Beruflich Dokumente
Kultur Dokumente
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
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
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.
, ..,
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
TECHNIQUE OF ORGANIC CHEMISTRY
GENERAL PLAN
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
DISTILLATION
CONTENTS
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 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 DefinitionH
P
PI,
j)A,PU
p~ ! (1) Partial pres[mrr.s of indivicilml componcnts of mixtures
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
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.
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
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.
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'.
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
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'
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
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'",','
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
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
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:
Ji = NH'l'/Y
for the entire mixturc, and:
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):
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,
B. MISCIBLE
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:
~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
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
-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
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
(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).
: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
::;::;;
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-
0:::
0
(L
<!
>
='=
(.)
>
::;:
z
0
j:::
(.)
<!
0:::
u...
L1J
...J
a
:::;;
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
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~
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
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
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
] 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.
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
MOLE FRACTION MVC IN LIQUID
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
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)
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
::;;
Fig. 11. Dingram for 1.heol'eti(:ld plrLtos at. totall'eIhl:x for bnnzmHl- em·boll tetl'l1ehloririn
mixtuJ'PH.
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
,\@)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.
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
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.
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
(1-2) PLATE
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
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
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
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.
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
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:
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:
<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:
in which:
"M = ·~!c_:(:__a_-_l)
- a - rnc 2
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
::;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
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
...._------_.. _._ __.. ._-----'
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
MOLE FRACTION MVC IN STILL
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
n" 39 and 91
LID" I LID = 99
A B
o t 0
MOLE FRACTION MVC IN STILL
ig. Hl. Still-product composition Curves calculated hy Smoker equation for ideal bin-
ary mixture. Ci = 1.05.
oL---------------------------~
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
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.
8() Rose, Johnson, and Williams, "Stepwise Plate to Plate Oalculation of Batch Dis-
W
I-
<:(
_j
_j
j::
CJl
o
z
u
>
::;
z
o
i=
u
<:(
a::
LL.
w
_j
o
::;
I
MOLE FRACTION MVC IN STILL
:;\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
MOLE FRACTION MVC IN LIQUID
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
(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
MOLE FRACTION MVC IN LIQUID
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
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
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).
TABLE I
DATA FOR n-HiCI'TANE-MgTHYLCYCLOUIJJXANE TES'r MIXTURE" (ALL COMPOSITIONS IN
MOLle FRAc'rroNB n-lhn"l'ANB)
~--~--
VfLIlor Liquill 20
11])
l~o
, ·1
H.p" °e.
C{JHl!)n. (:OlIlIHl. (7!lU lI11n.)
.--------~
• 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
smaller errors. The dntn are from Bromiley and Quiggle. (2) The n-heptallc ant!
methylcyclolwxane used by Bromiley and Quiggle had the following characteristics:
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.:
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)
(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)
TABLEV
BENZENE-CARBON TETItACHLORIDE REFRA.CTIVE INDEX VERSUS PLA.TE NUMBER
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
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
----~----- ..-- ..
.~.,.-- ------.--~
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
All these methods use a differential approach n.ncl so are particularly appli-
cable to packed columns.
2. Packing Factor
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)
d:c
Integration gives:
l.
'~D dx
n = --------.---..-.----... -..-
y" - x" - iD (Xn - YII)
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-)
b = (0: - 1) (1 - f xv) + D
0' _.
T./
and B
2a
or:
ii
[ , dpjlljJ = KplVIZ/O
}I,
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.
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:
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
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:
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
(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
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.
t"T dT
-=.--=
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"
.._-_
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
W
.J
m
<t
a:
o
>
Lt
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
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)
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:
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
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:
and:
l' (011t/OZ) - (2cjJt/ II "') = 0
These may be combined to give:
~=~
VII,,, Z a-I
(a + 1) log y.
?f..:_
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
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
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.
')
(_]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:
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:
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
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
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
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 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:
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.
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._~_
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:
81 = 91.3
For the value of S when a:D = 0.40:
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
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
The derivation of tile equation for the distillation eurve for this case is
almost identieal with that for (,limp Ie distillation. Consider any instant
~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
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
PER CENT OF CHARGE DISTILLED
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
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
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
cr=W;/~~=~~/~ (23)
or:
rlA./A aeriE/B) (24)
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~
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
PER CENT OF CHARGE DISTILLED
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.
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 -
Hnd:
or;
log l',
100'
",.
8 = ------
(kH)1Z - 1
+ /.;1 (33)
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
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
H5(~ I} 5 x5
H4(~ I) 4
X4
H3(~ I} 3 x3
H2(~ I} 2
x2
H, (~ I}t x,
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
h = 71 0 ,1 + lIo.? + h + ." = I: h,
O,3
i=O.l
(hy tenths)
h, = IIix(i-o,oo)
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)
(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:
(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
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
(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:
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.
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.
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.
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 • •• ):
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)
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
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:
([: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
(70)
q = l/(a n - 1) (71)
!_ (~ -
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).
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)]
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
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
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
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.
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:
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
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.
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.
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:
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:
(86)
(
1-
Xvc
XDc
)(1 - X'D
X so )
TABLE XI
CALCUl>A'l'IONS OF AVERAGE REFLUX RA'l'IO
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)
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:
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
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:
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:
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
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
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:
The equation:
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
--"[-"
I
I
I
- "
., 1 /
,---
.
o
/
.~ I
/
1.0
LJ w
o
156 Crosley, "Batch Reetific[Ltion of Complex Mixtures," Am. Ohern, Soo. Chicago
Meeting, 1946.
142 A. AND E. ROSE
and:
xDJdxDR = (aARnX'A/~)/(XSR/~) = (aAn"XSA)/XSR
If holdup is negligible:
DXDA = :._ EX'A; DXDB = ~ EX.B; DXDR = - SXSR
or:
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
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
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
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.
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
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
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
TABLE XVI
EFFECT Ol!' HOLDur ON SIZE OF INTEHl\mDIATE FRACTIOJ>,'69
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.
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
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.
---- ~:~;;
,
/
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
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
5 THEORETICAL PLATES
w
..
> METHYL ALCOHOL OBTAINED IN ONE DISTILLATION
~...J
'0"/0
I
W
a:
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
.~.------
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
~
LJJ
.1071 .. ~••• 0.0001
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.
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
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
w
r.!)
a:
«
:I:
u
....J
g
l-
lL.
o
~
...;
n
",n
167 Rose and Welshans, Ind. EnO. Ohem., 32, 669 (1940).
164 A. AND E. ROSE
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.
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:
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.
(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
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.
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.
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
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-
(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.
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.
n = 200/I1T
n = 233/ ~7'
General References
Badger and McCabe, Elements of Chemical Engineering. 2nd ed., McGraw-
Hill, New York, 1936, Chapter IX, "Distillation."
174 A. AND ]~. ROSE
Pa1't 1. APPARATUS
I. INTRODUCTION
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
A. WE'1''l'ED-SURFACE COLUMNS
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
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
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."
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.
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
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)
\__ V
l_'-
",.
",.
-
~
"-
""'"
"'=
"---,-=
~/
I! HOLES 5 MM.
2 HOLES 4 MM. 01
~~~ ~1Q UIO OUTLET?i5 MM. 10.
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.
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-
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.
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
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
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.
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
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.
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'.
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
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
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).
194 A. L. GLASEBROOK AND F. E. WILLIAMS
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.
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"
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.
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
" ..
~
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
'6 Laughlin, Nash, and Whitmore, J. Lim. Chem. Soc., 56, 1396 (1934).
200 A. L. GLASEllROOK AND F. E. WII,LIAMS
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.
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
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
American Petroleum Ir-stitute, Meeting at Chicllgo, Il1., Nov. 11, 1946, p. 42.
202 A. J". GLASgllROOK AND F. E. WILLIAMS
f ~ FEMALE
MERCURY SEALED
NO-LUB JOINT
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.
~ ~ FEMALE
NO-lUB JOlrIT
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
.£
. ~
~ ..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
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
~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 ~
..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
Lt:I H,,:! 0 1 0
~~c:i~~
208 A. L. GI,ASEBROOK A.."N"D F. E. WII,LIAMS
* 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
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
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
(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'
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
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
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
2. ;3-mn. dhlllet(,r"
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
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
'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
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
TABLE XX
HSCTIPICA'l'ION CHARACT~aUSTICS OF BRL'UN Bl_lBBLJo;-C.U' CnU)"AIN"
* 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).
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
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.
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-
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
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
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.
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
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
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
)13 Willingham, Sedlak, Westhaver, and Rossini, Ind. Eng. Chem., 39, 706 (1947).
232 A. L. GLASEBROOK AND F. E. WILLIAMS
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.
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
values are given in Table XXV. The efficiency factor of the column under
various operating conditions is given in Figure 28.
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:
~
: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
Fig. 28. Eflieiency faetor n. speed of rotatIOIl for rotating concentric-tube column
at s(lvnraJ values of throughput. m
TABLE XXV
Holdup,
Throughput, liquid/meter length, Li;Uid holdup,
m!. liquid/hr. ml. mi. theor. plate"
2. Construction of Columns
A. COLUMN TUBES
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
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
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
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.
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).
242 A. L. GLASEBROOK AND F. E. WILLIAMS
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
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
JACKET
HEATING JACKET
- 25 MM. 0.0. COLUMN
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.
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
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
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
Producl Condenser
Joint
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
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
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
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
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.
TO TRANSFORMER
TO B.R
MANOMETER
25 MM. 0.0.
TOV;-~
7! 5 LITER
TO VARIAC
[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.
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
TABLE XXVII-A.
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
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
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,
American Petroleum Institute, Meeting at Chicago, Illinois, Nov. 11, 1946, p. 42.
II. ORDINAHY FHACTIONAL DIS'l'ILLATIO:-i 271
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.
114 Hall, S. A' I and Palkin, Ind. Eng. Chern., A.nal. Ed., 14, 6§2 (HI42).
II. ORDINARY FRAU'rrONAL DISTILLATION 273
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
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.
METAL A
TO POTENTIOMETER
METAL B
COLD HOT
JUNCTION JUNCTION
Fig. 50. Therllloelectric circuit for teJUIJemtul'c mCILSUI'CIllP.ut.
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.
760
w
a:
:::;,
(/)
~
0::
a..
a:
oa..
~
TI;M PERATURE
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
TUN
CHROMEL
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.
TO VACUUM
TO SYSTEM
8 STOPCOCK
TO-=-_.r---.~~
CON-
TROLLED
SYSTEM
DIETHYLENE
GLYCOL
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)
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)
[_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)
For Pyrex:
(3 = (3.6 X 10- 6)3 = 1.08 X 10-·
For mercury:
'" = 18.186 X 10- 5
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.
192 Willingham and Rossini, J. Research Nail. Bur. Standards, 37, 15 (1946).
292 A. L. GLASEBROOK AND P. E. WILLIAMS
======)J
============~~~=====~
Fig. 63. Longitudinal section of mltle and Fig. 64. Longitudinal seotion of male and
f('ma.le stanrlard·tltper joints. female spherical joints.
2. Lubricants
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
Diameter 1/4"
Seamless less tho n I.D.
steel of steel tube
tube
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
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
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
---~------
poncnts boil below their melting point (HubliulP), tllP diHtillation will hnvr
to be condlleted azeotropically to remove Rueh materialH as liqllirl fLzeo-
tropes.
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
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
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
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
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.
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
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
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
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
317
318
(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
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
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
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.
(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
~ 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
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
(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
"
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.
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.
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.
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
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.
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
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
TABLE VII
TERNARY SYS'l'EMS WITH W A'rER AS Orn~ COMPONJ~N'r33
Temp.!
Components °C. Component~
°°°
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
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
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
TABLE IX
POLARITIES EXPU!;:sSED AS DIPOL~: MOMENTS
. p. X 10" e.s.u.
Compound at 20°C.
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
::;;
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-
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
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
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
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
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.
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.
(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):
(3) Reflux ratio at condenser. The minimum reflux ratio for a binary mixture can
he readily calculated using the Fenske cquation:13
(17)
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
Mole fraetion
,,-hoptane in Bti II Rutin. Ractuul
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.
(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:
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.
Section Ji 10 15 21.
Section B <3 <4.5 5
Section D 3 4.5 7.0
6. Equipment Performance
A. PACKED LABORA'l'OHY COLUMNS
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.
PHENOL
CHARGE
ffi NON-AROMATICS
w3: TO FIELD
z~
w)-
:lit:
..J
0>
w
1-0
u
w
It:
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.
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.
3. Selection of an Entrainer
A. CHEMICAL CONsIDEHA'rIONS
\
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
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.
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
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
~
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
.
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
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"
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.
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
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
MOLE FRACTiON OF A
CONDENSER
8-IOmm,Q,D.
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
STEAM DISTILLATION
GENERATOR FLASK
B. CON'l'INUOUS DIS'l'II,LATION
DISTILLATE
REFLUX
DISTILLATE
(AZEDTROPE)
FEED + FEED +
ENTRAINER ENTRAINER
MAKE-UP
BOTTOMS
BOTTOMS
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
C. EQUIPI'ImN'f PlDltFOHMANCE
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
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
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
General References
Extractive Distillation
THEORY OF EXTRACTIVE DIS'I'ILLA'rION
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.
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
Azeotropic Distillation
THEORY OF AZEOl']WPIC DIS'rILLA'l'ION
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
389
3HO A. AND E. ROSE
IV. LIQLTEFlIW G..\SES AND LOW-BOILING LIQUIDS 39}
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
~ ~ ~~'"
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
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
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.
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
-~ 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
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
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.
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.
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.
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.
FLEXI8LE LEAD
TUBING
CONDENSER-
/
DETAIl. OF UDPER
PAIlT Of CONDENSER
LARGE DEWAR
TUBE WITH
FRACTIONAT-
ING COLUMN
INSIDE
BAFFLE
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,
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"
so that the temperature on the outside surftlee of the vacuum jacket was
the same as that inside.
40() A. AND E. ROSE
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~.
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).
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
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.
... 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.
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
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
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
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
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
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
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
~:~~~~: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
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
B. INTERPRETATION OF DATA
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.
TABLE V
CALCULATION o~' CUT POINT BY METHOD m' MARTIN
C,H,in
Volume, segment.
liters Mean telllp.) °C. "a[o. %
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
<.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
MILLIMETERS PRESSURE
Fig. 24. Compressibility (Jol'l'eetion factors for deviations from Boyle's IIIII' (at 25°C.).
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
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
)] ,
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 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% 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
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.%
...,
: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'
1--.1...-,r-- 3
~6
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.
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.
A. MICRO COLUMNS
D. PRESSURE COLUMNS
n. SIMPLE DISTILLATION
TABJ~E XU.
SEH'ARA'l'ION TEMPERATURES AND CUT'rING PRESSURES EMPLOYJilD
WITH WARD HIGH-VACUUM ApPARATUS
l':r;cflsHl'e,
SCPll.ration mm.Hg: No.5 No.4 No.3 No.2 No.1
-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
~
/
-61 -64
v
o 80 160 240 320 400 480 560 640
AMOUNT OVER,mm. Hg
Fig. 31. E;.,.'perimental simple distillation curves. 111
'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
~
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 ~
'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
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).
IV. LIQUEFIED GASES AND LOW-BOILING LIQUIDS 459
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
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
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
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.
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.~
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
2. Rectification
A. COUNTEIWURRlDN'l' CON'l'AC'l' nmC'rIFICA'l'ION
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 "
!BEFORE SEPARATION
i- I
: : 1
! AFTER SEPARATION kI !
I
·.;. . n:·;·,-l
I
3
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
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
Fig. G. Simple horizon tl11 stills for Fig. 7. Modified Cbisell flasks.
entruincr-VtlCUUlll distilbtioll.
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).
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
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
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
. ~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~
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.
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
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.
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
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
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
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
SYMBOL INDEX
495
496 E. S, P1TIRRY
RYlllbol Definition
-~------.----------------------
I. INTRODUCTION
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-
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
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
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
(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
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
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
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
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
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
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."
T
C\.I
C\.I
----,.--,.
11
lb.~ ~48-l
1 Jf1~-75-~
(A) (8)
Fig. 2. cICold-finger" RWIR «(limcnsions in lllillirnotel's).aa
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
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.
VI. DISTILLATION UNDER HIGH VACUUM 513
FORE PUMP
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
....
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.
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
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
\\"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
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
-- ..,,--'------------"'-----
'
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
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.
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
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.
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
):
....... 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!
_._---"-- . ""---_.
"'-'~"--'----''''-'---~-'''''''----'-~
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.
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
-r-
o
'"
L.'--_ _ _ ___'
~mi~~ASBESTOS
1----
60 -~-_j ~ \""-MICA AND
WIRE
(A) + INSULATION
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".
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
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.
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)
~
~~~#~~~~~------~~
A B c
Fig. 21. Almquist tuhullll' st.ill. 65
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
PREHEAT TUBE
RESERVOIRS
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.
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
DISTILLATE
RECEIYER--
RI:::SERVOIR
• --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
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
E. BHEGER DI~SIGN
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),
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
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 £
(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:
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:
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)"
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
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)
(22)
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)
TABLE XIII
A vs. t D
2P1AI 2-VPIAI
1] = -: /-=" = -~/-c::.-='-
v 21fPl 'V 211'
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).
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
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--
'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
---
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)
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.
(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.
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) :
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
VACUUM REFRIGERANT
CHAMBER
! VACUUM
j I
CHAMBER
_,..VACUUM VACUUM
PUMP PUMP
SILVERED
VACUUM
JACKET
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
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
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
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
(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:
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:
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.
VI. DISTILLATION UNDER HIGH VACUUM 553
TABLE XVII
CRITICAL FLOW RA'l'E (LAMINAR TO TURBUI,ENT FLOW) AS A FUNCTION OF P
FOR TUBES 1 CM. IN DIAMET~JR
1 0.001 185,000
10 0.01 18,500
100 0.1 1,850
1000 1.0 185
10,000 10. 18.5
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-~
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~ __ •
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
2. Oil-Vapor Pumps
A. GENERAL
B. CONDENSA'rION PUMPS
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~
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
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
~ ~~
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.
HIGH VACUUM
- OIL-VAPOR PASSAGE
_-INSULATION
CONDENSED-OIL
RETURN _ _ _
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
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)
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
- --_ ....
Cold Cathode
I
--
McLeod
------1
Pironi
----- - ~-....,--
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
THREE-WAY
srOPCOCK 'iI',C\J\JM
RtFERENCEr-----~ CHAMBER
\lACUUM
VACUUM
CHAMBER
SCALE
SCALE REFERENCE
VACUUM
OIL
RESERVOIR
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
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
'"
: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.
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.
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
VACUUM
CHAMBER
BAFFL~ '"
PLATE~~ --
TO VACUUM
PUMP
j TO VACUUM
PUMP
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.
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.
VI. DISTILLATION UNDER HIGH VACUUM 587
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-
"...
=
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
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)
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.
_~PINCH CLAMP
(Al
l
GLASS TUBING WITH
FLANGED END
RUBBER TUBING GLASS (OR METAL)
METAL Fl,lllllr,E
(B l [C)
'ig. 4.6. Gln.8s-to-glass connections: (A) small diameter; (B) intermediate diameter;
(C) large diameter.
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.
·-'.'(;~TU'''G
- SPACER
FLANGE
VACUUM OR
PRESSURE
FOR TESTING
TIGHTNESS
~~=d/
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
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.
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:
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
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
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
SUBLIMATION
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
W
0:
:::>
I- B
<t
0:
W
0-
::;;
W
I-
LIQUID
m.p.
D
PRESSLJRE\mml-
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
TAl~LIi] I
V.APOIll'Ill~SSUllDJ (n' Sn~n) OlWANIC CO~Il'()!TN])A .'>-'1' 'J'!·T-N MmUI'ING POIN'~'
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.
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
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°,
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
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
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).
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
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
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
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
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
c-_====
-'~.'
--.~.......
.. .
}
I
~~-.~"-'--.:.,~,,-.
,,_-'''-
--
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 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
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
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
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
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
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
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
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 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
Fig. 27. Vacuum sublimator with liquid- Fig. 28. Vacuum suhliml1tor with
cooled conclcnser.147 liquid-cooled condenser.14S
149 Bolstad and Dunbar, Ind. Eng. Chern., Anal. Ed., 15,46<1 (1943).
(i38 It. S. TIPSON
-
Fig. 29. Horizont.al vacuulll sublimator with liquid-cooled condenser.119
D. J~N'l'RAINER--VAcum\1 SUBLIMA.'l'ION
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
V. EXAMPLES OF SUBLIMATION
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
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
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
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
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
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