the condit
which lead to turbulence. Ii; was first demonstrated by ReynoldsOn
if a certain critical veloeiiiy, /J." is exceeded, the pressure drop as cal cuI:
by the Poiscuille equation is too smnll. A dimensionless number, Il, cr
the "Heynolds number," wat! defined whieh is used to determine the tn
tion value of htminar to eritiCltl velueity;
Me = RJ.l.1 pl' = 2Bf//pd
For glass, R hltS a value of 1000 and for met111 hetween400 and 500.°6
Using the value WOO and substituting for air at 25°(;. a value of .
1.845 X 10- 4 , and a value of p = 1.5G X lO-Q g. pel' Dubie eentimeter,
micron pressure, one obtains:
/J.e = ----a:---
2.3M X 1()6
em. pel' seoond
The velocity is also equal to the volume of gas flowing per second div
by the area, or:
95 Taylor, A Treatise on Physical Chemistry. 3rd ocl., Van Nostmnd, New York,
p.177.
96 Ruckes, Ann. Physik, 25,"'983!(1908). McAdams, lIellt Transmission, 2nd
McGraw-Hill, New York, 1942,_Chap. 5.
Combining equations (60) and (Gl):

Me = 4Q/rrc[2 = 2.365 X l05/d (52)
Therefore, at the critical velocity, Q = 185,OOOd micron liters per second,
where rZ is the diameter in centimeters.
Perhaps a better understanding of this value can be obtained by referring
to Table XVII, in which Q for a tube 1 cm. in diameter is converted to
liters per second, F. Comparing the speeds of some of the more common
laboratory and industrial mechanical pumps (Table XVIII, p. 564), it can
readily be seen that turbulent flow may occur at the initial stage of exhaust
for small-diameter tubes, but becomes less likely as the pressure is lowered.
TABLE XVII
CRITICAL FLOW RA'l'E (LAMINAR TO TURBUI,ENT FLOW) AS A FUNCTION OF P
FOR TUBES 1 CM. IN DIAMET~JR
PI' P lllm . Ii'
1 0.001 185,000
10 0.01 18,500
100 0.1 1,850
1000 1.0 185
10,000 10. 18.5
IV. VACUUM PUMPS

1. Backing Pumps or Fore-Pumps
Not many years ago, a vacuum pump t.o the organic chemist brought to
mind a water aspirator. The laboratory-size 'Watel' aspirator serves its useful
purpose for Buchner-funnel filtrations or Claisen-flask distillations but
generally is not suitable for high-vacuum dist.illations. The laboratory
water aspirator has a pumping capacit.y of the order of 20 cc. per second
producing an ultimate vacuum of from 8 to 12 mm. mercury, depending on
the water temperature and pressure. Compared to this is the smallest
Cenco mechanical pump, the Hyvac, which has a rated speed of about 100
cc. per second, and the Welch Duo-Seal 1395 mechanical pump with a speed
of about 1.50 cc. per second at 100 fl.. A list of the more common mechani-
cal pumps and of their rated performances are given in Table XVIII. u7
A. MECHANICAL ROTARY PUMPS
The rotary oil-sealed mechanical pumps are generally used for laboratory
and pilot-plant operations. These pumps and others which exhaust
97 Sullivan, Rell. Sci. Instruments, 19, 1 (1948).
564 J. C. UECKER
TABLB XVIII
Hm'AHY OrrrSEJAI.ED MECHANICAl. PUMI'R!I'I
Pumping
Free air ~)HH,;d at
No. disfllace .. Ill-I 10-2 10-' Limiting
Pump
of
atftgc3 Rp.m,
m.cnt,
I'/sec.
Bun.,
I./s"".
nun ..
1./""0. '_. __
unn.,
l./.e!l.
.-
preS:3ure,
nun.
C'erwo
1'l'OSSOVIl,C 1 000 0,58 O,GO O,g,1 g X 1O-~
HYVlLC 2 350 0,18 0,10 O,Ol) 0.08 2 X 10-'

Megllvlw 2 32fi 0,52 0.30 0.28 0.20 5 X 10- 6
Megavac 2 (\00 O,U5 0,78 0,52 0,37 7 X 10- 6
HYPol'vae-20 2 42fi 3,30 1,00 l.80 1. 50 7 X 10- 6
HypervtLe- 25 2 570 4.40 2.00 2,50 2.10 5 X 10- 6
HY[Jel'vllC--lOO 2 '150 Hi. 00 12.00 11.00 11.00 2 X 10- 6
Welch
Duo-Seal 1·\00 2 450 0.35 0.20 O,Hi 0,12 5 X 10- 5
Duo-Seal 1-105 2 300 0.56 0.37 0.31 0.24 2 X 10- 6
Duo-Seal H05 2 525 0,07 O.GO O,GO 0.50 4 X 10- 6
Duo-Setll 1403 1 375 1.70 1. 30 0,58 4 X 10-1
Duo-Sonl H06 1 300 0,56 0.38 0,20 5 X 10-3
Duo-Soni 1404 1 300 0.5(i 20 X 1O-1a
DUO-SOILI 1410 1 450 o 35 20 X lO-l.
Duo-Selll 13H5 1 1725 0,20 25 X 10- 3•
Kinnell
CVD 55() 2 525 7.2 iUi 4,7 3,8 1 X lO-'
CVD 8fllO 2 550 21.7 lfl.3 13,\) 11.1 1 X 10-'
VSD 55() 1 450 (U 3.\! 1.5 5 X 10-1
YSD 778 1 3GO 12.7 7.!l 3.3 5 X 10-1
VBD 8811 1 3GO 21.0 13.7 5.7 Ii X 10-1
DVU 8810 1 450 52.0 33.0 14.1 5 X 10-1
DVD 12814 1 415 103.0 fH.O 28.2 15 X 10-1
DVD 14018 1 3GO 14'i.O \)2.0 37,8 5 X 10-·
DVD 141418 1 3GO 22(l.O ]44.0 5H.O 5 X 10- 3
DVD 181420 1 3GO 320.0 205.0 82,5 5 X 10-1
Stokes
Microvac 14t1-E 1 350 '1.7 3.2 2,1 2 X 10- 3
Mir.rovac 140-1~ 1 500 7,1 4.8 3.2 2 X 10-3
Miel'ovltc 148~~-; 1 370 15,0 10.3 6,8 2 X 10- 3
Mitll"ovnc 140-11; 1 385 28.3 HI. 2 12.7 2 X 10- 3
Miel"()vac 212-E 1 3811 .5,*,0 30,8 24,3 2 X 10- 1
Microv!tC 412-11; 1 4.00 111,0 75,5 50,0 2 X 10- 3
Microvac (lI2-I<; 1 ,125 238.0 lfi2,0 107.0 2 X 10- 3
Beach-Russ
Class D RP15 1 200 7.'1 5.5 :3.3 2 X 10-1
RP30 1 200 16,0 11,0 (l.G 2 X 10- 3
UPSO 1 200 25,0 18.5 11,0 2 X 10- 3
!tPIOO 1 200 50.0 38.0 22,0 2 X 10-'
HP250 1 200 130,0 08.0 58,[) 2 X 10-3
IU)375 1 200 185.0 135,0 82,0 2 X 10- 1
RP750 1 200 370,0 270.0 WO,O 2 X 10-'
a Mnllufacturer's gU!1l'11ntee. Limiting pressure is lower !;lum this value.
against atIIlot-)pheri(~ pressure lUll ui::ieu 11i::i the primtll'Y vacuum pump and
111'ecommonly rei'el'l'eu to ltJ:> "backing pumps" or "fore-pumps" . The
basic features and operating principles of present-day rotary pumps, as
given by Sullivan,97 follow. Certain features are common to all these
pumps.
VI. DIS'l'ILLATION UNDER HIGH VACUUM: 565
"A rotor, H, rotates illi:iide a Htatof, X. Lille or are!t of cuntact G, between rotor
and stator, and contact K, hetween vane V and stator divide the volume between
rotor and stator into two chembers (in pumps with two vatlc~ instead of one thi:;
volume is divided into three chambers). Through the motion of the rotor one
chamber is expanding and the other contracting. The expanding chamber is con-
nected to the intake, I, of the pump and the contracting chamber to the exhaust, E.
Oil films at G and K seal the pump. A valve, D, usually, but not in all pumps,
spring loaded, seals the contracting chamber against a flow of gas inward. This
valve is oil sealed. The free air displacement of a rotary oil-seal PUlll p is a fUIlction
of the volume between the rotor and the stator and the number of reYolutioIls the
rotor makes in unit time. In single-vane pumps the free air displacement, expres-
sed in liters per second, liters per minute, or cubic
feet per minute, is the product of the number of
revolutions of the rotor per time uuit and the
volume between the rotor and stator (Fig. 34).
A sliding, oscillating seal is provided for the v~ne
in the top ~ection of the Rtator."
!-._--!--.....)«
Even though mechanical pumps with f1'e,,11
uncontaminated oil can produce vacuums ill D
the low micron range, as a practical mattel·
single-stage rotary pumps find their most ex- G
tensive use in distillation as hacking pumps
above the 100 iJ. region. In general, their
pumping efficiency falls off appreciably with
decreasing pressure, particularly below 100 p.
:Multistage mechanical pumps have been Fig. :J4. Cellco Hyvllc pump.9'1
designed to lower the ultimate vacuum and
at. the same time increase the volumetl'ie (.~apacity at lower pretlsures.
However, with the improved designs of vapor pump!' now lW!1iluble, it is
eOllsidered good distillatioll praetiep to rely on vapor-booster or vapor-
coudenRntion pumps to maintain vaCUUDUi below 100 f.!, and Oll mechanical
pumps or ejectors to eompress from there to atmospheric pl'emlUre. This is
even more apparent when one considers that it is not unc:ommon for a
small laboratory still to require a pump with a speed of 50 to 100 liters per
second in the ~ to 10 p. range. As can be seen from Table XVIII, if a me-
chanical pump alone were used to maintain the vaCUUlll, a In.rge commercial
unit would be attached to a relatively small piece of laboratory equipment.
All mechanical rotary vacuum pumps arc sealed with a low vapor pres·
sme lubrieating oil. 1n new pumptl a sealing oil with a viseosity in the
range of SAE 10 to 20 is cOllllllonly used. As the pump wears alld the
cleanmees gmdm,lly inerease, lL hetwiol' oil in the range of SAE 20 to 30
can be substituted. Hydrocarbon oils are produced and sold under various
trade names for mechanical pump oil use, usually with the descriptive
566 J. C, HlGeKER
name mechanical vacuum pump oil. For special applications, sealing oils
other than hydrocarbons have been used.
In addition to the dccrease in volumet,ric efficiency with lower pressure:,;,
which is an inherent characteristic of mechanical pumps, resulting from
clearanr:e limitations, the vapors which are HbeI'ated either before or during
the distillation contaminate the sealing oil and impair the pump's opel'a~
tion. Any attempt to maintain the rat.ed volumetric capacity of a me-
chanical pump at low pressures in connect.ion with organic distillations is
futile.
There are several ways to cope with this problem. One simple method is
to change pump oil as required. A vacuum gage between the mechanical·
pump and the oil-vapor pump will indicate when a change is needed. On
large installations, a continuous system of renewing the sealing oil and re-
claiming the contaminated oil by suitable means has been successfully used.
Cold traps, which will be discussed later, assist materially in condensing
volatile contaminants which would otherwise reach the fore-pump. An-
other scheme is to intl'ociuee in the VlLcuum line a hot :hone whose tempera-
ture is sufficiently high to decompm;c or "crack" scmicondellsible vapors
TABLE XIX
OPERA!l'ING RANGES OF VACUUM PUMPS
I
Legend
1
*-"-_ Optimum Range
ie-----"'" Extended Range
Possible by
Special Design
Oil-Condensation Pumps
I ----I
Oil Vapor Booster Pumps
1--- --~----I--l
Oil Vapor Ejector BOoster Pumps
1--- I.' --r-----I
Roto ry Mechanical Vacuum Pumps
1--- ------ ----- I I
Ejectors - Steam and Hydro -steam
1--- ------ ----- --
I .
I-f,er ASPlr~~o~ __ •
,. 0.1 100 1000

mm. 10-4 I. 10 100 760
to carbonaceous residues 01' to none on den sible gases. A helix of resistance

wire suspended in the vacuum line and operated at a voltage which main-
tains a dull red heat has been successfully used. o8 On largel' installations,
multistage steam ejectors or hydrosteam ejectors are used as backing
pumps. Vapors as well as "permanent gases" are readily handled by ejec-
tors, thereby eliminating the contamination problem.
A diagrammatic representation of the ranges covered by present-day
pumps is shown in Table XIX.
B. STEAM AND WATER EJECTORS
Vapor-ejector fore-pumps are usually operated with high-pressure steam

(60 to 250 p.s.i.g.), although other actuating vapors have been tried. 99
A typical single-stage ejector is shown in Figure 35. 100 Such units are
10
8 4 5 6 7
Fig. :35. Basic one-stage steam-ejector assembly; (1) diffuser; (2) air chamber;
(3) steam nozzle; (4) steam chest; (5) nozzle plate; (6) sucHotl; (7) discharge; (8)
stel1m inlet; (9) nozzle throat; (10) diffuser throl\t~!Oo
joined in series to produce almost any degree of vacuum from atmospheric

to pressures in the micron region. A five-stage ejector can be made to
produce a "shut-offl1 vacuum of between 25 and 50 IJ.. As a matter of
steam conservation, intercondensers arc interspersed between some of the
stages. In such cases, to remove the condensed steam to the atmosphere,
the ejectors and condenser are generally placed about 34 ft. above a water
basin, sometime called a "hotwell," into which the end of a barometric leg
is submerged. When such height is not available, the water can be re-
moved by a discharge pump which requires about an 8- to 10-ft. head for
its operation. By and large, the barometric leg is the more common choice.
98 Morse, Rev. Sci. Instruments, 11,277 (1940).
Q9 Work and Haedrich, Ind. Eng. Chern., 31,464 (1939).
100 Standards of Heat Exchange Institute. Steam Jet Ejector and Vacuum Cooling
Section, Part 1, Stet\m Jet Ejectors, Heat Exchange Institute, New York, 1938, p. 4.
568 J. C. HECKER
Barometric condensers are generally used on folU'- and five-stage ejec-

tors, sometimes on three stages, and mrelyon Rmall two-stage assemblies.
Commercial-size steam ejectors are ordinarily too large for laboratory ap-
plication. However, small units CU,Il be made which find many applica-
tions in vacuum distillation. 10 I
Steam jet ejectors operate on a mass-velocity-principle. The propelling
steam expands adiabatically through a divergent nozzle, converting its
pressure energy to kinetin energy. The mass of high-velocity steam is di-
rected through a mixing chamber and into a convergent-divergent diffuser.
In passing through the mixing chamber the steam entrains a definite mass
of gas or vapor being evacuated. In imparting velocity to the gas, the
steam is slowed clown, and the resultant mass enters the diffuser where the
kinetic energy is converted to pressure which is considerably higher than
that in the vacuum ehamber. ']'herc is no condensation of actuating vapors
dming the compression eyde. The high expallRion ratio of the steam in a
four- 01' five-stage ejoetOl' may accelerate the steam to Mitch numbers of 9
to 11. The eooling effcet from this cxpansion in numy cases produces ice
on the nozzle month as well as in the (liffusel' throat. The ice formation
alters eritical dimensions and interferes with the ejector's operation. The
condition is alleviated by adding steam jaekets to the nozzles and diffusers.
Owing to the high velocity there is praetien.lIy no lateral eseape of steam or
entrained gas or back diffusion from the clischarge opening. POI' this
reason, pressures thousands of times lower thtw the vnpor pressure of water
B.re produced with steam clisefuLrging into ltml throngh the entraining vae-
uum chamber.
A hydrosteam ejeetor consists of a water ttHpimtor di:;wharging into at-
mospheric pressure with one or more Rteam ejeetors joined in sories to pro-
duee successively lower preHsllres. A motor-driven eentrifugaI pump cir-
culates water through the Hspil'ator, suffieiently llew "makeup" being added
to condense the steam from the ejector stages. Such units are compaet,
relatively free from trouble, and are useful for degassing purposes as well aH
backing 1mits fot' oil-booster pumps.
2. Oil-Vapor Pumps
A. GENERAL
Mercury-vapor pumps are not as common in organic high

lation as oil-vapor pumps and therefore consideration will be to
pumps using organic compounds as boiler fluids. A description Of 1l1etcul'y
"
101 J. D. Pedersen, Jeannette, Pennsylvania.

VI. DISTII"LATION UNDER HIGH VACTTUM 5G~
pumps is given by Dushman 102 and their constl'Uetion is discusRed by Barr

and Anhol'n.I 03
Rince molecular distillations are conducted at pressures in the 0.1 to
10 J.I. region, a standard condensation pump which relies principally on the
diffusion principle is in most cases not entirely applicable. The terms
"condensation pump" and "diffusion pump" are used synonymously, al-
though the present-day high-vacuum pumps are more nearly of the conden-
sation type, as designed hy Langmuir,1O·j than of t.he diffusion type as pro-
posed by Gaede. los
B. CONDENSA'rION PUMPS
In the discussion of steam-ejector pumps it was pointed out that the

actuating vapor (steam) was not condensed during the compression cycle.
As distinct from this, a condensation pump relies on the fundamental
principle that the actuating vapor, after having imparted its momentum to
the gas molecules, is rapidly condensed, the condensing zone being main-
tained sufficiently low that no appreciable re-evaporation and hence
random motion of the oil vapor takes place. Therefore, the primary dif~
ference between the operation of a condensation pump and an ejector pump
is that in a condensution pump the gas moleeulos acquire a forward momen-
tum as a result of collision with oil-vapor molecules, whereas in the ejector
pump the gas is entrained in a high-speed stream and rapidly moved along
with it into a compression zone. One could be considered pal,ticle motion
and the other mass motion.
A section of a typical oil-condensation pump is shown in Figure 36, One
or more stages operate from a common boiler, which lllay or may not have
partitioning zones depending on whether the pump is fractiona,ting or non-
fractionating. The actuating fluid is vaporized from the boiler, the vapors
conducted to vapor-directing jets by appropriate chimneys, and the vapors
projected in the [ure-pressure direc(;ioll. Molecules diffui"Iing into the stream
are propelled forward. The oil vapor condenses on the eold wall and returns
to the boiler to repeat the cycle. More efficient operat.ion is obtained from
mUltistage pumps, and, when extremely high vaeuums are needed, frac-
tioilia.ting
,;;1':
pumps are used.
'm!\TI, Scientific Fonndations of VaClwm TeChnique. Wiley, New York, 104n,

pp. 98.
103 J3arr and Anhorn, Scientific and Indllstrial Glass Blowing and Laboratory Tech-
niques. ,lhstrulTIcnts Publishing Co., Pittsburgll, Pennsylvania, 1949, pp. 161-178.

104 Lan$hmir, Gen. Elec. Rev., 19, 1060 (IOl6); .T. Franklin Insl., 182, 719 (1916).
10. Gaede, Ann. Physik, 46, 357 (1915); Z. tech. Physik, 4, 337 (1923).
570 .T. C. HECKEH
C. BOOS'rmR PUMPS
Prom the description of the ejector pump tl,lld the eondensation pump, it
would appear that j,11e1'e il:l a sharp line of demarcation between the two
me(~hunisrns. That is not exactly the ease. The type of pump which op-
erates in the intermediute region is called a booster pump because it was
first used to "hoost" the gases from the fore-pressure side of condensation
pumps to [I pressllre level eltHily reached and maintained by a reasonable
size of mcchanicaJ fore-pump.
A pump pl1iiterned after the typical umbrella-jet eondensation pump
but altered to give maximum speed in the 10-3 to 10- 4 mm. mercury range
has been designed. A 1iypica1 pump of this type ill similar to the one shown
in Figure 30. ChumeterisLic performance curves llJII are given in Figures
rllGH 'VACUUM CONNECTION GAS MOLECULES ':::,
I OIL'JAPOR~
lOW· PRESSURE REGION
r",°l. -' ,,
--/ ·'1 ,
I
I .)--Ist STAGE JET
I I
I ~_!..-OIL VAPOR CONDENS·_S
I : ON COOLED WALL
I
I
I ,
I
I I
I
I
~_- FORE'PUMP
CONNECTION
Fig. 36. Schematic

sect.ion, oil conden-
sation pump.
100 Glass-Metal Diffusion Pump, Type GM-220, marketed by Distillation Produc~s
Industries, Rochester, New York.
VI. DISTILI,ATION UNDER HIGH VACUUM 571
27 5 111
2.75 amp.
2.75 amp.
I
250 - --- - -- .. -- -
225
0 ~
-:_-~~~~L-~\ ---
3.25 amp.
,
,
200
2.50amp. ~
1/[/
f.! /1/
"
V-- 1\\ ,\( .
"
,,
1/ 3 .00~
I.) ~ 111/1/ \ l\ ~ '.
\
,
~
In
.....
175
2.250m p'-;r; (/ 11
amp.
f; :31.25 amp. 1\ 1\ 1\ ~ " , ,
,
~
~
ci'
150
1/11 fit !i ill
1---
" \ 1\ I\~!\
\
,, ,
,, ,
\
, ,
! IIIJ 1/1-- al~p.1 ~ 1\ 1\1\1\
125
IJJ
IJJ 3.50 , ,,
5i ,
It (I/;
100
,,
k.'o'o amp.~ ~ 1\ 1\
~ '
75
/1/ (II III ~

'"
l\
50 1-
~ ~~
FORE-PUMP; - 0.5 Ifs f-,
-~
25
Vf 71P --- 12.0 lis
PUMP FLUID: Octoil
--.....:::: t:-=::: ~ ~~
10 5-
fill 1 I III11 I 10- 4 10-3
HIGH VACUUM, mm. Hg
Fig. ;)7. Typical high-vacuum speed curve for It vnpor-boostcr ]lump: glttss-mctal
diffusion pump, type GM-220.10G
280
I I I I ,I 1
HIGH VACUUM 1 10- 4 mm. Hg
260
V 2.50 amp. _'- .1
FORE-PUMP; 0.5 lis
PUMP FLUID; Octoil
-
2,75 amp.
240
'I 3.00 amp. I
220 3.25 amp. 1 I
\ "\ 3.50 amp.
.,u 200
....." VI 1\ \ 3.75 amp.
~ 180 -/
/' "'\ _-
~
0 160
2.~5a~p.
\ i\ 1\ \\
\ \
IJJ
W
a..
f/)
140 -1 1
2.00 amp.
, 1\
120 \ \ 1\
100 \ \, \ \ \
80
\
60
20 40 60
1
80 100 120
1\
140 160 180 200 220 240 260 280 300 320 340
1\
FORE PRESSURE, microns
Fig. 38. 'rypical fore-pressure breakdown CUl'ves for a vllpor-hooRter pump: glaRs-
metal diffusioll pump, Lype Gl\1-220.10G
572 J. C. HmJJ(BR
37 and 38. Sueh pumps nrc modified condensation pumps and as such
have an inherenL upper limit La their optimum operating range, usually a
top limit of a micron or two.
D. OIL-VAPOH B.mC'l'OR PUMI'S
Oil-vapor pumps have been designed whieh combino the principles of

the ejector and condensation pumps and are eapable of optimum perform-
ance in the ru,nge of 1 micron to several millimeters of merenry. An ex-
HIGH VACUUM
- OIL-VAPOR PASSAGE
_-INSULATION
CONDENSED-OIL
RETURN _ _ _
Fig. ;3!l. GlnsR booster Jlump (vapor ojeetol').
ample of an oil-vapor booster pump which eombines the condensation and

ejector principles is shown in Figure 80. This type of pump is generally
preferred to t.he umbrella-jet t.ype for distillation work above 1 J.!. Per-
formance etll'VeS similar to. those in Figures 37 and 38 will indicate their
optimum pumping range. Laboratory-size pumps of the type shown in
Figure 80 have speeds of 25 to 30 liters pel' second in the 1-10 j.J, range.
VI. DISTILLA'rION UNDER HIGH VACITUM 573
E. RI<JLEC'l'ION OF VAPOR PUl\{P-PERFORMANCE SPIWIFICA'rroK';
It is obvious that an oil-vapor pump should be chosen whose maximum

or, at best, optimum range covers the anticipated operating pressure of the
vacuum still. To assist in this choice we can deduce from the previous dis-
cussions that three values can be used to define the relative performance of
vacuum pumps:
1. E:vhaust PresSl(re or Fore-TJ1'essllre Breakdown-Both Dynamic and Siotic: As
indicated in Tahle XIX, there is a definite limit to which vapor pumps can compress
gas. They do not discharge to atmosphere directly but require a "fore-pump" or
"backing pump." The foro-pressure into which a vapor )Jump can discharge the
compressed gas is a characteristic. of the design and type of pump, and, in general,
the maximum fore-pressure will result under zero throughput, or itt the so-called
static fore-pressure. It is important in evaluating a )Jump to compare values on a
comparable basis-Jynamic under load, fmd static under no load. Within limits
the fore-pre~sure is related to the heut input, as shown in Figures 37 and 38.
2. Ultimate l'acltum: This is the lower liU1it of pressure which nwy be attained
in a finite time ill a closed vessel connected to thl) pump. 'With most types of
pumps the ultimat,e vacuum depends on the fOTe-pressure I1S well as on the vapor
pressure of the actuating oil. In actual measurement its value will also depend on
whether fL refrigcrfLte<t trap is used and if so. on its temperature.
3. Speed·-[T~!lallll Expressed in Liters per Second: One value of thc speed is usu-
nlly not sufficient to characterize pump performance. Curves sue It as shown in
Figures 37 and 38 indicate whether the pump has 11 broad or narrow pumping zone
and also the effect of fore pressure on the capacity. The speerl can be expressed
mathematically as:
- (dP lelt) = (SIV) (P - P.)
in which P is the pressure at any time, t; Po, the ultimate vacuum attainable by the
pump; V, the volume of the space being evacuated; and EI, the pump speed.
There are several ways of measuring the speeds of pumps and certain pre-
cautions must be taken to obtain reliable values. 107 Unless one is equipped
to make and to interpret the measurements it is probably just as well to
rely on the values of an accreclited manufacturer.
There are other less tangible but equally important factors in selecting
a pump for vacuum-distillation application, such as:
(1) The ability of the pump to handle semicondensiblc or other contaminants
which may reach the pump during a distillation. (2) Operating efficiency, which is
usually of secondary importance in laboratory work hut should not be overlooked
for larger installations. (3) Maintenance, which includes the durability of the pump
as well as its ease of inspection and cleaning.
101 Dayton, Ind. Eng. Chem., 40, 795 (1!)48). Dushml11l, Scientific Foundations of
Vacuum Technique, Wiley, New York, 1949, pp.l05-110.
,J. C. HIGCKfm
l~. UQUlDS Fon OlIrVAPOH PUMP BOILI,ms
The propel' choice of n,n operating fluid for vapO!' pumps depends on one
or more of the following ehanwteristics:
(1) Stubilitu of the fluid on prolonger\ heating at a hoiler temperature (and pres-
sure) necessary to produce copious vapor ebullition. (2) VU1JO], pl'e88U1'e in the
range neee~S[Lry to permit the production of the desired ultimate vacuum. (8)
Physical P1'OlJC1·ties such us being fluid at the tem perature of the condense!', wetting
ability, viscosity, etc. It is also desirnHe that the fluid have a low heat, of vaporiza-
tion, althongh this is not a critical property. (.1-) Relut-ivc nnl'cuctivity to permanent
gases (particularly oxygen), metals, water, D-nd va.pors whieh Illay be releasod during
the disti1lation and evontually roach the pump. .
TABLB XX
OIl... Vi\[·OR PUMP l!'l.UIDR (OPT[MUM Hi\NGm)
0.01 n.l 1 10 100 !flOO w.OOo I'

Material lO-' 10-< 10 -, 10- 2 10- 1 1 10 IUrn •
Organic ester8
Di-2--ethyllwxyl sebi\clltc
(Octoil-S)" .............. x
Di-2-ethylhexyl phtlmlnte
(Octoil)" ................ x
Amyl sebacate (Amoil-S)" ... x x
But.yl SebIlCl1tc ............. x x
Amyl phthalate (AIl1oil)" ... x
Butyl pht,hal::\te .. '" ....... x x
H lIdrocarllOns
Litton mOICf:lllnr luhricnnt" .. x
Alliezou" .................. x
Alliezon B' ................ x
Myvane 20" ............... x x
Myvttnc 10" ............... x x
Myvnnc 5" ................ x x x x
O'l'(/ann,qilicon cornpmuuis
DC Silieone No. 703" ....... x x
DC Siliconc No. 702'1 ....... x x
Chlorinated hyrlTOcarlJOns
Moclor 1254' ............. x x
ArDolor 1248' .............. x x x
N lll'coil-1 01 ................ x x
Miscellaneolls
Tl'i-p-crcsyl phosphn.tc ...... x
Glycerol ................... x x x
" Distillation Products Industries, Rochester, New York.
b Litton Engineering Laboratories, Redwood City, Clllifol'llil1.
< Metropolitl1n-Viclwl's Electrical Company, Ltd., Manchester, Englf~lld; marketed
by ,Tames G. Biddle Company, Philadelllhia, Pennsylvania.
d Dow Cornin~ Corpol'lLtion, Midlltn(, Michigan .
• Monsanto C emical Company, St. Louis, Missoul'i.
I NILtional Research Oorporatioll, Cambl'idge, Massllchusctts.
VI. DISTILLA'rION UNDER HIGH VACUUM 575
Even though the specifications seem rather stringent, llsually one or more
fluids can be found which are satisfactory. In Table XX are listed some
of the more common fluids with the properties relevant to making a pump-
oil choice. When mixturcs a,rc given, suc~h as hydrocarbons, the vapor
pressure is governed principally by that of the lightest constituent.
Laboratory oil-booster pumps are ordinarily not made fractionating or
self-purifying since it is a relatively Cl1SY mutter to change the pump oil as
desired. However, on large installations for continuous operation it is
advantageous to have the pumps fractionating. Arrangements should be
provided for removing heavy tars or residues which collect in the boiler, as
well as the light ends which are continuously swept towRrd the fore-pl'essme
region by the gas flow through the system.
An oil-vapor pump is relatively foolproof. However, there are certain
opcrat'inU precautions that should be taken. In general, pump oils, even
though organic liquids, can st,Llnd considerable abuse. This is no excuse,
however, for their being mistreated when 11 little reasonable care will
greatly extend the life of the oil. It is good practice to cool the pump boiler
to 50 to lOODe. below normal operating temperature before exposing the
fluid to air. And it is generally desirable to boil or distil the pump fluid at
pressures not greatly in excess of the normal operating boiler pressure. For
condensation-pump fluids this is in terms of tenths of millimeters of mer-
cury pressure and, for oil ejector booster pump oils, in terms of centi-
meters and tens of centimeters. Thermal switches or pressure-operated
switches can be incorpomted to give automatic protection to the boiler
fluid. The heat input to the boiler should be adjusted according to the
manufacturer's recommendation for optimum performance. The mere
darkening of a pump fluid is no reat-lon for a change to fresh oil. Color, of
itself, is no criterion of ability of an oil to pump wen. The need for change
should be governed principally by the performance of the pump, both
as to ultimate vacuum and to speed. A dark, messy-looking fluid may
behave even better than t,he original charge, whereas a elear, colorless
fluid contaminated with a light-boiling, difficult.ly removable contaminant
may be ready for discarding. Oecasionally during the degassing cycle or
during the removal of t.he light fractions, constituents reach the pump and
condense on the cold diffuser wall. This is particularly true when solvents
have been used for cleaning the still between runs. Also, the cooling water
should be turned off during exposure of the pump to atmospheric pressure,
since moisture from the air might otherwise condense on the cold inner walls
of the pump when the humidity is high. The fluids can sometimes be suc-
cessfully purged of low-boiling contaminants or water by boiling the fluids
for several minutes with t.he eooling w[~ter off. This opemtion must, be
watched dosely to make sure the liquid is not all vaporized into the fore-
576 J. C.l!ECKEH
pressure urm. Also, in the C~Lse of a water-cooled glass pump, the condenser
should be kept full of water so that excessive thermal shock will not occur
when eold water rURhes against the glass seals.
V. VACUUM GAGES
1. General
A chart of the various gages which eover the pressure ranges enr:ountered
in high-Vl1ellUm distillation is given in Table XXI. From the list one
TABLE XXI
OI'BH~'l'ING n.AN(tE~ or,' VACUUM G-A(JBH
~.-
1
Legend
-- -- --
1 -1
Mercury fV'onometer
Usual Ranges I
Bourdon
k- - -- - Extended Range ------, I
Possible by Oil Manometer
Special Design ~-- - I
1-- -
Alphatron
-- .... _- --_,.-
I I"
Bimetallc StriP
.
1-- --I
Thjrmistor
--- ---I
1--- -----
I
Thermocouple
I , 1
Thin Wall Deflection

- - - -I
1-- - - --- ---I
- --_ ....
Cold Cathode
I
--
McLeod
------1
Pironi
----- - ~-....,--
I'- 0.1 I 10 100 1000

'11m, 10 -- 5 10 4 - 10 - 3 -
10 2 10 I - 10
Atm,
might think that a "foolproof" gage could be selected upon which full reli-
ance could be placed I1nd that readings as indicated by a pointer on a dial
could be accepted as the "true" pressure. This is not always the case.
Most vacuum gages rely on an indirect method of determining the pressure
and do not determine the density directly nor "count" the number of
moleoules per unit volume. It is helpful to have an understanding of the
principles of opcration and precautions tobe taken in the use of the gages
which have been accepted as most pl'£wtict),l. In interpreting gage readings
one should know whether condensible gases or vapors affect the readings,
VI. DISTILLA'l'ION UNDER HIGH VACUUM 577
and whether the sensitivity depends on the nature of the gases or vapors
being measured.
2. Fore-Pressure Gages
The U-tube mercury manometer (or an open tube submerged in a well)
and the Bourdon-type gage are used for fore-pressure measurements from
atmospheric pressure down to about 10 mm. (see Chap. V, Figs. 5 and 14).
Such gages are relatively trouble-free and their accuracy is sufficient for
indicating the degree of vacuum during the evacuation cycle. A special
Bourdon gage 108 has been designed which can be used in the range of 1 to 20
mm. If greater sensitivity is required, an oil manometcr can be used. A
nonviscous, low-vapor-pressure organic fluid, usually a diffusion-pump oil,
is used to fill the tube. The level difference in actual measured milli-
meters can be converted to millimeters mercury by multiplying scale 11 mm.
by density oil/density mercury. It is sometimes convenient to make a scale
calibrated directly in millimeters mercury. In this case one scale division
in millimeters mercury equals density mercury/density oil. The reference
vacuum can be a small rotary mechanical pump which will give an ultimate
vacuum of 25 fJ- or less. In Figure 40A is shown a U-tube model with a
stopcock arrangement for equalizing the pressure in the two limbs during
"pump-down" or "outgassing" periods. In Figure 40B the volume of the
well is made large relative to the volume of the measuring tube, so that a
fixed scale can be used. Oil-filled manometers must be "outgassed" each
time they are exposed to atmospheric pressure. This is done in a matter
of minutes and is usually not objectionable, particularly if the system is to
be maintained under vacuum for !1ll extended period of time. The oil
manometer is relatively free from trouble, measures total pressure, and is
most useful in the range of 0.1 to 20 mm. The range has been extended
by the use of a sloping U-tube in place of a vertical arrangement. 109
The bench-mounted McLeod gage is generally not a practical instrument
for high-vacuum distillation application. The portable tilting type llO •
shown in Figure 41 is useful as a fore-pressure measuring device. The er-
rors in regard to condensible vapors which are inherent in the standard
McLeod gage are not avoided by the simplified design. However, the in-
strument is portable and very convenient to usc.
The gage is connected to the system by a rubber tube which should be
kept as short as possible and be thick-walled vacuum tubing. The gage is
mounted on a swivel which, on rotating 90 0 , cuts off a definite volume, V,
at the system pressure, P, and compresses this volume to a higher known
108 Wallace :tnd TiernaIl ProductR, Inc., Belleville, New York.
100 Hiekman, Chem. Revs., 34, 82 (1944).
110 Flosuol'f, Ind. Eng. Chent., Anal. Ed., 17, 108 (1045).
578 J. C. HECKER
PV value. The gage is calibrated in miCl'QIlSj models are available which

COV(~l' the ranges of 0-700 J..! uncl 0-5000 iJ.. A similar device called a Vaeu-
stat is described by Yarwood.lIl
THREE-WAY
srOPCOCK 'iI',C\J\JM
RtFERENCEr-----~ CHAMBER
\lACUUM
VACUUM
CHAMBER
SCALE
SCALE REFERENCE
VACUUM
OIL
RESERVOIR
Fig. 40. Oil nmnollloters: (A) U-tubej (B) Hingle-tube well.
CHARGING POSITION
READING POSITION
Fig. ,no 'rilting-typo pOl'tl1hle :J\rlcLcDll gage. 1l"
tll Yarwood, Hi:gh Vaclmm Techm:q'Uc. 2nd rev. ed., Wiloy, New York, 1945, p. 39.
VI. DISTII,LATION UNDER HIGH VACUUJl.I 579
3. Fine~pressure Gages
Of the gages listed in Table X,",,{I, those which have received greatest ac-
ceptance for high-vacuum distillation measurements below 100 J1- are:
Ph'ani, thermocouple, cold-cathode (Alphatron), and portable McLeod.
The characteristics of these gages are shown in Table XXII.
TABLE XXII
VACUUlII GAGES FOR MOLECULAR DISTILLATION
Useful
Type Principle range Advanta!l;es Disad "l1n tagcs
Pirani Loss of heat from 0.1 p. to (1) Portable (1) Filament contam-
hot wire propor- 0.3 (2) Continuous ination causes zero
tiollll1 to pressure mm. reading shift; must be
(3) Attach directly periodically 1'ecal-
to apparatus ibrated
(2) Sensitivity de-
pends on nature of
gas or vapor pres-
ent
/I
Thermo- As above except O.lp.to (1) As for (2) above
couple thermocouple 0.1 (2) Less sensitive than
junet.ion used mm. Pirani
It
"Alphl.1- Positive ion ellrrent, O.lp.to (1) (1) Precautions at-
troll" proportional to 10 (2) It
tending use of rlL-
pressure; radium mm. (3) It
dium; not serious if
eapsllle source of (-0 Wide range operating instruc-
electrons tions are carefully
followed
Portable Compresses known o to (1) Portable (1) Intermittent read-
Mc- volume to known 5000 p. (2) Convenient ing
Leod PV (2) Hg contamination
(3) Pressure readings
are low if conden-
sible vapors present
The Ph'ani and thel'mOCOlbpZC gages depend on the change of heat conduc-
tivity with pressure, which, at low pressures, increases linearly with pres-
sure. Such gages, in general, are operated in such a manner as to maintain
a constant energy input to the warmed element. The element consists of a
filament or ribbon of some metal (such as tungsten, nickel, or platinum)
which has a high temperature coefficient of resistance and which is not at-
tacked by the gases or vapors to be measured at the temperature involved.
As the pressure is raised and lowered, the hot element will lose heat at a
varying rate and thereby suffer a temperature change. Therefore, such
gages resolve themselves into means of measuring the temperature of the
hot element. In the thermocouple type, the temperature is determined by
means of a thermocouple spot-welded to the heated element. In the re-
sistance type, the Pirani being an example, the temperature is determined
580 .J. C. RECTum
by mel1Aul'ing tho ehnngo in l'CSistnIlcC, 01' tho potolltial l12 is measured

which must be supplied to the Whent8tone bridge to keep the resiBtrmee and
henee the temperature of the wire constant. The ehange in length call
also be llsccUor LhiR purpOHe.
VACUUM
CHAMBER
COMPENSATING PRESSURE
REFERENCE MEASURING
TUBE TUBE
I
[
I
I
I I
I / POWER-SUPPLY
/ / CONTROL BOX
/ I AND METER
/
I
/
I
I
/
I
/'
I
/
/ .
/
L ____________ J
Fig. 42. Plrani gllgP.
A schematic diagram of a resistance-wire gage, Pit'ani type, and calibra-

tion plots are given in Figures 42 and 43, respectively. A Wheatstone-
,bridge arrangement is used for measuring the resistance ehanges .. To com-
m Dushman, Scientific Ponn,zation" of VaMmm 'l'echnique. Wiley, New York, 1949,
p. 317.
VI. DIS'l'ILLATION UNDER HIGH VACUUM ,581
pensate for room-temperature and bridge-voltage variations, a second tube

duplicating as closely as possible the characteristics of the measuring tube
is sealed off at an ext,remely low pressure and balances the opposing arm of
the bridge cireuit. The calibration of the gage is dependent on the thermal
conductivity of the gas in the system.
1.0
I -
0.9 I /
ARGON/ / CARBON f'OXIDE
0.8 I
II
0.7
/ /Ar
'"
:t
E
E 0.6 I f_ ~ / W~TER VAPOR
1
w-
et:
::l
(/)
0.5 _j_
/
! ~~ . . / HELIUM
I
/ ~'/
(/)
w
0: ./ AdETYLENE
Q_
w
0.4 L
::l
0:
I-
0.3
j
II
f) ~ /V
/' '/
/ [§ V ---
!----
I-- HYDROGEN
r/ ~ -:::-
0.2
~
O. I /_
o
~V
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0,9 1.0
mm. SCALE READING
0- 0.75
Fig. 43. Cnlibration curves for various gases with Ph'ani gage, type PG-1A, Distillation
11 roducts Industries. Bridge voltage 3 volts.
The thermistor-type of gage has been described by Becker, Green, and

Pearson.u s These semiconductor element.s have a high negative tempera-
ture coefficient of resistance. A linear response was obtained in a vacuum
gage between 1 and 1000 fJ.. A gage of these characteristics would be adapt-
able to high-vacuum distillation technique.
In performing a distillation, organic vapors diffuse i,a the gage, deposit
on the hot element of heat-conductivity gages, and alter their characteris-
tics. The gage calibration drifts. This "zero shift," as it is called, is
characteristic of this 'type of gage in the presence of organic vapors. The
gage contamination can be lessened by attaching the gage to the dawn-
113 Becker, Green, and Pearson, Trans. Am. Inst. Elec. En(],·s., 65, 711 (1946).
582 J. C. HECKER
stream side of a cold trap or baIfie, although this is likely to givc readings
lower than actually exist. in the evaporating zone.
If the Pi rani or therrnoeouple gage is to be relied on for periods of weeks
or months, a means for a periodic calibration should be provided. A test
rack can be assembled from the following essential components:
(1) A condensation pump capable of producing 10-'1 mm. or lower.
(2) A mechanical backing pump.
(3) A cold trap with outlets for attaching a gage or !l,ages.
(,4) A McLeod gage which can be read to a miel'oll in tlw mnge of 1 to 20 }.l.
One or more gages along with the McLeod is attached Co a high-vacuum

manifold connected to the condensation pump. Seventl points on the
faulty gage are checked against the McLeod, ~md the IJimni is reset accord-
ingly. The pressure can be adjusted by [tdmitting a small continuous
amount of air through an adjustable leak valve. A Rhor!; pieee of thick-
walled rubber tubing with a fine wire inserted through the opening and with
a screw pinch clamp compressing the tubing agaillfit the wire makes a
simple, satisfactory, adjustable leak valve.
The Alphatl'on developed by Downing and Mellen ll1 opr.mtes on the
ionization-gage principle, the ionizatioll being produecd by alpha particles
as distinct from elet\tron bombll,rdment in the hot-fih.ment gage. This
gage is rugged and suffers little from eont:uninntion. However, since the
gage contains a radium source, certain precautions have to be taken in the
use of this instrument. This is not a serious consideratioll if the manu~
facturer's opemting instructiolls are followed.
The gage should be connected as near the distilling zone as possible to
obtain an indication of the true pressure. Contamina.tion difficulties
sometimes necessitate its being plaeecl away from the still, but in such cascs
allowance should be made for possible pressure drops between the gage and
the distilling zone.
VI. COLD TUAPS AND BAFFLES
As pointed out previously, during most ditltillations of organic materials,

vapors which are dissolved as impurities in the original material, or vapors
genemted by thermal decomposition during the distillation, often fail to
condense in the still. This is particularly true when the condensing zone is
kept warm to assist the flow of viscous condensfttes, or to prevent the pre-
cipitation of solids from saturated solutions. If vapors so liberated from
the distilland are not condensed by some cooling means between the still
111 Downing and Mellen, Rev. Sci. Instrltments, 17, 218 (1046); Electronics, 19, 142
(1946).
VI. DIS'£U,LA TION UNDER HIGH VACUUM 583
and the condensation pump, they are likely to be wholly 01' partially con-
densed or absorbed on the cold condensing zone of the pump.
Since the vacuum attainable by a condensation pump depends on the
vapor pressure of the operating fluid, it is obvious that the performance of
the pump will be impaired in proportion to the type and amount of con~
taminating vapor that reaches it. Annoying "bumping" of the boiler
usually accompanies a contaminated pump, which also affects the speed
and degree of vacuum attainable. With the heat input adjusted properly
and with fresh pump fluid in the boiler, evolution of vapor takes place with-
out the bumping and bubble formation which ordinarily accompany a boil-
ing liquid.
In addition to protecting the pump, one frequently wishes to examine all
the constituents of the distilling liquid and therefOl'e endeavors to trap
any vapors which do not condense in the still. Of the many shapes and
forms of cold traps, the one shown in Figure 33A is particularly suited for
high-vacuum distillation.
The refrigerants most commonly used are liquid nitrogen, liquid ail', or
pO\vdel'err dry ice. Since liquid air has a varying composition, resulting
from the faster vaporization of nitrogen, its temperature 1l5 may vary be-
tween ahout - 183 ancl- 196°C. Liquid nitrogen is now readily available
in commercial quantities and is preferable to liquid air since 2i more can:..
stant lower temperature of about - 195°C. is obtained, and, in addition,
the possible explosion hazard of liquid ail' is obviated. An imploding trap
containing liquid air can produce an explosive reaction when connected to
a condensation pump 01' to a still containing hot oxidizable organic com-
pounds.
A liquid-nitrogen trap Can effectively remove water vapor, carbon di-
oxide, and relatively heavy organic V!Lpors. However, such gases as ethane,
ethylene, methane, and carbon monoxide have appreciable vapor pressures
:1t liquid-nitrogen temperatures and are not adequately trapped. By re-
ducing the pressure over liquid nitrogen, lower temperatures can be pro-
duced, as shown in Table XXIII.
In cases in which the extreme low temperature of liquid nitrogen is not
TABLE XXIII
TEMPERATURE VERSUS PRESSURE OF COMMON REFRIGERANTS
PreRsure, mm.. . .. 760 76 7.6 0.76

No" .............. -HJ6"0. -210°C. -220°C. -227°C.
CO: ........... - 78"0. -103°C. - 12°C. -137°C .
• International Critical Tables. Vol. III, McGraw-Hill, New York, 1928, p. 204.
b Ibid., p. 207.
115 Dodge and Dunbar, J. Am. Chem. Soc., 49, 5111 (1927).
584 J. C. HECKErt
needed, a trap temperature of -78°C. CiLn be produced with solid carbon

dioxide. Even though dry ice is readily obtained there are occasions when
a cylinder of carbon dioxide gas, but no dry ice, is at hand. Carbon dioxide
snow can be obtained by releasing carbon dioxide gas into a cloth bag at
atmospheric pressure. The vapor pressure of water at dry-ice temperature
is about 1 }l-, and since molecular distillations are generally performed in the
low~micron region, water vapor cannot be effcQtivcly removed by a dry~ice
trap. To provide better heat transfer between the dry ice and the trap
wall, a liquid, such as acetone, alcohol, or butyl Cellosolve, is added to the
powdered dry ice to form a mush.
As shown in Figure 50, it is common practice to luwe two cold traps. In
performing a distillation, it is advisable to delay cooling the traps until a
pressure of around 100 )J. is reached. However, if it is known or suspected
that the still charge contains solvents or volatile contaminants, the second
trap is chilled and a pre-evacuation or uegussing period is allowed. At the
end of this period the vacuum is broken and the trap warmed and drained
before the actual distillation is begun. If the solvents or vapors are ex-
cessive, both traps can be filled during the degassing cycle and drained be-
fore attempting to attain the desired dil:ltilling vacuum. In any event, it
is advisable to fill the second trap first aJtel' a vacuum of around 100 M is
reached, get the vapor pumps into operation, and then fill the first trap.
If the first trap is filled when the pressure is high, vapors which are con-
densed at this high pressure will be slowly liberateu as the vacuum gets bet-
ter, prolonging the time needed to attain the ultimate distilling vacuum.
If copious quantities of liquid materhls are being collected in the cold
thimble, a bottom vacuum-jacketed section is added, as in Figure 33C.
The re-evaporation which would occur as liquid drops to the bottom of the
trap is lessened if the vacuum flask is added. A short piece of heavy~wall
rubber tubing, with pinch clamp, eonnceted to a small bottle permits the
periodic draining of collected fluids and their removal to the atmosphere.
The conductance or "speed" of cold tra,ps of the general design shown in
Figure 33B has been discussed by Dushman.1l6 It should also be observed
(equation 49) that at temperatures lower thiLn 25° the conductance will be
lower than at 25°C. The conductanee will be:
V Tr j298C2b
in which Tr is the temperature of the refrigerated trap and C26 its conduot-
ance at 25°C.
For liquid nitrogen at -196°C. (77°1(.), the conductance for any gas is
V77/298 = 0.52 timeS' that for the same gas at 25°0.
Il! Dushman, Scientific FOlmriations oj l'ac!~1tm Technique. Wiley, New York, 1949,
p.105.
VI. DISTILLA'l'ION UNDER mGR VACUUM 585
An obstruction or series of obstructions in the vapor path which will in-

duce multiple collisions of the molecules with a surface maintained suf- .
ficiently cold to condense the impinging vapor moleeules is called a baJlle.
A right-angle bend or a cold trap could be considered specific baffle designs.
In some instances a baffle is used to prevent the condensation pump oil from
diffusing upstream into other parts of the apparatus. A baffle designed by
VACUUM
CHAMBER
BAFFL~ '"
PLATE~~ --
TO VACUUM
PUMP
j TO VACUUM
PUMP
Fig. 44. Metal cold traps.
More, Humphreys, and Watson 117 is shown in Figure 44A. The inner con-
tainer for refrigerant is surrounded by two helical ramps fastened to the
cold thimble. Another common type of baffle is shown in Figure 44B.
The conical stack of washers of decreasing diameters has tubing fastened
thereto, through which cold water or a suitable refrigerant can flow. In
some instances, when continuous operation is. desired, the tubing becomes
the expansion zone of a Freon compressor system.
117 More, Humphreys, and Wat~oIl, Rev. Sci. Instruments, B, 263 (1\)37).
586 .r. C. IIECKBH
Mechanical baffles, as db,;tinct from cold traps, [1re not commonly used in
small laboratory im;tallations. Theil' importi1uc:e in pilot-plant ttnd eom-
mercinJ installations is apparent.
VII. LEAK DETECTION
As vacuum stills are rmtde more complex, the ehanee for leakH increaHcH.
Systems which nre assembled with gnsketed joints, metal welds, rotating
seals, and so on are rarely absolutely tight. A eertain amount of air in-
leakage is usually tolerated, the degree depending on (1) the standard of
tightness established, and (2) the balanee between pump oize and leakage
rate. One of the trying diffieultieH in operating Vlwuum equipment, is too
often the time-eonsnming len,k hunting which pbguos mnny an operator.
There are various time-honored methl)ch; of apprcmc:h and useful instru-
ments which can shorten the hours oft on consumed in fruitless search.
Whenever possible, the system should be put under pressure and the large
leaks deteeted an(1 closed prior to evacm~t.ion.
A summary of variations of the pressure technique is given in Table
XXIV, Part A. Internal preSSLlres of 5Ib. or more are desirable, although
less Pl'OSRlll'e can be m;ed for large leah;. Tn instanees in whieh tlw vessel or
system cannot he put uncleI' prel'lsuro, one of tJw variom; vaellum methodH
outlined in Table XXIV, Part 13, ean be used. In t1tt.empting to find leaks
by smearing the suspeeted [1ma with vacuum grease or painting with Glyp-
tal, one should rememher that. it is very difficult to fw;e gbss or repair a
metnl wold after the sealant has been drawn into tL pinhole or small crack.
For glass apparatus the Tesla-coil is still n nsd'ul tool. A hand instru-
ment can be obtained from Heientifk supply houses. The system is put
under moderate vaeuum (0.01 to 1 mm.) and the probe of a Tesla~coil
played on the su~pccted joint. The most minute pinholes will permit the
direct passage of a spark to t.he interior of the vessel, whereas a uniform
glow will result over a nonlenky portion of the apparatus. If desired, the
leak can be proved by applying acet.one, cttl'hon tetl'l1chloridc, diethyl
ether, etc:. on the suspected spot and sparking the apparatus with the coil.
A characteristie glow of the vapor used will result when the solvent is
drawn into the vacuum chamber. Compounds containing chlorine give a
greenish glow and hydrocarbons, diethyl ether, and water vapor a greenish-
gray glow, while air gives a red or pink discharge. Precautions should be
taken in using a high-frequency discharge, for an intense spark can itself
puncture a hole in thin glass. The outgassing of a glass system can be ac-
celerated by intermittently ionizing the gas by means of ~L high-frequency
discharge. Obviously, the spark test cannot be used for metal equipment.
The halo(fcn-ion rletcctm· lIR (see Table XXIV B) is more sensitive than
the soap-bubble test. It, is based on t.he observation that heated platinum
shows a sh:up increase in positive iOll. emission when in contact wit.h a gas
or vapor containing halogens, In using the device the system is put under
a positive pressure of a halogen-bearing gas or a mixture of such gas and
air. The gas escaping through a leak is drawn by means of a self-contained
fan into and through the sensitive element. If it is not practical to put the
system under a positive pressure, t.he sensitive clement can be sealed in the
vacuum system, probably in the fore-pressure line, and the system probed
with halogen-bearing gas. When a leak is approached, the halogen gas is
"tmcked" through the leak and quickly detected by the platinum electrode.
As an alternate location, the device could be attached to the discharge post
of the mechanical fore-pump, with suitable means being provided to trap
out entrained oil globules, and the same probing procedure followed. Typi-
cal halogen compounds which can be used are chloroform, carbon tetra-
chloride, and Freon. An amplifier cnn be substituted for the microH,m-
meter and the sensitivity of the instrument increased.
A simple and less sensitive halogen detector is an acet.ylene burner 01' gas
bunsen burner, equipped with a copper plate supported in the flame. The
ail' intake to the burner is through a rubber hose which is moved about near
the suspected lea,k, as above. Halogen-bearing gas or vapor escaping
through the leak of the Hystcm under pressure is drawn into the flame, im-
parting a bright green color from the reaction of the halide with the hot
copper plnte.
The varying sensitivity with composition of gases of the Ph'ani or ther-
mocouple gage can be put to practical use in leak hunting. The sensitivity
is highest for gases of high thermal conductivity, such as hydrogen and
helium, q,nd lowest for "heavier" gases. 'Therefore, if f~ gl1S or Vl1por which
is either "lighter" or "heavier" than air is played on a sllspected leak while
the system is under vacuum, a leaky spot will produce a sharp deflection
of the Pirani gage. Either natural or illuminating gas can be used in place
of hydrogen or helium as a probe gas. Also, propane, acetone, or carbon
tetrachloride vapor sprayed onto the suspected joint have been used. This
method is limited to leaks of such size that u vacuum of between 5 and
200 J.t can be established, which is the most useful range of the Pirani gage
for this purpose. Thermocouple gages can be used for this purpose, al-
though the Pi rani is more sensitive.
The most sensitive of all instruments for detection of leaks is the mass
spectrometerhel~'U'm detector-Yo A modifieation of the original mass i:lpee-
118 White and Hickey, Electronics, 21, 100 (1948).

119 Nier, Stevons, HustraJid, :Lnd Abhot.t, ,T. Applied Phys., 18, 30 (1947), Jacobs
PTJrl 'l·"hr ,1vir! 1" '2' (1 0 '7)
588 J. C. HECKER
It., 1<:) OOlr':l 1<:) 0 10lQ

00
o o IQ H~ 0 o
, 1<';)
"...
=
0
S oj
'S ~
<>
....
S '1
.0 .~ <a ...-<
J ...
0
'"CI 0
p..
<1l
.~ ""
'" '"
'"CI
~
] §' S =
6
...::!
:-SA
~.~ 1p::
0"
'+'<
rJ3 r>:..w .w ~
8
0
""
'~
td
~
-g "i o:l ] g
·s
z z <>~ "d
~5 <>
j z '§ ]N
~
0
..9
~ ~. ~~
~:E 0< l~
~ g ~~
I
,...; M ~.,..; .Q .at..: 00 oj 0

.....
VI. DIS'l'ILLA'fION UNDER HIGH VACUUM 589
8..... 8,.....
......I I
,..;
t
r.."8
lJ::.t!A;
c4 cO
590 J. C. HECKER
, o
'f]
;a o
...,
V
o
88 00
00 o
.r,
OC'l
,..,1 ........
00
,....
1
,..,
1 .r,
........
1 1
00 oH oo .....
o
""0
I ..... 1 rl
.r,
'Q ,_.
1 ...,
1 o
V
VI. DIS'rILLA'rION UNDER HIGH VACUUM 591
trometer of Aston and Dempstel'12(J "tuned" to helium ions aiTords the basis
for the detecting; instrnment. If a singly charged positive ion of molal.'
mass 111 is accelerated by a negative potential gradient of E volts and pro-
jected into it magnetic field of strength of Ii gauss, it is bent into a circulal'
path of radius r centimeters, according to the relation:
1'2 = 2070(EMIl-P)
This follows from the fundamental relationship:

e/'Ii/' = 2E/IJ2I· 2
Therefore, for 11 fixed value of IE and H, ion~ of different mass will have dif-
ferent radii. Eleetrons from an ion source, on striking helium and other
gas atoms, knock out electrons thereby forming positive ions. By means
of a. strong potential graclifmt, the positive ions are quiekly pulled hom the
ionizing zone, both to aceelerai;e them in n path at right t1ngles to the field,
MECHAN ICAL HELIUM (OR OTHER VACuLJ M SYSTEM HELIUM (OR OTHER
ROUGHING PUMP LEAK DETECTOR) UNDER TEST PROBE GAS)
Fig. 45. Typical lenk detector - vacuum system arrangemont.
as well as to remove them from the high electron density where recombina-
tion with electrons might aceur. In passing through the magnetie field, the
iOllS are sorted out according to their mass and charge. A collector plate is
placed to receive only the helium ions, which when present cause a positive
c:nrrent to flow through :11l amplifier circuit, and thence to a meter.
'Fig'Ul'e 45 shows an arrangement of a vacuum system and a leak detector.
The leak detector has self-contained var:uum pumps which maintain n.
pressure lower than thn,t, needed for the ~uspeet syRtem during test. The
:,;ystem under test is first exhausted with its own fore-pump and, when
the pressure is sufficiently low, the throttle valve admitting gas from the
system to the leak detector is opened. The degree of opening will depend
on the vaeuum in the system under test. A fine jet of helium is played on
120 Aston, Phil. Mag., 38, 707 (H)19). Dempster, Ph1l8. Rev., 11,316 (1918).
592 J. c. HECICmn
the suspected joints. A leak will almost instantly be revealed on the output
meter of the detector. A "squeu.ler" device can he hooked into the circuit
to permit one person to hunt leaks at locations where the meter is not
readily watched.
Helium was chosen for the probe gas for the following reasons:
(1) Helium is a rare gas in the atmosphere (1 part in 20(),OOO) i therefore a low
background level is established. (2) The helium ion has a distinct and illdividltal
location on the ion spectrum and thereby eliminates the possibility that an ion of
any other gas might be mistaken for helium. (8) Its rate of diffusion through the
leaks and through the vacuum systems is rapid, being exceeded only by hydrogen.
(4) It is plentiful and readily available in this country.
The instrument will detect one pad; helium in 200,000 parts of air and,
on occasion, this can be increased seveml times. Under such dilutions the
helium has a partial pressure of about 5 X 10- 11 mm. mercury. The
quantity of helium flowing through thc spectrometer tube corresponds to
about 1.5 X 10-6 micron liters per second. In comparison with pumping
speeds in tens and hundreds of micron liters per second, the size of leaks
that can be detected are really infinitesimal. The application of the mass
speetrometer to the detection of leaks has been described by Worcester
and Dough1;y 121 and by Thomas, Williams, and Hipple. l22 The proper use
of the instrument and the optimum balance of exhaust speed versus volume
of the system have been analyzed. ln For rapid response, a speed of ex-
haust comparahle to or greater than the volume of the chamber under test
is desirable.
VIII. VACUUM FITTINGS
Even though it is desirable to eliminate as many temporary joints as

possible, Tubber-tubing eonnect.ions and rubber-gasket flanged joints have
their place in assembling vacuum-distillation equipment. In joining glass
tubes, rubber tubing C[Lll be successfully used (see Fig. 40). Small-diameter
(< 1 cm. O.D.), good quality, heavy-wall, rubber tubing is recommended.
A standard red rubber tubing with a 3/ln-in. I.D. and a lh-in. O.D. is
adaptable to 8-mm. Q,D. glass or metal tubing. The wall thickness should
be such that the tubing will not collapse when vaCuum is applied. By
leaving a gap of a half inch or so between the two ends of the tubes being
joined, a pinch clamp can be used to seal off sample bottles, gages, conden-
sate from a cold trap, etc.
Worcester and Doughty, Trans. Am. Inst. Elec. Engrs., 65, 946 (1946).
121
us Thomas, Williams, and Hipple, Rev. Sd. In8trument8, 1'7, 368 (1946); Westing-
hou8e Engr., 6,108 (1946). "
lU Jacobs and Zhur, J. Applied Phys., 18, 34 (1947).
VI. DIS'rILLATlON UNDER HIGH VACUUM 593
POI' joining tubes of intermediate diameter (Pig. 46B), between 1 and 5

n., rubber tubing of 2-3 mm. wall thickness can be used. There should
~ no gap between the ends of the tubes, or the rubber may "suck in." In
)th this case as well as in that of the smaller diameter tubes, the inner
ameter of the rubber tubing should be in the range of 5-15% smaller
Lan the outer diameter of the tube over which it slips. The elasticity of
Le tubing will determine the amount of stretch that can be accomplished;
_~PINCH CLAMP
HEAVY-WALL GLASS (OR METAL)

~UBBER TUBING TUBING
(Al
l
GLASS TUBING WITH
FLANGED END
RUBBER TUBING GLASS (OR METAL)
METAL Fl,lllllr,E
u[j' ilII2I2 UIlIIIIIIIZIl1722Tl1

~~~~
GASKET
(B l [C)
'ig. 4.6. Gln.8s-to-glass connections: (A) small diameter; (B) intermediate diameter;
(C) large diameter.
he tighter the better. In general, soft rubber (durometer <45) is to be

)referred to the harder grades (durometer >45). The inside of the rubber
leeves and the outside of the j Dining tubes should be smea,red with castor
dl, vacuum grease, or other low-vapor-pressure lubricant (see Table XXV).
['his serves the double purpose of lubricating the parts, thereby making the
ube insertion easier, and sealing the minute air gap between the rubber
:leeve and the tube. If the surfaces are dry they will invariably leak;
;herefol'e, joints of this type, and, in general, all rubber joints, should never
)e made without a sealing lubricant.
POI' large diameters (I.D. >5 em.), standard flanged joints as shown in
8'igure 46C are recommended. These can be obtained commercially in
rarious sizes. Rubber or synthetic gasket material of durometer hardness
594 J. C. HECKlm
TABLE XXV
SgALING BY LUl.lHrCAl'INO COMPOUNDS WITt! VA.POl\ PumsslJIu"s LESS 'rlIA.N 1 MICRON
AT 25°C.
---~.-----' --_
GOUlpOllUd Pl'(lIJm'til~S
CUlll!(ms!ttioll and booster' pump fluids Liquid !It room tempemtUl'e, low viscosity
Clts~or oil Liquid at room tcmpCl'I1ture, high viscosity
Celvltc8ne-lightn M.p. about OO·C.,.thin grease
Celv!l.cene-medium" M.p. about 12()OC., bct,ween ·"light', ami
l'hen,vy"
M.p. [,bollt l30°C., more viscous than
"light"
Mvvaeene-S" M.p. grol1tm- Limn 300 DC., li~ht, tacky grease
siiicolle grease b M.p. gl'e:11,ol' Lilltll ;~OO DC., light, taeky gl'e!lS!l
Apiczon I.' Thin grease
Apinzon M lmel N' More viscous tlmn "I."
Apiezoll Q' Cmphito mixl1(1 with low-vapor-pressure
pamiHn oil l'Oshh;_es. Semisolid at 25°C.
Used like put,(,y t.o SOld joints, etc.
Apie?OIl W' WltX. Useful for wlLxing joints in ghLss 01'
mckll. Softens nt 60-70 ·C. Soluble in
xylene
de Khotinsky ecmont<i Mh:LIlJ'c of IIhclhl[J ltnd piteh. SoftCIlR at
[)O°e.
Picein" HtLnl blnek WIIX (rubber, shellac, bitllmen).
Softens lIt 50°C.
Glypt!Lllaequc)·j Dl'i(lH to a lHtl'd tilm. Useful for p:tinting
threaded joints in m(\tal pipo lines
Shelhu: Addition of butyl phtlmiaLo til:! plastici?er
proven ts (,r:wking; of the ,ll'ied film
" Distillation Produe(;s IJl(tll~trie;;, Roohcstcl', New York.
I,Dow Corning Corpol'utioll, Midl!11111, MiehiglLll.
, MetrCl]lol.ilml-Viekot·s Eleetl'icml COUlPILllY, Ltd., MlLll(:hcKtnr, EnglllIlcl; murketed
by JlLmes G. Biddle COIl1PllllY, PltillldelplJill, l'ullnsylvltuill.
a CenLml ScientiJir: Cmnpnuy, Chicago, IllilloiR.
, Sehradel' !Ind Ehlel'~, New York, New York (iIllporters).
j Genel'llllGlectric Company, Sohellectady, New York.
between 50 and 70 i~ generally URed. Ahm, the scheme shown in Figure

·17B is applicable.
GLASS OR
METAL
TUBING __ ~
·-'.'(;~TU'''G
- SPACER
FLANGE
METAL BODY RU BBER SEAL

(A) (S)
Fig. 47. Gla~s-te-Uletlll joints: (A) slllnll tt\hing; (B) large tUbing.
In making glass-to-metal joints, 1),' gasket or rubber-tubing connection of

some sort is commonly used. The refinement of a graded glass-tO-lnetal
seal is rarely used in assembling high-vacuum distillation apparatus. Glass

tubes can be joined to metal tubes by the scheme shown in Figure 46A or
in Figure 4GB. An assembly shown in Figure 47A can be used for inserting
glass tubes, such as a Pirani gage, into a metal chamber. Modifications
of this general scheme have been described elsewhere. 124 A rubber section
is squeezed by a compression nut against the tube and the side wall, thus
sealing both smfaces. For large diameters, a gasket joint snch as shown in
Figure 47B can be used. The metal flange Can be the base of a vacuum
chamber or a separating washer between a glass-to-glass, metal-to-glass,
or metal-to-metal section. A special molded gasket is ordinarily used for
this purpose.
Flanged metal jlJ1:nts can be made in many different ways. Some of the
common patterns atVshown in Figure 48. With reasonable care in as-
sembly and alignment the simpler designs are frequently satisfactory.
VACUUM OR
PRESSURE
FOR TESTING
TIGHTNESS
~~=d/
(AI (RI (el
Fig. 48. Metal flange - gasket assemblies.
Phlinflat-sUijaceflanges uan be used successfully if the surfaces can be drawn

up with uniform tightness against a lubricated gasket. A flange with a
single recessed groove into which is inserted a square or rectangUlar gasket
is a refinement over the flat surface. A double recessed groove is a further
refinement. The inner gasket provides the vacuum seal while the outer
one forms a confined space which can be evaeuated with an auxiliary vac-
uum pump, ail' pressure or probe gas pressure applied therein, and the
response of the vacuum gage on the system noted. This is particularly
helpful in leak hunting in complicated systems with many flanged joints.
Typical examples. of sliding and rotating shaft seals are shown in Figure 49.
In all mtses it is advantageous to have an oill'eservoir on the atmospheric
side of the seal. For tL vertical mounted shaft, as shown in Figure 49A,
suffiuient height is allowed on the bearing and seal housing to afford space
for an oil pool above the oil-retainer seal. 1<'01' horizontally mounted shafts
(Fig. 4gB and C) a double seal 01' packing; is commonly used with an oil-
filled space between the two.
The circular gaskets of round cross section known as "0" rings can he
adapted to the examples given and to many other designs. The manufac-
124 Llnll'itsell and LauritHen, ]lev. Sd. Instruments, 19,919 (1948),
596 ,1. C, HECKER
turer gives information on the proper selection of sizes for any particular ap-
plication.
As a final precaution, dry joints 7/,s1J,ltlly Zetdc, If it is undesirable'to
lubricate the mating surfaces, as a second choiee the edgc to the atmosphcre
should be oiled or smeared with a suitable vacuum grease,
ATMOSPHERE
. ROTATING OR
SLIDING SHAFT
(A) VACUUM
ATMOSPHERE
ROTATING OR
SLIDING SHAFT
PACKING NUT _
DETAIL OF
LANTERN RING
LANTERN RING ~.y~:rr====r1:; OIL-CUP OPENING
PACKING
VACUUM {CI
li'ig, 49. Sliding fLnd rotating Heals,
. Fore-pressure lines are generally of such sille that standard pipe can be
used. In making up threaded joints on standard pipe, it is advisable to
smear the.threads with a. sealing compound, preferably of a nonhardening
type. After the joint is tightly drawn together the contact between the
VI. DI8TILLA'.rION UNDER HIGH VACUUM 5\)7
two pieces can be coated with GIyptul if the joint is found to be leaky on
test. A well made threaded joint can be tight and is confOiderably simpler
than welded pipeline construction.
In selecting valves for metal systems, the choice will depend to some ex-
tent on the pressure region in which they are to be used. If a valve is to
be inserted in the high-vacuum line (0.1 to 10 /J.), a valve with low imped-
ance should be selected, that is, one in which very little reduction in cr088-
sectional area takes place when the valve is open. Sorne impedance in the
fore-pressure line can be tolerated. Valves whose packing glands have been
replaced by a sylphon bellows arrangement have decreased the possibility
of leaks from that source. When glass stopcocks are used, one should select
those with large bores and preferably with oil-sealing cup arrangement at-
tached.
IX. INTEGRATED SYSTEMS
Having discussed the basic components which go into a distillation sys-

tem, we can now consider the construction of a typical laboratory assembly.
Let us refer to Figure 50 and take t.he example of a batch pot still with an
outer tube 12 cm., an inner tube 6 cm. in diameter, and a distance of 12 cm.
between the base of the· inner tube and the side-arm vacuum connection.
It would be nice to know. the total gas load so that a speed of exhaustion
could be mathematically determined. Since there are many factors in-
volved, such as liberation of dissolved gases, thermal decomposition, and air
inleakage, a quantitative value of the gas load is rarely known for all the
samples which will be examined. A few trial runs in a given piece of equip-
ment will soon establish whether sufficient pumping capacity is available
at the distilling zone. Pressure measurements at the still will give a direct
indication of the effective pumping speed.
Lacking data or experience, a first approximation can be made by pro-
viding a still of evaporating surface of A square centimeters, with a speed
of exhaust of approximately 1/4 A micron liters per second. In the ex-
ample chosen, the still has an evaporating area of about 80 cm. 2 and, by
the rule of thumb above, an exhaust speed of 20 micron lit.ers per second
should be available at the still. Furthermore, if we assume the distillation
will be done at 2 Il, a speed of 10 liters per second would be equivalent to 20
micron liters per second.
As experience is gained, sizes of pumps and connecting tubing can be
estimated for any given size still without a complete mathematical analysis
of every detail of gas flow through 'the system. However, as an exercise in
demonstrating a method of approach and the effect of tubing and obstruc-
tions in high-vacuum systems, an analysis of the distillation train in Figul't'l
598 J. C. HECKIDR
50 is presented. As a simplification, it is assumed that the gas is ail' at

25 C. and t.he numerical formulas as given in nn earlier scction apply. First
Q
of all, dissolved air liberated from the still charge will find resistance in en-
p= 70
S =0.28
Fig, 50, Typieal ])I'()SSUJ'(! - gaR flow l'(lbUOllShip in II distillation system. 1', pressure
in miC1I'Onl; of nWl'eUry, S. gas flow ill liters lXU' second.
i;ering the space between the condensing tube and the outside wall. Since
the upstream area, A', is not large compared Lo Lhe downstream area, A,
the aperture conduet[LllCC is:
Po = 11.7.,1 = 11.7?r(S2 - 32) = SUO
Hence: P1 = 590 X 2.8 = 1650 litel'f:i PCl'.sl)concL
The gas will next find resil,tance in passing between the walls of the still
and the condenser. The conductance. along this concentric path is:
F2 = 1l.7A (rdl)x = 590 X 5!tZ X 1.07 == 202 liters per second
VI. DIS'l'IUATION UNDER HIGH VACUUM 599
Molecules will find little difficulty in entering the side arm from the
still; the area of approach is comparable to the area of the opening.
Therefore, the so-called long-tube formula will apply (even though l <
50d) since no end correction Ileed be included.
Fa = 12.2 (da/t) = 12.2 (125/12) = 127 liters per second
Proceeding next to the trap, we can use equation (42) for the design shown
in Figure 33A. Hence:
F~ = 130 liters per second
This is the conductance at 25°C. If the trap is to be filled with dry ice,
and assuming the gas flowing through is at the temperature of the cold
thimble, -78°C., the conductance will be V195/298 X 130, or:
F4 = 105 liters per second
In flowi.ng from the trap to the side tube, no end resistance need be con-
sidered since the area of approach is comparable to the tube entrance area.
The connecting tubing between the trap and the pump dome has:
F6 = 12.2 W/l) = 12.2 (125/10) = 153 liters per second
The dome shown will have a negligible resistance. However, were a
bame included an additionall'eduction in speed of 50 to 80% could be ex-
pected. For this example, we shall assume that the pump either with or
without a bame is a unit with a speed of SIl liters per second. Having es-
tablished the conductance of the various components betyveen the evapo-
rating surface and the oil booster pump, we can now arrive at a size required
to fulfil the prescribed condition of 10 liters per second at the distilling
zone. Solving for F in the equation 11F = l/Fl ... liFo:
F" liP",
1 1650 0.0006
2 262 0.0038
3 127 0.0078
4 = 105 0.0095
5 "" 153 0.0065
Total ............ 0.0282
gives: F = 35.5 liters per second

Since the speed of exhaust, Sa, has been estimated to be 10 liters per
second the required speed of the oil-booster pump, Sn, can be obtained
from the relationship:
600 J. C. HECKER
1 1 1 1 1 1
+:p or ·-:· = 0.1 - 0.0282 = 0.0718
88 8/1 1I'
or: 8[J = 13.9 liters pCI' sceond
Therefore, a booster pump (or pump plus baffle) with n speed of 13.9 liters
per second will exhaust 10 liters per second from the distilling zono through
this interconnecting system. A reduction in effective speed of 39% through
interconnecting tubes and baffles is not unusual in high-vacuum practice.
From the spccifieations and performance eUl'ves of the pump selected,
the fore-pressure into which the gas will be eompressed can be determined.
For this example we shall take 70 JL. Up to the first oil-vapor pump intake,
t.he flow is essentially molecular, but aft.er passing through t.he pump and
being compressed to a fore-pressure value of 70 JL into a tube 1.4 cm. in
diameter, the flow is in the intermedil1te zone, Prl = 98 > 2. In compress-
ing the gas, the volume flow rate dec!'el1ses to 1.4/70 of its value on entering
the pump, or to 0.28 liter per second. In cont,rl1st to this the micron liter
flow rate of 20 miel'On liters per second is eonstant throughout the system
(less any air inleakage along t.he path or deeomposition vapors from the
oil-vapor pumps).
Assuming the first oil-vapor booster pump is connected to a second oil-
va,pol' booster pump by a 20-cm. length of 1.4-em. 1.D. tuhing, from Figure
2 we can ealculate that the conduetanee would be about 3.7 liters per sec-
ond. The second vapor pump, size SR, ean be determined from:
1 1 1. 1 1 1 1 1
-=-+--
B SE F
or - = - - - = -
SE S
- -
F 0.28 3.7
= 3.57 - 0.27 = 3.30
or: BE = 0.30 liter pel' second

the required speed of the oil-vapor ejeetor pump.
From the specifications of the pump selected, the fore pressure against
which it will operate cnn be determined. For this example, assume that in
going through the pump the gl1s is compressed to about 200 }.t, thereby re-
ducing the volumetric flow rate to about 0.10 liter pel' second. Assuming
we use a 30-cm. length of 8-mm. tubing to join the second vapor pump to
the mechanical pump, the Pd factor then becomes HiO and F = 0.65 liter
pel'second. The pump size is determined as before from:
1 1 1 1 1
SM = ;q_ - F = 0.10 - 0.G5
or: SM = 0.12 liter per second
the required mechanical pump speed to remove the gases to the atmos-
phere. As an indication of the size of rotary pump, a Cenco Hyvac has a
VI. DISTILLA'l'ION UNDER HIGH VACUUM 601
speed of 0,10 liter per second and a Cenco Megavac a speed of 0.30 liter
per second in the 200 J.I. pressure region.
If the oil-vapor pump could compress gas to twice the fore pressure, a
fore-pump of only half the size would be necessary. There have been efforts
made along this Ene during t.he past several years, resulting in the design
of oil-vapor ejector booster pumps which will operate into fore pressures in
the 1-10 mm. mercury range. Also, oil-vapor pumps have been developed
which combine the action of the first and second oil-vapor pumps, one such
pump replacing two. A pump of this general design is show11 in Figure 39,
As a generalized summary, an approach to the design of a laboratory
high-vacuum distillation system can be made along these lines:
(1) Calculate or estimate the evapor:;tting area in A square centimeters and
thereby est.ablish an exhaust speed of 1/4 A micron liters per second.. Determine
the operating vacuum, J.L, and establish the liters per second by dividing 1/4 A
micron liters per second by J.L, the pressure in microns.
(2) Layout the interconnecting piping, trap, etc. between the still and the oil-
vapor pump. Calculate the resulting conductance and the pump size required.
If less than 50% of the rated pump speed is utilized, open up the vacuum lines to
cut down the resistance.
(3) Select an oil-vapor pump with optimum pumping characteristics at the de-
sired operating vacuum, and preferably one with a high fore-pressure breakdown.
(4) Continue as in (2) fOl' the next stage of intel'connecting piping, and so on to
the mechanical fore-pump.
As a practical matter, for both vapor and mechanical pumps the next;
largest sized pump available is used in each case, the extra capacity being
useful for shortening pump-down time and for handling traces of unpre-
dictable air inleakage. In fact, after having arrived at the speed required
at the still from (1), a vapor pump having at least 50% more speed can be
chosen and the connecting tubing between the still and pl1mp so chosen to
maintain the desired conductance.
A schematic diagram, as in figure 50, on which one Can mark the various
pertinent data, the flow rates, and the pressures at different points-the
latter depending in large part on the pumps selected-will be of assistance
in making an over-all evaluation of any contemplated system.
General References
GAS FLOW A'l' mmUCED PRgSSURES
Chapman and Cowling, The Mathematical Theory of N on-ttniform Gases. Cam-

bridge Univ. Press, London, 1939.
Dushman, Scientific FoundatiDns of Vacuum Technique, Wiley, New York, 1949.
Jeans, An introduction to the Kinetic Theory of Gases. Macmillan, New York,
1940.
n02 .r. c. ,H1DCKl!jlt
KOJllH1l'Cl, Kinetic Theory oj Gases, With an Intl'od'Uction to Statistical Mechanics.

1;lcGl'I1W-Hill, New York, 1938.
Loob, I{iniilil: Theory of Gases. 2nd ocl., MoGraw~l-lill, New York, 193'1.
Taylor, A 'l'reatise on Physical Chemistry. 3rc! ccl., Van Nostrand, New York,
1942.
(H<:NloJUAI, VACUUM 'l'IWHNIQUJa
Calvin, Heidolbnrp;er, Reid, Tolbert, and Yaukwieh, /.sotopic Carbon. Wiley,

New York, 19·19.
DUlloycr, Vacltum l'racil:C(!. Van Nostl'!tud, New York, 1920.
Bspe ane! Knoll, lVcrlc.~t()jJ"cmul£J del' I1ochvalcuurn-tcchnilc. Springer, Berlin, 193G.
Heproducod by Edw!mlH, Ann Arhor, Mich., 19M.
Farkas and Melvillo, RX]lO'l'imental 111 cthod.~ in (iliB BCl.lction.~. IVlacmilbu, Lon-
don, 193!).
.Tn!1lHLIlancla, Hiah Vacua. V!111 Nost,mncl, New Ymk, 1\)'17.
1(ayo, II iah Vacua. Longnutlls, Groen, London, 1027.
IClemeno, Die Bchandlun(J und Reindantell'ml(J lion Gasen. Almdem. VerlagR-
gesellschaft, Leipzig, lUBR. Hcproclucecl by Edwards, Ann Arhor, Mieh., 194:~.
Martin and Hill, VaCtlll1lL J>1'llctice, ]\1:elboUl'IlC Unlv. Pres::;, Melbourne, 1047,
MOllch, ValcuU'lntechnilc in Laboratol'ium. Wagner, Weiltl(Lr, 1937. Hepl'odllCocl
by Eldwarcls, Ann Arb(lr, Mich., 1944.
Sandel'~on, Vacu1l?n Manipulation of Volatile ComlJOunds. Wiley, New YDrk,
1948.
Strong, ct al., 1'1'occdll1'cS l:n JExpcl'irrwnlal Ph!Jsi(~.~. Prentice-Hall, New York,
1943.
Yarwood, lIiah Vctcnwn 'lTe(Jhniljuc. 2nd ed., Wiley, New York, 1045.
CHAPTER VII
SUBLIMATION
R. STUART TIPSON, Mellon Institllte uj Indllstrial Research
I. INTRODUCTION
When cl'ystah: vaporize without melting and the vapor, upon cooling,
condenses directly to crystals, the process is called sublimaHon. The ini.tial
solid is the sublimand, and the product is the sublimate. The procedure of
vaporizing a melt and directly condensing the vapor to crystals will be cal-
led q1lasisublimation. Theoretically, any compound which distils without
decomposition may also be sublimed at appropriate temperature and pres-
sure; however, sublimation may be immeasurably slow, even under opti-
mal conditions, if the vapor pressure of the crystals is very low.
An important diifel'ence between distillation and sublimation lies in the
fact that, in sublimation, vapor molecules reach the solid-gas interface
principally by a peeling-off of surface layers, whereas in distillation this is
supplemented to a much greater extent by diffusion and convection in the
liquid distilland. Moreover, sublimation as a fractionation procedure suf-
fers from the supposed impossibility of producing reflux. In distillation,
the liquid condensate can flow by gravity, and it is therefore possible to
bring about eountercurrent, contact distillation or a series of multiple re-
distillations. It appears that no satisfactory countercurrent, contacting,
vertical sublimator has been devised, presumably owing to the mcehanical
difficulties involved. Fraetional recrystallization by repeated single sub-
limatiolls may be compared with repeated fractional recrystallization from
a. solvent (see Volume III, Chapter VI, this sel'ies).
Direct crystallization from the gas phase may be much more efficient
than distillation; e.g., volatile impurities which would be dissolved in a
l'iquid condensate but which are not dissolved or appreciably adsorbed by
the crystalline sublimate can be removed. rrhus, separation of volatile
crystallizable eompounds from nonvolatile substances, ttIlcl from volatile
substances which do not condense under the conditions of the sublimation,
is often readily aecomplished. Separation of compounds which are of com-
parable vapor pressure is usually not readily achieved because of the neces-
sity for repeated single sublimations, and even this treatment may not
503
604 It. S. 'I'IPSON
suffice. When applieahle, sublimation may give a high yield of very pure,
crystalline pl'oduc:t, and the proeess is often rapid.
U. GI~NEUAL CONSIDERATIONS
The vapor pressure, p, of a pure heat-stable eompouncl increases with the
temperature in a(J(Jonbnee with the Clausius-Clapeyron equation: 1
dpldT = LI1'V
whcre L it:! the btent hoat of sublimation, V the difference in volume be-
tween Vltpor and solid, and T the absolute temperature.
In the phase diagmm (Fig. 1) the curve CP depicting the relation be-
tween tempemture Hnd vapor pres8ure,2 nt equilibrium between the solid
and its vapor, is known as the sublimation curve i its upper limit is the triple
point, P, nt whieh solid, liquid, and vapor eoexist in equilibrium. At P,
the Intent heat of sublimation 3 of the solid is equal to the h1tent heat of
fusion plus the latent heat of evaporation of the liquid.
On heating at atmoo;pheric pressure, a solid will eventually melt and then
boil if the vapor pressllre is less than 1 atmosphere at the melting pointj
it will sublime if the vapor pressure reaches 1. ntmosphere at a temperature
below the melting point. Thus, if X (Fig. 1.) is 760 mm., the crystals sub-
lime at Z. For example, the triple point of cyanogen iodide 4 is at 146 0
and 993 mm.; hence it may be readily sublimed nt 760 mm. Below the
triple-point prCfJl:mre, a substnnee is not stable in the liquid stnte. Curve
PY represents the behavior of the supereooled liquid.
Depending on the information available, the vapor pressure at any par~
ticular temperature, T, may be estimated by means of the Clausius-
Clttpeyron equa.tion or the following empirical formulas. In a univariant
system, the vapor pressure, p, is npproximatelyG given by
log p = 7.53 - (2.95T. + 4.59T k )/T
where 'l'k is the boiling point nnd 1'. the melting point, 01':6
1 Hume-Rothl!ry, Phil. MIL(j., 28, 465 (1!l39). Milosavl(\viu, C. A., 41, 4347, 5766;
Compt. rend., 224, 731, 1345 (11)47). Compare Elgin et al., Chemical Engineers' Hand-
boole, McGraw-Hill, New York, 1941, p. 624.
2 namS[LY and Young, Trans. Roy. Soc. London, 175,37 (1884). Striibin, Chem. App.,
16,139 (1929).
3 Wolf and Weghofcr, Z. phy.9ik. Chem., B39, 194 (1038). Dunken and Wolf, ibid.,
B3S, 441 (1938). Kit!trgorodskil, C. A.', 40, 6921); Acta Physicochim. U. R. S. S., 21, 379
(1946).
~
Ketelaar and Kruyer, Reo. trav. chim., 62, 550 (1943).
vall Liempt. Z. an()l'g. allgem. Chem., 111,280 (1920).
6
6 Duclaux, Compt. rend., 214, 78 (1942). Compare Verschaffelt, C.· A., 21, 1907;
Bull. classB aci. Acad. roy. Belg., 12, 641 (1926).
VII. SUBI,IMA'l'ION 605
(1/(3) log 1T = n log B + CO(O - 1)(1 + a,3 - 20)(20 +a - 1)-3
where 1T is ratio of vapor pressure at the t.emperature T to the vapor pres-

sure at the triple point; nand C are constants; /3 is ratio of heat of sub-
limation to heat of liquid vaporization at the triple point; B 5s ratio of
temperature T to the triple-point temperature; and a is [(2 Tc/T /) - 1 ],1.,
where To and T t are the critical and triple-point temperaturcs, respectively.
The tempcrature at which the vapor pressure equals 7HO mm. is often
referred to as the subliming point. It is as characteristic of a substance as
the melting point [md may be changed by the presence of another gas.
W
0:
:::>
I- B
<t
0:
W
0-
::;;
W
I-
LIQUID
m.p.
D
PRESSLJRE\mml-
Fig. 1. Effect of pressure on subliming point (CP),

melting point (PD), and boiling point (PB).
Unlike the melting point but like the boiling point it may vary enormously
with the pressure in the system. The temperature at which deposition
of sublimate becomes noticeable is often recorded in the literature. This may
not be the true subliming point, since under the conditions employed the
vapor pressure of the solid may never reach 760 mm. The value observed
may also depend on the apparatus used. For example, sublimation of in-
digo did not occur7 when the distance between sublimand and condensing
surface was 2.5 em., but when it was 0.01 to 0.1 mm. a sublimnte colleeted. g
The practical subliming point has been defined7 as the lowest temperature
at which a sublimate discernible under the microscope is obtained when the
substance is maintained for twenty minutes at that temperature in a stand-
ard, specified apparatus (see Fig. 6 and Table II), Ten minutes was
chosen9 as the time period in determining the subliming points of certain
7 H. Hoffmann and W, C. Johnson, J. Assoc. Offic. Agr. CherniRt,~, 13,367 (1930).

8 Kempf, Z. anal. Chem., 62, 284 (1923).
& J. W. Brown, Tran8. Roy. Soc. Can., III, 26, 173 (1932).
606 n. S. 'l'IPSON
amino ac:ids nnder vacuum. In a group of 700 organic: compounds, the

differcnee between the suhliming point, and the melting point increascd
linearly with increasc in the melting point.. 11I
In order t.o melt a compound whieh sublimes at atmm;pheric pressure) it
mnst be heated at eleva,ted pressure. For r,xampln, hnxar~hloroethanel1
melts at 187 0 • At 185 0 , it has a vapor pressure of 1 atmosphere. Hence,
it rettclily vaporizes without melting, and, on eooling, the vapor crystallizes
without first giving liquid (see Fig. 1). On heating under pressure, hexa-
chloroethane melts and distils. The vapor pressurcR of a number of com-
pounds at the melting point are given in Table 1. Measnrement of sub-
limation pressures follows the prineiples ~Llld toehui(l1leS of vapor-pressure
meaflurernents in geneml. 12
TAl~LIi] I
V.APOIll'Ill~SSUllDJ (n' Sn~n) OlWANIC CO~Il'()!TN])A .'>-'1' 'J'!·T-N MmUI'ING POIN'~'
MeltinlT, V 11pur pressure fl. t

Comp0I1IHl point.,oC. melting lJOint, nun. I-Ig
----
Toluene ...... , .. , ....... . -\)5 0.001
1J-Nitruhour.aldohy( It, .. . IOn ()'(ll
Benzoic achl ......... , ....... . 120 (i
Nnphtlmlmw ............. , .... . 70 7

Acdio acid ................. . \ll. t \)..1Ji
Hyuro(,[uinoue .............. . Hi\) H.l
Benzene ......... , ........... . 5.5 80
CalIlphor ........ , ........... . 179 370
Hl.)xaehl()n)l)tJll~Ill· ............. . 187 ' 800
Acotyleno. . . . . . . . , ............ . -8Ui BI2
Cyanug(JlI iDdi!lp ............... . Hfl 1)93
Carbon dioxido ................ . -li7 5.1 atm.
The heat of sublimation, i.e., the onergy required to overcome the co-
hesive forces of the crystal [md transform it to vapor, depends on the nature
of the compound. In general, each chemical group eontributes to the
inter'molecular forcos aeeording to its Ioeation in the moleeule, its polarity,
and its contribution to the van del' Waals attraction, particularly its abil-
l'I{ofier and Dos(!r, Die Chemie, 55, 13 (1042). L, Kotler lint! A. Kofler, M-ikro-
M elhoden ZU?· Kennzf!ichnun!l or(Jani.~cher Sto.{fe unCI' Sto.tJacmische, Univcl'sit[i.tsvedag
Wllgner, Innsbruck, 1948.
11 Staedei, Be)"., Il, 1735 (1878).
12 Coolidge, J. Am, Chem. Soc., 45, 1637 (1923); 46, 680 (1024). A. S. Coolidge
and M. S. Ooolidge, ibid., 49, 100 (1\)27). Compare Thomson, "Determiuution of Vapor
Pressure," in Weissberger, ed., Physicalll.Iethods oj Organic Chemistry, 2u(1 ed., Int.cr-
$[)ience, New York, Part I, 11)49, Chap. V.
VII. SUBLIMATION (\07
ity to form co-ordinate jinks, e.g., hydrogen bonds. A few Hverage valuesl~
are: -COOl-I, 8970; -OR, 7250; =CH 2 and -CRa, 1780; and
-CH2- , 990 calories. Energy supplied to overcome molecular cohesion
also increases the interna~ vibration in the molecule, and if, at any bond, the
vibrational energy exceeds the binding energy, the bond will break and the
molecule decompose,14 Theoretically, any substance can be sublimed 01'
distilled unchanged if the energy which the molecules must acquire to over-
come the cohesive forces and reach the condensing surface does not exceed
the heat of disruption of the least stable bond.
In order to obtain maximum velocity of sublimation the surface area of
the sllblimand should be as great as possible, i.e., the material should be
finely powde~·ed. 10 This may also reduce the rcta,l'ding action of less vola-
tile impurities and the danger that a dust of impure material, arising from
decrepitation 16 caused by the presence of mothe)' liquor 01' solvent of crys-
tallization, may reach the condensing surface. The sublimand should be
spread in a thin layer, except in entrainel' sublimations with thorough per-
meation by the entraincl'. It may well be agitated mechanically. The dis-
tance from sublimand surface to condenser surface should be shortY
Direct formation of crystals from the V!1pOr pbase depends on the proper-
ties of the compound, the presence of nuclei at the eondensing surface, the
pressure, and the temperature, of the sublimand and sublimate, respec-
tively. Simple symmetrical molecules like quinone, anthracene, and naph-
thalene readily yield beautifully crystalline sublimates. In sublimation,
as in crystal growth from solution, the number, shape, and size of the result-
ing crystals depend on the rate of nucleus formation 18 relative to that of
crystal growth and on the rate at which material is supplied. Hence, sub-
limation of pure crystals may be directed toward the preparation of many
small crystals or of a few large ones. In common practice, three kinds of
sublimate are e1\eountered: c:1ke, powder, and macro crystalline sublimates.
While some substances, e.g., menthol, benzoic acid, and mtphthalene,
13 Dunkel, Z. phydk. Cheln., 138, 42 (lU28). K. H. Meyer [mel Mark, Del' Auf/w(,
del' hochpolymeren 01'{jrLni8chcn Nailll'stojJe" Almdcl1l. VCl'lagsgescllschaft, Leipzig, 1930,
p. 32. J. W. Hill, Science, 76, 218 (1032). K. H. Meyer, Natural and Synthetic High
PoZ'!Jm.llrs, Intersuience, New York, HJ42; 2nd ed.,1950.
" McDonald, J. Franklin Inst., 221, 103 (11)30).
16 Hulett" Z. physik. Chem., 37, 385 (Hl01).
10 Richards, Z. physik. Chern., 46, IS\) (1\)03).
17 E:rafft and Dyes, Bel'., 28, 2583 (1895). Krufft and Weihmdt, ibid., 29, 1316, 2240
(1896).
18 Kaischew and Stranski, Z. physik. Chem., B26, 317 (11-)34). Lang;muil', Phys. Rev.,
2,329 (1!)13); 8,149 (1916); .T. Am. Chem. Soc., 38, 2221,2250 (1916); Tran8. Fara-
day Soc., 17, G07, 621 (1921-1922). Volmer et aZ., Z. Phy,~ik, 5, 31, 188 (1921); 7, 1,
13 (1921); z, phI/sa,. Chem., 102, 2!i7 (1\)22).
608 H. S. 'fIPSON
readily yield individual cl'ystals no matter what the conditions are, other
compounds arc <1iHicmJtl9 to obtain in this form.
The rnte of eondensation 2D obviously depends on the rate at which the
vapor is :mpplied to the condenser and on the temperature difference be-
tween the condenser walls and vapor. On striking; a solid surf~tee, a vapor
molecule may either condense or rebound. Under vaeuum, the ratio2l
of the number striking to the number condensing is close to unity for non-
polar molecules sueh us naphthalene, but for benzoin acid (dipole moment
0.8. X 10- 18 e.s.u.) it is pl'obiLbly less than 0.3, l1lul for camphor (dipole
moment 2.85 X 1O-.1~ e.s.u.) its value is only 0.17. The condensing mole-
culos mn.y not be immediately frozon into immobility, but probably move
more or less freely in a la,yer OIl the condensing surface for a short time.
The lowor .the condenser tempemture, the more rapidly will the particles
be arrested. COllsequently, to avoid the deposition of amorphous Of micro-
crystalline materinl, the tcmpemture of the eondellfling surface should not
be too low. The most perfect crystals will, ill general, result if the tempera-
ture of the cooling surface is only slightly below the melting point, 1:.e.,
the maximum compatible with direct condensatioll to crysbl11s. This is
importl1nt in the mi(lrosublimntion of compound~ to get discrete euhedral
crystals of measurable angles. 22 Slow growth is oonc1ucive to perfection of
shape. Formation of hard coatings may result if the condenser tempera-
ture is considerably below the melting point. Thus, water cooling, as com-
pared with air cooling, usually favors deposition of a hard coating of sub-
limate difficult to remove.
It is obvious thai, the thickness of the sublimate cO!tting on the walls,
and therefore the shape and relativ(\ size of the condenser, will ltffect the
character of the suhlimfLte. InstnllfLtion of blLffle~ impedes passage of the
vapor through the cundensing chamber and increases the condensing sur-
face. Brushes or swinging lUlImnm's arc employed industrially for removal
of deposits from the condensing surfaces. The introduction of a cold gas
into the condens[],tion chamber often cnutles rapid eondensation and deposi-
tion of snowlike sublimates.
HI. TYPES OF SUBLIMATION
1. Simple Sublimation
Simple SUblimation, involving diffw,ioll of the vapor to the condensing
surface at atmospheric pressure, is limited to substances with a relatively
Edel' and I-has, Mikroch,emie, Ernieh fi'eswchr., 43 (1930).
II
~G
Langmuir, Phys. Rev., 2, 329 (1913); J. Am. Chern. Soc., 35, 122 (1913). See
Miyamoto, Tran.s. Faraday Soc., 29, 794 (1933); Herzfeld, J. Ohern. Phys., 3, 319 (1935).
21 Alty, Proc. Roy. Soc. London, A161, 68 (1037).
22 Shead, Proc. Oklahoma Acad. Sci., 15,86 (1935); 16,87 (1936).
VII. SUBLIMATION 609
high vapor pressure. In order that direct crystallization from the vapor
may occur, the temperature and pressure at the condensing surface must lie
below the triple-point temperature and pressure.
2. Entrainer Sublimation
One way of bringing about sublimation of a substanee which, on heating,
would ordinarily melt and then boil, is to introduce an inert gas or vapor
as an entrainer, so that the partial pressure of the sublimand vapor is
brought below the triple-point pressurc and the sublimand vapor is swept
along with the entrainer. On 'suitable cooling, the vapor will then crystal-
lize directly, without first liquefying. Thus, rZ-camphor (triple point
179°C., 370 mm.) is readily sublimed below 179° in a current of air at at-
mospheric pressure. This entrainer technique was suggested by Liebig. 23
It is particularly useful for heat-sensitive compounds with fairly lolV vapor
pressure. With respect to volatile impurities, the function of the entrainer
is, in a sensc, similar to that of the solvent in recrystallization from solu-
tion.23a In such processes, when the entrainer is saturated and diffusion
in the solid state is not the rate-limiting step, the time t (hours) required
for transfer of weight W of volatile substance in the sample taken is: 24
t = [W X 760 X 22.4 X (1 + aT) l/vMR
where a is the thermal coefficient of expansion of gases, T the temperature
(°0.) in heating chamber, v the vapor tension (in mm.) of the substance
at TO, M is molecular mass of the substance, and R the rate of entrainer
flow (in liters per hour). The entrainer must be well distributed over the
sublimand surface or, preferably, permeate the finely divided powder.
The more intimately it comes in contact with the sublimand, the greater
the rate of vaporization up to a certain value.
The choice of the entrainer depends on the chemical and physical proper-
ties of the sublimand. Thus, air may be employed as the entl'ainer for
benzoic acid,25 phthalic anhydride, 01' naphthalene,26 which are inert to
oxygen. For the sublimation of salicylic acid, a 6% carbon dioxide phts
air mixture is used. Sublimation in steam may be employed for substa,nces,
23 Liebig, Ann., 101,49 (1857). Compare Jaeger, U. S. Pat. 1,852,782 (1932); James,
Chemistry & Industry, No. 34, 595 (Aug. 20, 1949).
23a Sec Volume Ill, Chapter VI, this series.
24 Fuller, Chemis("'Analyst, 29, 6 (1919).
23 Cole, C. A., 17, 1483; U. S. Pat. 1,445,870.

26 Societe pour l'exploitation des Procedes Abderhalden, C. A., 23, 3034; French
Pat. 649,974 (1928). See also, Jackson, C. A., 17,1355; U. S. Pat. 1,446,564; Tim·
brol Pty., Ltd., C. A., 33, 5710; Australian Pat. 107,040 (1939).
610 R. S. 'l'IPSON
e.g" ,B-naphthol, camphor, benzanthroIlc,27 or anthracene, which melt above

100° and are practically insoluble in and not decomposed by water. A
dry sublil1,late may be obtained directly with water vapor t1l; t1tmosphel'ic
pressure as the entrainer, if the condenser is kept above 100°. The same
prinejple may be employed with other entraincrs.
The temperatUl'e of the entering entrn,iner should be close 1;0 the vapori-
zation temperature of the sublimand. If the entminer temperature is
lower, the vapor is cooled and premat,ul'c condensation results. If the
entrainel' temperature is much higher, the vapor bceomes superheated
and the advantages of low-temperature vaporization of eompounds un-
stable to heat may he lost. In practiee, sliuht28 superheating, to prevent
premature crystallization before the vapor rcnches the condenser, is ad-
visable. In the snblimation of salicylic :1oicl, maximltl dispersion of the
powdered sublinmnd is ensurcd 2iJ by whirling it with an inert gas, e.g., air,
nitrogen, or carbon dioxide, to form a doucl; this is then treated with ad·
ditiollHl inert. gas hot enough to effect sublimation. Alternatively, the
powder may be directly treated with hot eni;miIlCl· 311 so thnti the substaIlce
is volatilized while it is being suspended.
Entminer sublimation usually givcs feathery flakes, tL fine light crystal.
line powder, or fl, fluffy snowlike sublimate ("flowers"). Crystallization
occurs in the vapor space of t.he eondenser rather than on it.s walls because
the entraincl' transfers heat from the vnpol' to the walls, and tends to whirl
up crystallization Iludai. The greater the percentage of entrainer, the
greater is the likelihood of "snow" formation.
3. Vacuum Sublimation
Sublimation of hent-sensitive compounds fl'eqnently heeomes possible
if the pressure in t.he system is reduced, and vaeuum sublimation is often
the most convenient; method for isolating or purifying an organic compound.
It often takes but a few minutes to libera.te a volatile eompound from non-
volatile substances which may otherwise be difficult to remove. Although
oxalic acid tends to deeompose to formic add and carbon dioxide on sub·
limation at 760 mm., it is readily sublimed 31 at 10 mm. The conditions of
molecular clistilhttion-a high vacuum in whieh the distance from sub-
limand to condenser is less than the mean free path of the vaporizing mole~
eules-often give excellent results with compounds of low vapor pressure.
The theory and equipment for sublimation in a high Vacuum are closely
27 Lyford, C. .'1.,22,1255; U. S. Pat. 1,662,070.
28 Field, C. A., 22, 1366; U. S. Pat" 1,662,056.
.. Comte, C. A., 29,1.438; U. B. Pu.t. 1,987,282.
20 Livingston, C. A., 29, 1438; U, S. Pl\t. 1,987,30l.
U G. Klein !lnd O. Werner, Z.phllsiol. Cheln., 143,141 (1925).
VII. SUBJ,IMA'l'ION 611
related to those for molecular distillation (sec Chapter VI). Quinine,a!

morphine,32 maleic acid,83 and glucosea 4 may be sublimed under moderately
TABLE II
EFFECT ON SUBLIMATION POIN'!' CAUSED BY DIMINISHING THE PRESSURE"
RATE OF SUBLIMATION
Ra.te of sublimation .. t
0.5 to 1.0 mm .•
SublimlLtion temp .. Subli·
'C., DbB. mand Sublimate
:M.P .• temp.) 'l'illle, Yield,
Compound 'C. At 700 mm. At 0.5-1 JUm. °C. hr. %
Naphthalene ........... 79 36-38 25 50 0.5 86.2
Iodoform .............. 119 43-45 30-34 75 0.5 96.7
Caffeine (anhyd.) ....... 233 72-74 36-39 150 0.5 99.9
Theobromine ........... 348 146-149 110-114 210 0.5 99.8
Benzoic acid ............ 120 43-45 25 80 0.75 99.9
Hexamethylene-
tetramine .......... " 45-47 25 90 0.75 99.3
Quinine (anhydrous) .... 175 157-160 99-103 165 1.0 99.3
!3-Naphthol. ........... 122 43-45 33-35 75 l.25 99.6
Saccharin .............. 224 84-86 59-53 150 l.6 99.9
Acetanilide ............. 113 66-58 34-36 70 2.25 99.8
Cinnamic acid .... 132 58-60 62-56 90 2.25 99.7
DL-oAlanine ............. 295 136-137 59-63 180 2.25 99.6
Phthlllic anhydride ...... 129 60-52 27-30 80 2.6 99.5
Coumarin .............. 68 40-42 30-33 50 2.75 100
Urea .................. 132 59-61 49-52 95 3.25 99.2
Barbital ............... 188 66-68 43-46 115 3.25 99.6
Vanillin ................ 80 47-49 33-35 55 4.0 99.5
Anthracene ............. 215 77-79 28-31 100 5.0 99.1
Cholesterol ............. 145 No sub. 38-40 130 7.0 99.5
Alizarin .. , ............. 285 71-73 34-38 180 9.0 99.1
Isatin ................. 200 78-80 47-50 110 10.0 99.7
Cinchonine ............. 260 164-167 82-86 170 19.25 99.6
Acetyisalicylie acid ...... 135 77-80 52-55 105 2l.0 99.7
Cocaine .. , ............. 96 No sub. 48-51 85 35.0 99.6
Atropine ............... 114 No sub. 60-04 95 42.0 99.2
• Based on Hoffmann and .Johnson, J. Assoc. OfJir.. A!71'. Chemists, 13, 367 (1930),
Compare lUari, C. A., 25, 2880; Ann. chim. applicata, 21, 127 (1931); Edel', Schweiz:.
Wochschr., 51, 228, 241, 253 (1913); Hortvet, J. Assoc. Offic. AgT. Chemists, 6, 481
(1923); Hubacher, Ind. Eng. Chern., Anal. Ed., 15, 448 (1943).
b Coudenser temperat.ure about 20°,
high vacuum without decomposition, and dipcutaerythl'ito}35 is best puri-

fied in this way. Lauric acid sublimes~8 at 22°, myristic acid at 27°, paI~
.. Krafft aud Weilandt, Ber., 29, 2240 (1896) .
.. Hansen, Ber., 42, 210 (1909).
34 J. W. Hill, Science, 76, 218 (1932).
36 Ebert, BeT., 64, 114 (1931).
(j12 R. S. ~'n'SoN
mitic acid at 32°, stearic acid at 38°, alizarin 36 at 45°, and strychnine 36 at
103°. Table II shows the effect of diminished pressure 0.11 the sublimation
point of a number of compounds. Most. of the a-amine, acids are subli-
mable 37 under high vaCHum, without decomposition. The preparation of
t.hin films of such paIn,!, compound~ as purines, pYl'imi,lines, and amino
acids by sublimation onto quartz slides in a molecular still is used 38 to pre-
pare samples for the st.udy of their ultraviolet absorption\pectra.
Compounds which distil at atmospherie pressure may be sublimed under
suitable vacuum. Naphthalene, which melts at 79 0 and h~ts It boiling point
of 218 0 at 760 mm., sublimes mpidly at 70° at. 13 millimeters. The vapor
pressure of i3-naphthol at its melt.ing point (122°) is 2.5 mm.; hence, for
sublimation, t.he system is evacuat.ed to a pressure belcinv ~.5 millimeters.
If simple vacuum sublimation of compounds with H low vapor pressure
is slow, judicious admission of an entrainB'1' may speed up t.he pt'Ocess, just
as it speeds up simple sublimation. Introduction of entl'ainer will tend to
increase the pl'essure in the system.
4. Quasisublimation
Some compounds such as camphene hydrate 3D htwe been first obtained
in cryst.alline form only by crystallization from the vapor derived from the
melt. Presumably, nonvolatile impurities are left behind during the treat-
ment, and volatile impurities, retarding erystallization from the melt or
from a solvent, do not hinder erystallization from the vapor phase. De-
pending on the chemical and physical properties of the compound, any of
the above-ment.ioned methods used for sublimation may prove suceessful.
It is a common industrial procedure to melt the crystals, vaporize the
liquid, and recrystallize t.he compound directly from the vapor phase.
When a subst,ance with H. fairly high triple-point pressure, such as camphor
(triple point 179°C., 370 mm.), is heated in a vertieal sublimator-for
inst.ance, a basin covered wit.h [l cold glass funnel--the heavy vapor lies on
the SUbstance, the partial pressure over the compound rises to the triple-
point pressure, and it melt.s and eventually boils. However, if the hot
vapor near the condensing surftwe is diluted with hot air so that the partial
pressure remains below the t.riple-point pressure, the vapor condenses di~
rectly to erystals on the condenser. Similarly, on boiling certain substances
in a ret.ort, sublimate is observed above the vapor during the period while
vapor is still rising and some air remains in. the retort. As soon as the air
3& Kempf, Z. anal. Chem., 62, 284 (Hl2:1).
O. WArner, Mikroohemie. 1,33 (1923).
37
38.T. F. Soott, Sinsheimcr, and Loofbburow, Scie.nce, lQ7, 302 (1948). Comparo
Christ, Button, and Botty, ibid., 108,91 (1948).
3' Ascha.n, Ber., 41, 1092 (1908).
VII. SUBLIMA'l'ION 613
has been expelled, the vapor condenses as liquid in the neck of the retort.
However, if the vapor is passed into a large chamber containing hot air,
sublimate again \forms until the point is reached at whieh the partial pres-
sure of the vapor exceeds the triple-point pressure, when it aglLin condenses
as liquid. In Ol;her that direct condensation to crystals may be accom-
plished, it is necessary that the partial pressure of the condensing vapor be
reduced below tte triple-point pressure. This may be brought about by
diluting the vape:r. Except that the material is first melted, the procedure
is similar to entrdiner sublimation (see Sect. IH2). A process eombining
entrainel' quasisublimation with rapid condensation in the presence of addi-
tional cold entn~iner has been descl'ibed 40 for compounds whose vapor
pressure in the se,Jid
,. state is less than 10 mm. (see Sect. IV2B).
IV. DESIGN AND OPERATION OF SUBLIMA.TORS
A sublimator is essentially a chamber in which a temperature difference

can be maintained between t,wo surfaces. The sublimancl is placed on the
hotter surface, and the sublimate collected on the cooler surface, which may
be situated above the vaporizing surface, to one .side, or below it. The main
advantage of the horizontal sublimator is that, with proper use, sublimate
cannot possibly return to the sublimand vessel. The kind of apparatus
chosen is determined by such factors as (1) the stability of the sublimand on
heating; (2) its vapor pressure and ease of vaporization; (3) the size of
the sublirnand sample; U) the relative importance of obtaining the maxi-
mum possible yield of sublimate; and (5) the physical form of sublimate
desired. Certn,in pieces of equipment designed for distillation can be used,
with little or no modification, for sublimation; in this category are some of
the molecular stills (see Chapter VI).
The sublimand may be heated 4L by any of the methods described in
Chapter II of the present volume and Chapter I, i'Heating and Cooling,"
in Volume III of this series. Care should be taken that the temperature is
not too high in relation to the pressure in the system. Especially under
high vacuum, the sublimand may exhibit a behavior reminiscent of a boiling
liquid, viz., a dancing movement of the individual particles 42 or a vibration
of the whole sublimand mass,43 because it has been so heated that the vapor
.0 Robertson, J. Chem. Education, 9, 1713 (1032). Alzikovieh, C. A., 42, 3744; .T.
Applied Chem., U. S. S. H.., 20, 460 (U)47).
41 Krafft and Bergfeld, Ber., 38, 254 (1905). McDonald,.T. Franklin Inst., 221, 103
(1936). E. Beckmann, Nat1t7''II!issenschaften, 9,305 (1921). Kempf, Z. anal. Chem., 62,
284 (1923). 1(l\r1'er and Rosenberg, He7v. Chim. Acta, 5, 575 (1()22). Hubacher, Ind.
Eng. Chem., A.nal. Ed., 15,448 (1943).
,2 Krafft and Dyes, Ber., 28,2583 (1895). Arctowski, Z. anorg. Chern., 12, 417 (1896).
{8 Spring, Z. physik. Ch~m., 15, 65 (1894).
()14 H. S. T1PSON
pressure of the compound locally exceeds the pressure existing in the ap-
paratus. To avoid tmnsfer of dust of crude Ruhlimancl to the condenser
surface, !1. plug of glass wool 01' other porous merlinm may he inserted he-
tween the vtLporizing and condensing surfaces.
Cooling devices will be considered individually, since the choice depends
on sneh faetol's as the physical form desired for the f·mblimate.
1. Vertical Sublimators
A. SIMPLE SUBLIMA'l'ION
Air-Cooled Condenser. Perhaps the simplest method H of subliming

a small sample is to place a thin layer on a glass plate resting on 11,11 as-
bestos sheet heated with a microflame. A second glass plate, which acts
as condenser, is tilted 0.5 to 1.5 mm. nbovc the sublimand by mea,ns of a sup-
port, 3 to 4 mm. high, inserted uncleI' one (mel. In determining the mini-
mum temperature at which a visible sublimate results, it has been recom-
mended 4fi that the sublimand be heated electrically for dose temperature
control und that the condensing surface be placed 0.01 to 0.1 mm. from the
sublimand surface. For sublimation in an inert gas or in vactw, tho whole
(Lpparatus may be enclosed. As it rule, the sublimlttion is better conducted
in all ellclosecl space. 1'hus, one may plaeo the sublimand on it watch
glass, eover it with another watch gluss, and heat the lower glass gradually. ~o
The two glasses may be of the same size, ground to {it eaeh other at the
edges, and held together by mean::l of a metal clamp.17 A fIlter paper may
be placed between the glasses to catch falling sublimate.
One of the first uses of microsublimation 4B was the systematic examina-
. tion and identification of such alkaloids aH stryehnine, morphine, and atro-
pine. A little of the material under study was placed in a depression in
a platinum foil, covered with a glass slip, and then cautiously heated. The
apparatus was later impl'oved 49 hy using a sublimation eell eomposed of a
H Oddo, Gaz. chim. Unl., 23, 313 (lSfl3). 'I'UJllIlttIlll, Schwciz. WochNchr. Pharm., 48,
749 (1910); C. A.. , 5,2095; 6, 914, 2529; A.polh.-Zl{]., 26, 3tH, 812 (1911); 2'1,494,507,
515 (1912); C. A., 6, 910; Bel'. plum!!. Ge,q., 21, :312 (HJ12).
'5I{empf, Z. anal. Chern., 62, 284 (1023).
46 Gorup~Besancz, Ann., 93, 265 (1855).
41~I{olbe, llandwol'terbuch del' reinen wul anaewandten Chemie. Vieweg, Bl'tIUllsch-

weig, 1850, Suppl., JAg. 1-4, p. 425. .
~8 Helwig, Das J.")ili/croslcop in del' 'l'oxikologle. Beitraye zur mikroskoJiischen lmd mikro-
chemi8chen Diagnostfk, Zabel'll, Mainz, 18C5. Cornplll'(j Pfeil, An{]etv. Chem., 54, 161
(1941 ).
40 Guy, Phann, J., viii, 711) (18tiC-1867); ix, 10, fiR, 106, 195, 370 (1867-1868). Wad-
dington, ibid., be, 40H (1867-1868). Blyt.h, J. Chem. Soc., 33, 313 (1878); Ber., 11, 996
(1878). Liidy, C. A., 27, 2; Pharm. Zentralhalle, '13, 209 (1932).
615
glass ring supported by a glass disc and coverecl with a.noth.er gla.ss disc or,
preferably, a microscope slide, so that the sublimation might be observed
under the microscope. 5o The cell "'as placed on a brass plate bored to ac-
cept a thermometer and heated with a microfhme or electrically. The first
fractional micl'Osublimation,51 of a mixture of alizarin, flavopurpurin, and
isopurpurin, was conducted in such a cell having !t lead ring. An asbestos
ring was used in subliming caffeine from an alcoholic-potash extract of tea
leaves. 52 To diminish the chance of crystallization on the walls, a brass
ring, 1 cm. in diameter and 1 mm. high, is recoIDmended53 for compounds
which do not attack the alloy. In another design,54 cylindrical wells of
various depths and diameters are driiled in the top of the metal heating
block; sublirnand may be placed in each and covered with a glass slide.
A modification known as the "apophorometer"ii" consist$ of an electrically
heated platinum ribbon on which the sub lim and is pltwed; this is encased
in two watch glasses ground to fit. The whole may be enclosed in another
vessel for sublimation in an inert gas or under diminished pressure. A
rather similar apparatus Gn has a built-in, circular thermometer for measur-
ing the temperature of the sublimand. Certain organic dyes may be de-
tected by microsublimation57 in an apparatus of this kind.
For work on a somewhat larger scale,5R the lower vessel may be a porce-
lain dish, und a funnel or beaker may be employed as the condensing sur-
face. A covered porcelain crucible, heated in a sand lmth, has also been
used.6~ If the sublimate shows a tendency to fall off the condenser into the
sublimand, a sheet of paper or cardboard, punctured with pinholes, should
be placcd between the lower and upper vessels. A modification60 of this
device employs a steam coil for heating a moist sublimand, and a dry subli-
mn,te results.
Larger amounts nmy sometimes'be sublimed by spreading the substance
50 Kofler, Plwl'ln. Monalsh., 13,81 (1932); C. A., 26, 3871; Arch. Pharm., 270,298
(1932). Schiirhoff, C. i1., 26, 4507; Arch. PhCLTIIl., 270,363 (l!J32). Deininger, Pharll!.
Ztg., 78, 31)2 (193'3). Kofler and Doser, Die Chemic, 55, 13 (194-2).
61 SChUIlCk and Roemer, Ber., 13,41 (1880).
62 Kley, Rec. tmv. chim., 20, 344 (1901).
63 Shead, Proc. Oklahoma Acad. Sci., 16, 87 (1936). Rosenthalcl', C. A., 27, 451;
A.polh. ZIg., 47, 1358 (932) .
• , H.. Fischel', C. A., 32, 4276; Dcut. Apoth.-Ztll., 53, 361 (1938).
65 Joly, PI·OO. Roy. Irish Acad., 2, 38 (1891); Phil. Mag., 25, 301 (UJ13); 27, 1
(1914); Chem. News, 107,241 (1913). Fletcher, C. A., '1, 3580; Sci. Proc. Roy, Dublin
Soc., 13, 460 (1913).
50 L. Fuchs, M ikrochim. Acta, 2, 317 (1937).
57 ICutzelnigg, Mikrochim. Ada, 3, 33 (W38).
68 F. Michel, Chem.-Ztg., 36, 138 (1912).
r;~ O. Fischer and B:epp, Ber., 22, ?57 (1889).
60 F. C. Koehler, Chem.-Ztg., 39, 122 (1915).
616 R. S. '.rIPSON
evenly on the bottJflln of an Erlenmeyer flask, loosely stoppering it, and

immersing it to a depth of an inch in 11 heat.ecl oil bath. Salts of organic
buses,O! C.Il., N-methyl-a-pyridone hydroehlol'icle, have tllUt-l been sublimed.
A test tube may be used in It similar way. To avoid eontaminntion of sub-
Iiml1te by residual sublimancl during withdrawal at the end of the sublima·
tion, !L snugly fitting seeoncl teKt tube 62 of slightly smaller eliameter, whose
bottom has been ent off (Fig. 2), may be inserted. The sublimate is de-
posited on the w[Llh, of the inner tube. Another solution to this problem is
the 118e63 of a ground-glass joint as showll in Figure 2. This tube is l'eeOIll-
mended for use in the determination of
benzoie neid by sublimation.
Liquid-Cooled Condenser. If the
temperature of the condensing surface
might rise to the melting point of the
sublimate, it becomes necessary to cool
it. Cooling is nJso indic[Lted for quan-
titative experiments and for sublima-
tions Itt room temperature. For micro
WOl'k,64 it may be sufficient to put a drop
of water or 11 damp cloth on a flat glass
plate or watch glass used as the con-
densing surface. Xanthine bases were
thus readily sublimed from plant ma-
terials. Pieces of dry ice mlLy also prove
satisfactory as coolant.
Fig. 2. Simpievcl'ticnlBublim!1tors. G2,03 ]'01' prolonged Bublimations and for
those on a macro scale, a stream of
cold liquid65 is preferable for cooling. In 1111 carly device,oo a ICcolcl"finger"
condenser is inscrted through the neck of 11 eonicnl funnel which rests on
the rim of a vessel containing the sublimalld. The rim i-lerves to collect
. liquid condens!1te, and a porous plato resting on the dish catches any dis-
lodged sublimate. Depending on the sublirnand, a seal between rim and
61Decker, J. pl'akt. Chem., ISS, 28, 222 (lSH3).

62Llewellyn, Chem. News, 97, 108 (lU08).
63IIcllville, Analyst, 61, 104 (1936).
6. Nestler, Z. Unlm·such. Nahr. 1L. GenU881n., 4, 2S0 (11101); 5,24,5 (1902)j 6,408
(1903); Bel'. delLI. bolan. GI18., 19, 350 (WOI). Fmnk, Z. Unlersuch. Nahr. u. Genu88m.,
6, 880 (1903).
6& IJandolt, Ber., 18, 5G (1885). R Philippo, C. 11., 6, :.\:342; 8, IOn; 10,1059; Mitt.
Lebe1~sm. JlVg., 3, 41 (1912); 4,!3M (lOla); 6,177,2:1:3 (HlI5)j 7,:37 (HlW). Milller·
Iloessly, Mill. Le/)ensln.li1l(J., 6,251 (1015). Villi Zijp, Pharm. Weekblad, 64,916 (1927).
J!1cqucmain, Bull. soc. chim. France, 53, 638 (10:33).
66 Hertkorn, Chem.-Zlg., 16,795 (1892). Nicoiaysen, ibid., 25, 1031 (1901).
VII. SUBLIMA'fION 617
Fig. 3. VertimLl suhlimatol' with Iiquid- Fig ..f. Vertical sui1Iima.tnj' with liquid-
cooled condenser.07 enoled enndens(~l'. tl'
Fig. 5. Vertic[\1 sublimator with liquid- Fig. 6. Vertic[\l sublim[\tor with liquid-
cooled condenser. 60 cooled condenser.71
618 H. S. 'l'!PSON
funnel may be provided by mercury, pttmHin wax, or plaster of Paris. The

appttl'atlls iR readily modified for entminer or vacuum sublimation. A
simpler fonn 67 is depioted in Figure 3. If a tesl; tuhe iH URGd as the sub-
.Ii'ig. 7. Vcr Lienl tlntmincl' suhlillWtOI'.71' 1<'i.l(. 8. Vert.ieal ent.raincl' sublinllLtol'.12
Fig. 9. Vel·tical cntrainer sulllimator.73 Fig. 10. Vertical entraincr subliml1tor.H

67 Morton, Laboratory Technique in Ol'ga.n:ic Chemistry. MClGraw-Hill, Ncw YOl'k,
1938, p. 215, Fig. lIt).
liming vessel (Fig. 4), three glass pl'ongs68 may be sealed onto the tubular
condenser to prevent contact with the walls of the tube on withdrawal.
For quantitative work with small amounts, the sublimate is washed off
the condenser, the solution evaporated to dryness, and the residue weighed.
:For direct weighing of the sublimate,60 the apparatus shown in Figure 5 was
developed. The condenser consists of a spiral tube which, after the subli-
mation, is transferred to the weighing tube for protection of the sublimate.
The inside of the coil is now dried, a.nd the assemblage weighed. It is use-
ful for determination of caffeine in coffee and tea, and also for benzoic acid
estimations. Another device llses a condenser having at the bottom a
concave face corresponding exactly to the curvature of an adhering watch
glass on which the sublimate collects. 7o If a cake-type of sublimate results,
the watch glass plus deposit may be removed at the end of the sublimation
and weighed. An appal'atus71 employing a microscope slide as condcnsing
surface is shown in Figure 6.
B. ENTRAINER SUBLIMA'l'ION
Air-Cooled Condenser. A simple apparatus for entrainer sublima.tion

is depicted in Figure 7. In a device (Fig. 8) used for the sublirnation72
of anthracene in a current of air, the snblimand is heated and its vapor is
carried by the ail' current into the funnel, where it crystallizes. (3-Naph-
thol was sublimed 73 in the apparatus shown in Figure 9 at the rate of almost
1 g. per minute. It"was heated until almost molten and a current of ail'
passed over its surface either by air pressure on the flask side tube 01' by suc-
tion at the upper end of the condensing tube. The latter was plugged with
cotton to minimize loss of sublimate.
The apparatus shown in Figure 10 was designed 70l for the sublimation of
compounds difficult to vaporize. A vigorous stream of entrainer is passed
through the tube and over the heated sllblimand. The sublimate collects
on the inner walls of the funnel aml between two sheets of paper. The
lower of these is supported on a glass tripod, the upper between beaker and
funnel.
68 von Fellenbel'g, C. A., 26, 4733; Mitt. Lcbensm. H1fg., 23, 97 (1932).
69 Exner, .T. ilssoc. a.flic. Agr. Chemists, 1, 208 (1915).
70 Benveguin, C. il., 21, 1033; Mitt. Lebensm. Hyg., 17, 315 (1926). Compare E.
Philippe, Mitt. LelwnslIl. Hyg., 3, 41 (1\)12).
71 H. Hoffmann, .Jr., ami W. C .•Tohnson, .T. Assoc. a.ffic. Agr. Cherni.~ls, 13,367 (1930).
71e E. H. Northey, private communication. Comp:1l'e Fieser, Experiments in Organic
Chemistry, Heath, New York, 19:35, p. 232.
72 Robertson und Deakcrs, J. Chem. Education, 9, 1717 (1932).
73 A. J. Blliley, Ind. Eng. Chern., Anal. Ed., 12, 194 (11)40).
H Baeyer lIlla li'ru.ude, Ann., 202, 164 (1880).
620 n. s. 'rJl'SON
Liquid-Cooled Condenser. Devices of this kind arc not shown since

they differ from those jUHt described only by the preHonee of 11 liqllid-
cooled conden.gel' (see Sect. IVIA) above the sublimanrl surface.
C. VACUUM SUBLIMA'l'ION
The techniques Hnd equipment for producing moderato and high vacuums
are discussed in Chapters V and VI.
Air-Cooled Condenser. Probably the simplest apparatus consiHts of
a flask, containing the sllblim1Llld, which is ev::wulitcd and scaled and
then gradually heated wit.h a free flfL1lle. The method appears t.o have
been first; employerPfi for t.he sublimation of indigo n.t 30 to 40 millimeters.
Urea was similarly sublimecl,7" Another simple device is a vertical t.ube,
heated in an oil hath or in a well bored in a metal blnck,77 and conneeted to
a vacuum pump. The sublimate colleets on the cooler wnlls of the tube,
but is oHon difficult to remove. If Imnging deposits are formed, they may
be dislodged by slight vibration ~1Ild ,viII fall back onto the residue. A
loose asbestos pIng may therefore be inserted in the tube so that it is a few
millimeters inside the hoating block. Owing to the mueh gre:.Lter voln.tility
of certain nitrilcs, t.hey were readily sop::trablc from the rcln.tecl (tmidcs in
this appamtuH at. 0.1 to 1 mm. preHHuro, at tmnp(1l'lLtures near the melting
points.
So that the sublimat.e might be 1'1l!tdily aeeesHible after the sublimation,
it ground-glass joint has been tlse(l to (lonneet. the eondOllSOl' with the
vaporizer. To eatch falling deposits of sublimate, a porous asbestos plate
was interposed between it and the sublimand (Fig. 11). It may be omitted
if the sublim::t1;c collects as H hard deposit. The sublimand, plate, and
ground joint were llcated in an nil' bath. The advantage of heat.ing the
joint IS that if lcnlmge, either acciclentiLI or deliberate (see Sect. IVID),
occurs there, the entming enLmineI' is at the same temperature as the subli-
mtmd. In this appamtus,78 sueh compounds as indigo, and mono~ and
dibromoqllinizarin,7~ which by other methods sublime with difficulty and
considerable loss, were readily sublimed. An apparatus of 25-mm. internal
diameter suffices for the sublimat.ion of 1 to 4 g.; one GO mm. wide was used
for the sublimation of 1B g. of indigo in 3 hours. The device is also useful
for fmctioMl sublimations of compounds of different volatility, and for
determination of the sublimation temperature. The color of the vapor and
7. von Sommnruga, :Ann., 195, 302 (1879).
76 Bourgeois, Bull. 800. chim., 7, 45 (1892).
77 Ashley et al., J. Chem. SOD., 103 (1942).
78 Riiber, Her., 33,1(155 (HlOO). Compare Gettler, Umberger, and Goldbaum, Anal.
Chern., 22, 600 (1950). I '
79 Liebermann and Eiiber, Ber., 33, 1658 (1900).
Vll. ~UBLIMA'l'lON 621
its spectroscopic ab~orption may be readily ascertained. 11'or quantitative

sublimations, the two parts of the apparatus may be weighed before and
after a sublimation. Incidentally, the appara,tus is also u~eful for the de-
terminat.ion of solvent of cryst.allization.
In high-vacuum work, the ground joint must be greased and preferably
not heated. It should therefore be placed80 near the top of the tube, where
it is unlikely to become clogged with sublimate. This apparatus is useful
c-_====
-'~.'
--.~.......
.. .
}
I
~~-.~"-'--.:.,~,,-.
,,_-'''-
--
Vig. 11. Vertical vaCUUlll Pig. 12. Vertical vacuum

sublimator.78 mi crosubJimator.B'
if two sublimates are expected. For example, on subliming !t mixture of

benzoic acid plus o-benzoylbcnzoic acid, each formed a separate band on the
cool part of the tube and could be scraped out separately. Quantitative
separation of certain types of compound could be made at 10 j.(- provided
the temperature. of the sublimand was very accurately controlled (see Sect.
V). A thermostatically controlled heating block, drilled t,o accept tubes
of several different diameters, was therefore designed.
Although vacuum microsublimatiOll was first employed many years ago,Sl
gO Hubachel', Ind, Eng. Chern., final. Ed., 15,448 (1943).
81 Rosenthaler, Ber. pharm. Ges., 21, 338, 525 (1911). Compare 'l'unrnalln, ibid., 21,
312 (1911); Joly, Phil. Mag., 25, 301 (1913).
()22 R. s. ~rIl'S(JN
the sublimate was not collected on a microscope slide and was therefore not
always suitable for mjcroscopic examination. In order to overcome this
problem, a microscope slide wai:! bent ati right angles neal' each end,82 to form
legs, and placed over tihe suhlimalld in H horizontal tube, whieh was then
evacuated and heated in the under section of n modified Pregl heating
block. The sul>limate iihon collected on the under side of the microscope
slide. By providillg a suitable opening nenl' the sublimttlld end of the tllbe,
provision could be made for admission of 11 slow :,;tl'eam of dry cntrainel'.
In a Homewhat similar nppnratus,R3 :mblimatei:-l eollcotecl on a mieroscope
Hlide held 0.1 to (l.Ol mm. above the sllrbc() of the subJimand, which is
hetttecl elociil'ienlly. It may be used under va.euum or with an atmosphere
of in(1]'t gaR. An exeellent deviee s.j for the Hublimations'l(t of OJ) mg. of mtt-
L()rial onto a slide or round COVOI' gla::;s is tImt HhoWll in Figul'(~ 12.
Liquid-Cooled Condenser. IiJxtl!l'nal Cool'ina. Vertical, vacuum sub-
limlLtion with external eooling8" of the eoncienHing surfaec may be conducted
by placing the :mblimnnd in n small teR'(i tlllle whidt is covered with a piece
of parchment paper or a porous plato and then inserted in a large test
tube. This is evac\lat,(~d, and heated below while being eoolnd above.
Tho npparatus mny he used eithol' for micro- or mael'o8ublimntions. If
the evacuating Lube is rcplac:ecl by a drying tube, it may also be employed
for sublimn,tion at n.tmospherie preHslll'e. Hda;ted deviees,86 in which the
sublimate may be eoUocted on a cover git1RS 01' may be prevented by means
of a wire serGen from falling back into the crude sublimand, lULVe been de-
~cribocl.
Another form for direct sublimatioll onto a miero:-lcopo cover glass con-
sislisB7 of an elcctrically heated, thermostntieally controlled metal cylinder
on whieh is plu,eed a glass ring eOllt~tining the ~\Ublimancl, H, glass slip, and a
wiLter-cooled metal eylindol'. By ehanging the gInSi:! slip periodically,
fractional miel'osuhlimatioll may he ohttLined.
Internal Cooll:n(J. One of the simplest devices for vacuum sublimation
cOllsists88 of l), test tube cooled, for oX[Lmple, with running wat~r and placed
82 Schoellcr, Z. angll'Ill. Chem., 35, 500 (1022).
83l{empf, Z. anal. Chern., 62, 284 (11)23). [(oncr fmcl Dombach, Mikrochemie, 9,
345 (1931), II.. Fischer, ibid., 15,247 (1\)34).
Edcr, C, A" 7, 2832; Sahwciz. lVol!hschr., 51, 228, 24:1, 253 (1013).
84
8~"
Erler, Arch. Pharm., 253,14,17 (1015). Wltgollllal', C. A., 21, (}75; Phal'ln. Week-
iliad. 64, 10 (1027). Eclet· and IIlIas, Mikl'Ochemic, Emich Festschr., 43 (1930). Eder,
C. A., 8, 398; l'iel'teljahrs.~chr. n(l.tur!01,.~ch. Ge.q. Z111'ich, 57, 201 (1012); C. A., 6, 1204;
Apolh.-Ztg., 26, 831 (lUll).
a. Sckworzow, Z. angew, Chem., 20, 109 (1907).
86 Viehoovol', J. A.sSQc. Offic. A(J1". Chemists, 6,473 (1923).
87 B. L. ClarkOltnd HOl'mu,nce, Ind. Bng. Chem., Anal. Ed" 11,50 (1939).
88 Nicolnys!lll, Chem.~Zl(!., 25, 1031 (1901). Bl'euseh, Z. physiol. Chem., 22'7, 242
(1934).
VII. SUBLIMA'I'ION 623
inside a larger test tube containing the sublimand. When the larger tube
is evacuated and heat.ed, the sublimate collects on the smaller tube, but
some of it is likely to drop off when the inner tube 'is removed, unless the
deposit is compact. An apparatus of this kind has been employed B9 for the
simultaneous vaporization and irradiation of ergosterol to give a vitamin
D-containing sublimate.
The sublimator may be provided9o ,vith a ground-glass joint; it is then
useful for fractional sublimations. A similar tube, having a fiat bottom
on which the sublimand is spread in a thin
layer, is vall\able for quantitative microsubli-
mation. The distance from sublimand to con-
denser tip is only 10 mm. This device has been
employed for the estimation of 2-methyl-l,4:-
naphthoquinone, acetylsalicylic acid, phen!L-
cetin, phenobarbital, nicotinamide, salol, and
sulfanilamide in pharmaceutical tablets. Other
components, e.g., starch, sucrose, lactose, tal-
cum, magnesium stearate, L-cystine, and D-
glutamic acid, do not suhlime at 150°C. and
10 M.
In the apparatus91 shown in Figure 13, the
rubber stopper and innermost glass tube may
be removed if it is desired to cool the condensing
surface to a low temperature, e.g., with dry
Fig. 13. Vertical vacuum
ice-chloroform. An apparatus of this kind sublimator wHh liquid-cooled
has been found suitable92 for the sublimation of condenser,91
100-g. samples of certain compounds at 0.025
mm. It has also been used93 in the preparation of gas-free liquids, by
quasisublimation of the liquid onto the condenser, cooled with dry ice-
ether. The dissolved gases are not condensed but are drawn off and sepa-
rately collected. The sublimate is then melted, allowed to drop back into
the still, and again quasisublimed. It is essential that the temperature
of the condenser be low enough that the vapOl: shall condense directly as
solid since, were liquid to condense, it might redissolve some of the undesired
gaseous impurities. Furthermore, the vapor pressure of the sublimate
must be negligible at the temperature of the condenser, so that 10SB of
sublimate shall be inappreciable.
8! Whittier, U. S. Pats. 2,106,779 and 2,106,780 (H)38). Compare Milas, U. S. Pat.
2,117,100 (1938).
1111 Huba.cher, Ind. Eng. Chern., Anal. Ed., 15,448 (1943).
91 Viehoeve\·, J. Assoc. Offic. Agr. Chemists, 6,473 (1923).

!2 E. Tiedemann, WiS8. Veroffentl. Siernens-Konzern, 5,229 (1926).
O. Hibben, Bur. Standards J. Research, 3, 97 (1929).
624 R. S. 'l'Il'SON
A micl'osublimator of this same type is providedo4 with a small well for

the sublimnnd. It is alleged that the sublimnte is then deposited on the
bottom of the condenser insteitd of OIl the walls of the still and the upper
part of the condenser. A similar sublimator,V& having a vacuum-jacketed
cooling element is suitable for semimicrosublimations involving use of dry
ioe, liquid ammonia, or liquid nitrogen as eoolant. With sueh a sublirna-
tor,g6 cholesterol has been sublimed from bloocllipides at 88° and 0.003 f./..
If sublimators having ground-glass join1is are to be used under very high
vaellums, it is necessn.ry to lubricate the joints with special grease,U7 and
clanger arises thlLt the sublimate may become eontaminl1ted with the grease
during the sublimation 01' on withdmwing the eondenser at the end of tl~e
sublima1iion. An apparatus having a large flange jointpa instead of a tap-
ered joint was therefore devised (Fig. 18, Chap. VI). This was Int.el' sim-
plifiecl9u by providing a sublimand well which eould be heated in an oil
bath and by sm1ling the water-cooled condenser into the upper dome (Fig.
10, Chnp. VI). In another modification,lOO an electric hot plate was em-
ployed and the clorne was replaced by a tightly iitting iron cover through
which passed a tubular condenser, arranged horizontally. Benel1th this
was placed a glass eonclenser pinto for collection of sublimate.
So that the distance between condenser and sublimancl surface might be
adjustable to suit the pal'ticuhLl' sublimanrl and to avoid the danger of
conttLminat.ion by grC(Lse inherent in the UHe of tapered ground-glass joints,
it is recommendcd 101 that a rubbm' stoppel' plaeed on It stout glass flange be
employed instead of a ground joinli. Fnrthermore, It PYl'ex glass cloth
may be employed to cat.eh any falling sublimate, and the eondenser may
he pushed down until it is almost touching t.his glass cloth The eondenser
is bulb shaped t.o d(~ter deposit.ion of sublimate in t.he upper parts of the
still. A rather similaT, but perhaps less flexible, dosign ha.s also been de-
scribed. 102 Onto the edges of the condenser, having a convex lower sUl'~
ftwe, are sealed t.hree small pieces of glass rod. These are bent inward at
right angles to form a support for a metal screen fine enough to cat.ch all
crystals falling on it. Since many sublimates att!1ck metal, a sealed-in,
coarse sintel'ccl-glass plate rnight well be substituted for the metal sereen.
91 Riegel et a/., Ind. Eng. Chern., Anal. RIZ., 15, 'tI7 (1M3): flee Fig. 17, Chnp. VI.
" Marbel'g, J. Am. Chem. Soc., 60, 150\) (1938).
16 A. E. Koehler, E. Hill, nnd Fenmey, F'ederaNon Proc., '7, No.1, Part 1, 165 (1948).
17 Burch, Pl'oc. Roy. Soc. London, 123A, 271 (Hl29). Morton et 0,1., Ind. Eng. Chem.,
Anal. Ed., 11,4130 (1939). Bu,iley, ibid., 14,177 (11)42).
'I
Carothers ami Hill, J. Am. Chem. Soc., 54,1557 (1932).
89 Strain and Allen, Ind. Eng. Chem., Anal. Ed., 7, 443 (H)35).
101 Bailey, Ind. Eng. Chem., Anal. Ed., 14, 177 (1942).
101 Morton et 0,1., Ind. Eng. Chem., Anal. Ed., 11,460 (1939).
112 Nelson, Ind. Eng. Chem., Anal. Ed., 14, 153 (1942).
VII. SUBLIMA.TION 625
To avert deposition in the holes of the screen or sintered glass, the upper
surface of the heating bath should be slightly above the level of the screen.
Simpler devices for catching a falling sublimate consist of a watch glass
sealed onto or mounted under the condenser end.
Certain stills suitable for distillation of liquids loa require modification for
use as sublimators since the place of deposition of a sublimate must be ac-
cessible whereas a liquid will flow from the condensation area to an acces-
sible point. A still has been designed for the sublimation l04 of compounds
which readily drop off the condensing surface. It consists essentially of an
alembic in which the positions of sublimand and sublimate are the reverse
of those normal for distilland and distillate in distillation. The sublimand
is introduced through a filling tube into an annular receptacle surrounded
by an electrically heated oil bath. The material volatilizes and strikes the
cooling surface; here it collects, and then falls off into the receiver, not
back into the sublimand. Material which collects as huge crystals falls off
readily and the operation need not be interrupted. The cooling surface is
entirely surrounded by warm space so that sublimate collects at no other
part of the apparatus. The short distance between the annulus and the
condenser is advantageous.
For very difficultly volatilizable compounds it is essential that the con-
densing surface be close to the surface of the sublimand. The gap should
be large enough to accommodate the flow of uncondensed vapor, arising
from decomposition and leaks, and yet be small enough to be commensurate
with the mean free path of the sublimand vapor molecules. For rapid
pumping, the side arms for connection to the evacuating system should be
wide. Molecular stills (see Chapter VI) may be employed for fractional
sublimation of a mixture whose componcnts differ in volatility or molecular
size, or both. Thus, a thin layer of the sublimand is spread on the bottom
of the sublimator105 and the condensing surface, only 2 to 3 mm. from the
solid, is cooled with running water or dry ice-acetone. The sublimator is
then evacuated to less than 10-6 mm. mercury. Under these conditions,
1,2,5,6-dibenzanthracene (m.p., 260°) sublimes readily at 14:0° and slowly
at 100 0 • A still of this kind has been employed in the study of pyrolysis of
high polymers. loa A number of sueh sublimators, macro or micro, may be
attachcd l07 to a manifold, 40 mID. in diameter, maintained at 10-5 rom. and
provided with an auxiliary degassing line maintained at 10-3 mrn. If the
103 Hickman and Scmford, J. Phlls. Chem., 34,637 (1930); SPC Fig. 15, Chnp. VI.
104Kleipool, Chem. Weekblad, 43, 123 (1947). Compare F. C. Koehler, Chem.-Ztg.,
39, 122 (1915).
1~6 McDonald, J. Franklin [nst., 221, 103 (1936).
lOG Madorsky and Straus, Chem. Eng. News, 26,948 (1948).
lOT Riegel tit al., Ind. Eng. Chem., Anal. Ed., 15,417 (1943).
626 It. S. TIPSON
sublimators are electrically heated, sublimations mD,y be allowed to run for

long periods of time without attention.
The microsublimator described by Eder 10R (see Fig. 12) has been modified
in a number of ways, e.G., by encasing most of the thermometer with a
water-cooled condenser on which sublimate collects. lOo The thermometer
bulb reaches into the sublimand in the small well. With this device, it was
possible to determine the "initial temperature of sublimation," i.e., the
temperature at which sublimate was first observed at a given pressure while
the temperature was being raised 10 per minute. When two Hueh con-
densers were placed ill the ends of a, U-tube, containing the sublimand and
bearing a ground joint at each end, a device for fraetional macrosublimation
was obtained. With the system suitably evaeuated, one non denser was
kept cold until all of one component had sublimed. Then this sention was
closed off and the other condenser was cooled, the temperature of the stib-
limand being gradU!111y raised until sublimation of the second component
was 11PPl'eciable.
In another variation,11O the t.hermometer is repb.eecl by a water-cooled
tubular oondenser. The apparat.us is useful fol' the qualitative sublima-
tion of plant eonstituents. A simplcr form, with no ground joint, was also
described. By flattening the lower end of the condenser, III and touching it
with a drop of anhydrous glycerol, a Oewel' glass could be made to adhere.
The sublimand was placed in the small well. Alternatively, this was filled
with sand 01' fine iron filings on which was planed a copper foil or small
glass dish containing the sublimancl; wit,h this armngement, the well could
be dispensed with. By changing the glnss rlise before each rise in tempera-
ture, it, was found possible to eonduci; fmetional miel'mmblimations for,
microchemical charaeterization.
In a modifiC!1tion,lIO the lower end of the condenser is (lurved downward
and is placed over an annulus which serves for the colleetion of any liquid
condensate, formed, for example, in treatment of plant products. The
bulb containing the sublimand is inserted in an electrically heated oil bath
80 that the middle of the ground, restricted nede is level with the oil surface.
If no liquid condensate forms but some sublimate collects on the walls of the
upper part, this restl'ieted neek may, after the 8ublimntion, be plugged with
a ground-glass stopper so that the sublimate may be dissolved out, Should
the sublimate tend to collect as hanging deposits, a metal gauze or perIo-'
rated disc may be placed on the annulus. If a small sample of sublimate is
Eder, Schweiz. Wochschr., 51, 228, 241 (1913).

108
Illari, C. A., 25, 2880j Ann. chim. applicata, 21, 127 (1931),
109
110 Viehoevel', J. AS8oc. O;ffic, Agr. Chemists, 6,473 (1923).
1110. Werner, Mikrochcmie, 1,33 (HI23). G. Klein and O. Werner, Z. physiol. Chem.)
143, 141 (1925).
627
deflired for microscopic examination, a microscope COvel' glass mDy be

placed across the annulus, just below the end of the condenser.
Figure 14 depicts an aPPl1ratus which may be used 1l2 for quantitative
or qualitative macrosublimations and for miC1'08Ublimation onto a micro-
scope slide. Glass dishes of various sizes, containing the sublimand, may
be placed on a spring support. For micl'osublimations, a glass capsule
with a narrow base is placed in the centml hole in this spring support. If
desired, a porous diaphragm, c.g., lOO-mesh platinum gauze, may be placed
on the dish. The disadvantage of this apparatus is that t.he sublimate
must be washed off the condenser. For com-
parative results, two such sublimators may be
heated in the same bath, cooled with the Same
cooling system, and evacuated by the same
pump.
D. ENTRAINER VACUUM SUBLIMATION
Air-Cooled Condenser. If a poorly fitting

ground-glass joint is deliberately employed in
the apparatus previously described 113 for
vacuum sublimation (Fig. 11), hot ail' enters
the 8ublimator, is directed onto the surface of
the sublimand, and acts as entrainer. The
simple apparatus shown in Figure 15, usedlH
for purification of phthalimide and benzoic acid
by repeated sUblima.tion, 'consists of a suction
Fig. 14. Vertical vacuum
flask attached to a glass funnel inverted over a subJimator with liquid-cooled
dish in which the material is heated. For the condenser .112
sublimation of small quantities, a watch glass
may be used; for larger amounts, the funnel may be placed on a piece of
perforated cardboard 01' aluminum sheet resting on a porcelain dish con-
taining the material. For continuous sublimation, use of a strong, al'ohed
aluminum disc, packed on the funnel with pl'emoistened asbestos paper, is
recommended. A small cotton wad is placed in the side tube of the suction
flask to minimize loss on evacuating.
In the apPD,l'atus ll5 depicted in Figure 16, the entrainer is admitted at
any desired speed and the center tube is used for connection to the vacuum
pump. A piece of paper or other porous mat~rial is placed as shown, to
prevent subliml1te from falling onto the sublimand and to prevent spattel'-
112 Hortvet, J. Assoc. Olfie. All!'. Chemist.~, 6, 481 (H)23).
113 Riihel', Ber., 33; 1655 (1900).
111 John and Fischl, J. pralct. Chem., 110,282 (1925).
116 lJicpold~r, Ch~m.-Ztg., 35, 4 (1911).
628 Lt. 1::1. 'l'IPSON
ing of sublinuLnd outo the Huhlimnte. An excellent dovice llG for preventing
8ublimate from fnllillg; bade into the sl1blirn:1I1d cOllsi::;t;s of a. double bend in
the tube betwoen tho two as shown in Figure 17. SublimtUld is introduced
into the U-t,nhc through its straight limb aud is heated in nn ail' bath or oil
bath. Tho elltrnilwl' enters through a. bnftlo system insertod in the stmight
1imbi this prevents the furmation of vortices which would <.ml'l'y sublimate
against the rlil'octioll of tho ontl'ainer current. At tho top uf the bent limb
Fig. 15. Vllrtical ell- Fig. 111. Vertitml cntminer-· Fig. 17. Verticu.) ClI-
trainer _. vaCUUIll 81lhli- VILllllllIU sublimu.tor,ll6 trainer - vacuum subli-
matOl'.IH llllLtor.118
is placed a l)orfol'at(\r[ pOl'cehin disc covered with filtel' plLpcr and hold ill
place with a piece of wide rubb~r tubing. This prevents loss of sublimate
when suction is ftppliou to
the tube above this plato. The internal pressure
is kept at ablJut 40 mm. The nppamtlH:i WttS foun.d useful for lu~stening the
sublimation of thioxnnthone, whieh deeomlloRed during simple v~tCuum sub-
limation bcoause of the slowness of the In,tter process.
A rather Himilar U-tube sublimutol' was used ll7 for purifying 2,3~benzal1-
threne and pCl'ylene, and preparing unit crystals of these materials suitable
for determination of crystal structure. Tho pre8sure in the system was
under 1 mm" and dry argon was admitted in a constant stream ut consta,nt
temperature, The uevice was (~lso used fo~' fractional sublimation since,
110 Christopher, Proc. Chem. Soc., 27, 236 (1911).
117 Hertel and Bergk, Z. phY8ik. CheJn" B33, 319 (1936).
VII. SUBLIMA'rION 629
by careful regulation of the temperature, two zones were obtainable from

certain two-component sublimands.
Liquid-Cooled Condenser. Because of their similarity to vertical
vacuum sublimators previously described (see Sect. IVle), differing only
in the admission of an elltrainer, devices of this kind will not be discussed.
2. Horizontal Sublimators
Retort-type sublimators differ from distillation devices of this kind in that
the tube connecting the vaporizing chamber with the condensing chamber
mllst be short, wide, and readily cleanable, so that the possibility of clogging
is minimized. For preference, this tube should be maintained at such a
temperature that no crystallization occurs in it. The condensing tube or
chamber must also be wide and voluminous.
A. SIMPLE SUBI,IMNl'ION
Air-Cooled Condenser. The simplest form is a horizontal tube sealed

at one end. The sublimand is placed near the sealed end and heated;
the sublimate collect.s in the cooler, unsealed encl. Thus, trioxymethylene
was readily sublimed 1l8 in a bomb tube heated in a furnace with the un-
sealed end projeeting from the furnace. The apparatus has bePll improved
by using a standard-taper or spherical ground-glass joint for connecting
the vaporizer to the cylin(lrieal, horizontal condenser.
Enlarging the sealed end of the tube to give a true retort permits the sub-
limation of relatively large amountis of material. The retort side arm may
be inserted 119 in the neck of a bell jar resting on a glass plate. The subli-
mate collects in a porcelain dish. The apparatus may also be employed
for vanuum sublimation if the bell-jar rim is carefully ground to fit the
supporting glails plate. Industrially, to prevent blonking by sublimate, the
eonnceting tube extends clown into the vaporizer, is enlarged where it
le!LVes this chamber, and is heat insulated outside the vaporizGl'.12o The
condensing chamber may be equipped with a rotating drum from whieh the
sublirmtte is mechanically scraped, in fmctionl'l, into a receiver.
For miel'ofmblimation, the sublirnand is introduced 121 into a glass tube,
e.(j., 200 mm. long and of 7 mm. external diameter, closed at one end. The
closed cud is placed in a metal bloele having a t,hermometer in another hole.
The block is heated with a. mieroburner, and it is possible to observe with
lIB Tollens. Ber., 15. 1828 (1882).

119 R. Wright, Chern. News. 103, 138 (lDll).
116 11'. Bayer and Co., German Pats. 332,196; 334,B(i\); 343,319 (1919).
121 F. Pregl. Quantitative Organic Microanalysis. Translation by FyJeIUall. Churchill,
London, 1024, p. 176.
630 H. S. 'l'Il'SON
some preeisiol1 the temperature at which sublimation int,o the eool part of
the tube be('.omes l1ot.iecnhle. AftfH' sublimllt.ion is complete, the section
of the tube cont:tilling the Rllblimate iR out off, nne! tho Rublimnte removed.
A 8ubstamc (loutailling moisture may be heated to 12()O and thl\ water
which condemms nt. It (l~ip;. 18) may be evnporn,ted by Imm;l\l:\ of the micrON
flume c. With (1xtl'mndy 81llaU samples,122 tho mierotnbe is thon drawn
out; to !1 filler bore at b t1lH11!he temperature of the bloek is grn.dnally raised
until sublimation oecUl'S Hnd tlw l:\ublimatn (:olleets at b.
Fig. 18. Hori~ol) tal ]i'ig;. ID. I1oriwut:d Hllblimato[' with Iiquid-
microRublimlLtor. m ooolo(i ('.OJ)(i(\IH;er.'~3
Liquid-Cooled Condenser. A simple horizontal I:mblimat(}l·!2;! which

may be cooled with running 'water is shown in I"igure 19. The sublimand
is placed in a crucible which restH ill the hole of the annulus a.nd is covered
, with an inverted dish, funnel, or bettker.
Very readily subliml1ble compounds may be I:lllblimed 12 '1 ill a desiccator
heated above by menns of all cleetde light bulb held in phwo with plaster of
Paris. The sublimnn(l is placed in It beaker. A sm:ond beaker may (;on-
tain phosphorus pentoxicic for de~dccfLtion; normally, it should be removed
prior to sublimation. The top of the desimmtor may be insulated, and the
bottom ttnd walls, which S{WVe ttH the couclenl:ling surfneo, appropriately
cooled.
B, IDNrl'RAINlm SUIILIMA TION
Air-Cooled Condenser. The simplest deviee of this kind eonHists 126 of

a long tube to one end (A) of which is Mtached the ROmee of cntrainer,
to the ot.her a bubhle countor. The subHmnnd is pbccd in the tube neal' A
Benedetti~Piehlel', Ind. Eng. Chern., Anal. Ed., 2, 309 (1!)30).
122
123 Brilhl, Ber., 22, 238 (1880). Tseng and Hu, C. A., 28, 5718; 8cicnce Quart. Rall.
Univ. Pelcing, 4, 327 (1934).
u. Cornog and Olson, Ina. Eng. Chem., Anal. Ed., 11, 551 (H130).
12& Volhard, Ann., 261, 380 (1801). Honigsehmid [<{HI Birckcnhll.{lh, Rer., 55, 4
(1022),
VIr. SUBLIMATlON 631
and is suitably helL ted. The sublimate collects in the cooler end of the tube.
In another apparatus,I26 the sublima,nd is placed in a spiral glass tube
which is then inserted in a thermostat kept just below the melting point.
The nil' current enters through soda lime and drying tubeR, passes over the
sublimalld, through a heated tube and stopcock encased in brass tubing,
and thence into a wide horizontal tube where sublimate collects. The en-
tminer then passes through a calcium-chloride tube and into a,n aspirator.
With this apparatus, the sublimation of naphthalene and of mixtures of
naphthalene with f3-naphthol was studied.
Industrially, the sublimalld vapor, e.g., naphthalene or benzoic acid, is
led by means of an entrainer from a heated chamber,127 which may be
f1 rotatable, horizontal cylinder heated with a steam coil,128 into the
eoncleuHing chamber. The sublimand is sometimes melted, and a series

of fractions of sublimate may be collected.
The simplest form of horizontal semimicrosublimator consists of a
straight glass tube into which is introduced a boat containing the sub-
stance. A current of air or other indifferent gas is then passed through
the tube, which is heated under the boat and cooled beyond it. En-
trainel' microsublimatioll may be conducted as previously described (see
Sect. IV2A) except that a longer tube, having a capillary constriction
for the first third of its length, is used. Next to the constriction is placed
a layer of asbestos and then the sublimand. Both ends of the tube are
left open and protrude from the block. The end nearer the sublimand
is connected to the source of inert gas.
For cntrainer quasisublimation, with cooling by means of cold air, the
compound is placed 129 in a retort and heated until molten, but not boiling,
c.g., until its vapor pressure is about 100 mm. The vapor is blown, by
means of compressed air directed onto the surface of the melt, into a wide-
l"ig. 20. Horizontal en-

trainor quasisublimator
with air-cooled conden-
ser.129
126 Perman ~tlld J. H. Davies, J. Chem. Soc., 91, 1114 (1907).

117 Andrews, Conover, John, and Ruth, Brit. Pat.. 179,091 (Hl21); Ohern. Zenlr"
93, IV, 830 (11)22); Selden Co., Swiss Pat. 93,810 (1921).
12B Cole, C, A., 17, 1483; U. S. Pat. 1,445,870.
129 Robertson, .T. Chetn. Eclltcatiolt, 9, 1713 (1932).

632 n. S. 'rIPSON
mouth bottle whose bottom has been removed (Fig. 20). A eotton bag is
tied over the end of this receiver, and the mouth of tJlC bottle is stuffed with
cotton to pl'event escape of material. The aerated vnpor tends to con-
dellse as a mist, but n. current of air, ehilled by passage through a copper
eoil immersod in ice water, is dil'o(Jted into the reeeivel' and a fluffy, el'ystal-
line proclu<:l; results. N aphtlmlell<l, heaLed to about 140°, aUlI benzoic
acid, heated to about 170°, are rendily purified in this way.
Liquid-Cooled Condenser. A simple appanltus 130 for cntl'ainer suh-
limation, Jllwing a water-cooled condenser, is showIl in Figure 21. The
sublimnnd is heaterl by means of an aluminum hloek and the stream of
vapor is heaLed with a Sl~COIlCI block before passing into the condensing
ll~r
~y
Fig. 21. Horizontal cutrainer sublimator Fig. 22. Horizont!tl clltmincr sublimator
with liquid-cooled cOlu.lenscr .l30 with liquid-cooled condenser. us
vessel. The physieal character of the sublimate can be varied by chang-

ing the speed of the stream of dry entrainer. A sublimator of this kind
may be assnmbled 131 from It drying pistol plus a few supplementary pieces.
A rather similar steam sublimator 132 has been described.
A horizontal sublimlLtor, 8uitrLbie for quantitative sublimation 133 (Fig.
22), consists of a vaporizer tube and a condensing tube. A plug of glass
wool is stuffed into the narrow end of the vaporizer, the sublhnand is in-
troduced, a plug of glass wool is inserted into the wide end, and the whole
weighed. It is then attached to the condensing tube which is cooled with a
stream of iee water, a slow current of dry air is drawn through, by suction
on the side arm· of the condensing tube, and the vaporizer is heated elec~
trically. After sublimation is complete, the condensing tube is detached,
dried, and weighed. Camphor, naphthalene, and benzoic acid sublimed
l~O.De Bruijn, Chem. Weekblarl, 37, 249 (194.0).
l3l Adickes, C. A., 38,2853 i Chem. 'l'qch., 15, 173 (1942).
111 D. Craig, Ind. En(j. Ohern., Anal. Ekl., 9, 56 (1937).
133 Fuller, Chemist-Anal~$t, 29, 6 (HI19).
VII. SUBLIMA'l'ION
633
readily in this apparatus, giving a 99 to 100% recovery. A convenient

rate of entrainer flow was found to be 1.3 liters per hour.
The maximum speed of sublimation cannot be attained with the en-
trainer sublimators so far discussed because the en trainer tends to pass only
over the surface of the sublimand and does not become completely saturated
with its vapor. This may be overcome 134 by a sealed-in, coarse sintered-
glass plate interposed between sublimand and sublimate. The finely
powdered sublimand is loosely packed on one side of the plate and held
in place by means of a porous asbestos plate (Fig. 23). The entrainer
then has access to all of the sublimand; a useful speed is 1 to 2 cc. of en-
trainer per second. The sublimand and sintered plate are conveniently
heated in a metal block which should be covered with an asbestos sheet,
since otherwise the upper part might not get as hot as the lower. The
Pig. 23. Horizontal entrainer sublimator with sintered-

glass plate.134
temperature is preferably kept some 10° below the melting point. The
condensing tube may be closed with a solid mbber stopper and cooled ex-
ternally with running water instead of by an inserted condenser as shown.
Efficient cooling is essential in quantitative work.
C. VACUUM SUBLIMA'l'lON
The horizontal vaeuum sublimator is essentially a retort having a

cooled condenser; the pressure in the apparatus is diminished, by IIleans
of a pump, until below the triple-point pressure of the sublimand. The
maximum effect is obtained if the material in the retort is only just below
the melting point.
Air-Cooled Condenser. Sublilllators of this kind may be conveniently
grouped according to whether the souree of heat is stationary, movable,
or graded.
Stationary Source oj Heat. The sublimand is shaken down into the closed
end of a tube 135 which is then held horizontal and tapped until the compound
spreads out into a horizontal layer. After the tube has been evacuated and
the open end sealed, the end containing the sublimand is pbced horizon-
tally in a furnace, so that the other end projects into the open air. The
134 Solt.ys, lIfikrochemie, Emick Feslschr., 275 (1930).
11& Volhunl, Ann., 261,380 (1891). Knocke, Ber., 42, 206 (1909).
634 R. S. 'rIPSON
sublimate then .collects in this cooler part of the tube. To ensUl'e a con-
stant teIIlpCr[,tnre for the sublimand in smaller-scale work, the sublimand
end of the tube may be enclosed in n heating jaclcetJ:l6 containing a liquid
of suitable boiling point. The liquid is heated to boiling, under reflux
(Fig. 24). An Abdel'halclen drier may be used for this purpose,
In early experiments on sublimation u,t, pressures of about 0.001 mm.,
an all-glass apparatus devoid of ground joints was employed. 137 In case
the sublimand might not have been thoroughly dry and free from volatile
impurities, a condensing or absorption trap was connected 138 between the
sublimate receiver and the pump. Quinine sublimed readily at a bath
temperature of 170-180°. Morphine, indigo, alizarin, anthracene, chrys-
ene, camphoric acid, caffeine, theobromine, and codeine were also subli-
_.----,..-._- "_
Fig. 24. Horizontal VI1CUUlll subli- Fig. 25. Modified retort for VIWllUm :mblima-
mator. taG tinn.13Q
mabie in this apparatus. The sublimator was improved 139 by the introduc-
tion of two ground-glass joints, as shown in Figure 25. This rendered it
capable of being readily disassembled, a feature of particular value in frac-
tional sublimation employing a stationary source of heat. The retort is
so shaped that it can be heated in any kind of bath, including a hot-air
oven. The joint by which it is attached to the condensing tube, and the
first few inches of the latter, should also be heated. At pressures of 0.5 to
17 mm., such compounds as alizarin and indigo were sublimed quantita-
tively. The apparatus was also found useful for the preparation of certain
acid anhydrides from the corresponding acids, e.g., maleic anhydride from
maleic acid.
A sublimatol' of totally different design,140 useful for macrosublimations,
HI Krafft and Bergfeld, Ber., 38, 254 (1905). Hansen, ibid., 42, 210 (1909).
137 Krafft and Wflilandt, Ber., 29, 1316, 2240 (1896); 32, 1623 (1899), Krafft and
Dyes, ibid., 28, 2583 (1895).
118 Compal'e Biltz, Ber., 45, 3662 (1912).
131 Kempf, Ber., 39, 3722 (1906); Chern.-Ztg., 30, 1250 (1906); J. pmkt. Chern., 78,
201 (1908). Compare N. L. Smith, Anal. Ohern., 20, 1252 (1948).
HO Morer, J. Am. Ohem. Soc., 34,550 (1912).
VIr. SUBLIMATION 635
is flhown in Figure 26. It consists of a large bell jar carefully ground to

fit a glass plate. The joint is never heated and may be lubricated, since
sublimate does not come in contact with it. On t.he glass plate rests a
large crystallizing dish which serves as receiver for the sublimate. A hole
Fig. 26. Vaouum subliInatorYC
is bored in this plate for introduction of the evacuating tube. Two corre-
. sponding holes are bored in the plate and this dish, through which are pas-
sed wires for an electric heater supported by a wide glass tube which pre-
vents access of sublimate to the wires. The bottom is cut out of a glass
crystallizing dish; t.his is then inverted and placed on the hot plate to pre-
vent sublimate from reaching the latter. In the hole in the crystallizing
dish is placed a platinum or glass dish containing the s1.lblimand. In
subliming such compounds as salicylic acid it is important that the material
should not come in contact with metal. The apparatus gives rapid, ef-
ficient sublimation of such compounds as naphthalene and benzoic acid.
636 R. S. TIPSON
Movable Heating Zone. By slowly and uniformly moving the sublimand

end of a horizontal tubular sublimator into the heating zone,Hl e.g., a tubu-
lar electric heater, or hy periodically moving the source of heat back along
the tube,W e.g., about 6 cm. each time, preferably with simultaneous in-
crease in temperature by a few degrees, it has been found possible to obtain
some degree of separation of two substances, through the deposition of more
or less distinct bands of each material in the cooler part of the tube. These
may show gradation in color, and their melting points and other properties
may readily be checked.
Graded Heating Zone. By use of a graded heating zone,142,143 most read-
ily obtained eleetrically, a great improvement in fractional sublimation
results. Sublimand is volatilized· at the closed end of the tube and the
vapor is allowed to condense fractionally by controlling the heat through-
out the length of the tube. Such a device is shown in Figure 21, Chapt~r
VI.
Liquid-Cooled Condenser. External Cooling. Quasisublimation, fol-
lowed by true sublimation, has been used in freeing a sublimable substance
from dissolved lH gas, and in separation 14b of 3 substances normally gaseous
at room temperature (Ch. IV). The mixture of gases is cooled until
liquid, e.g., in [I, vessel cooled in liquid nitrogen contained in a Dewar
flask. The two more volatile gases are then allowed to vaporize off partly
and are collected !LR liquid in a second cooled vessel. This liquid is again
allowed to vaporize and the vapor is condensed to solid in a third cooled
vesRel. Fillltlly, this solid is sublimed under diminished pressure, the oc-
eluded gaseous impurity being mainly removed in the process. Quasi-
sublimation has also been employed146 for isolation of crystalline p-xylene
sublimate from a liquid mixture of n;,- and p-xylenes at a pressure below
10-6 mm. mercury. A long evacuated glass tube containing the mixture
was slowly and uniformly pulled through a temperature gradient main-
ta,ined by a thermostat system. The temperature of the hot end was
-15 0 and that of the cold end was -70°. In t.his way, crystals of the p-
xylene were obtained in the warm end, and liquid m-xylene plus some p-
xylene was distributed along the gradient (see Fig. 2, Chapter V).
An apparatus 141 for true sublimation under diminished pressure is shown
1~1 Elbe and Scott, Ind. Eng. ahem., Anal. Ed., 10,284 (1938).
142 Morton et al., Ind. Eng. ahem., Anal. Ed., 11,460 (1939).
143 Almquist, .T. Biol. ahem., 120, 635 (1937). Morton et al., Ind. Eng. ahem., Anal.
Ert., 11,460 (1939). Hickman, ibid., 14,250 (1942).
144 McI{elvy and C. S, Taylor, Nail. B'(lr. Standards U. S. Sli. Technol. Papers, 18, 679
(1923).
H& R. W. Gray, J. Chem. Soc., 87,1606 (1905).
H6 Elbe and Scott, Ind. Eng. Chem., Anal. Ed., 10,284 (1938).
147 Hedley, Chemistry &: Ind'(lstry, 44, 752 (1925).
in Figure 27. It consists of a brass tempOl'lltllre equalizer A, with ther-

mometerwellE,andailtrgecavity into whieh fits tnbe C,whose upper portion
is cooled by the cylincil'ieal water jaeket B. Inside C Tests a removable
glass Rlceve D. C is closed by [L rubber stopper carrying a glass tube con-
neeted to the vacuum pump. The sublimand is phtced on the bottom of
C; on heating A, it volat.ilizes awl is collected on sleeve D. The tempera-
t.ure at which sublimation stmts mny be accurately determined, and the
sublimate may be removed intact on sleeve D. The constituents of a sub-
limable mixture may be removed in stages in a fractional SUblimation, D
being replaeed with a replicate at each stage.
Fig. 27. Vacuum sublimator with liquid- Fig. 28. Vacuum suhliml1tor with
cooled conclcnser.147 liquid-cooled condenser.14S
The horizontal vacuum still~4B shown in Figure 28 is not subject to clog-

ging of the delivery tube. Loss on repeated sublimation is eliminated.
'l'h.e sublimand is placed in one flask, vacuum is applied, and the sublimand
is gently heated until sublimate collects in the other flask, which is cooled
by runn:i.ng water or dry lee-chloroform. If l'esublimation is necessary,
the flask which had held the sublinu1nd may now be replaced by a clean
flask to act as condenser, and the compound resublimed, reversing the
direction of vapor How. Should it be desired to collect two fractions of
sublimate without releasing the vacuum, the two-bulb flask shown may be
used as the receiver; bY' turning it through 1800 and cooling it, either bulb
may be used as condenser.
Equipment of this kind may also be employed for sublimation (')f ice-
the so-called "freeze-drying" (see Volume III, Chapter VIII, this series).
149 Bolstad and Dunbar, Ind. Eng. Chern., Anal. Ed., 15,46<1 (1943).
(i38 It. S. TIPSON
Internal Cooling. A self-explanatory design 140 of horizontal vacuum subli-

matol' is shown in Figure 29. The apparatus has been modified 150 (Fig. 30)
so that the temperature and pressure may he accurately measured. The
condenser consists of a bent metal rod, one end of which is inserted into
the sublimator and held in place by air-tight rubber tubing; the other,
-
Fig. 29. Horizont.al vacuulll sublimator with liquid-cooled condenser.119
Fig. 30. Horizontal vacuum microsublimator with liquid-cooled condenser. I,.

longer end dips into a receptacle containing a freezing mixture. To the
flat end of the rod inside the sublimator is fixed a cover glass by means of a
drop of glycerol. The sublimand is placed in a small platinum boat held
onto the thermometer bulb with platinum wire. The thermometer may be
110 Prins, Chern. Weekblad, 9, 343 (1912). Fl'eudenberg et al., Ann., 494, 57 (1932).
Helin and Vanderwerf, Anal. Chern., 21, 1284 (1949).
150 Wagenaar, Z. anaL Chem., 76, 224 (1929); 79,44 (1930); C. A., 24, 993; Phal'ln.
Weekblad, 66, 1121 (1929).
VII. SUBLIMA.'l'ION 63~)
pushed in or out as desired. The part of the tube to be heated is encased

in copper gauze and then heated with a microflalllc. After 10 minutes at
25 mm., good yields of sublimate werG obtained with tartaric and succinic
acids at ('lOC>; citric acid, malic acid, gallic acic1, morpl1ine, and narcotine
at 1000; and mannitol, apomorphine, and veratrine Ht 140 0.
Belated devices!"! are useful for drying materials by subliming out the
water as ice ("lyophilization"-sce Volume III, Chapter VIII, this series).
D. J~N'l'RAINER--VAcum\1 SUBLIMA.'l'ION
Air-Cooled Condenser. Industrially, such compounds as pyrogallol

are sublime<l under diminished pressure with only slight admission of
Fig. 31. Horizontal entrninel' - VaCUl!lll sublimatol".16;;
air. UIlller these [,onditions, a snowlike sublimate is not obtained.

Vacuum is applied at the top of the condenser chamber, which lies along-
side the retort, and a trap is installed to prevent entranee of sublimate
into the vacuum pump. Pivotocl baffies are arranged horizontally, one
above the other, in the condensing chamber, which is provided with a dis-
charge door at the bottom.
For small-scale \wrk, the sublimand may be placed in a small boat which
is then inserted in !1 horizontal tube. Inert gas is passed in very slowly at
one end of the tube, its speed being regulated, e.(/., with a stopcock. Sue-
tion is applied at the other end of the tube. Tho section containing the
sublimand is then gradually heated in H,n air bath or by means of a tubular
electric heat.el'152; sublimate eollects in the cool end, to which suction is
applied.
A preheated entrainer is employed in the apparatus 153 shown in Figure 31,
which is particularly useful for hastening the sublimation of difficultly vola-
tilizable compounds. A thin layer of sublimand is spread in the wide sec-
1fi1 Campbell ::md Pl'essmnn, Sr.!€/tce, 99, 285 (1\)44). Compl1re Pome~ and hving,
ibid., 101,22 (1945). Strickler and Sch!1ffer, ibid., 107,71 (1948). HolzIU!1n, ibid., Ill,
550 (1950).
15' Pith:_, .J. Chem. Education, 23, 403 (1\)40).
m Kempf, Chern.-Z1u., 42,1(1 (HllS).
640 R. S. 'l'IPSON
tion of the tube which, together with the coil and the first ground joint, is
heated in a hot-air bath. For the sublimation of compounds which are
so readily decomposable that they may only be eA"Posed to the minimum
sublimation temperature for a short time, a kind of "flash" sublimation
may be employed. The dry, finely powdered sublimand is added periodi-
cally in small portions, without breaking the vacuum, from a receptable
connected above the vaporizing chamber by means of a ground-glass joint.
Additions of sublimand should be started after the apparatus has been
evacuated and heated to the desired temperature. The device 154 shown in
Figure 32 ensures complete condensation by cooling the entrainer with a
water-cooled Liebig condenser after most of the sublimate has been de-
posited in the air-cooled chamber. The en trainer may be prepurified, and
its rate of flow is readily regulated.
An apparatus (Fig. 33) made from the hardm;t glass is usefup55 for ma-
terials which do not melt but are sublimable at high temperatures, e.g.,
phthalocyanins. Crystals over 1 cm. long were preparecl by this means f01"
x-ray study. The sublimand is introduced, through B, into A, and the
apparatus then fitted together as shown. A pump and gage are attached
at B, and a stream of inert gas, e.Ij., carbon dioxide, enters through the
capillary at 0, preventing deposition of sublimate in the vertical arm.
For the phthalocyanins, A was heated to red heat, and the sublimate col-
lected at D, whieh was kept at 400 0 by heating the wire gauze. Unsub-
limable impurities remained in A, and volatile impurities passed beyond
D toward B.
Liquid-Cooled Condenser. The horizontal entrainer-sublimator with
sealed-in, sintered-glass plate (Fig. 23) is readily transformed 156 for opera-
tion under dimini~hed pressure. The capillary through which entrainer
enter's must be carefully prepared; if too wide, the sublimate is carried
too far; if too narrow, substance may sublime backward against the en-
trainer stream. For many compounds at a pressure of 20 mm., the subli-
mation velocity is about three times as great as at ordinary pressure and
the same temperature. The sublimate obtained under diminished pressure
is compact, whereas at ordimtry pressure it is fluffy.
The apparatus has been modified for microsublimation (Fig. 34). The
sintered-glass plate is sealed into the middle of a tube, 14 cm. long, of 12
mm. outer diameter, conveniently heated in a Pregl heating block. At
230 0 and 20 rom., strychnine gave 13 mg. of sublimate per hour, whereas
at the same temperature and ordinary pressure, only 4 mg. of sublimate
l64 Guthier and Payer, ibid" 48, 807 (1924).

166 Barrett, Dent, and Linstead, J. ahem. Soc., 1719 (1036).
166 Soltys, Milcrochemie, Emich Festschr., 275 (1030).
VII. SUBLIMA'I'ION 611
were obtained in an hour. .A fnrthel' modifi('atiuni"T has bC'f'll devised

which makes it pOHsiblc to change the l'eceinr w,; often as (lesil'!:'c[ during the
sublimation, thui:i permitting tl~e i:mlation of a scricii of fmetinn" of subli-
mate.
Fig. 32..Hol'izontal entminer - vacuum ~mhlill\:.ttol'y4
J"ig;. 33. EntrairWl'-VILCUUIll sublimation at Fig. :H. Horizontal cntr:tllJ('I'-V:lCUUnl

high temperature. I '5 rnicr08ubUllItltor with liquid-eooler! 0011-
dellser. lij6
V. EXAMPLES OF SUBLIMATION
In the foregoing, examples of aetual sublimations have been citetl to

illustrate the various devices and procedure::;. A few others, mostly of an
analytical nature, are appended. Examples of direct sublimation of a
eompound from crude, natuml product are the following: separation of
157 HUrka, C. A., 37, 3302; MikrlJc/icmie fler.lIfii.:l'ochim. Acta, 30,193 (1942).
642 R. S. 'l'IPSON
caffeine from dried cofiee or tea leaves, ferulic acid from asafetida, santonin
from artemisia, gentisin from gentian l'oot, hydrastine from hydrastis
rhizome, and cinnamic acid from Sumatra benzoin.158 Canthal'idine may
be sublimed from the dried cantharides insect or from powdered American
blister beetles,169 particularly after moistening with chloroform plus suf-
ficient hydrochloric acid to render the material distinctly acid. The sub-
limate obtained from glycosides usually consists of the aglycon; if the
glycoside is moistened with mineral acid prior to sublimation, the subli-
mate is almost certain to be the aglycon. This may often be employed as
a rapid, simple means of idcntification 160 for the glycoside.
Such pharmacologically active substanees as caffeine, strychnine, mor-
phine, barbituric acid, and phenacetin injected into animals are recoverable
from the central nervous systcm by sublimation. 161 Fatty acids, choles-
terol, and cholesterol esters may be sublimed from blood, brain, etc. In
some cases, it may be nGcessary to pl'epurify the material. For instance,
in determining vanillin in vanHla beans l62 by vacuum sublimation, the
compound is extracted into a solvent, and the residue sublimed after evap-
oration of the solvent. This method is also said to be preferable in estimat-
ing the caffeine in coffee. A similar procedure is used in the determination
of Baccharin in foods/ 63 e.(j., ice cream. Extraction with diethyl ethel',
and evaporation of the latter, gives a material from which the saccharin
may be sublimed. Benzoic acid in ketchup may be similarly determined.
Sublimation of ice is a useful procedure 164 for drying substances at low
temperature. The stability of biological products is greatly enhanced by
thorough drying under such conditions. 166 Palatable dehydrated milk,
IfoS Van Itallie, Pha1'ln. J., 112, 31 (1924). Compare Zapotocky and Harris, J. Am.
Pharm. Assoc., Sci. Ed., 38, 557 (l949).
169 Viehoev(J]' and Capen, J. Assoc. O:{fic. A.gr. Chemists, 6,489 (1923).
160 R. Fischer, C. A., 32, 722; Arch. Pkarm., 275, 516 (1937).
IGl E. Keeser and J. KeeseI', C. A., 24, 5066; Arch. exptl. Pathol. Pharmakol., 147,360
(1930).
1'2 HO!'~vet, J. Assoc. Offic. Agr. Chemists, 8,559 (1925).
163 Oakley, J. Assoc. Offic. Agr. Chemists, 28, 298 (1945).
164 l!']osdorf and G. W. Webster, J. Bioi. Ohem., 121, 353 (1937). Flosdorf and Mudd
J. hnmunol., 34, 469 (1938). Flosdorf, Stokes, and Mudd, J. Am. 'Med. Assoc., 115,1095
(1940). Flosdorf, Hull, anti Mudd, J. Immunol., 50, 21 (1945). Flosdorf, Drug & Cos-
meticInd., 57, 188 (1945); J. Chem. Education, 22, 470 (1945); Modern Packaging, 19,
No.3, 133, 164 (1945). Bradish, Brain, and McFarlane, Nature, 159,28 (1947). Brad-
ish, Chern. Products, 10, 60 (1947).
1a6 Craigie, Brit. J. Exptl. Pathol., 12, 75 (1931). Elser, Thomas, and Steffen, J.
Immunol., 28, 433 (1935). Flosdol'f and Mudd, ibid., 29, 389 (1935). Reichel, U. S.
Pat. 2,066,302 (1936). Scherp, Flosdorf, and D. R. Shaw, J. Irnmunol., 34, 447 (1938).
Flosdorf, Boerner, Lukens, and Ambler, Am, J. Clin. Palkol., 10,339 (1940). Mudd and
Flosdorf, New England J. Med., 225, 868 (1941).
vn. SUDLIMATIOK 643
fruit juiee:>, and meat 10r, may be prcpared. 167 At tCIllI1erat,l1l'es of - (iO to
-85°, the rate of vaporization of ice 168 is 10- 4 to 10-6 g. per square centi-
meter per second. Thus, for example, blood plasma or serum should be
desiecatecl 169 below _10° until praeticnUy no moisture remaills, and finally
freed from residual traces while the temperature is allowed to rise to +37 0 •
In fractional sublimation, some degree of preferential vaporization and
recrystallization at any given pressure may be effected hy close control and
graded change of the temperature of t.he heating and cooling surfaces, pro~
vided there is a pronounced difference in vapor preSSllre of the components
of the starting mi:xiure. Thus, eaffeine is sepnl'n,ble from theobromine by
fractional sublimation. 'With a stationary Bouree of heat, the temperature
of the retort may be gradually raised, and SllcceSf'ive frnct.ioml of sublimat.e
collected. These are progressively poorer 01' richer in the desired eOIll-
pound. If a horizontal t.ubular still is used, It movable heating zone (see
Sect. IV2C) may lead to the same result. 'rhus, after no more sublimate
collects in tl1e cool end of the tube, tIle tempcl'l1ture of the heater mllY be
raised, the cold end of the tube moved further from the heater, and a second
fraetion of sublimat.e collecteclYo At the end of the sublimation, the vari-
ous fractions may be carefully scraped ont by lUeans of a long metal spa-
tula, or the tube may be cut. An industrial method of achieving fractional
sublimation is to heat the mixture gradually, e.g., electrically, and collect
a series of sublimates on a moving condensing surface.17! Pl'Ogressi\'e
condensation may also be brought about by passing t.he vapor through n
series of condensers at. successively lower constant temperatures. Indus-
trially,172 in the separation of such mi:;..iures as anthraquinone plus anthra-
cene, or phthalic anhydride plus naphthalene, this may be achieved by
dividing the boxlike condensing ehamber, through whieh the vapors pass,
into sections by means of parallel gauze screens from 'whieh aceumulated
crystals may periodically be shaken by t.he impact of swinging weights;
eaeh section is provided with a cooled baffle. Alternat.ively, the condeml-
ing chamber may be cylindrical and subdivided into a series of' concentrie
lOG Flosdorf, Meat, 22, No.5, 27, 58 (1945).

lij7 Flosdorf, Food Intis., 17,22, 98, 100, 102, 104, 106,108 (19,15); Chem. Eng. Prog-
Te.~S, 43, 343 (1947); Freeze-Dryina (Drying by Sublimation), Heillhold, New York, 19M).
las E. Miescher and Tschudill, C. A., 40, 2720; IIelv. PhYB .•1cta, 18, 456 (1945).
160 Flosdorf (md MlHhl, "Large Settle De8ie~lLtion of Blood Substitutes from the Frozen
State," in Mudd and Thalhimer, eds., Blood Substitutes and Blood 7'ran8fusion. Thomas,
Springfield, 1942.
170 Kempf, .J. prakt. ClWIIl., 78, 213 (1908).
171 Fletcher, C. A., 8, 1942; Brit. Pat., 29,537 (1912).
172 C. E. Andrews, U. S. Pats. 1,324,716; 1,324,717 (1919). The Selden Co., J. M.
Selden, and C. G. Selden, C. ii., 16, 1341; Brit. Ptl.t. 173,723 (1H20). C. A., 16,2242:
Brit. Pllt. 174.013 (1920). H. D. Gibbs, C. A., 18, 1071; U. S. Pllt. 1,484,260.
644 R. S. TIPSON
cylincler~173 by gmlze screens on which travel brushes for removal of crys-

tals. Tho chambers increase in capacit.y from the center out and the vapors
are lcd first into the innermost chamber, which may aetually contain the
vaporizer. The purest product collects in the innermost chamber, the
illlpurcst fraction in the outermost, or vice versa.
The following examples illustrate fractional vctGU'tlm sublimation. I-Hy-
c1roxyanthmquinone (0.200 g.) gave 0.l(i5 g. of sublimat,e after 0.33 hour
at 130 0 and 9 /.I. j tho same weight of 2-hydroxyanthl'aquinono gave only
0.063 g. of sublimate after 24 hoUl's at the same temperature and pres-
sure. Hence, on koeping a mixture of the two isomers at 100° and 10 /.I.,
the sublimate l74 was essentially the I-hydro.ll..JT compound, which was col-
lected and removed. The temperature of the sublimand was then raised
to 180 0 and the sublimate mainly consisted of the 2-isomer. Separation
was much sharper than by fractional recrystallization from a solvent.
Similarly, salol is readily and almost quantitatively separated from a 50: 50
mixture with phenacetin by first subliming out the former at 35 0 and 11 /.I.
for 2 hours, and then subliming the latter at 120 0 and 13 /.I. for 0.25 hour.
Again, certain soft drinks contain 175 both benzoic acid and saccharin; sub-
limation at "15-55° and 1 to 2 mm. during 3 hours gives the benzoic acid,
which is collected and weighed; treatment at 14.5-160° and 1 to 2 mm.
during 3 hours then gives the saooharin.
Propel' control of the preS81£re in the sys(;em will also aid in fractionation.
For example, a series of sublimates may be isolated at progressively lower
total prcssnres, the temperature of suhlimand and conrlenser being kept
constant.
Fractional microsublimation is useful in the examination of drugs 17n and
in pharmacognosy.177 Thus the characters of the orystalline sublimates
from many drugs have been ascertained, and the behavior of mix.tures of
two, three, or more components has been studied. 178 Of 92 such mixtnres,
61 were completely Tecognizcd by this method.
173 The Selden Co.,.r. M. Selden, and C. G. Selden, C. A., 16, 2242; Brit. Pat. 173,789
(Hl20). E. Q. Adams, C. A., 18,2; U. S. Pat. 1,470,950.

174 Hubacher, Ind. Rn(J. Chem., Anal. Ed., 15,448 (1943).
176 Oakley, J .•issoc. Offic. A(Jr. Chemists, 28,298 (1945).
178 Dezani, Chern. Zentr., 1916, I, 1044; Glom. farm. chim., 64, 30'1 (1915). Seifert,
C. A., 35, 2673; Dent. Apoth.-Ztg., 55,576, 584 (1940); C. A., 37, 2882; Dent. Apoth.-
Ztg., 56, 600 (1941).
177 Kandersteg, C. A., 38, 2787; Schweiz. Apoth.-Ztg., 82, 61, 81 (1944).
178 Weismn.nn, C. A., 29, 8237; Phann. Ada Helv., 10, 125 (1935),
General References
Gartshore, "Sublimation," in Chemical Engineers' H andboo/,;. lI.IcGraw-Hill,
Iew York, 19,11, pp. 1471-1478.
Kempf, "Sublimieren," in Houben, eel., Die Methoden del' ol'lJnlli.~chen Chemie.,
;rel ed., I, 503-707 (1025).
Kofler, L., and Kotler, A., Mikl'O-Methodell zur K.ennzeichnUlllJ orlJwlischer Stojfc
wel StoJfgemische. Univer~itiitsverlag Wagner, Innsbruck, HHS.
Vogel, "Die heterogenen Gleichgewichte," ill Masing, efL, Ilandbltch del' JIf;tn/l-
Jhysik. Vol. II, Akadem. Verlagsgesell~chaft, Leipzig, 1\)37. Reproduced by Ed-
yards, Ann Arbor, 194-1.
Wilke, "Sublimation: Its Applic(J,tions in Chemical Processing," Clwn. lnds.,
i3, 34-38, 122, 12,[ (.July, 10·18).
Subject Iudex
A Azeotrope (clJnlilllll.'d)
Absorption, ill pUl'ificlLtioll of g!lse(Ju~
lIlId PI'PSSUl'l!, :'158, 40·(
sepamtillll, 3tili
mixture, 10
Azeotl'op~ forlllatiull, :tml lJydrogen bond,
Absorptioll apparatus, Orsat, .j.Hl
359
Acetic acid, drying, 3B3, 3(\8
Az{'otropic distHllltioll, 10, 12, 317, 31fl,
Acid cOlltmninunt in still, effect of, 300
350-387
Acitivity coefficient, 20
appumtu8, 373
in extractive distillatiou, 323
application, 38T>
Adiabatic eOllllitiollS, 177. See also In-
c;tleulatiull, 381
sulation.
[!olullln, 380
Adsorption, preferential, by activated
cOlltinu()w" ;~n, 378
charcoal, ,lID
laboratory, 373
"\git!Ltioll,226. See :<180 Stirring.
proeetiul'e, 373
Ail', dissolved, 401
teHting;, 368
liquid, as coolant, 389, 583
Air bath, 266 theorY,3Si
Azeotl'opic mixture. See /izentro!,e.
Alcohol. Soe Ethanol.
AzeotropiRIU, 31 fl
Alembic, 476, 511
Almquist Rtill, 532
Alpha (relative volatility), (l
B
Alphutroll, 579, 582 Backing pump, 5G3, 5i3
Ammonia as coolant, 411 B[lck pressure. Sl~l' Pl'c,~S!tl'e dr(Jp.
Amplified distillation, 52(\ Baffle, high vacuum system, 555, 585
Analytical disti1ll~tion, 6G, 178, 202, 217, Bal:mce, gas density, 417
225,256,2D5,312 Ball packing, 207, 20\)
high vacuum, 517, 526 Barometric condenser, 568
low-temperature, 411, 4Hi Barometric leg, 567
Aniline point, 335 13:Ltch distillatioll, cht,rge, 100
Aperture, in high vacuum system, 552 composition, 50
Apophororneter, 615 Batch distillation, 4, 5
A. Pt. (aniline poiut), 335 curves, fl7-U5
Ascarite, 419 H.E-T.P., 153
Association, molecular, 4!J, 35!J, (\07 determination, 56, m
Assumptions, uswtl simplifying. Sec lnrge runs, 220, 223, 2DG
USA. Humber of plates, 1G2
Atmosphere, definition, 541 operoting line, 57
Automanometel', 490 reflux ratio, 153
Auxiliary agent. Sec Chaser, "Entrainer, relative volatility, 153
Solvent. sample,4G
Azeotrope, 27-2H, 222, 302, 319, 35G small l'lln~, 2\)6
extractive distillatioll, 321 tot~Ll moles in still, !J9, 122, 132
former. See Entrainer. variables, 145, 169
heterogeneous, 320 Batch rectification unit, 175
homogeneous, 319 Batch size and holdup, 302
647
SllH.TECT INDEX
Bead~, gbRs, to prevcut humping, 402 Brinn as coolant, 411.

Bend,dfeet on high vacuum, 557 Bruun column, 221
Berg nnd James equatiolls, 130 Bubble-cap column, 219, 220
Bcrl snddles, pncking, 207 Bubble-cap plate, 219
Bicarbonate to prevcnt bumping, 402 Bubble plate, 176
Binary systems. See lIIi;ttI!l'e.~, binal'Y. Bumping, 231, 301
Binding energy llnrl deeomposition, 607 prevention, 477, 478, 492
Boiler, definition, 4 vacuum dist.ilIation, 474,478, 4\)1
Boiling point(s), 6, 310, 475 Bumt, gas, 413, 41.5
ill codist.ilbtioIl, 17 Butlldiellc determination, 416
CUI've, 97 Butll-Heli-Grid pMking, 3\)4
dwintion from idealit.y in extractive Butane, tllkp-off rtLte in distillation of,
distill a tion, 333 424
difference alld ehoice of column, 2H7
guses aud volatile liquids, 447-456
c
100y-tompemture distillation, determina- Cake sublimate, 607
tion,411 Caplwity. Sne Throughput.
maximum, mixtlll'C, 27 Capillary to prevent bumping, 478, See
minimum, mixl:ure, 27 also Bll'mpin(/.
mid number of plates, 172 Cm'bon dioxide as coolant, 411
[llId preRsure, 464, 47"1 Cm'bon monoxide determination, 416
rmd reflux ratio, 173 ClLrlJOrunrlum column canting, 184, 187
ILnd relative volatility, 22 C!1rborunrlum packing, 207
Boiling point -numpoRitiun diagram, 7 Carlson-Colburn equation, 2'1
Boiling point-per ccnt dist.ilhltiou din,- Carothers Htill, 531
gram, 11 Carrier oil in high VI1.c.uum disti1\nt.ion,
Boiling stonPR, 477, 402. S[!n also Bu.mp- 523
ing. Cruswell equation, 76
Boiling tube, ·192 C!1rtesian diver Il1rLTlo&tat., 283
Boil-up rate, 201, 26ft See alRf) Through- C(Lseade distillation, 470
pu.t. Caseade fractioIll1.ting molecular still, .53H
automatic ('ontrnl, 207 Cent,rifugal still. See Still.
cI11cull\tion, 271 Chitin packing, 207, 209
and COIUlllll efRcit'Hlcy, 2!)8 Changcr, receiver. See Receiver cha1lger.
determination, 271, 2i8 Channeling, effect on H.E.T.P., (in
ILnrl Pl'OSSUl'Co, 2\)\l effect of panking on, H)9
and pressure ch'op, 270 Charcoal, antivated, preferential adsorp-
Booster pump, 570 tion by, 419
Bottoms, 53 Charge, definition, 4
Bomdon gage, 577 Charging procedure, 304
Bowman-Briant equation, 88 Chaser, to avoid peroxide explosion, 300
Bowman equation, 88, 1.51, 166 effect on frantional distillation, 217
Boyle's law, deviations from, conectione and holdup, 147
for, 4:30 with internal heater, 267
Brass packing, 208 low-temperature distillation, 428
Break, in distillation curve, and composi- Chaser-holdup ratio, 303
tion, 312, 313 Chemical analysis in interpretation of
calculat.ion, 165 disti1lation data, 314
Breger Rtill, 539 Chilton-Colburn transfer units, 77
SUBJECT INDEX
ClaisrTl flask, 478 Column (rOiltinued)

Clapeyron equation, 10,22,172, GO-I Oldershaw, 222
Clark equtttion, 2;' oprn tube. Sel' CO/Illlm, !}/lip!!J ill/)e.
Clau~ius-Cbpeyroll equatioll, 1\1, 22, 172, p:u.:ki)d, Il, 8, ·W, 175, lU~, 28;')
n04 pxlr:u:tive flistilhltirm, 354
Clpaning the still, 300 II.E.T.P_ determiuation, 80
Codistillation, III, 17,320,358, 3U-37ij point composition, (H
amI moleeuhr weight, 17 theory, 31, 75
Cohen equntioll, fH paeking. See Packin!].
Coil condenser, 251 perforatrd pbte, 220, 222
Cold air jacket, 402 performlllwe, differential cf!lll1tionH, 75
Cold finger coudenser, 251 plate, fl, 176, 21!J
Cold finger still, high vat:uml1, 500 eontinuoU8 di,;tilhtioIl, ·17
Cold junctioll, thcl'll1o()ouple, 278 cxtmctivc diRt.ill:Lt iolt, 35,1
Cold trap. Sl'e Trap, cold. Podbielni:tk, 380, ·103. Sec also
Collection, tbtilbtc, 307 Column, H!lper-Cal; IJI/d-Rol)ol;
low-tempL'mtut"e distillatioll, 413 Pw;/;in(/, }]eli-Chid; 8mlli-RofJot;
Column, ·1, 12, 17:\ 178-240 S 1/ pr.lfractimwloJ'.
Bruun, 221 pressure, low-temperat.ure - rlistillatinn,
bubble-cap, 210, 220 442
ehoice, 17G, 22.'i, 29G effect on separation, 8 l
c
cleaning, 300 rotating, 220

concentric-tlll)e, 178, lfH, 484 ('o]lf:tmtric-tuhc, 231
multiple, 1\)8 eonc,227
reflux tiistri\mt,joll, 108 Simons, 413
rotating, 231 micro, 441
vacuum di :;tilla tion, ·185 spinning-band, 230
spacers, H)7 twisted, 231
therm\Ll reetifi('ILtion, ·173 spiro-I, 183. See I1lso Column, wtl'e-coil.
two tube, l!H cupped, 188
construction, 235 pitch change, 186
(1orrosion, 69 screen, 184
dead space, errors, 428 tube, 178
definition, 4 low-temperature distillation, ·101
distorted-wall, 183, 193 Stedman, 210
efIieienny, 75, 177, 170 Supercool, 405
empty tube, 80, 178, I7\) temperature gradient, 83
film-tn)C, 175, 178, 1\)8 theory, aO-\)7
helix-packed, 1\)!) therlUltl rectification, 472
Hyper-Cal, 25\), 268 thw:1ued, 180
insulation. See Insulation, .Turkel. types, 178
jacketed. See Jacket. unimmbted, SO
Inborat.oI'Y, 190, 201 vaeUUIn lli~tilll1tiou, 186, 231, 4G8,
low-temperature distilhtion, 304 ·170,47ll
micro, low-temperature distillation, Vigreux, 11)3
4'11 wet.ted-surface, 175, 178
moving-pal·t, 226 Widmer, 184,245,4.84
non-adiabatic. See Column, um:nsu- wire-coil, 183, 190
lated. double, 193
650 summC'f INDEX
Compensating jacket. See J ackct, heated. Contact redification. See Redificctiion.

Complct.ion of dist.iIlatioIl, indications of, Contaminants in sWI, 300
309 Continuous distillation, 4, 45
Composition, change and separation H.E.T.P. determination, 47
sharJlness, 127 Coolants, 411, 583
distillate, calculation, 16 Cooling. See I1lso Condenser.
simple distillation, 7 intel'llal, 622, 638
initial, effect on distilllatioll curve, l().'S external, 622, 63(i
liquid, simple distillation, 7 subliultLtion, 608
and plate number, 44, 93 Corm] condenser, 25lJ
still and product, 59, 67, 10-1 Corrosion, Rffoct on H.E-T.P" 69
totnJ reflux, 63 Cox dmrt, 20
vapor, 31, 77 Critical solution t,emperature, 334
cl1lculation, 16 Criticnl temperature, 605
simple distillation, 7 Crosley equation, 142
vapor-liquid relation, ~4 Cl'ystrtl shape, subliml1tioIl, 607
Composition - boiling point diagmm, 7 Crystal size, sublinllLtil'm, 607
Composition -per cent distillation dia- Crystalline distillate. See Solid d£sUllaie.
gram, n CrystuUine sublimate, 607
Compressibility correction, 430 CrystlLllization, fractional, 9
Concentric-tube column, See Coll.mn. C.S.T. (critical solution teIllpemture), 334
Condensate partitioning. See Partitionin(l. Curve, distillution, 10, 11,97-145,310
Condensation (chemieal), by alkali in Cut point, 429 '
pl1cking, 301 Cuts, 3
Condensation, fractional, 250, 442
partial,4 D
and preS~lUl"e, 474 Dalton's law, 16, 108
Condensation pump, 569 Dead spuce in eolull1n, errors, 428
Condenser, 3,249-260,346 Dee:~nter, eontinuous, and column, 372
air cooled, 614, 619, 620, 627, 629, 630, Decomposition, lwoidltnce, 299
633,639 by high vacuum distillation, 505
auxiliary, 249 hazltrd index, 507
barometric, 568 Definitions, l-lJ
cold finger, 251 Degassing, 4\11
coil, 251 Degrees of freedom, phase rule, 15, 18
Corad,259 Dehydmtioll, by acid contaminant, 300
Liebig, 3, 251 Density, liquid, 68
liquid cooled, 616, 622, 632, 636, 640 and distillttte composition, 97
low-temperature distilliLtiou, 408 Dephlegrnator. Sell Condense,·, pm·tinl.
partial, 4,176,249,259 Desiceants, 301, 419, 489
vacuum distillati.on, 475 Detwiler-Markley still, 537
temperature, sublimation, 608 pewar flask, 402, 489
totsl, 4, 249, 251 Diethylene glycol manostat, 284
low-temperature distillation, 413, 441 Differential distillation 8. See also Simple
vacuum distillation, 480, 486 distillation.
Conductance, gas flow, 542 Diffusion, in vapor-reflux contacting, 175,
Constant-.boiling mixture. See Azeotmpe. 178
Constant-.yield oil, 519 fractional, 10
Contacting, vapor-liquid, 175,219 molecular, 499
SUBJEC'£ INDEX G5l
Diffusioll t:oofliuieut, Jllolceuhr vapor, SO, Eliminution tempcmture, 517

80 Empirical HuhliIl1atioll urluatioll, GOil
Diffusioll pump, 488, 5G\! BrHt of distillation, indielLtions of, 3D!)
Dilatometer, 417 Elll'ichillg s('r:tiou, -J.
Dipole 1ll0llltlllt of wIV(Jnt::;, a3\) EnrieJullcnt, 3, 17l:i
Distilbhilit,y in hip;h vaeUUlll, 518 Elll'ichnwnt fnetnl', a8, ().1
lJistilland, bateh si:w, 302 Entmincl', 320
l'harging, 302 amount, 382
ehaser, 302 hailing point, 3G3, 31i;j
definition, :3 ('hoke,35!)
pretreatment" 300, 417, 4\)1 nOllsekdivc, 30:l
lliHtillatu, llefinitioll, :3 Jll'cheated, li3!1
analysiR, \)7, 312 rCllloval, 3Gli
('ulleetioll, 807, ·U8 "elective., an3
('omposiLioll, ('al('ubti(J1l uf, 32 vacuum dist.illatioll, 478
('urve, \)7-145, :)10 Entminr'l' q\l['Hi~nblimllt.ion, (i;{ 1
Di~till:1tion, definition, 2 Entraiu()r sublimation, (3D!), GIn, (jao
rll'()['eSS(~s (atmosphpril' :llld high Ent.rainer sublimation eqUtltioll, 500
vacuum), 4f1S Ellt.milllnent, 485
Distorted-wall colullln, 183, Hl3 Equilibmt.ioll time. See 1'ill/(.' of ('lllililm/.-
Dividing. See P([l'iilionillfl. lion.
Doeksey-1ItLy c(lnat.ion, 8(\ 1~(lUilihrium, (~h8e!l('1! ill high v:u'uum
Driving force, 88 distillation, 4HS
Dry ice, 304, 583 lltta.imnellt, a~ driving force, 5, 22
Dry ice-at:etonc, 48IJ in praeticI11 distilhtion, 15
Dry iee-carbon tetrnchloridll, 4S!J primary, in practical distillation, 30(1
Drying, column, 310 and relative vollltility, 11
freeze, 637, 642 ,md theoretical phte, 6
Drying ugent. t:;ee Desic!:anis. vapor-liquid, determination, 8
Dry still pot, 303 Equilibrium ('Olnposition, 25-28
Duct, high vacuum system, ,554 Equilibrium distillation. Soe Flash dis-
Dufton spiral packing, 400 tillation.
Duhem equation, 24 Equilibrium ratio, 30
Diihrillg'S rule, 10 Equilibrium st.ill, 8, 26, 325, 3G8
Dump packing, 108 Ethane fractioll, sepuration, 423
Etlutnol, anhydrous, b}r !lzeotl'opic dis-
tillation, 366, 385
E1JIIlhltol' tube to (woid bUlUping, 4\)2 Ethylene fnwtion, sepamtion, 423
l~dgeworlh-,Tohn:,;tone equation, 133, 138 li1vupol'lltioIl, definition, 2
Eflil·,ieu\\y, (',olnmu, 177. Sec also Scpara- Evaporative distillation, 4!)7
I/:on, $hal'PI!C8~. Evaporator, falling-film still, 532
plate, 0, 11,30, 140, 2IU, 223 Exhaust.ing scetiOIl. Sec Strippill(J sec-
l'I'lllrphrcr, 31 lioll.
amI \\'eir height, 225 l'~xlUlustspeed of vaclllltn elHlmber, 5·t3
l'cetificat.ioll, 181, IIl!J ExplLllsiou of gllSCS, 600
EfHdencyfactlll', 147,182,202,225 of glass, eoefIieient" 28!J
high vacuum distillation, 504 Explosion, peroxide, 300
Elimirmtin!l r:\ll've, 508, 518 E;xtmction, fm.cUonal, 10
Eliminution distillatiou, fi 17 Jiquid-liquitl, 338, 3(;7
pilot materials, 522 r~xtl'8.rtive distillation, 10, 12,317-360
652 SUBJECT'INlmx
ExtrU(~ tivc JiBtilbtioll (contz:n'ued) Forc-Jll'CS~UI'C gagc, 577

apll(l,mtuH, 341 }!'orc-pl'CSSUl'C line, tIi::I.lHeter, 562
applicatioll, 355 Fore-pump, 563, 57:3
of lIzcoLl'o]Jes, 321 Fraetiollul diffusion, 10
baldl, 3'11 Fl'IIetiolllt! distillation, 3, 10, 175-316
cilleulation~, 351 lIJlPl1mtll~, 175-2U4
cOlltinuou:l, 347, 3,19 Imtdl, us
cOlltl'ol, 350 multieomponent rnixtUl'l!8, 130
H.E.T.P., 352 ~tcpwisu (mlculatioll, lOLl
bbomtol'Y, 341, 346 theory, 104, 120,-129
procedure, ::141 Iimittltions, 106
phte COIUIllll, 354 laborlLtory, 178
Lemperature mcUSUn!Jlll'nt, 350 limitatiOllH, 317
theory, 322 low-tempomtul'c, 3lH
proeedul'D, 2n5-316
F Plll'po8c,2!l5
schedule, 308
Factor A (effieiemlY factor), 147. See also
var:uum,470
E.[Jir.ienc)J factor.
high, 507
FlIlIing-film still. Soo Still.
FI'llc1iiollal extraction, 10
Feed, definition, 4
FracliollllJ sublimation, G23, G21i, 637, Ilia
Feed rltte, 270
Fmetiolltttillg column. See Column.
li'ellske equlltioll, 33, as, 67, U5, 107, 115,
F1'I1ctiolllLting f[I01;ol', 104, 1GS
123, 124, 132
dcriV':ltioll, 39 Fmotionatioll, definitions, !l
Film-type l:olnmn, 175
Fraction cuttcr, 480
Fl'Q(ltion(s), definition, 3
Film thiekncss, film-type still, 50B
Flange in vacuum system, 595 yield, Cttlculatiol1, 133
and rcflux ratio, 138
Fhsh distilla.tion, 2
Flask, Cluisen, 478 Fraction size, 308
Fraction weighing, 304
improved, 479
Dewar, 402, 489 Fmmework, still mounting, 294
1"looding, 9, 192, 199. See al~o Pre- Frocze dl'yiug, 637, G42
flooding. Freezing point determination, 417
, Flow, free mQlecular, 551 Fritted-glass plate, 223
intermediate, 5'15
Knudsen, 467 G
luminal', 467 Gage, Bourdon, 577
low pressurc, 045 fine prcssure, 570
moleeulrtl', 545 fore-pressure, 577
rate, liquid reflux, 47 ioni?u,tion, 582
Rtreamlinc<l, 178 MeLcod, 490, 577
turbulent, 1\)\1, 457 Ph'ani, 513, Sn), 587
high vacuum, 562 pressure. See Gage, vacumn.
units, 542, 561 vacuum, 4S9, 576
viscous, 545, 56(} location, 400
laminar, 560 GIL8, inert, introduction ill vaeUUlll dis-
1<'10worH, 8uhhmllto, 610 tilhltion, 478
Flowing-film still. See Still. liquefied, distillation. See Low"temper-
Foaming, 491 ature distillation.
Fore-pressure breakdown, 573 trelltment'of samples, 418
SOBJECT INDEX t}53
Gas allalysis by low-temperature distilla- Heli-Grid p:wking, 198, 211

tion,3S\) low-tempemture di8tillatioll, 3\)7
precision, 433 IIeli-P!1k packing, 205, 209
Gas eXllllllsion, 600 Helium'leak detector, 587
Gasket, high vacuum system, 592 Helix-packed column, HJ\)
Gas law, Hi Henry's law, 18,91, 111
nas pressure, total, 16 H.E.'l'.P., 6-13, 30-97, lOn, 115, 145, I·Hi,
Gas sample, pretreatment, 417 147, 166, 169, 17(i, 170, 180, 185 .•
GIIS sampling apparatuR, 417 200, 211, 221, 223
Gauge. See Gage. blLtch distilll1tion 153
Gauze packing, 184, 210 -composition diagram, 4J
Glass coefficient of e:l.-panSiOll, 280 determination,32-5\), ti:'HiS
Glass plwking, 203 batch distillation, fiG, III
Glass wool to avoid bumping, 4\)2. See continuous distilhtioIl, 47
also Bumping. graphic and tahul:.U' methods, 41
Gould still, 531 IUulticomponentsystems, .'is
Grease. See LubricaNon. nomograph, 56
packed column, 80
H partial reflux, 67, 82, III
plate-equivalents proecdlll'e, !.i8
lb.lugl'll ionlcllk detector, 587 pressure effect, tin
HI17,al'd index of thermal deeompmsitiull, procedure, 36
507 sample, 45
Heat b:dlLllce 111H1 sePllmtion sharpne~~, st.epping-offpJ'o(,t'~~, 37, 67
46 test mixture, 38, 43, 70--74
Hcat dCt'.ompoHit.ion, avoidance hy high total reflux, SO, \)1
vacuum distillation, 505 with USA, 47
Heat exchange, 175, 177,271 without USA, 52, 86
Heating, 262-273 and diameter of column, flD, 82, 85
control, 268 different liquids, 8
economy, 465 different mixtures, 74
external, 264 different operators, 201
internal, 304 and height of columll. ~iH, Xii
with chaser, 261i Heli-Grid packing, 807
jacket. See Jacket, hmtted. limitations, 68
low-temperutul'e distillatioll, 407 low-temperature distillatiull, 3\)5
steam, 465 partial reflux, 82
sublimation, 633 determination, 33, 45
well,265 and rate of distillatioIl, 82
Heat IOSH inlahomtory distillation, 2G.! nnd rehtive volatility, Hl8
Heat ratio, 474 totall'efiux, 81
Heat of sublimation, latent., 604 Hit:kml1I1 still, 528, 535
Heat of vaporization, 19, 173 High frequency discharge for leak uetel'-
Height equivalent to a theoretical plate. t.ion, 586
See H.E.T.P. High vacuuIU. See Yrwnum distillt!tion,
Height of separatioll :,;tage, 88 high.
Height of a transfer unit. SeeH,'J'.U. History of distillntion, 1-2
Helical packing, 199, 237, 484 of high vacuum distillation, 4\)6
diameter, 200 of vacuum distillatioll, 465
glass, 203 Holdup, 5, 12, 145, 176, 178,181,191,199,
low-temperature distillation, 401 217, 220
metal, 200 above given plate, 57
corrosion, 202 batch distj)]ation, 57,121,146
054 SUBJECT INDEX
Holdup (Continued) I
und hatdl size, 302
Ideal gas IaIL SeePt'I:(r:cl {IllS law.
and boil-up I'ILt.O, 12\1
Immiscihle system. Spo llU:!:llll'c, IJill ilJ'!/,
('u.l(:uhtioll, 112
imm.zscib{", lind Jl!i;clll,)'A, In Ii li.ico 1Il-
limitnt.ioIls, IH
jlonenl, immisci/,/c.
nnd chaser, 147
Implosion, 2(;1, ii28
column segment, 88
Inertin. offeet, of holdup on HC'IJal'atioll,
(JompositioIl, 116
152
condenser, 121, 251
Infrared an:.dy~i~ of distilbte, 312
dctcl'min:1tion, 112, 117
Inhibitor, polymcirization, 301
by mn,teri III bal:1flCO, 11 \)
Insulation, (iO, 180, 182, 1\14, 1\)8, 238,
IlOJlVOiati]cmatcri:d, 118
402,484. Sec also Jackel.
ent.ire eolumn, 113
thic:kness, 2:30
uqm~tioIl, 116, 123, 181
In tcrehangenble joillt. Boo J oill/.
and equilibrium, 12\1
Intermediate flow, 54.5
individUltl (',omponnllts, (i'~, 1.12
Intermolecular fo['(:es, off!'c:!; 011 azeotrope
inert.ia effeet on separation, 152
formation, 350
mass, 112
effect on extnwtive distillation, 350
mole~, 112
effect on sllblimntion, 600
negligible, 104
Interpret.at.ioll of pract.ical distillrLtioll,
packed column, 113
310
und per cent distilled, 147
Ionization gagc, 582
plate (real), 112
Isobutylene det.erminatioll, 416
plate colllmIl, 112
Isothermal dist.illation, 44(j
recovery, 303, 300
Isotopes, separat.ion, 94
and sharpness of separ:ltioIl, 147
spiral columll, 184
static, 118, 3D\) .J
theol'etic!tl plat,o, 113 .Jacket. See also Insulation.
theory, 120-129 cold air, 402
top plnt.e, 57 column, 178
totRl weight, 118 compensating. See .Jacket, hea/ed.
volatile component, 112 heat.ed, 182, 241, 245
volume, 112 automatic, 248
Holtlup-chul'ge ratio, 302 qunrtz, 241
Holdup-throughput ratio, 117 temperature, 247
Hotwell for ste/tnl ejector, 567 vnetlum, 176, 17\), 182, 184, 198, 215,
Hot zone, high vacuumsystelll, 566 222,239,402,404,484
H.T.IT., 77,14(;. See also If'!fJ. '['.1'. expansioIl, 240
Hydrate removal, 4 H) silvered, 230
Hyd-Robot" 218, 3\)0 Joint in distillation system, 292, 400
Hydroc:!tl'boll analysis, cut points, 431 flanged,5\)3
Hycll'oe!Lrboll distillation, Jow-tompOl'lL- Illetal, 595
ture, 420 glass-to-metal, 5U4
Hydrogen bond, and aZeotroJle forJl1:ttion, ground glass, 620, 624
31S\) high vacuum system, 5\)4
and heat of sublini,ttioll, 607 threaded, 596
Hydl'oPt~roxidll oxplosion hl1zarci, 300 v:teuum disti1lntioll, 47(i
H'ypel'-Cal eolumn, 250, 208 vacuum sublimRtion, 624
Hj'VilC lJUlnp, l:l6ii JUllctioIl, reference, thermoeouple, 278
SUBJEC'l' INDEX
K Lubrication, high v(~cuum systelll, 593

joints mul stopeocks, 203
Kettle, definition, 3
Kinetic theory, molecular distillation, 502
Knudsen equt1tion, 54.5
Jl.Iacl'oerystalline sublimate, 607
Knudsen flow, 467
Manifold for multiple sublimation, 625
L Manometer, 477, 489, 577. See also Gage.
backpressllrc, 200, 267, 2on, 272, ,'ill
I,uminur flow, 467, 560 oil, 577
Langmuir equation, 501 Jl.Ianometer fluid, 272
I,atent heat of sublimation, 60-1 J\[anostat, 281, 488
IA"ttent heat of vaporization, 1H, 173 bauk pressure, 272
Leakage, and polymerization, 301 bubbling, 283
and pressure control, 291 bulb area, 280
I,eak detection in high vacuum system, 586 Cartesian diver, 283
Lewis equations, 76 (liethylene glycol, 28-1
Lewis method, I-I.E.T.P. deterl11iulLtion, 67 fine adjustment, 200
Liebig condenser, 3, 251 low pressure, 284, 285
Liquefaction, 38\! mercury, 282
Liquid, gas confining, 417 nonelectrical, 281
gas-free, by qU[1sisublinmtion, 023 :'lensitive, 283
Liquid air as coolant, 389, 583 sulfuric acid, 282
Liquid density, 68 (mel temperature illJaugc, 287
Liquid nitrogen as coolant, 411, 583 tempemture-compcnsated,288
Liquid return rate, reflux, 2\!8 VftpOr pressure, 28(i
Liquid-vapor composition diagram, 7, Mantle, hetlling, 265
25-28 insulation, 182
Loss, percentage, in distillation, 315 Marslutll-Pigford equations, \)6
Low-boiling liquid, distillation. See Low- Mass Rpectroll1eter for leak deteetioll, 587
temperaiute distmatian. Mass velocity of V(LPOI', 78
Low-pressure distilhtion. See r acnllll! }'Iatehett~Levine still, 52\)
distillation. MaterillJ IJal:tnee, for phte l'fticulations,
Low-temperature distillation, 1\)1, 24-1, 4fl
260, 273, 380-461 Rayleigh pl'oeeciul'c, \I\)
automatic, 427 :\ht.cri:tl exchang(1 in fmc:tion:tl llislilla-
batch size, -121, 441, 444 tion, 175
(',alculation, 428 l\Iatel'ifLl tmnsfer, in paeked eolumn, 1\)\)
(~ondenser, 408 in wetted-surfuec column, 178
errors, 427 l\!c;C(lbe-Thicle procedure for still-prod-
inverted, 442 uct eomposition calculation, 5B, 12D
limitations, 431 graphic, 50, 117
low pressure, 442, 44.4 McLeod gage, 490, 577
micro, 441 McMahon paeking, 204, 208
packing, 3!H McMillan liquid [lir distributor, 410
precision, 431 McMillan pac:king, 400
prcssur,e, 442 Mean free jlltth, 463, 546, 625
principles, 434 undeflected, 548
procedure, 419 Melt, qU(1sisublimlltion froUl, 603
simple, 443 Mercury, coefficient of expansion, 28\!
special apparatus, 441 Mercury mllllostat, 282
total condenser, 413 Mercury scal, 407
656 SUBJECT INDEX
MereUl·y vapor pump, 568 Moncl packing, 208

Metal packing, 208 Mounting, still, framework, 294
Metll!>Tle fruction, sepumtion, 422 Multicomponent mixture, Imteh distilla-
Microbllr, definition, 541 tion curve, 13U
Miel"o columIl, Simons, 441 composition rehLtions, 1)1
MicrodistilIlltion, low-temperature, 441 distillate composition, 140
Micron, definition, 541 H.E.T.P. determination, 53
Microsublimation, 630, 638 Hayleigh equation, 108
Miscible system. See Mh;l,w·e, binaru, Murphree plate efficiency, 31
miscible and Ati;vture, rnultirornponenl,
misc£ble. N
Mist trap, 48fi Nickel paeking, 204
Mixing in wetted-surfaee (·,olumn, 178 Nitrogen, liquid, as coolant, 411, 583
MixtureCs), definition, 6 Nonideal system in extructivc distilhLtioll,
binary, immiscible, 15 322
miscible, 18
partly miscible, 29
o
classification, 15 Oil, cOIlstunt-yield, 519
close-boiling, separation, 94, 172 pump, 488, 565, 574
distillation of, 5 Oil bath, 264
ideal, 18 Oil immersion pump, 487, 563
volatility, 20 Oil mllnometer, 577
multicomponent, immiscible, 29 Oil vapor pump. See Pump.
miscible, 30 Oldel'shaw column, 222
batch distillation, 139 Open tube column, 86
partly miBcible, 29 Operating line, batch distillation, 57
nonideal, 23-25 Operating-line equation, 47
volatility, 20 Orsat absorption apparatus, 410
Molecular ussociation, 49, 359, 607 Othmer still, 326
Molecular collision and viscosity, 545 Oxygen determinu,t,ion, 416
Molecular dUtll" common guses and VllPOI"S,
p
544
Molecular diffusion, 499 P!cckcd column. See Column, par-Iced.
Molecular diameter, 86 Packing, 4, 75, 175, 198
common gases and vapors, 54'1 ball, 207, 209
Molecular distillation 497, 541. See [Ilso Berl saddleR, 207
Vacunm distillation, high. brass, 208
Molecular flow, 545, 551 Buta-Heli-Grid, 394
Molecular still, 497 Carborundum, 207
sublimation, 613, 625 chain, 207, 209
Molecular velocity, 86 choice, 299
Molecular weight, average, fi8 corrosion, flf)
in codistillation, 17 density, 208
. and distillation, 497, 500 drying, 310
Mole fraction, in Bowmall equation, 88 Dufton spirals, 400
in Cohen equation, 94 dump, 198
and equilibration time, 129 gauze, conical, 210
in Fenske equation, 38, 351 spiral, 184
on given plate, 92 glass, 203
in operating~line equation, 47 helical, 199, 237, 484
in RayJeighequation, 102, 121j 140 diameter, 200
and relative volatility, 6,21,318 glass, 203
in stepwise calculation, 127 low-temperature distillation, 401
SUBJ~CT INDEX (;57
Packing Phlegm, dcfillHion, ~i

helical (continued) Physical propertic"q, a14. Spn also Boilina
mct:1I,200 point, Refracti!'!! index, et.(·,.
corrosion, 202 Pig rcc:eiver changPI', 480
ITcH-Grid, 19S, 211 Pilot mat(]ri[~1 for elimination ciistiIl:1tioll,
low-temperature distillation, 397 522
Heli-Pak, 205, 209 Ph'ani gage, 513, 570, 587
low-temperature distillation, 394 Plate (real), 4, 170, 210
McMahon, 204, 208 bubblo-eap, 2H)
McMillan, 400 fritted-glass, 223
metal,208 metal, 22.1
miscellaneous, 209 perforMed, 219
Monel,208 theoretical, 6, 7, 8, ::10. See al~o
nickel,204 H.E.T.P.
preformed, 210 height equivulent to. See H.E.T.P.
prouedure, 237 Plateau iu distilhttioll curve, 310
protruded, 69, 208 PInto column, !), 176, 211)
Raschig rings, 207 continuous distillation, 47
shape, 199 extractive distillation, 3[)4,
size, 199, 235 Plate composition, 49, 64
Stedm[ln, 210 differential equutions, 125
support, 236 stepwise cnleuhtion, 126
surface area, 76 Plate efficiency. See Efficiency, plat~.
tamper, 237 PIt.te equivalent, 56, 298
type, 146 Plnte-equivalents procedur(', calculation,
wire-coil, low-temperature distillation, 68
3m) Phtte number :uul eompositiou, 44, 93
Packing factor, 75 Plates, infinite, 111
Partial condenser. See Conden,~cr. and relative volatility, 320
Pnrtiul pressure, 14, Hi, 18 theoretical, Ilumber of, 13. See al~o
Pnrtial reliux, 5, 106, 176 H.E.T.P.
H.E.T.P. determination, 67, 82, fll hatch distillation, 162
sepamtion, 5\l and boiling points, 172
stilI and produet composition, 5n determination, 32-59,6.'\-68
Partitioning, 176. See also Reflux ratio. partial reflux, 33
COIltinuous, 251 hydrocarbon analysis, 4:~5
flow-regulating, 251 minimuIU, 164
micrometer, 253 ll,lld reflux ratio, 171
intermittent, 2.51, 2.54 !\nd relative volatility, 171
stopcock, 252 Plate-to-plate calculations, 4{l
theory, 131 Plate-to-plate compositioll, 50
timing, 251, 37H Plotting distillation curve,310
vacuum distillation, 484 Podbielniak column, 389, 403
Path, ulltieflp.cted, 4(19. See also Mean Podbielniak still pot, 406
free path. Poiseuille's law, 560
Penn-State still, 327 Polarity, solvent, :~38
Perfect gas law, 16 Pole-height equations, 166
Perfect simple distillation, 8, 30 Polymerization, avoidance, 300
Perforated plate, 219 . Potentiometer for temperature measure-
Peroxide explosion hazard, 300 ment,278
Phase rule, 15, 358 Powder sublimate, 607
Phase separation, 366 Power ratio, heater, 263
658 SUBJECt' INDEX
Precipitation, fractional, 10 1'I'Op),lolle, t.ake-off rate in distillation of,

Preferential adsorption by [wtiVltted char- 42·1
con1,419 Prot.ruded packing, GH, 208
l'reflooding, and condenser capacity, 250 Pump, backing, 50:3, 573
and Heli-Grid packing, 217 Dooster, 570
low-temperature distillation, 422 choice, 573
and manometer, back-pressure, 272 condensation, 50n
alld packing support, 2:n diffusion, 488, 5GB
procedure, 30.5 fore, 563, 57a
and l'8ctification effieicncy, H1D mcchalli cal, 5G3
test procedure, 36 ml~rcury vapor, 568
Preformed packing, 210 moderate vacuum, 487, 5(i3
Pressure, and azeotropi~m, 358 oil-immersion, 487, 563
and boil-up rate, 299 oil vapor, 568
control, 280, 494, See also I1famosia!. fmet.ionating, 575
avoiding oscillation, 291 liquid, .574
cascadc, 282 precautions, 575
diminished, See Vacuum disl'ilZation, oil vapor ejector, 572
moderate, solvent recycle, ;345, :1-16
effect on condensation, 474 steam ejector, 507
on separation, 84 Toeplel', 417
011 vaporization, 474 water aspirator, 487
exhaust, 573 water ejector, 567
low, See VaclLwn distillation, moderate, Pump oil, 488, 565, 5H
low-temperature dist;i]lation, 413, 442 color, 575
partial, 14, 16, 18 Pump speed, 573, 597
reduced. See Vacuum rlisiillat'ion, mod- Purit,y of distillat.e, determilmtioll, 307
erate.
and temperature, 280 Q
total gl1S, 16 QU!1ekcllhush-Steenbock still, 536
vapor. See Vap01'pI'CS8ure. Qm1rtz j!1cket, 241
Pl'essure column. See Column. Quasisublimution, 603, 612
Prcssure drop, and boil-up rate, 270 entrain(Jr, 6:n
Bruun column, 222
definition, 9 R
empty tube column, 181 Haoult'slttw, 18,20,21, 90, 91
and feed l'I1te, 270 Raschig ring~, packing, 207
and flooding, 9 Rate, boiluj), See Boilttp rate.
and 1LE.T.P., 69 of distill atiOIl, \)
Oldel'shaw column, 225 high vacuum, 500
plate column, 176, 220 of sublimation, 008
and total preaBure, 468 take-off, 178, See I1lso Part£Uoninr/.
vacuum distillation, 484 Ratio, reflux. See Reflux ratio.
wetted-surface column, 178 Rayleigh equation, 99,104,120, 13H
Pressure gage, See Vacuum gage. algebl'!1ic solution, 100
Pressure regulator, See M anostat. graphic solution, 102
Pressure units, 541 multicomponent mixturo, 108
Pretreatment of distilland j 300, 417, 491 nomograph, 104
Probe gas, leak detection, 586 Receiver, definition, 3
Propane, take-off rate in distillation of, Bupport, 294
424 vacuum distillation, 480
SUBJI-;CT INDEX G59
Receiver ehauger, 480, 486, 48(; RefrigeratioIl, ·108, 5s:.~

ball-and-soeket, 48:~ Relative volatility, 20
Rtopcock, .182 batch distillntioll, 145,1.5:3, Hi\)
Hectiiication,3. Sec also ,separation. [Lnd boiling point, 22
contact, ·1, 87, ·109 (~nd concentration, 27
countercurrent, 40n deJillitioll, ti, 318
thel'mal, ·1, 85, 87, 470 equation, 21
theory, 47:3 fmd H.E.T.P., 8, 168
ltcetific:1tion unit, butch, 17;j multicomponcnt mixture, HI, 108
l~ectifying sectioll, Ii 5 Itnrillumber of plttteR, ] 71
Hedistillation, multiple, 470 :Lnd pressure, '1114
Hcduced ]lre~,ure. 8co VIWIHun distil/n- and sepamtion "harplless, 11
tioll. and vapor-liquid equilibrium, 2li
Hpflux, ·1, j 75, 17(; Hcsidue, and holdup, 30()
lillitt~. See Partilll njlux. low·tempcratum distillation, ·120
flow l:atc, liquid, ·17 Hesistanee, gus flow, 5-l2
iufinitfl. See 'l'o(i!l reflllx. Rcsistaneo thermometer, 2·jS, 27\), 580
low-temperature distilhttioIl, 421, 425 Hetort, 3, 47G, 629
minimulll, 61, 111 Retul'll-to"~till proeedul"c, ,16
partial. See Partial rejlit.l:. Riegel still, 529
totul. Soe 'l'oIILi l"(!jlll.T:. Rosin oil, holdup determiu[Ltioll, un
Reflux dividing. See Partitionill(l. Rot!lIueter, 3·15
Reflux partitioning. See Partitionin(j. Rotllting column, 226, 227, 2:n
Reflux liquid mtc, 8(1
Reflux ratio, 5, 13. See also PartitioninlJ.
batch distillation, 145, 153, 169
s
and boiling points, 173 Safety, vacuum distillation, 476
calculation, 166 Salt bath, 26,1
choice, 307 S~llting out, 367
control, 176,260, :ng Sample, gas, pretrcatmen t, 417
determination, 13, 200 H.E.T.P. determinatioll, 45
and distillate composition, 131 still, 32
aud distillation curve, 308 S[Lll1pling apparatus, 33, 417
empirical, 307 Rand bath, 2M
gradual increase, 132 Semimicro distillation, 479
and H.E.T.P., 146 Scmimicro falling film still, 530
low-temperature distillation, 425 Semipermeable barrie!", 508
and number of plates, 171 Semi-Robot, 3111
and plate efficiency, 14G Sepal'l1ting powel', definition, 6
and separation sharpness, 11,45 Separation, 90. Bee also Rectification.
stepwise increase, 131 dose-boiling Ii quids, 3
and time of distillation, 12 holdup inertia effect, 162
and timer ratio, 252, 255 liquids from solids, 2
and yield fraction, 138 other than by distillatioll, n
Reflux temperature, measurement, 273 partial reflux, 59
Refractive index, ·13, 97, :307, 312,494 pressure effect on, 8·1
and composition, 313 sharpness, 11, 13, G7, 107, 146, HIll, 225
H.E.T.P. determination, 41, 70 and composition change, 127
low temperature, 417 aud holdup, 147
and temperature, 314 ::Jtandard, i 70
Uefl'igerant~, 411, 583 total reflux, 53
660 SUBJECT INDl~X
Separation factor. See Emichment Start of fraetional distillation, 305

factor, theory, 94
Separation stll!!;e, height of, 88 Steady-state distillatiulI, 47
Separation steps, number of, 76 composition, 59
Shoek wave ill V:1por flow, '168 Stoam distilbtion, Hi, 17, :320, 358, :374-8
Shot bath, 20,1 Steam vaporization efficiency, 377
Side arm, properties required in modemtB Stearic acid, holdup determination, 11\)
vacuum, -180 Stedmltll calculations, 160
Silicone, 476 Stedman column, 210
Simple distilhLtion, 2, 98, 389 Still, alembic, high vacuum distillntion,
perfect, 8, 30 528
Simons column, 413 Almquist, 532
micro, 441 boiling point, high vacuum, 511
Sintered glllsB to avoid bumping, 492, Breger, 5:39
See ltlSO Bumping, Carothers, 5:31
Slit, effect in high vacuum system, 558 centrifugal, 500, 508, 513, 540
Smoker equation, 53, 67, 107, 117, cone-type, 509
182 laboratory, 514
Smoker procl~(lure for Htill-product emIl- eJeaning, 309
position calculation, 59 cold finger, 509
SIlOW Bublimnt,e, 610 eoncentric tube, 4B9
Solenoid part-Hioner, 255 cOlltltmirmnts, 300
Solenoid vnlve for j1l'eSSUl'e control, definition, 3
292 Detwiler and Markley, 537
Solid distillate, 299,479, 480 equilibrium, 8, 26
Solution temperature, critical, 33-1 azcot,l'opie distillation, :368
Solvation in extractive distillation, 338 Othmel',820
Solvent distillation. See Extractive dis- Penn-State, 327
tillation, extractive distillation, 343
Solvent for extractive distillation, :U8, falling-film, 500, 523, 532
320, 325 Hemimicl'll, 539
(lhoice, 338 flowing-tUm, 500, 508, 61:1
concentratioll, 324 Gould, 531
dipole moments, 339 Hickman, 528, 58,'\
efiectivenesH, 333 high vacuum, 498
polarity, 338 cascade, MO
preheating, a43 laboratory, 527
recycle, 344 micro, ,531
removal by vaouum distillation, 401 horizontal, 477
Sorel graphic proceuure, 41) Matchett and LHvine, 52\)
Sorel-Lewis calculation, -17 molecular, 4.97
Sorel method for phtte-to-plate calcula- sublimation, G13, (125
tions,67 mounting, 294
i:'lpinning band column, 2:30, 2:31 pot-type, 528
Spiral column, 178, 188, 184, 188, 190 Quackenbush and Steenbock, 5:36
pitch chang!', 186 Riegel, 529
;::ipiral tube oolumn, low-tempentture Strain and Allen, 531
disti1latioIl, 401 Taylor, 538
Stability of charge, 299 tubular, 532
Htandard distillation, 170 vacuum distillation, 476
l:!tandard -taper joint. See Joint, vertical, 483, 485
SUB.JEC'l' INDEX OG[
Still hearl, 3, 4, 175, li6, 3n. See also Superfmctionator, 22fi

Condenser, Part.ilionill(l. ::4uperhcating, 475
Cot'ad,250 local, -lD2
Still mounting, framework, 2lH sublimatioll, fill)
Still pot, li5, 2nD Support, for distillation column, 294
concentric tube, 527 packing, 2:3G
cy lindrimtl, 2m, 264 Surface, li1lllid-fiI11l, 175
high vacuum, 409, 527 Surface tension and r,olumn l'hoice, 225
low-temperature distilhtion, 40G Surge tank, 2\)1
Podbielniak, .JOG Symbol index, xviii---xxvii
Still Rample, 32 • high Va(~llllm di>'tilhtioll, ·105, 5·14
Stirrer seal, 505
Stirring,2:U
magnetie, 493 T
Stopcock, high vunuum syRtem, 507
Take-ofl'. See Partitioning.
ordinary distillation, 2D2
Tamper, URC in p:wking, 2:37
Strain-Allen still, 531
Taylor still, 538
Streamlined flow, 178
Teflon to avoirl bumping, -1\]2. See also
Stripping section, delinit.ion, 4
Bumping.
equations, 5:3
Tempcmture, critical, 605
Sublimand, 603
eliminatioll, high vaCllUIll, i\ 17
Sublimate, G03
gradient, effect on flow, 8::1
macrocrystal1inc, 607
llleasuremoIlt, 2(12, a07
Sublimation, 60a---M5
extractive distillation, WiD
condenser temperature, (i08
high vaCUUlll, 510 .
cntrainer, 609, 61ll, 630
low-temperature rlistillatioll, 411
equation, 609
reflux, 273
vacuum, 639
vacuum distillation, 4m
equation, empirical, 604
and pressure, 280
flash, 640
solution, critical, 334
fractional, 623, 626, 637, 643
Terminology, 1---9
heat of, 604
Tesla coil for leak detection, 586
high temperature, 640
Test mixture, choiec, 35
latent heat of, 60·1
H.RT.P. determination, :~8, 43, 70-N,
low temperature, 636
298
multiple manifold, 625
Theoretical plate(s), 6, 7, 8, :30. See also
simple, 608, 629
H.E.T.P.
steam, 609
height equivaJentto. See H.E.T.P.
vacuum, 610, 620, 633, 644
number of, 13
entrainer, 627
tletermination, 32-59, 65--68. Slle
Sublimation CUl'Vtl, 604 also H.E. T.P., deiponination.
Sublimation pressure, 606 partial reflux, 33
Sublimator, 613 hydrocarbon analysis, 435
horizontal, 629 Theory of azeotropic distillation, 357
U-tube, 628 of distillation, 1---174
vertical, 614 limitations, 13, 15
Subliming point, 60.',) of extractive distillation, 322
practical, 605 of high vacuum distillation, 501
Sulfurio acid manostat, 282 of vacuum distillation, 467
Supercool column, 405 Thermal exposures, various stills, 507
662 SUB,JECT INDfo;X
Thermal h[Lzard, rerluetioll by vacuum Totlll reflux (continued)

distillation, 505 ordinary distillation, 17G
Thermal rectificatioll. Sec al;;o Rcct~fim and part.itiuner, 25li
tion. separation, 68
column, 472 time of, 306
Thermistor, 581 Tower. See Colllllln.
Thermocouple, 2'17, 275, 411 Transfer, vapor-liquid, 178
calibration, 278 Transfer unit, height of. Hoc H.T. U.
cOllstruction, 276 Transformer, 263, 2!l'1, 345
leads, 279 Trap, cold, 304, 4SIl
multiple junction, 277 high vacuum system, 555, 582
Thermocouple gage, 579 hot, 566
Thermocouple well, 27ti va.cuum distill!ltion, 488
Thermometer, 275 Triple point, sublimation, 604
partial immersion, 275 Trouton's rule, 22, 172
resistallce, 248, 279, 580 Tube, concentric, high vacuum system,
total immersion, 275 558
Thermometer well, 260, ,i7!) high vacuum system, 553, 5(11
Threaded column, 186 Tubing, precision-bore, 197
Throughput. See also Roil-'II1) rain. resistance, 54:3
bntch distillation, 145 rubber, high vaouum system, 592
Bruun column, 221 Tubular still, 532
and column efficiency, 221 Turbulence and column efficieney, ]\J\l,
definition, 9 230
empty tube colu.mn, 179 Turbulent flow, 199, 4ti7, 562
and H.E.T.P., 146 Two-phase system, distillu.tion, 301
Oldershaw column, 225
plate column, 220 u
n.nd plate efficiency, 146 Udder receiver clULllgcr, 'ISO
and SCIJa!'a tiol! sh arpness, ] 2 Uninsulated column, 186
and time of dh;tillation, 17(1 Unobstructed path distillation, 41)7
and turbulence, 1!l9 USA (usual simplifying u~sumJltions), '17.
Throughput-holdup mtio, 147 106
Time of distillation, 12, U5, 176, 22:3
calculation, 133 v
low-temperature, 424 Vacustat, 578
Time of equiJiUl'lLtioll, ]29, 141>, 215, 221, Vacuum, ultimate, .573
223 Vacuum distillation, columns, 18fi, 2;~ I,
Time of extractive distilbtion, 342 ·168,470,47!l
Time of pl'cfloocling, 305 entrainer, 47S, 403
Timer, 255, 345 high, 'lH5-!l02
Time of sublimation, 601l analytical, 517, 52!l
Time of total reflux, 30G application, 513
Toepler pump, 417 connections, 502
Total condenser. Sce Condenser. cyclic, .511
Total reflux, composition, still and prod- laboratory, 601
uct, 63 plumbing, 551, 592
definition, 5 simple, 509
H.E.T.P. determination, 32,47,80, 91, symbols, 495, 5·1'1
107 technique, 50\)
SUBJEC'r INDEX
Vacuum distillation, high (conthLllfY/) Vllpor pressure (COlltilllll"d)

theory, SOL immiscible Rystem, 15
moderate, ,1(l:)--1!J.l partbl, 5
apparatu~, -t7li stellIll distillation, 17
eondcnsPl', 480, 48G allll temperature, 1H, 20, 21, 302,
cut point, 4n:! 604
partinl cOlldc:Jlser, 475 and vapor compositioll, 1·1
pro,:eclurc, ·Hll and volatility, a18
~pintl eolumll adjustment, ISO Vapor pressure mtio and n:l:ltivc, vol-
theory; {tl7 aWity, (i
and pressure d.op, 1i(i Vapor rate, 4!l
and relative volatility, 12 Vapor sampling, :~2
safety, ·17(i Vapor Ycloeit,y. See'l'h)"l)lIghl'lll.
temperature IllCa~UI'mncnt, 4fl~~ nUl:-;~, 78
and thermal rectificat.ion, 8:), ,liD radially lwmagf,d, 81
V lLCUUlll littings, SD2 Vapor vis(lo;ity, 8::1, 181
Vll,CUUm gage)s, 48\), 5ill Vari:L!). Sec 'l'rrw.,8fonncl'.
locfLtion, {DO Veloeity, mole(~lt1a.r, 86, 50:3
Vacuum jaeket. See Jacket, V(LCIIIW~. vapor, maSH, i8
Vacuum sublimation, lilO, 020, G3:3, fH4 radially averaged, 81
entraincr, G27, O:3!l VertiefLl still, ·183, -185
Vaouum system, high, 540-602 Vigrnux IJolumn, H13
eaJeull1Lions, ,5m Vis(!nsity, Jicluhl, 181
integrated, 507 aud molecular collisioll, 5-15
pump speed, 5\)7 vapor, 83, 181
Valve, diaphragm, low-iemperaturc dis- Viscous flow, 5-15, 560
tilbtion, :3D-1 Volatility, fl, 20, 31.8
high vacuum H.)stmn, 55fi, 5\.17 rc~htive.See Relalive volatility.
]lurtitioning. Sec PartitiOllill(j. with steam, 2U. Sec l~lso Sleam diN
throttling, 201 tillf!tioll.

Van del' \Vaals fOl"l~eR, (JOG Voltage, heater, 2(J8
Van Laar equation, 24
Vapor oomposition, :31, 77 w
Vapor dHIusivity, {6D W factor, 551
V npol'iztttion, ·!7·1 Warm-up time. See 1'-imc of eqllililml-
latent. heat of, H), 1T:I tion.
partial, ,b Wateh glaRS Hublimator, 014
Vapor-liquid eOIll]lositioll diagnuD, 7, Watm', dissolved. Sce Wet material.
25-28, 37 Weathering, low-templlrature distilhtt,ioll,
Vapor--liquid cquilibrium, detcmllina- 407.
tion, 8 Weighing fractions, 30-1
Vapor-liquid equilibriuIll relation, 1 t Weight traction, 120
Vltpor pressure, analysis by, H7 Weir height and plate eHicicncy, 225
nnd l,zeotropism, 370 Westhaver and Kuhn eqUlttions, 7n
and boiling point, 10 Wet material, 221, 225, :301
eodistillation, 17 low-temperature distillation, 4H)
and composition, 24, 30 Wetted-surface column, 175,178
definition, 5 Widmer colUlIln, 184,245,484
cquution, 10 Wire-coil column, 183, 190, 193
ideal mixturc, 18 low-tempemture distillation, 3Dn
664 SUBJECT INDEX
y z
Yield fraction, calculation, 133 Zero shift in vacuum gage, ill presence of
and reflux ratio, 138 organic vapors, 581
Cunlulative Indexes, Volulnes I-V
AUl'nORINDEX
PAGES
Anuerson, J. n. See Swietoslawski, W.
B:Liley, E.n. Sec Nichols, J. B.
13:.Lle, IV. F., and Bonner, J. F., ,Jr., Radiollctivity . .... I, Part II 1U27-W90
Bauer, N., Density ......................... . ' I, P~LI't I 25:3-296
Bauer, N., and Fajans, K., Refractometry .... . I, Part II 1141-1240
Boekelhcidc, V. See Noyes, W. A., ,Jr.
Bonner, J. F., Jr. Sce Bale, IV. F.
Bowmall, J. R., and Tipson, n. S., Distillalion I/.miel Modl'l"IIW
Vac"lltlln ................................... . IV 463-4\)4
Brockway, L. 0., Electron Diffraction .... . , .. , ., ......... . I, PlLl't II 1109-1140
Broughton, G., Solvent Rcmot'al, ElJap(Jmtio1i, and Drying . .. . III 605-634
Carlson, C. S., Extractive and Azcolropic Distillation . .... . IV 317-387
Cassidy, H. G., Adsorption and Chromatography . ............ . V 1-327
Craig, D. Sel} Craig, L. C.
Craig, L. C., and Craig, D., E.ctractiof! (Utd Distliblliioll ..... . . III 17HHl
Cummins, A. B., Filtration ..... ......... " ......... . III 485-603
Donnay, J. D. H., CTystalloclw1lZi('(!I.:171aly.~i~ .... ..... . I, Part I 1017-10:39
Egly, R. S., Heating and Cooli:ng ..... ....... , .... . III 1-U8
Faj[Lns, K. See Bauer, N.
Fankuchen, I., X-Ray Diffmctum .. . , ... . I, Plll't II 1073-1108
Geddes, A. L., Di.ffusivity ... ............................. . I, Part I 551-619
Glasebrook, A. I,., and Williams, F. K, Ordinary FractlonalDis-
tillation: Apparatus .................................. . IV 175-2\)4
Golding, H. B., Centrifugin(/. . . . . . .. . ................... . III 143-170
Harkins, W. D., Properties of M onolayers and Dl!plex Films. , .. . I, Part I 427-485
Harkinll, W. D., Surface and Inte11acial Tension .. ... , ...... . I, Part I 355-413
Heller, W., Polarimetry .... ................... ' .......... . 1, Part II 1491-1610
Hecker, J. C., Distillation under High Vacu'um: The Yacum!!
System .............................................. . IV MO-M5
Hofmann, M. P., Mixing: Sppcial Equipment .. " .. . III 131-'1'12
Jelley, E. E., Microscopy . ............................... . I, Part I 847-981
Komarewsky, V. 1., and Riesz, C. H., Catalytic Reactions . ... . II 1-78
Mark, H. See McGoury, T. E.
McGoury, T. E., and Mark, H., Viscosity . ................ . I, Part I 327-353
Michaelis, L., Magnetic Susceptibility . .................. , .. . I, Part II 1885-1926
Michaelis, L., Potentiometry . ................... , ......... . I, Purt II 1713-178·1
Moore, D. H., Electrophoresis • ............................ I, Part II 1085-1712
Miiller, O. H., Polarography ..... .......... , ............ , .. I, Part II 1785-1884
Nichols, J. E., and Bailey, E. D., Ultracmtrijllge . .......... . I, Pltl't I 621-730
Noyes, W. A., Jr., and Doekelheiue, V., Photochemical
Reactions . •.....•..••................................. II 79-141
Peacock, M. A., Crystal Form . ......... , ................. . I, Part I 983-1015
665
666 CUMULA'l'IVII, INDEXES, VOLUMES I-V
VOLUMg PAGES
Perry, E. S" Di8tillation u.nda IIIOil YaWl/III, , , , . , , IV '195-540
R.iesz, C, I-I, See Komarllwsky, V. I.
R.ose, A., and Hose, E., Distillation, 'l'}WOl'!l, , ..... IV I-II{
Rose, A., and Hose, E., Distzllation of Liquefied Gases ((/!(} Low-
BoiUna Liquids . ... , ........ , , , ...... , ........... . IV 389-461
Rose, E. See Rose, A.
Rushton, J. H., Mixing. , ...... " "" , . , , , , " , , .. III 99-130
Shedlovsky, T., Con(l1l.Clometl'l1 . .. , . , , , . I, Part II 1651-1683
Skau, E. I.., and 'Wakeham, n., ~Idhn(, nnrl Frneziny 'l'cllljJwa·
tures .. , . , , . , ...... , , .... , .. , . , ...... , ............... . I, Part I 49-105
Smyth, C. P., Dipole Moments." ...... .... . I, Part II 1611-1650
:Stauffer, R K, Dialysis and Electrodialysis .. III 313-3(i 1
Stewllrt, D. \Y., Mass Spectrometry . ....... . 1, 1'ILrt II H191-2058
SturtevlluL, .J. M., Culorimel'l'l/ .... ......... . 1, Part I 731-84.1
StmteVl<llt, ,J. Iv1., 'l'empcmtw'e Control. . . . . . . ........ . I, Part I 29-'[8
StllrteV!1ut, ,J, M., Temperature 1I1eas1l1'B'IlIc1It . .............. . I, Part I 1-27
Swann, S., .Jr., FJ/ectrolutic Reactions .... .................... . II I-t:1-208
Swictoslawski, W., and Anderson, ,J. n" Boiling and Coruien.m-
tion 'l'emperatures . ..... , ......... , ...... , ............. . I, PInt I ]07-140
Thomson, O. W., Vapor Pressu1:e ... ,.. . . . '" .. ' .. I, PCIl't I 141-251
Thomson, O. "V., Paracha?' .. . ,. . . . . . . . . . . ... , ... . I, Part I -H3-427
Tipson, R. S., C1'ysta1l1:zalion (/ind Recrl,~t(Ll~i4a.l-i()n .. ...... , .. . HI 3(:\3-4840
Tipson, R. S., Sublimation . .. , . , .................... , IV nO:3-M5
Tipson, n.. S. See also Bowman, ,J. R.
VoId, M . .T. See VoId, It. D.
VoId, R. D., and VoId, M. J., SolnbUil!l ... " .. .......... , .. I, Part I 2!)7-:{2!l
Wagner, R. H., Osmotic Pres.nal'! . . , ...... , ............... . I, Part I 487-541l
'Wakeham, H. See Sleau, E. L.
'WeRt, W" ColO1'imei,1'Y, Photometric Analysis, ji'lu01imetl'lI, ant!
'1)m·bid~melry . .. , ... , .............. , ............. , , , .. . I, Part II 1899-1490
West, W., Spectroscopy and Spectrophotometry . ............ , . I, Part II 1241-1308
Williams, F. E" Ordinary Fractional Distillation: Procednre .. . IV 295-:UH
William., F. E. See also Glasebl'ook, A, L.
SUBJECT INDEX
Adsorption (II. O. Cassidy), . , .... , ........... , ... . V 1-327
Azeotropic DiRtillation (C. S. Carlson) _......... , ......... . IV 317-387
Calorimetry (J. M. Sturtevant,) ... , ... , ... , ....... , .. , .... . I, Part I 7:n-845
Catalytic Reactions (V. 1. KomlLl'cwsky and C. H. IUesz) ..... . II 1-78
Centrifuging (H. B, Golding) ... , .. , , ... , .............. , , .. III 143-170
Chromatography (II. G. Cassidy) ............ , ...... , , ... . V 1-327
Colorimetry (W. West) .......... , , , .......... , .... , ..... . I, Part II 1399-1490
Condensation Tempemtul'cs (W. Swietoslawski and J. R. Antlcr-
son)., ............. , ........ , .... , .............. ' , ... . I, Part I 107-140
Conductometry (T. She{lIoYsky) .... , , : ................... . I, Part II 1651-1683
Cooling (R. S. Egly) .................. , , ... , ............ . III 1-98
Crystal Form (M. A. Peacock) ..... , .... , ... , , ........... . I, Part I 983-1015
Crystallization (n. S. Tipson) ...... , , ... _....... , .. , ..... . III 363-48,1
Crystallochemical Analysis (J. D. H. DOllllllY)., .. ,.", ..... . I, Pl1rt I ]017-1039
cu~rurJA'l'lVB INDEXES, VOLUMES I--V 667
VOLu~m PAGES
Density (N_ Baner) ....................... , ............. . I, Part I 253-29(j

DbJrsis (R. E. 8tauffer) ....................... , ........ , III 313-361
Diffraction, Electron (L, 0, BrockwtLY). , , . , ......... , ... . I, Part II 1109-1140
Diffraction, X-Ray (1. l"ankuchell) .... . I, Part II 1073-1108
Diffusivity CA. L. Geudes) ................. . I, Part I 551-619
Dipole Moments (C. P. Smyth), ...... . I, Part II 1611-1650
Distillation, Theory (A. and E. Rose), . . . . . . . . . . ....... . IV 1-174
Distillation, OrdinHl'Y Fractional: Apparatus (A. L. GlnHebrook
and F. E. Williams). ", .. , , . , ..... , .. , ................ . IV 175-29-1
Distillation, Ordinary FraetioIl:LI: Procedure CF. E. Williams) .. IV 295-316
Distillation, Extractive and Azeotl'opic (C. S. Carlson), . , . , . , IV :317-387
Distillatiou, of Liquefied Gases :LIlU Low-Boiling Liquids CA. :mll
E. Rose). , . ,,' .... , . , .............. , .. , , ... , , , ...... , . IV 389-461
Distillation, und{~r ]vloder:J.tc Vacuum (,J, R Bowman and R. S.
Tipson) .......... , .... "., ..... , .... , ... , ...... ,., .. . IV "163--19-1
Disti)btion, under High VammI1l (E. S. Pel'l'Y)· ............ . IV 495-540
Distillation, under High Vacuum: the Vacuum System (,I. C.
Hecker) ................ , .. , ................... , .... . IV 540-645
Distribution CL. C. Craig and D. Cmig), ... , ............. . III 171-311
Drying (0. BroughtoJl) ................. , ... , ......... . III 1305-(j;3-!
Duplex Films (W. D. Harkins). . . . . . . . .......... , . 1, PILl't I 427-·1-85
Electrodialysis (R. Eo Stauffer) ..... , . . . . . ....... , . III 313-361
Electrolytic Reaetions (S. SW;Ulll, Jr.) .. , , ........ . II 1-13-208
Electron Diffraction (L. 0. Brockway). , ....... , .......... . 1, Part II 1109-1140
Electrophoresis (D. H. 1\loore) .... ,....... . ....... . I, Part II 1685-1712
Evaporation (0. Broughton) ............. . III 605--(}34
Extraction (L. C. Craig and D. Craig) .... . III 171-:311
Extmctive Distillation (C, S. Carlson) ... , . IV 317-387
Filtmtion CA. B. Cummins) ............... . III 485-G08
Fluorimetry CW. West) ............... , ... , , . , .......... . I, Purt II 1:3fJ!H4()0
Fractional Distilhttioll: Apparatus (A, L. Gbsobrook [LIltl F. Eo
Williams) ............... , ...... , .... , ... , .. , ......... . IV 175-291
Frl].ct.ion;t! Distillation: Procedure (If. E. Williams). , ........ . IV 295-310
Fr(;ezing Tempemtul'es (E. L. Slmu and H. ·Wakeham) ...... . I, Part I 4!H05
Heo.ting (It. S. Egly) .......................... , .. , . , III 1-98
High Vl10uum Distillation (E. S. Perry) ................... . IV 4.95-,,)40
High VaeuumDistillation: The VacuumSyst!'IU (J. C. Heeker) IV MO-645
Illterfac:iul Tension (W. D. Harkins) .. , .. , ........ . I, hut I :355--11:3
:\Iagnetic Susceptibility (L. Michaelis) ....... . I, Pltrt II 1885-1926
Mass Spectrometry (D. W. Stewart) ............... . I, Part Il 1m)1-2058
l'.relting Temperatures (E. IJ. -SkUll and H. Wakeham) ...... , I, PtLrt I -1D-105
l\Iicl'oscopy (E. E. J ellcy) ...... , ..... , .......... , . , ..... . I, Part I 8-17-981
Mixing (,J, H. Rushton) ........................... , ..... . III 90-130
.1vIixing: S]lceiul Equipment (M. P. Hofnmnn) ..... , ....... . III 131-1·12
Monobyers CW. D. Harkins) ....... , . " ... , ., .... , ....... . I, Part 1 427-485
Osmotic Pressure (R. H. Wagner) ......... , ., .. , ......... . I, Part I 487-549
l';trachor (G. W. Thomson) ......... , .......... , ......... , I, Part I 41:3-426
Photochemical Reactions (W. A. Noyes, Jr., and V. Boekel-
heide) ...... , ...... , , .......... , ... , ... , ....... , .... , II 79-141
668 CUMULATIVE INDEXES, VOLUMES I-V
VOLUMT·} PAGES
Photometric Analysis (W. West) ............... ' ......... . I, Part II 1:lO£1-14DO
Polarimetry (W. Heller) ....................... , ......... . I, Pm't. II H01-IGlO
Polal'ogmphy (0. H. MUJler) ............................. . I, ParG II 1785-1884
Potentiometry (1,. Miehaeli:;;). . . . . . . . . . . . . . . . . . . . . . .. . .. . I, Part II 171:3-1784
B.atlio!wtivity (W. F. Bale anti J. F. Bonner, .JI'.) ........... . I, Part II 1£127-1\1\10
Recrystallization (R. S. Tipsou) .......................... . III 3Ga-48o!
Refraetometl'Y (N. Bauer and K. }Cajuns) .................. . I, Part II 11-11-1240
Solubility (R. D. VoId and M. J. Vold) .................... . I, Part I 2\l7-:32l\
Solvent Removal (G. Broughton) ...................... , .. . III GO.'i-(j:)4
Spectrophotometry (W. West) .......... ' ................ . I, Part II 1241-13\)8
Spectroscopy CW. West) ....................... , ......... . I, Part II 12,11-1398
Sublimation (ll. S. Tipson) .............................. . IV OO:3-G45
Surface Tension (W. D. Harkins) ......................... . I, I'art I 355-413
Temperature Control (J. M. Sturtevant.) ................ , .. . I, Part I 29-48
Temperature Measurement (J. M. Sturtevant) ............. . I, Part I 1-27
Turbidimetry (W. West) ................................ . I, Part II 13!l!H490
Ultracentrifuge (J. B. Nichols and E. D. Bailey) ........... . I, Part I (i21-no
Vlwuum Distillation: Moderate Vacuum (.T. JL Bowman
und R. S. Tipson) .................................... . IV 4G3-494
Vilcuum Distillation: High Vacuum (E. S. Perry) ......... . IV 495-540
VlLcuum Distillation: The High Vacuum RYRtem (.T. C.
Hecker) ................................... , ......... . IV 540-6>15
Vo.por Pressure (G. W. Thomson) ......................... . I, Pm·t I 141-251
Viscosity ('1'. E. McGoury and H. Mark) .................. . I, Part I 327-353
X-Ray Diffl'[lrtion (1. Fllnknchen) ........................ . I, Part II IOn-l108

35168

Hochgeladen von

Dokumentinformationen

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

35168

Hochgeladen von

Copyright:

Verfügbare Formate

JNDl.\N Au I~WULTU.

Volume I: Physical Methods of Organic Chemistry

Copyright, 19,1'11, by Interscience Puhlishers, Inc.

INTERSCIENCE PUBLISHERS, INC., 250 Fifth Avenue, N ew York 1, N. Y.

Volume I (Sct'-oml Edition-in two parts). l>hysicullVlcLhods of Organic Chcm~

slawski and J. R. Anderson, "Determination of Vapor Pressure" by G. W.

l)refacc . ............ , . , ..... , . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . \lii

Chapter I. Theory. Continued

Chapter I. Theory. Continued

Chapter II. Ordinary Fractional Distillation. Part I. Apparatus. By A. L.

Chapter II. Ordinary Fractional Distillation. Part 2. Procedure. Continued

Chaptm·[V. Distillation of Liquefied Gases and Low-Boiling Liquids. By

Clwpicr Fl. Distillation under High Vacuum. Part 1. Distillation. Continued

Clwpter VII. Sublimation. By R. S'l't.<Al1:I· TIl'SON . . . . . . . . . . . . . . . . . . . . . . . . . . . Goa

Symbol Definitions Page

(1) Empirieal eOll:st[Lnt, Vtlpor~pr(1SS\Il'e equat,ion ............... . 1!1

Symbol Definitions Page

f' (1) First derivative of a funetion (f) . ...................... '... 121

k (1) Proportionality constant

Symbol Definition. Page

) (1) Vapor pressures of pure subst[\llces

p, (1) Pressure of mereury at '1" ........... ,,.. ,. , . , ,, ,.. ,. , , ' , , 280

Symbol Definitions Puge

1", (1) Temperature chosen as standard for mHnostat.. . . . . . . . . . 289

(1) Volatilities, various components .......................... 20,318

w (1) Weight of volatile substance, entraincr sublimation equation.. G09

S~'lllhvl DcJinitjolls Page

XA (1) Mole fraction component A in liquid, multicomponent mixture 108

SymlJOI Dcllnitions Page

X.a (1) Mol" fmetion compOllunt C in st.ill. . . . .H2

cYlllbol Delillitions Pagn

!/" (1) Composition of vapor (mole fraction more volatile component)

Symbol Definitiolls Page

PI (1) Density of lic[1li. I, hol.lup equations ..... . 181

Britain by Coffey.' These efficiently employed heat in bubble-plate

A. DEFINITIONS m' GENERAL 'L'ERMS AND THE APPARATUS

6 See Science, 103,138 (1946).

In continuous distillation (usual in large-scale operations) the material to

B. DJ..:lcIXl'I'IO;\:-; OF Y.\I([.\ llLI';S

Reflux Ratio, Holdup, Vapor Pressure, and Volatility. <hw of Lhe

The magnitude of the change in eompositioll fot' UHn th(~u]'()1i('al pbte

lls\Lally expressed in termK of the q1HlIltity of liquid equivalent to the vapor

2, Relation of Distillation to Other Separation Processes-Fractionation

is tho flcparation of iRoLopi(~. UJ'alliurn HlloridoR b.v diffllsion barriers.14 In

20 40 60 80 100 120 140 160 ISO 200

;\ pl)<lr :';('pal'aLiml iH l"f~l)I'(,~plll<'d !JY ('111'\"('.1 of Fi).!;IIl'!'~' 1';\'('11 IllI' \"f'r."

binary mixhm' as indicai('d by the di,;t.ilhttiol1 ('·Iln·f' n of Fig\lm 2. The

It is essential to emphasize that the theory of batch distillatioIl is a Ilew

n. VAPOU-LIQUID COMPOSITION RELATIONS

All separations by distillation depend upon the difference in composition

The simplest mnong the above elasses is represented by tL mixture of two

lllHl, from the ftllldllITlPlllal gas law,;:

foJ' a single ['.umponent. Thus tlH~ partial pressure of any component in

Il'l IhJfl f)G X lOLl 59:3G 1

Steam iJi8l'ilialioll. Clldi8! iila! ilill. If j he' alnluHpltC'l'i(, Pl'",,:,\IH'P IS

aecuracy of all sneh calculations depends upon the assumption of sufficient

Ideal Mixtures. Raoult's Law. With liquids that are completely

in which PI is the partial vapor pressure of component 1, PI is the vapor

where 'YL is defined as PyI!PIXl.

yo\:ttility v~tl\\e divided by ttnot,ltel'. The htl'glll' vobWity iH cOllvontioually

indicntes tbtLt, when rela.tive volttWity cqunls unity, no sepamtion is pos-

26 Rose, Ind. $nIO. Chem., 33, 1i9a (UJ41).

If t.he average boiling point is about O°C., t.his becomes:

log a = (8.9) -=-=-+