Rates of Reaction in a Recycling System-

Dehydration of Ethanol and Diethyl Ether

Over Alumina
JOHN B. BUTT, HARDING BLISS, and C. A. WALKER
Yale University, N e w Haven, Conn.

A new method for measuring isothermal rates of heterogeneously catalyzed gas reactions is
described. The method involves a differential reactor incorporated in a recycling system and
affords ease of control, occuracy, and many rate measurements in one run.
The results for the reaction system investigated show that four reactions are important: the
dehydration of ethanol to ethylene, the reaction of ethanol to form diethyl ether, the dehydration
of diethyl ether to ethylene, and the reaction of diethyl ether to form ethanol.
It is also shown that the rate controlling steps are the monomolecular surface reactions for the
two dehydrations and for the formation of ethanol from diethyl ether, and the bimolecular
surface reaction for the formation of diethyl ether from ethanol.
The values and the temperature dependence of all pertinent rate and adsorption constants are
determined and reported, and they are shown to correlate very well the'data up to 80%
conversion.

The measurement of the isothermal
rates of heterogeneously catalyzed gas
reactions is ordinarily accomplished
directly in differential reactors or in-
directly in integral ones. The former is
a powerful method but difficulties are
met in analyzing small differences in
concentration across the short bed.
The use of integral conversion data is
limited by the necessity to hypothe-
size a rate equation before interpreta-
tion is possible and by the experi-
mental problem of maintaining isother-
mal conditions in the reactor.
A recycling differential reactor, first
proposed by Dohse ( 5 ) , affords ease
of temperature control, accuracy of
analysis, and direct measurement of
rates. A new reactor based on Dohse's
original idea but extensively improved
and modernized has been developed
and used in this research. Reactants
and products are continuously circu-
lated through the catalyst, and samples
of the reaction mixture are withdrawn
frequently for analysis. The extent of
reaction per pass through the catalyst
bed is very small; consequently main-
tenance of isothermal conditions is not
difficult. Conversion data, and thus
rate data, are determined at many
points in a run from the sample analy-
ses, so increased accuracy of the rate
data is assured. A run may be con-
tinued until very high conversions are
attained; hence it is possible to meas-
ure the rates over an unusually wide
range of conversion in one run.
A somewhat similar device was used
by Smith, et al. ( 1 3 ) , with consider-
able success for ion exchange systems.
Perkins and Rase (10) and Polotniuk
and Dobrovol 'skii (11) have reported
the use of reactor systems involving
recycling operations. These devices
however involve recychg the reactants
and products in a steady state meas-
urement of rates at a given conversion
and are not similar to the system de-
scribed here.
The dehydration of ethyl alcohoI
over alumina catalyst was selected for
study because such rates have not been
well-defined and because a moderately
complicated array of simultaneous and
consecutive reactions is involved. Di-
ethyl ether is an intermediate, and the
rates of its reactions over the same
catalyst were also measured.
R A N G E OF V A R I A B L E S
The variables investigated in this re-
search were as follows:
Temperature: 274", 294", and 314°C.
Reactant material:
1. Pure ethanol
2. Pure diethyl ether
3. 84 mole % ethanol-16 mole %
ethylene
4. 84 mole % ethanol-16 mole %
water
Initial pressure:
1. 250, 520, and 740 mm. Hg
(ethanol reactant)
2. 65, 150, and 250 mm. Hg (di-
ethyl ether reactant)
3. 620 mm. Hg (ethanol-water and
ethanol-ethylene reactants)
Range of conversion: 0 to 90%
Vapor flow rate: 0.24 to 1.1 cu. ft.
(standard conditions) /min.
The catalyst used in all experimen-
tal work was Harshaw Type AL-1404
high purity alumina.
EXPERMENTAL
Apparatus
Figure 1 is a schematic diagram of the
recycling reactor system. The major por-
tion of the system was constructed of 35-
mm. glass tubing, with 3-mm. capillary
tubing used for inlet-outlet lines A and
M and manometer C . The recycling
pump D was a Vanton Flex-i-liner pump
with a silicone rubber liner and 316 stain-
less steel body block. The glass system was
connected to the pump with gum rubber
tubing protected with high temperature in-
sulating tape. Catalyst bed J consisted of
11.64 g. of high purity alumina in the
form of '/8 x '/s in. cylindrical pellets.
A 1/16 in, shielded iron-constantan
thermocouple G was used to measure
catalyst bed temperature. This thermo-
couple was mounted such that it could be
moved over the 3/4 in. bed depth, al-
lowing the measurement of the catalyst
bed temperature profile. Temperature of
the bath K and of the reaction mixture

Page 42 A.1.Ch.E. Journal March, 1962

entering the recycling pump were also
measured with iron-constantan thermo-
couples. Bath K contained molten 312
tempering salt heated by means of a
1,000- and a 500-w. resistance heater.
Temperature was controlled to within
Is
0.l"C. with a bimetallic thermometer
and relay system connected to the 500-w.
heater.
Flow of the reaction mixture was
19.r
counterclockwise through the system,
such that in the U-tube reactor section
the down flowing gas was preheated to
the reaction temperature before flowing
up through the catalyst bed. Feed to the
system was introduced, and samples of Fig. 1. Recycling reactor system.
reaction mixture were withdrawn for
analysis through sample ports E or F,
fitted with serum stoppers, by means of a It has been pointed out by Brey and 2. Prior to taking conversion data on
hypodermic syringe. The inlet-outlet lines Krieger ( 1 ) and other workers that dif- beginning the next run, the catalyst was
A and M served as vacuum and inert gas ferent conditions for the activation of stabilized by recycling the feed material
systems, respectively. These were isolated alumina will result in different levels of to be used at the desired pressure for ten
from the main reactor system by a valve activity. The study of catalyst activity as minutes. T'o equipment was then evacu-
system (B, 0, and N ) and included a a function of the method of activation ated and recharged with reactant for the
trap L for the removal of condensible was not of interest in this work; therefore experimental run.
material from the vacuum system. To all the catalyst used was activated in the
same manner. The method of activation The stabilized activity was essentially
prevent possible condensation or adsorp- constant as evidenced by reproducibility
tion of the reaction mixture in the system consisted primarily of maintaining the
alumina at 300OC. in a nitrogen atmos- of the results. Check runs were made fre-
all exposed surfaces were heated with quently to verify this.
flexible heating ta es wrapped around the phere for 24 hr. before use. The activity
K
glass tubing and t e pump. of the alumina catalyst was found to be
constant and reproducible after approxi- Analysis
Procedure mately 4 hr. use with this method of
activation. For intermittent operation of The reactions of ethanol or diethyl
Operation of the recycling reactor sys- the equipment it was found that two ether over alumina catalyst yield a re-
tem involved the recirculation of a known procedures were necessary in order to action mixture consisting of ethanol, di-
amount of reactant through the catalyst reproduce catalyst activity if more than ethyl ether, ethylene, and water. Previ-
bed at conditions of temperature and pres- an hour elapsed between two consecutive ously it had been very difficult to work
sure such that a very small conversion of runs. These procedures were as follows: with systems such as this because of t!ie
reactants to products was obtained per unique analytical procedures involved. It
pass through the bed. The reactor system 1. The catalyst was kept in a nitrogen was found that the methods of vapor
was operated in a batch fashion by intro- atmosphere at the temperature at which fractometry afforded a convenient and
ducing the reactant and following the the next run was to be carried out. accurate method for obtaining analyses of
reaction by means of concentration, time
of reaction information for the constant
volume process. Owing to the small con-
version occurring per pass in the catalyst TABLE
1. SAMPLEEXPERIMENTAL
DATA
bed it was easy to maintain isothermal Run: Ethanol feed
conditions in the bed while making a run. Temperature: 294 C.
To make a run at a given temperature Initial pressure: 740 mm. H g
level the experimental apparatus was
Initial moles charge: 0.05106 g.-moles
evacuated to 0 to 1 mm. Hg. absolute
and the recirculatin section isolated from Total
g
the remainder of t e apparatus. The re-
cycling pump was started, and liquid
Sample Time, Molar composition, g. moles pressure,
mm. Hg
no. min. ether ethanol water olefin
ethanol or diethyl ether reactant of
weighed amount was injected into the A33-1 3 0.00696 0.03490 0.00942 0.00246 776
system with a hypodermic syringe. On A34-1 2 0.00436 0.04090 0.00578 0.00147 760
complete vaporization of the reactant in A34-2 10 0.01498 0.01785 0.01811 0.00314 783
the system the initial conditions of pres- A35'-2 9 0.01437 0.01845 0.01802 0.00399 787
sure and catalyst bed temperature were A36-1 4 0.01027 0.02850 0.01234 0.00408 796
noted. After this initial period of opera- A36-2 12 0.01563 0.01595 0.01942 0.00378 796
tion small samples of the reaction mixture A36-3 20 0.01405 0.01095 0.02317 0.00665 830
were removed from the system at speci- A37- 1 5 0.00757 0.02970 0.01216 0.00303 784
fied time intervals for analysis, and the A37-2 14 0.01596 0.01515 0.01998 0.00402 797
system pressure and catalyst bed tem- A38-1 1 0.00238 0.04310 0.00546 0.00314 783
perature were recorded for each sample. A38-2 8.5 0.01277 0.02195 0.01620 0.00342 787
Samples were withdrawn simultaneously A38-3 16 0.01603 0.01315 0.02162 0.00559 817
from E and F (Figure 1 ) in several runs. A39-1 3 0.00694 0.03690 0.00743 0.00349 790
No dserence was ever found, so only E A39-2 13 0.01534 0.01689 0.01884 0.00355 792
was used thereafter. When the desired A40- 1 6 0.00992 0.02720 0.01394 0.00402 774
conversion range had been studied, the A40-2 15 0.01608 0.01360 0.02145 0.00559 819
recycling pump was stopped and the re- A41-1 2 0.00393 0.04120 0.00588 0.00196 768
action mixture evacuated from the ap- A41-2 9 0.01282 0.02190 0.01631 0.00349 791
paratus. The system was then purged A41-3 17 0.01534 0.01432 0.02147 0.00618 830
with nitrogen in re aration of the next A53- 1 19 0.01717 0.01308 0.02077' 0.00361 793
run. These metkog of experimental A53-2 27 0.01712 0.01089 0.02305 0.00599 825
operation were used for all conditions of A54-3 36 0.01730 0.01168 0.02326 0.00567 820
temperature, pressure, and feed material A544 43 0,01733 0.00975 0.02410 0.00631 831
investigated in this research. A54-5 51 0.01751 0.00863 0.02486 0.00729 846

Vol. 8, No. 1 A.1.Ch.E. Journal Page 43

this mixture. The rapidity of analyses
eliminates any diKiculties which might be _J
0 06
encountered in storage of samples for 0
slower methods, and the high sensitivity z - R U N A-34
of the method allows the use of samples a - RUN A - 3 9
I
small enough not to disturb the constant t- 0.04
- RUNA-54

volume nature of the recycling reactor w

system. Indeed it would be very difficult cn
to carry out kinetic investigations with w
the recycling reactor without a rapid, _I
0 0 02
convenient method of analysis of very z
small samples.
I
In this investigation the fractometer used Q
for the analysis of reaction mixture samples 0:
a 0
was equipped with a %-in. column 2 m in 0 LOO 2 00 300
length and packed with Carbowax 1500.
Output from the thermal conductivity T I M E , MINUTES
measuring device of the fractometer was
recorded on a recorder modified for 0 to Fig. 2. Reproducibility of experimental data.
1 mv. full scale deflection and equipped
with Disc mechanical integrator. A car-
rier gas pressure of 20 lb./sq. in. gauge
and column temperature of 112OC. were
used for analyses. At these conditions,
with dry helium as the carrier gas, good
of moles and not from any considera- 2CzH,0H * C,H,OC,H, + H20
tions of volume. (B)
resolution of all components except water C,H,OC,Hs * C,H,OH + C,H,
was obtained, elution of components from Measurements in the Unsteady State
the coIumn being in the order ethylene, (C)
ether, ethanol, and water. A study of the
Most rate measurements are made
at steady state conditions, but these
C,H,OC,H, -&2C,H, + H,O (D)
results of thirty analyses carried out on a An analogous set of reactions was hy-
standard sample indicated that the meth- are at conditions varying with time.
od was accurate within an average devia- Since rates are influenced primarily by pothesized by Langer and Walker (8)
tion for each component analysis of less the distribution of adsorbed species on to explain their results with dodecanol
than 0.5% (based on the total number the catalyst, and since these must dehydration.
of moles of sample). change during the course of a run, Cope and Dodge ( 4 ) give a good
there is a question as to whether the summary of the equilibria in these re-
rates could be affected by the time re- actions. Under the conditions prevail-
DATA quired for necessary adjustments on ing here the equilibria in (A) and
the catalyst surface. There is no evi- (€3) above are far to the right. The
Approximately 120 runs were made,
dence of such an effect, and it appears equilibria in ( C ) and (D) are not
the detailed results of which are given
that the adjustments on the catalyst known as accurately, but they are
by Butt ( 2 ) . A sampIe of the data is
occur very rapidly. This is best indi- probably well to the right. In the
given in Table 1. It should be noted
that the fundamental measurements cated by the following comparison. longest time and highest conversion
made were the compositions of the re- Runs were made with both pure etha- runs made these reactions were still
circulating gas streams at various nol feed and an ethanol-water mixture proceeding, although very slowly.
times. Conversions and reaction rates composed of 84% ethanol, as men- Effects of Materials of Construction
were calculated from these. It should tioned previously. A comparison of It was feared that pump materials,
also be noted that the conversion per conversion data for the mixed feed rubber serum stoppers, and sealing
pass is always small and that there can runs with those for pure feed indicates
cement might possibly exert catalytic,
be no appreciable concentration gradi- that the measured rates of reaction
inhibiting, or adsorption effects on the
ent throughout the volume of the re- under equivalent conditions of partial
reactants or products. No evidence of
cycIing system; that is this is essen- pressures and temperature for the two
this was found. Runs without catalyst
tially a batch reactor. cases are equal. Thus the rate of ad- exhibited no conversion, and runs at
justment of the catalyst to the un- different flow rates which should ex-
steady state adsorption of water and
hibit different adsorption behavior (on
OBSERVATIONS ABOUT THE DATA ethanol appears to be extremely rapid such surfaces as the pump) showed
A N D THE SYSTEM in comparison with the reaction rates
the same conversions. Reproducibility
measured.
Isothermal Behavior (see below) was always good.
The temperature measured over the Sample Volumes Considerations of Gas Phase Diffusion
length of the catalyst bed varied no The cumdative volume of sampIes
Runs were made at gas flow rates
more than 1°C. throughout each run. removed was always less than 1% of from 0.24 to 1.1 cu. ft./min. (stand-
The bath temperature was 1 to 1-1/2"C. the total volume. Accordingly any ef- ard conditions). At rates of 0.68 and
above the catalyst temperature in all fects due to sample withdrawal were 1.1 cu. ft./min. conversion data were
cases. The temperature of the recircu- neglected. identical within experimental error, in-
lating gases in other parts of the sys- Chemical Reactions and Equilibria dicating no effect of this variable. Cal-
tem, such as the pump, was often ap- With ethanol as the reactant ether, culations with the mass transfer corre-
preciably different from that of the ethylene, and water were found in the lation of Hougen and Wilke ( 7 )
catalyst, but this in no way affects the products. With ether as the reactant showed that the mass transfer rate was
rates. The number of moles of charge ethylene, water, and alcohol were pro- four times the reaction rate in the
is known, and the change in the num- duced. These can be explained with worst case. For most of the cases the
ber of moles throughout the reaction the following reactions: mass transfer rate was ten to twenty
is determined by the analysis. The times the reaction rate. All runs for
rates are determined from the number C,H,OH *
C,H4 H,O (A) + determination of rate constants were

Page 44 A.1.Ch.E. Journal March, 1962

carried out at either 0.68 or 1.1 cu. tions may be derived for a reaction in terms of the rate of ether appear-
f t ./min . system as complex as this one. Equa- ance. Thus a plot of the reciprocal of
tions were written for the cases of the initial rate so measured for reac-
Consideration of Pare Diffusion adsorption of reactant, surface reac- tion ( A ) vs. the reciprocal of pa, was
The effectiveness factor defined by tion, and desorption of product con- a straight line confirming the first
Thiele (14),calculated according to trolling. The results described below order surface reaction of Equation
the method proposed by Weisz (15), show that adsorption and desorption ( 3 ) . Similarly a plot of the inverse
was shown to be 0.9 or higher even at steps are not important and that only square root of the initial rate of reac-
the highest rates of reaction. These the surface reaction cases must be tion (B) vs. the reciprocal of pAo was
calculations were carried out with an considered. The equations were then a straight line, confirming the second
effective diffusivity based on a value simplified because ethylene was not order surface reaction of Equation
of 0.043 sq. cm./sec. for hydrogen adsorbed on the catalyst, as shown by (4). The slopes and intercepts of
with the alumina catalyst at 20°C., as a negligible effect on the rates pro- these straight lines gave preliminary
reported by Carberry ( 3 ) . duced by addition of ethylene to the values for the constants appearing in
reactants, and the chemical equilibria the rate equations. These methods are
Reproducibility
are favorable so the reverse reactions illustrated in Figure 3 for the data ob-
In several cases a group of runs was can be neglected. tained at 294°C.
made at identical conditions with sam- The simplified equations are then A similar approach was undertaken
pling at different times for each run. written as follows: with the initial rates of those runs
The resulting composition vs. time ( a ) first order surface reaction, re- using pure ether as a reactant. Thus
curves practically coincided as shown action A: the rate of reaction (C) was expressed
in Figure 2. in terms of alcohol appearance and

m
a n
Z (1)
1200 ( b ) second order surface reaction,
reaction B:
ZOW
KS2L Kn2 pi'
rr
4(1 + PAL + P A W + plc&)* 12000 8000 ;
ZOW (2)
a000 4000
i=
1.00
Initial Rates ; E
The f i s t quantitative step in the in-
It00
& terpretation of data on this reaction
no0
-
-1e system requires the special considera-
tion of initial rates. At the beginning
of the reaction no product terms can
-I x
( I

( PEI
10' , mm. np-'

be important, so the surface reaction

equations become: Fig. 4. Initial rate data, dehydration of di-
(a) first order surface reaction, re- ethyl ether at 294OC.
action A:
Fig. 3. Initial rate data, dehydration of ethanol
at 294'C. that of reaction ( D ) in terms of water
appearance. The analysis of these data
showed that both reactions (C) and
INTERPRETATION OF D A T A ( b ) second order surface reaction, ( D ) were monomolecular surface re-
reaction B: action controlled, as illustrated in Fig-
Rate Equations
In correlating kinetic data for heter- K s J k' pno2 ure 4 for ether at 294°C. The conclu-
ro =
ogeneously catalyzed reactions it is 4 ( 1 p A o K A ) ' +
(4) sions reached from Figures 3 and 4
must be considered tentative, since
necessary to consider the rate of re-
action to be controlled by one of a The values of initial rates of reac- errors in measurement and judgment
number of steps occurring in the cata- tion were determined from experimen- influence the locations of the lines.
lytic reaction process. These steps tal data as follows: The verification of these conclusions
customarily include fluid phase diffu- 1. A parabola was determined ac- depends primarily on the high conver-
sion, pore diffusion, adsorption, surface cording to the method of least squares sion data discussed below.
reaction, and desorption. It has already which related the number of moles of The consideration of initial rates is
been pointed out that fluid phase and reactant (alcohol or ether) to time. an excellent beginning, but owing to
pore diffusion were very rapid com- This was limited to the initial portion the transient nature of the recycling
pared with the reaction rate under the of the conversion data. system at start-up initial rate data are
experimental conditions here con- 2. The equation was differentiated not as accurate as could be desired.
sidered; thus they may be disregarded. at t = 0. Obviously the determination of con-
All further work was done with atten- 3. The resultant rate was divided stants from these data will reflect such
tion to adsorption, surface reaction, by the weight of catalyst. inaccuracies. Furthermore the absence
and desorption processes. The results obtained show that etha- of water in initial rate investigations
The manner of deriving rate equa- nol disappears according to both reac- makes impossible the determination of
tions, based on the adsorption theory tions (A) and ( B ) . The individual the water adsorption constant; thus it
of Langmuir ( 9 ) , has been well-de- rate of reaction (A) can be deter- was necessary to carry out an addi-
scribed by Hougen and Watson ( 6 ) . mined by noting the rate of ethylene tional series of initial rate evaluations
A considerable number of such equa- appearance and that of reaction (B) on runs in which water was added to
Vol. 8, No. 1 A.1.Ch.E. Journal Page 45
the feed material. These additional
data permitted the preliminary evalua-
tion of the water adsorption constants
by the methods outlined above.
Initial rate analysis then established
that the following rate controlling
steps were valid:
Reaction ( A ) : monomolecular sur-
face reaction
Reaction (B) : bimolecular surface
reaction
Reaction ( C ) and (D) : monomolec-
ular surface reaction
Over-all Rate Equations
In consideration of the reaction
process described by Equations ( A ) to
(D) it is necessary to combine the
rate equations for the individual reac-
tions in order to define the conversion
0.06
0 05
D = (1 + + P A ,+ PA) 1. Initial rate data were sufficiently
(6) accurate to define fairly well the val-
ues of the adsorption equilibrium pa-
Similar expressions may be written for rameters, thereby reducing considera-
other components, for both the ethanol bly the complexity of the trial pro-
and diethyl ether reaction systems. cedure.
2. The availability of conversion
Analysis of High Conversion Data* and data for each of the components ap-
Refinement of the Volues of the Constants pearing in the reaction mixture al-
iowed "the determination of the con-
The rate expressions which are gen- stants for each reaction independently;
erally applicable for calculation of con- thus the over-all conversion of the
versions beyond the initial region are reactant material could be utilized as
given by equations of the form of ( 5 ) . a check on the combined rate equa-
These equations are the final forms tion.
used in the correlation of experimental The adsorption equilibrium parame-
data at high conversions. It was found, ters for ethylene and water, determined
on examination of these data, that separately in the initial rate analysis,
even the qualitative trends of conver- were also checked by these methods
sion with time could not be predicted
satisfactorily with equations based on
adsorption or desorption, c o n k i n g
that these processes are not rate con-
trolling. The results of the initial rate
00200 -=CALCULATED
EXPERIMENTAL:
investigations then can be verified as 0 0175 0 ETHER
indicating surface reaction rates con- C, WATER

v) trolling in all cases. 0 0150

w 0 04
-1 Constants for the indicated rate
0 0 0125
5 0 03 equations calculated from initial rate
data are not the best values for fitting cn
5a 0 02 both initial and high conversion data. y 0
00100

(3
In order to determine the values of 00075
0 01
reaction rate and adsorption equilib- 2 4
0.00
rium constants which best fit both g 0.0050
initial and high conversion data ac-
TIME, MINUTES
cording to the surface reaction mech- 0 0025

anism, the following procedures were

Fig. 5. Comparison of experimental and cal- used: noooa -
0 4 I I2 I6 20 24
culated conversions, ethanol a t 294.C. 1. The required surface reaction
TIME, MINUTES
rate equations of the form of ( 5 ) were
integrated by graphical methods with Fig. 6. Comparison of experimental and cal-
of any component which takes part in the preliminary values of the constants culated conversions, diethyl ether a t 294.C.
more than one of the reactions. For ex- and partial pressures calculated from
ample if it is desired to write an ex- total pressure and molar composition
pression for the net rate of disappear- data.
ance of ethanol in an ethanol reaction with the high conversion data for the
2. The resultant conversion informa- runs with added ethylene and water,
system, the following must be con- tion was compared with the experimen- respectively.
sidered: tal conversion data for the whole range The final values of the constants
C,H,OH 4 +
C2H4 H,O (A) investigated. are given in Table 2 for the three
3. The preliminary values of the temperature levels investigated. Fig-
BC,H,OH 4 C,H,OC,H, + H,O constants were adjusted as indicated ures 5 and 6 illustrate the fit of experi-
(B) by the agreement obtained between mental conversion data for an ethanol
the experimental and calculated con- and an ether run (740 and 250 mm.
C,H,OC,H, 4 C,H,OH + C,H, version information. initial pressure, respectively) at 294°C.
(C) 4. This second set of constants was with over-all rate expressions such as
with these reactions now written as ir- then checked by the procedures out- Equation ( 5 ) . Similar agreement was
reversible. It is clear that: lined above and the process repeated obtained for runs at other tempera-
until the best agreement of calculated tures and pressures. It was found how-
Net rate of disappearance of C,H,OH
= Rate in ( A ) +
Rate in (B) - Rate
with experimental conversion had been ever that the over-all rate expressions
obtained. did not agree with alcohol conversion
in (C)
Since the appropriate surface reaction Two important factors made the above data in that small range of conversions
rates are controlling in ( A ) through procedure possible. greater than 80%. The data presented
(C) then in Figure 2 for example indicate that
*High conversion data must be dealt with in
this way for an array of reactions as difficult as the rate of conversion of ethanol at
this. The rates of appearance of alcohol, ether, this point becomes quite slow. The
I'=
KaL K A p A KBzL KA'PA~ water, and ethylene at high conversions can be
D 40' found from the data of course, but these cannot two possibilities to be investigated here
be related to the rates of the actual reactions in-
volved because of the requirements of the ele- are the effects of reverse reactions or
- KsiL KBps
mental material balances. Thus the high conver- changes in the reaction mechanism re-
D
(5) sion data cannot be used for the measurements of
rates; they can only be used for checking and sulting in a great decrease in reaction
modifying the generalized rate equations derived
where from initial rate data. rate. In view of the previous comments

Page 46 A.1.Ch.E. Journal March, 1962

on the equilibria of this reaction sys- were all found to be controlled by the The authors are grateful to Union
tem it is felt that the latter case is corresponding surface rates of reaction Carbide Corporation, Allied Chemical and
much more likely; a change in the on the alumina catalyst in the range Dye Corporation, and E. I. du Pont de
mechanism of one or more of the re- of temperatures and pressures investi- Nemours and Company for fellowship
actions has occurred, and this change gated. Thus reactions (A), (C), and support during the course of this research.
The support of the National Science
is probably associated with the amount ( D ) are monomolecular surface re- Foundation is also gratefully acknowl-
of water adsorbed on the catalyst. Un- actions and reaction (B) a bimolecu- edged.
fortunately the data obtained in this lar surface reaction. The corresponding
research are not of a proper form to rate equations, in terms of disappear- NOTATION
proceed with any verification of this ance of reactant, were written with
hypothesis. values of all the pertinent constants. E' = activation energy, cal./g.-mole
In addition the effect of temperature I(' = chemical reaction equilibrium
Temperature Effects on the rates of the reactions occurring constant
The variation with temperature of has been determined. In general all K = adsorptionequilibrium con-
the rate constants and adsorption pa- rates of reaction increase with in- stant, mm. Hg-'
rameters given in Table 2 can be creasing temperature. The activation K8,L = monomolecular surface reac-
represented in linear form by a semi- energy has been determined for each tion rate constant; reactions
logarithmic plot of their values vs. of the specific surface reaction rate A, C, and D, g.-moles/min.,
reciprocal absolute temperature, as constants. The adsorption parameters g: of catalyst
suggested by the Arrhenius equation. for ethanol, water, and diethyl ether K,L = bunolecular surface reaction
The activation energies evaluated from were found to have approximately the rate constant; reaction B, g.-
this correlation for the reactions in- same temperature behavior over the moledmin., g. of catalyst
vestigated are as follows: range of investigation, decreasing with L = concentration of active sites,
increasing temperature. The relative moles/unit weight of catalyst
Reaction (A) ;
E = 10,000
Reaction (B) ;
cal./g.-mole
values of these parameters indicate de-
creasing adsorption of component on
alumina for water, ethanol, and di-
P
p
r
-
= total pressure, mm. Hg
partial pressure, mm. Hg
= rate of reaction, g.-moles/
E = 25,900 cal./g.-mole ethyl ether, in that order. min., g. of catalyst
Reaction ( C ) ;
Subscripts
E' = 28,500 cal./g.-mole
Reaction (D); ACKNOWLEDGMENT A =ethanol
E = 16,500 cal./g.-mole E = diethyl ether
This paper is based on a dissertation 0 = ethylene
presented to the Yale School of Engi-
The temperature dependence of the neering by John B. Butt in partial fulfill- o = initial conditions
adsorption equilibrium parameters for ment of the requirements for the Doctor W = water
ethanol and diethyl ether (determined of Engineering degree.
experimentally) and for water [deter- LITERATURE CITED
mined from the data of San la Ville
( 1 2 )3 indicated similar behavior of all TABLE2. FINALVALUESOF REACTION 1. Brey, W. S., and K. A. Krieger, 1.
RATE CONSTANTS AND ADSORPTION Am. Chem. Soc., 71, 3637 (1949).
three substances, decreasing adsorption PARAMETERS FOR ETHANOL AND 2. Butt, J. B., D.Eng. dissertation, Yale
with increasing temperature. The rela- D~ETHYL ETHERREACTIONS University, New Haven, Connecticut
tive values of these parameters at the ( 1960).
temperatures investigated indicate that Rate constant 3. Carberry, J. J., personal communica-
water is most strongly adsorbed on the g.-moles tion.
Tempera-
alumina catalyst, alcohol is next, and
diethyl ether least of the three. min., g. catalyst ) Value ture, 'C.
4. Cope, C. A., and B. F. Dodge,
A.1.Ch.E. lournal, 5, 10 (1959).
5. Dohse, H., Z. Physik. Chem., Abt. B,
KaiL 0.099 x lo-' 274 5, 131 (1929); 6, 343 (1930).
CONCLUSIONS (Reaction A ) 0.137 x 294 6. Hougen, 0. A., and K. M. Watson,
0.188 x lo-' 314 "Chemical Process Principles," vol. 3,
The results of this work verify the K.,L 0.627 x lo-' 274 Wiley, New York ( 1947).
utility of the recycling differential re- ( Reaction B ) 1.41 x 10" 294 7. Hougen, 0. A., and C. R. Wilke,
actor system as a convenient and valu- 3.03 x lo-' 314 Trans. Am. lnst. Chem. Engrs., 45,
able method for obtaining kinetic data. K,iL 0.054 x lo-' 274 445 (1945).
Conversion data for a wide range of (Reaction C ) 0.121 x 10-3 294 8. Langer, R. M., and C. A. Walker,
0.261 x lo-' 314 lnd. Eng. Chem., 46, 1299 (1954).
experimental conditions and for a com- 9. Langmuir, I., 1. Am. Chem. SOC., 38,
plex system of reactions were obtained KsiL 0.069 x lo-' 274
(Reaction D) 0.113 x lo-' 294 221 (1916).
and interpreted satisfactorily. The re- 0.182 x 10" 314 10. Perkins, T. K., and H. F. Rase,
actions occurring when either ethanol A.1.Ch.E. Journal, 4, 351 (1958).
or diethyl ether are passed over an 11. Polotniuk, V. I., and S. V. Dobrovol'
alumina catalyst are Adsorption skii, Zhur. Phys. Khim., 32, 2792
parameter (1958).
C,H50H + C,H, + H,O (A) (mm. Hg') 12. San la Ville, T.. Genie Chimiwe, 78,
102 (1957).' '
2C,H,OH + C,H,0C,H5 + H,O K A 1.35 x lo-' 274 13. Smith, S. B., A. X. Hiltgen, and
LOO x 10-3 294 A. J. Juhola, Chem. Eng. Progr.
(B) 0.73 x lo4 314 Symposium Ser. No. 24, 55, 25 (1959).
C,H,OC,H,+ C,H,OH + C,H, K E 1.00 x
0.80 x
10"
lo-*
274
294
14. Thiele, E. W., lnd. Eng. Chem., 31,
916 (1939).
(C) 0.61 x lo-' 314 15. Weisz, P. B., Z. fur Physik. Chem., 11,
C,H,OC,H, + 2C2H, + H,O (D) KW 1.51 x
1.22 x
lo-'
lo-'
274
294
1 ( 1957).
Manuscript received April 20, 1961; revision re-
The measured rates of these reactions 1.02 x 10" 314 ceioed July 6 , 1981; paper accepted July 6, 1961.

Vol. 8, No. 1 A.1.Ch.E. Journal Page 47