Ind. Eng. Chem. Res.

2000, 39, 1061-1063 1061

Circumventing an Azeotrope in Reactive Distillation

Jae W. Lee, Steinar Hauan, and Arthur W. Westerberg*
Department of Chemical Engineering and Institute for Complex Engineered Systems,
Carnegie Mellon University, Pittsburgh, Pennsylvania 15213

Using a McCabe-Thiele diagram, we explain nonintuitive behavior in a binary reactive

distillation column where we react and separate two isomers displaying a maximum boiling
azeotrope. Placing sufficient reaction on only two trays in the top section of the column, we
shift the top operating line to lie below the 45° line. On lower nonreacting trays, we then easily
step past the azeotrope. Once past the azeotrope, we even find that the column temperature
decreases as we step down the column.

Introduction
Azeotropes give rise to boundaries that limit the
separation possible for mixtures when using distilla-
tion.1-3 At times we can circumvent these limits by
combining reaction with distillation, a technique indus-
try uses to produce methyl tert-butyl ether (MTBE)4 and
methyl acetate.5 In this paper we use a visual ap-
proach6,7 we recently developed that allows us to see
exactly how to use an isomerization reaction to circum-
vent the azeotrope between p-ethylphenol and 2-phen-
ylethanol. The two isomers, p-ethylphenol (boiling point
218.8 °C) and 2-phenylethanol (boiling point 219.4
°C)8-10 display a maximum boiling azeotrope at a
composition of 0.55 mol fraction of p-ethylphenol, a
pressure of 1 atm, and a temperature of 228.5 °C. The
UNIFAC11 (universal quasi-chemical functional-group
activity coefficients) property set in AspenPlus predicts
this azeotropic composition and is consistent with
experimental values.8-10 Chemical equilibrium favors
the almost complete isomerization of 2-phenylethanol
(2-PHE) to p-ethylphenol (P-ETP) where the reaction
equilibrium constant is around 103 on the basis of using
ideal gas Gibbs energies of formation12 for each species:

C6H5-CH2CH2OH (2-PHE) w
CH3CH2-C6H4-OH (P-ETP) (1) Figure 1. A seven-stage reactive distillation column for convert-
ing 2-phenylethanol (2-PHE) into p-ethylphenol (P-ETP) and
simultaneously separating the product from unreacted reactant.
The Reactive Distillation Process The numbers are from a computer simulation using AspenPlus.
All compositions are for p-ethylphenol (P-ETP).
We introduce the staged reactive distillation column
shown in Figure 1 to allow reaction and distillation to sponding liquid composition. For our column, we assume
occur simultaneously for these two species. The num- the vapor stream leaving each stage is in equilibrium
bers on this plot, which we shall discuss shortly, are with the liquid leaving. For example, vapor composition
from a column simulation using the AspenPlus simula- y2 () 0.93) is in equilibrium with liquid composition x2
tor with the reflux ratio of 4 and under total condenser. () 0.86) in Figure 1, and this point will lie on the vapor/
The goal of this reactive separation is to produce pure liquid equilibrium curve at the point marked 2 in Figure
p-ethylphenol (P-ETP) and 2-phenylethanol (2-PHE) at 2. Figure 2 also features a set of “operating lines” that
the top and bottom by breaking the azeotrope. Figure 2 represent column material balances. Along these lines,
is a plot of vapor composition vs liquid composition (a possible compositions of the vapor and liquid streams
McCabe-Thiele diagram13) for the product, p-ethylphen- pass each other between stages (e.g., between stages 2
ol (P-ETP). It is on this plot that we show how reaction, and 3 y3 () 0.76) is on the first shifted operating line
strategically placed, allows us to bypass the azeotropic vs x2 () 0.86) between the points marked 2 and 3).
composition when distilling. Finally, for convenience, one also plots a diagonal “45°
In Figure 2, the S-shaped curve gives the vapor line” where vapor and liquid compositions are equal. In
composition of P-ETP in equilibrium with its corre- our earlier work, we showed how to adjust the slope and
the intersection of the operating lines with the 45° line
* To whom correspondence should be addressed. Phone: whenever we allow reaction to occur within the col-
412-268-2344.Fax: 412-268-7139.E-mail: [email protected]. umn.6,7,14,15
10.1021/ie990447k CCC: $19.00 © 2000 American Chemical Society
Published on Web 03/09/2000
1062 Ind. Eng. Chem. Res., Vol. 39, No. 4, 2000

shifted operating line, δ3, moves left by a farther 572/

810 ) 0.71 or to a point to the left of the origin. We
show no further reaction below stage 3. We continue
our tray-by-tray calculations by stepping between the
second shifted operating line and the equilibrium line
until we reach stage 6.
It is for these stages that we encounter the nonintui-
tive behavior we mentioned at the start of this paper.
Stages 4 and below have no reaction occurring on them.
Reaction above these stages shifts their operating line
downward below the equilibrium line which itself is
below the 45° line, allowing us to step past the azeotrope
as if it were not there. To the left of the azeotrope in
Figure 2, the product is heavier than the reactant (e.g.,
x4 > y4 in Figure 1), and yet is being depleted (e.g., y5 <
y4 in Figure 1) as we move down the column from stage
4 to stage 6. When it enriches, the temperature de-
creases.
We finish the diagram by including the bottom
operating line, which is for the stages below the feed
stage. We switch to it to step off stage 7. We find that
Figure 2. A McCabe-Thiele Plot for the column in Figure 1. This this stage is counterproductive as the column reverts
plot illustrates how the combination of reaction and separation
allows one to step past a binary azeotrope while distilling. to its intuitive behaviorswhich occurs when the operat-
ing line moves between the equilibrium curve and the
For this reaction and separation, the operating lines 45° line. The composition for the bottom product from
are nearly parallel in Figure 2 as the heat of reaction the column increases from below 10% to about 18%. In
is negligible. Equation 2 is a material balance and is an actual column, we should not include this bottom
the equation for an operating line above the feed stage. stage. We have included it here for illustrative purposes
Equation 3 indicates how reaction shifts the operating only.
line. Use of the commercial simulator AspenPlus verified
our results. We used the UNIFAC property set for
L DxD ξn physical property estimation (which, as we noted earlier,
yn+1 ) xn + - (2) predicts correctly the location of the azeotrope reported
V V V in the literature). The reaction turnover rates we used
ξn are insufficient to bring us to chemical equilibrium on
δn ) xD - (3) stages 2 and 3; thus, with the right catalyst they are
D entirely plausible.
We report all the flows, compositions, and tempera-
Reaction extent ξn is the total of the product A (in tures that the simulator computed in Figure 1. Note
moles/second) formed on all stages above and including especially that the temperature goes down as we move
stage n, which we can also call the sum of reaction molar down the column from stage 4 to stage 6, and the vapor
turnover flow rate until the stage n from the top. stream has more heavy species than the liquid stream
Compositions yn+1 and xn are the vapor and liquid on those stages, consistent with the construction in
compositions passing each other between stages n+1 Figure 2.
and n. L and V are the liquid and vapor molar flow
In short, we had available and used sufficient reaction
rates. D and xD are the top product molar flow rate and
on stages 2 and 3 to shift the column material balance
its composition. Each difference point (δn in eq 3 and
to cause this behavior. Graphically we used sufficient
points δ2 and δ3 in Figure 2) is a composition where a
reaction to shift the operating line below the equilibrium
shifted operating line intersects the 45° line.
line which itself was below the 45° line on stages 4
We supply pure 2-phenylethanol as a mixed phase in
through 6.
the feed stream. Our goal is to obtain almost pure
p-ethylphenol at the top and unconverted 2-phenyl- By far the most important use of the McCabe-Thiele
ethanol at the bottom. Without reaction, we cannot method for binary distillation is to provide scientists and
operate a binary column in such a way that top and engineers with a tool to sketch column behavior. From
bottom products are on opposite sides of an azeotropic these sketches, they can then see how best to design
composition. and operate such columns. We have shown here that
The top operating line starts where the top product our extensions that permit this plot to handle reactions
composition, xD, intersects the 45° line in Figure 2. By allows us to see when and how we can break a binary
material balance, compositions y1 and xD are equal; both azeotrope through the use of reaction. We needed only
are on the top operating line. Composition x1 is in sketches, not accurate drawings, to gain these insights.
equilibrium with y1; it is on the equilibrium line. We do
not allow reaction on the first stage so we find y2 on Literature Cited
the top operating line vs x1. On stage 2 (see Figure 1),
(1) Foucher, E. R.; Doherty, M. F.; Malone, M. F. Automatic
we allow a reaction turnover of ξ2 ) 240 mol/s. The Screening of Entrainers for Homogeneous Azeotropic Distillation.
intersection of the first shifted operating line with the Ind. Eng. Chem. Res. 1991, 30, 760.
45° line moves left by ξ2/D ) 240/810 ) 0.30, i.e., from (2) Wahnschafft, O. M.; Koehler, J. W.; Blass, E.; Westerberg,
0.95 to 0.65. We allow a reaction turnover of a further A. W. The Product Composition Regions of Single-feed Azeortopic
572 on stage 3. The intersection point for the second Distillation Columns. Ind. Eng. Chem. Res. 1992, 31, 2345.
 Ind. Eng. Chem. Res., Vol. 39, No. 4, 2000 1063

(3) Westerberg, A. W.; Wahnschafft, O. M. Synthesis of Distil- (11) Fredenslund, A.; Gemehling, J.; Rasmussen, P. Vapor-
lation-based Separation Systems. Adv. Chem. Eng. 1996, 23, Liquid Equilibria Using UNIFAC; Elsevier: Amsterdam, 1977.
63. (12) Daubert, T. E.; Danner, R. P.; Sibul, H. M.; Stebbins, C.
(4) Smith, L. A. Catalytic distillation process. U.S. Patent 4,- C. Physical and Thermodynamic Properties of Pure Chemicals:
307,254, 1981. Data Compilation; Hemisphere Pub. Corp.: New York, 1989.
(5) Agreda, V. H.; Partin, L. R. Reactive distillation process for (13) McCabe, W. L.; Thiele, E. W. Graphical Design of Frac-
the production of methyl acetate. U.S. Patent 4,435,595, 1984. tionating Columns. Ind. Eng. Chem. 1925, 17, 605.
(6) Lee, J. W.; Hauan, S.; Lien, K. M.; Westerberg, A. W. A (14) Hauan, S.; Westerberg, A. W.; Lien, K. M. Phenomena
Graphical Method for Designing Reactive Distillation Columns Is Based Analysis of Fixed Points in Reactive Separation Systems.
The Ponchon-Savarit Method. Proc. R. Soc. London 2000, in press. Chem. Eng. Sci. 1999, 55, 1053.
(7) Lee, J. W.; Hauan, S.; Lien, K. M.; Westerberg, A. W. A (15) Lee, J. W.; Hauan, S.; Lien, K. M.; Westerberg, A. W.
Graphical Method for Designing Reactive Distillation Columns IIs Difference Points in Extractive and Reactive Cascade II- Gener-
The McCabe-Thiele Method. Proc. R. Soc. London 2000, in press. ating Design Alternatives by Lever Rule for Reactive Systems.
(8) Lecat, M. Tables Azeotropiques: Azeotropes Binares Ortho- Chem. Eng. Sci. 1999, in press.
bares; l’Auteur: Bruxelles, 1949. Received for review June 18, 1999
(9) Horsley, L. H. Azeotropic Data-III; American Chemical Revised manuscript received December 20, 1999
Society: Washington, DC, 1973. Accepted December 21, 1999
(10) Gemehling, J.; Menke, J.; Fisher, K.; Krafczyk, J. Azeo-
tropic data, 1st ed.; John Wiley & Sons: Weinheim, 1994. IE990447K