Desalination 243 (2009) 305–313

Esterification of lactic acid with ethanol in a pervaporation

reactor: modeling and performance study
Kailas Wasewar*, Shyambabu Patidar, V.K. Agarwal
Department of Chemical Engineering, Indian Institute of Technology (IIT), Roorkee 247667, Uttarakhand, India
Tel. +91 (1332) 285347; Fax: +91 (1332) 276535; email: [email protected]; [email protected]

Received 10 December 2007; Accepted 7 April 2008

Abstract
The pervaporation (PV) reactor is a new technology for reaction and separation. Techno-economic studies are
showing that PV reactors have good market potential. The most common reaction system studied for the application
of PV is an esterification reaction between an alcohol and an acid in the presence of a catalyst. In the present paper,
modeling and simulation of a PV reactor for esterification of lactic acid with ethanol were studied. The experimental
data were used for the model validation and excellent agreement was found. The effect of several process variables
such as temperature, initial mole ratio of reactants, ratio of the effective membrane area to the volume of reacting
mixture, catalyst concentration, and flux on conversion of reactants were studied. The conversion of lactic acid was
enhanced in a PV reactor as compared to a conventional reactor. The PV reactor has a good potential for enhancing
conversion in reversible condensation reactions (especially esterification), generating water as a product. Optimum
conditions were observed at temperatures of 75 to 95EC for S/V of 10 m!1, for a catalyst concentration of 30 g/l for
a reactant ratio (acid/alcohol) of 2. The model can be used for the other esterification reactions in a PV reactor.

Keywords: Pervaporation reactor; Esterification; Lactic acid; Ethanol; Performance study

1. Introduction rate can be significantly increased when reaction

is coupled with PV, i.e. a PV reactor. Pervapo-
In recent years pervaporation (PV) has
ration involves the use of a liquid feed to produce
emerged as an energy-efficient and highly selec-
a vapor permeate and a liquid retentate. In per-
tive separation process for the separation of vola-
vaporation, the separation of two or more com-
tile products and for the dehydration of organic
ponents across a membrane is taken place by
chemicals. Also the productivity and conversion
differing rates of diffusion through a thin polymer
and an evaporative phase change comparable to
*Corresponding author. a simple flash step. A concentrate and vapor

0011-9164/09/$– See front matter © 2009 Published by Elsevier B.V.

doi:10.1016/j.desal.2008.04.030
306 K. Wasewar et al. / Desalination 243 (2009) 305–313

pressure gradient is used to allow one component when the other is in excess. The PV reactor tole-
to preferentially permeate across the membrane. rates the presence of water, which can be either in
A vacuum applied to the permeate side is coupled the reaction medium or as impurity of the react-
with the immediate condensation of the permeate ing reagent. There are upper and lower limits in
vapors [1]. Liquid transport in pervaporation the performance of reactor facilitation by PV.
takes place to the thickness of the membrane by Membrane permeability, membrane area and the
three steps. In step one, sorption of the permeate volume of the reaction mixtures to be treated are
at the interface between the feed solution and important operating parameters influencing the
membrane takes place and in second step diffu- reactor behavior. Operating temperature influ-
sion across the membrane due to concentration ences reactor performance through its influences
gradient of individual permeant carry out. In the on reaction rate and membrane permeability.
third step, desorption into the vapor phase at the Both conversion and reaction rates can be
permeate side of the membrane takes place. enhanced by using PV to remove water simul-
The most common reaction system studied for taneously from the reactor.
the application of PV reactor is an esterification A few studies are available on the PV reactor
reaction between an alcohol and an acid in the for the esterification of propionic acid with 1-
presence of a catalyst. propanol [3], acetic acid with ethanol [4,5], acetic
acid with N-butyl alcohol [6,7], acetic acid with
Cat benzyl alcohol [8], lactic acid with ethanol [9],
R1COOH  R2OH  R1COOR2  H 2O 
(1) acetic acid with methanol [10], succinic acid with
acid alcohol ester ethanol [9] and others [11,12]. It was found that
very little work on PV reactors for the esteri-
Esterification reactions are typically limited by fication reaction was available. All these studies
thermodynamic equilibrium, and face challenges ended with the experimental results showing the
with product purification. Commercially, they are kinetic and PV parameters.
carried out using either large excess of one of the In the present paper, attempts were made to
reactants or by removing one of the products by formulate the generalized model for a PV reactor
reactive distillation. The former is a relatively for the esterification of lactic acid with ethanol.
inefficient approach because it requires a large The experimental data were used for model vali-
reactor volume. As a result, reactive distillation, dation. The effect of several parameters such as
which favorably shifts equilibrium through the process temperature, reactant ratio, ratio of the
removal of one of the products, is becoming more effective membrane area to the volume of
common in plant-scale production. It is, however, reacting mixture, catalyst concentration, and flux
an energy-demanding operation and is not recom- on the conversion of reactants were studied for
mended when dealing with temperature-sensitive esterification of lactic acid with ethanol.
chemicals or biocatalysts. The afore-mentioned
difficulties can be overcome by PV reactors [2].
A water selective PV membrane is placed in
the reactor or in loop with the reactor by which 2. Model formulation
the equilibrium can shift to the right, thus reduc- Esterification reaction of ethanol using lactic
ing the excess reactants. Conversions exceeding acid to get ethyl lactate can be represented as:
equilibrium limits can be achieved by using PV to
remove water from the reaction mixtures, and that A  B  R  H
complete conversion of one reactant is obtainable Lactic acid Eth an ol Ethyl lactate Water (2)
 K. Wasewar et al. / Desalination 243 (2009) 305–313 307

The PV reactor is in semi-batch mode as water is Table 1

continuously removed from the reaction mixture Values and range of the various parameters used
using PV. The rate of water formation by the
Parameter Notation Value/
esterification reaction (2) can be written as:
and unit range

Initial mole ratio of R0 = CA0/CB0 1–3

rH  k 1C A C B  k 2 C R C H (3) lactic acid and ethanol
Ratio of the effective S/V (m2/m3) 0–30
A water-selective, hydrophilic membrane was membrane area over the
considered. The variation in the reaction volume volume of reacting mixture
Equilibrium constant for K = k1/k2 2–4
was neglected. Hence, the water balance in PV
temperature rangea
reactor can be: 50–95EC
Observed kinetic constant kobs = k1CC 0.0003–
d CH S for temperature range (m3/kmol min) 0.004
  k1C ACB  k2CR CH   J H (4) 50–95EC
dt V PV constants kPV1 0.0001
kPV2 0.000004
The relation between flux and feed water con- Reaction time t (min) 480
centration is nonlinear and not constant through- Catalyst concentration CC (g/l) 1–50
out the reaction. Hence the following modified Temperature T (EC) 50–95
relation can be used. a
Reaction rate constants for the esterification reaction are
taken from the available literature in Benedict et al [13]:
J H  k PV1CH  k PV2CH2 (5)  E   E 
k1  k0 exp    and k2  k0 exp   
 RT   RT 
The concentrations of ethyl lactate, ethanol and
lactic acid in permeate were negligible as com-
pared to water. Hence the material balance ethyl
lactate, lactic acid and ethanol can be:
 d CB  C C 
 k1CC  C ACB  R H  (9)
d CR d CA d CB dt  K 
 
dt dt dt (6) The reaction also occurs without the catalyst
  k1C ACB  k2CR CH  and the rate of the reaction with catalyst is
directly proportional to the catalyst concentration.
Eq. (6) represents the homogeneous esterification Hence, the overall rate of disappearance of com-
reaction rate equation without a catalyst. The ponent B can be given by:
effect of catalyst should be in the rate equation;
hence the reaction can be written as: d CB  C C 
  kobs  C ACB  R H  (10)
dt  K 
A B C  R  H C (7)
The value of kobs depends on the catalyst concen-
 d CB tration and reaction temperature. Also the balance
 k1C ACB CC  k2CR CH CC (8) equation for water becomes
dt
308 K. Wasewar et al. / Desalination 243 (2009) 305–313

d CH  C C  S
 kobs  C ACB  R H   JH (11)
dt  K  V

3. Solution methodology
The model equations are non-linear-coupled
differential equations which constitute the initial
value problem (IVP). The above differential
equations were solved using the RKF method in
MATLAB. For parameters such as process tem-
perature, initial mole ratio of acid over alcohol, Fig. 1. Comparison of model and experimental results of
the ratio of the effective membrane area to the lactic acid conversion for esterification with and without
volume of reacting mixture, and catalyst concen- pervaporation. (T = 95°C; Ro = 1.2; CC = 31.4 g/l;
S/V = 20 m!1).
tration flux, the equations were solved simul-
taneously. The values and range of the various
parameters used for simulations are given in Without PV
Table 1. Kinetic data and PV data were taken
from the available literature [13].

4. Results and discussion

With PV
4.1. Model validation
The model results were compared with the
available experimental data of Benedict et al. [13] Fig. 2. Comparison of model and experimental results of
for the conditions of temperature, T = 95EC, water concentration for esterification with and without
catalyst concentration, CC = 31.4 g/l the ratio of pervaporation. (T = 95°C; Ro = 1.2; CC = 31.4 g/l;
S/V = 20 m!1).
the effective membrane area over the volume of
reacting mixture S/V = 20 m!1; initial mole ratio
of ethanol and lactic acid = 1.2 and flux as per
Eq. (6) in kmol/m2min. The model results of the 4.2. Effect of temperature
conversion of lactic acid and water concentration The simulation results for effect of reaction
in a reaction mixture were compared with the temperature on conversion of lactic acid and
experimental results and are shown in Figs. 1 and water concentration are shown in Figs 3 and 4.
2, respectively. The temperature was varied from 50 to 95EC for
The proposed model results were in excellent a fixed value of catalyst concentration (CC =
agreement with the available experimental results. 31 g/l), reactants ratio (Ro = 1.2) and membrane
It was indicated that the PV enhanced the con- area to reaction volume (S/V = 20 m!1). It can be
version for the PV-aided esterification than for observed from Fig. 3 that conversion increased
the reaction without PV. The water content for with an increase in temperature but the change in
the reaction without PV was higher than for the increased conversion was not significant at higher
PV-aided reaction due to water removal by PV. temperatures. For 80EC and 95EC, the conversion
 K. Wasewar et al. / Desalination 243 (2009) 305–313 309

Fig. 3. Effect of temperature on conversion of lactic acid. T for curve 1: 50°C, 2: 60°C, 3: 75°C, 4: 85°C, 5: 95°C
(Ro = 1.2; CC = 31 g/l; S/V = 20 m!1).

Fig. 4. Effect of temperature on water concentration. T for curve 1: 50°C, 2: 60°C, 3: 75°C, 4: 85°C, 5: 95°C (Ro = 1.2;
CC = 31 g/l; S/V = 20 m!1).

was almost the same. The optimum temperature result water permeation flux was increased with
range was 75–95EC. An increase in temperature an increase in temperature. The reaction rate
induced not only an acceleration of esterification constants for the esterification were a function of
but also acceleration in PV. process temperature and were increased with the
Water production rate was higher at higher increase in temperature. The accelerating of the
temperature than in a lower temperature (Fig. 4) reaction rate constant with the increase of the
so the permeation parameter for water was also temperature for the forward reaction was faster
varied with the increased of temperature as a than the backward process.
310 K. Wasewar et al. / Desalination 243 (2009) 305–313

4.3. Effect of ratio of effective membrane area to tion of the water removal rate. The water extrac-
reaction volume (S/V) tion rate was high for a large surface area. When
the water removal rate is high, the equilibrium
The equilibrium shift of the esterification reac-
shifts more towards the right and greater con-
tion is depending on the amount of water in the
version will be achieved.
reaction mixture. As we increase S/V, the rate of
water removal will be more and therefore more
conversion. Also the cost of the membrane 4.4. Effect of catalyst concentration (CC)
depends on the required membrane area. Hence
Catalyst concentration may be an alternative
the membrane area and conversion should be
way to accelerate ester production. In view of
optimized to get the optimum production cost. As
this, the effect of catalyst concentration on the
the membrane area is small, the time required to
conversion of lactic acid was studied. The simu-
achieve a particular conversion will be more,
lation results for conversion of lactic acid and
hence higher operating costs. In the case of a
water content variation in reaction mixture during
large membrane area, operating costs will be low
PV process over various catalyst concentrations
but the capital cost will be more. In view of this,
are presented in Figs. 7 and 8. The catalyst
the effect of the ratio of effective membrane area
concentration was varied from 1 g/l to 50 g/l for
over the volume of reacting mixture on the
fixed temperature (T = 95OC), reactants ratio
conversion of lactic acid was studied. The effect
(Ro = 1.3) and membrane area to reaction volume
of the ratio of membrane area to reaction volume
(S/V = 23 m!1). For catalyst concentration above
on the conversion of lactic acid and water content
20 g/l, the final conversion of lactic acid was
in reaction mixture are presented in Figs. 5 and 6.
almost the same. The optimum catalyst concen-
The S/V was varied from 0 to 30 m!1 for 95EC
tration was 20 g/l.
temperature, 31 g/l catalyst concentration and
The conversion and water production rate was
reactant ratio of Ro = 1.3. The conversion
higher for higher CC. The effect was higher at the
achieved was a function of membrane surface
beginning than lower later during the reaction.
area; the conversion increases with an increase in
The variation of both the forward and backward
surface area. The time required to achieve a given
reaction rate occurred during the change of CC.
duty of conversion was also varied with surface
The water production rate was higher (Fig. 8) for
area of the membrane. Membrane area exerted no
a higher CC since the forward reaction rate con-
influence on reactive kinetics but caused a varia-

Fig. 5. Effect of S/V on conversion of lactic acid. S/V Fig. 6. Effect of S/V on concentration of water. S/V ratio
ratio for curve 1: 0 m!1, 2: 5 m!1, 3: 10 m!1, 4: 15 m!1, for curve 1: 0 m!1, 2: 5 m!1, 3: 10 m!1, 4: 15 m!1,
5: 20 m!1, 6: 30 m!1. (T = 95°C; Ro = 1.3; CC =31 g/l). 5: 20 m!1, 6: 30 m!1 (T = 95°C; Ro = 1.3; CC =31 g/l).
 K. Wasewar et al. / Desalination 243 (2009) 305–313 311

Fig. 7. Effect of in catalyst concentration on conversion Fig. 8. Effect of catalyst concentration on water con-
of lactic acid. CC for curve 1: 1 g/l, 2: 10 g/l, 3: 20 g/l, centration. CC for curve 1: 1 g/l, 2: 10 g/l, 3: 20 g/l,
4: 30 g/l, 5: 40 g/l, 6: 50 g/l (T = 95°C; Ro = 1.3; S/V = 4: 30 g/l, 5: 40 g/l, 6: 50 g/l (T = 95°C; Ro = 1.3; S/V =
20 m!1). 20 m!1).

Fig. 9. Effect of ratio of reactants on conversion of lactic Fig. 10. Effect of ratio of reactants on water concen-
acid. Ro for curve 1: 1 , 2: 1.25, 3: 1.5, 4: 2, 5: 3 (T = tration. Ro for curve 1: 1, 2: 1.25, 3: 1.5, 4: 2, 5: 3
95°C; CC = 20 g/l; S/V = 15 m!1). (T = 95°C; CC = 20 g/l; S/V = 15 m!1).

stant was higher than the backward one. Thus, the amplitude in water content. The water concen-
water contents in the reactor had higher maxi- tration in the reactor was lower for a higher Ro
mum amplitude for a higher CC during the during the process. It is well known that a suf-
reaction. ficient ratio of the alcohol to acid leads to a quasi-
complete conversion of alcohol even without PV.
So this method would be carried out at the cost of
4.5. Effect of initial molar reactant ratio (lactic
separation difficulties. Increasing the initial ratio
acid/ethanol, Ro)
when operating with PV may be the optimum
Figs. 9 and 10 depict the effect of initial molar performance condition.
reactant ratio on lactic acid conversion and water
in reaction volume. The reactant ratio was varied
4.6. Effect of flux
from 1 to 3 for fixed values of the other para-
meters. The higher conversion was observed for In the PV process, to study the applicability
higher ratios. The water production rate decreased for the esterification process, the important para-
with an increase of Ro and caused the maximum meter is the flux across the membrane. Flux
312 K. Wasewar et al. / Desalination 243 (2009) 305–313

depends on the type membrane used, the ope- 6. Symbols

rating temperature, surface area of membrane, the
CA — Concentration of lactic acid in
reaction mixture and its composition. Membranes
reactor, kmol/m3
with higher flux require less PV membrane area
CA0 — Initial concentration of lactic acid
and hence a lower production cost.
in reactor, kmol/m3
The effect of flux on the performance of the
CB — Concentration of ethanol in reactor,
PV reactor was studied by changing the flux to
kmol/m3
1.5 times to 20 times of base value of flux. It was
CC — Concentration of catalyst in reactor,
clearly observed that there was a slight increase
kmol/m3
in the conversion as increased in the flux; this
CH — Concentration of water in reactor,
was because of the reaction limitations.
kmol/m3
CR — Concentration of ethyl lactate in
reactor, kmol/m3
5. Conclusions
JH — Water flux through membrane,
The model for a PV reactor was developed for kmol/m2min
esterification of lactic acid with ethanol. The k1 — Forward reaction rate constant,
model results were compared with the experi- m3/kmol min
mental results and excellent agreement was k2 — Backward reaction rate constant,
found. The conversion of lactic was enhanced in m3/kmol min
the PV reactor compared to a conventional kPV1 — Empirical constant in Eq. (5)
reactor. The PV reactor has a good potential for kPV2 — Empirical constant in Eq. (5)
enhancing conversion in reversible condensation kobs — Observed kinetic constant, m3/kmol
reactions (especially esterification), generating min
water as a product. K — Equilibrium constant
The performance of the PV reactor was ana- rH — Rate of formation of water by reac-
lyzed by studying the effects of various para- tion, kmol/m3min
meters such as temperature, catalyst concen- S — Area of membrane, m2
tration, reactant ratio, ratio of membrane area to t — Reaction time, min
reaction volume and flux on conversion of lactic V — Volume of reaction mixture, m3
acid and water removal. Optimum conditions XA — Conversion of lactic acid
were observed at temperatures of 75 to 95EC for
S/V of 10 m!1, for catalyst concentration of 30 g/l,
for reactant ratio (acid / alcohol) of 2. The model References
presented can also be used for the other esteri-
fication reactions. The described model allows [1] K.L. Wasewar, Pervaporation: An energy efficient
the evolution of the reaction time necessary to technology for ethanol recovery and purification,
National Conference on Energy Management in
achieve a given conversion. The process para-
Changing Scenario, BITS, Pilani, India, 2005.
meters of temperature, catalyst concentration,
[2] S.Y. Lim, B. Park, F. Hung, M. Sahimi and T.T.
initial molar ratio of acid to alcohol, the ratio of Tsotsis, Chem. Eng. Sci., 57 (2002) 4933.
the effective membrane area to the volume of [3] M.O. David, T.Q. Nguyen and J. Neel, Trans. Inst.
reacting mixture can be changed in order to attain Chem. Eng., 69(A) (1991) 341.
the optimum of the PV–esterification coupling [4] Y. Zhu, R.G. Minet and T.T. Tsotsis, Chem. Eng.
operation. Sci., 51 (1996) 4103.
 K. Wasewar et al. / Desalination 243 (2009) 305–313 313

[5] K. Tanaka, R. Yoshikawa, C. Ying, H. Kita and [10] S. Assabumrungrat, J. Phongpatthanapanich, P.

K. Okamoto, Catalysis Today, 67 (2001) 121. Praserthdam, T. Tagawa and S. Goto, Chem. Eng. J.,
[6] Y. Zhu and H. Chen, J. Membr. Sci., 138 (1998) 123. 95 (2003) 57.
[7] Q.L. Liu and H.F. Chen, J. Membr. Sci., 196 (2002) [11] X. Feng and R.Y.M. Huang, Chem. Eng. Sci., 20
171. (1996) 4673.
[8] L. Domingues, F. Recasens and M.A. Larrayoz, [12] G.T.S. Gulik, R.E.G. Janssen, J.G. Wijers and J.T.F.
Chem. Eng. Sci., 54 (1999) 1461. Keurentjes, Chem. Eng. Sci., 56 (2001) 371.
[9] D.J. Benedict, S.J. Parulekar and S.P. Tsai, J. [13] D.J. Bendict, S.J. Parulekar and S.P. Tsai, Ind. Eng.
Membr. Sci., 281 (2006) 435. Chem. Res., 42 (2003) 2282.