2282 Ind. Eng. Chem. Res.

2003, 42, 2282-2291

Esterification of Lactic Acid and Ethanol with/without

Pervaporation
Daniel J. Benedict,†,‡ Satish J. Parulekar,*,† and Shih-Perng Tsai‡
Department of Chemical and Environmental Engineering, Illinois Institute of Technology,
Chicago, Illinois 60616, and Energy Systems Division, Argonne National Laboratory,
Argonne, Illinois 60439

Reactors coupled with membrane separation, such as pervaporation, can help enhance the
conversion of reactants for thermodynamically or kinetically limited reactions via selective
removal of one or more product species from the reaction mixture. An example of these reactions
is esterification of carboxylic acids and alcohols. Esterification of lactic acid (C3H6O3) and ethanol
(C2H5OH) is studied in well-mixed reactors with/without a solid catalyst (Amberlyst XN-1010)
in this paper. Rate expressions for homogeneous and heterogeneous esterification are obtained
from the experimental data using differential and integral methods. Experiments with a closed-
loop system of a “batch” catalytic reactor and a pervaporation unit reveal that fractional
conversions of the two reactants and yield of ethyl lactate exceeding the corresponding maximum
values in a reaction-only operation are obtained by stripping of the byproduct (water). The efficacy
of pervaporation-aided esterification is illustrated by the substantial gains in fractional conversion
of each reactant. A protocol for recovery of ethyl lactate from pervaporation retentate is proposed.
Simulations based on empirical correlations for kinetics of esterification and pervaporation reveal
the trends observed in experiments.

1. Introduction a product. Pervaporation is used to separate a liquid

mixture by partly vaporizing it through a nonporous
In recent years, there has been an increasing effort permselective membrane.10,11 The “feed” liquid mixture
to combine downstream/upstream separation with reac- is allowed to flow along one side of the membrane, and
tion to improve the process performance. In this regard, a fraction of it, the “permeate”, is recovered in the vapor
membrane technology has emerged as one of the viable state on the other side of the membrane. The permeate
separation processes. Because membranes allow selec- is kept under vacuum by continuous pumping or is
tive permeation of a component from a multicomponent purged with a stream of carrier gas. Maintaining a low
mixture, membrane reactors can help to enhance the vapor pressure on the permeate side, eliminating thereby
conversion of reactants for thermodynamically or kineti-
the effect of osmotic pressure, induces mass transport
cally limited reactions via selective removal of one or
through the membrane in this process. The permeate
more product species from the reaction mixture. When
is finally obtained in a liquid state after condensation.
multiple reactions are involved, the yield or selectivity
The permeate is enriched in the more rapidly permeat-
of a desired product can be enhanced by controlled
ing component of the feed mixture, whereas the remain-
addition of one or more reactants and/or removal of one
or more intermediates. Use of permselective membranes der of the feed that does not permeate through the
that are integrated into the reactor (the reactor walls membrane, the “retentate”, is depleted in this compo-
composed of a membrane, the so-called membrane nent.11 Although the concept of using pervaporation to
reactor) can permit this. Examples of processes where remove byproduct species dates back to the 1960s,14 only
such an enhancement in yield/selectivity can be ac- recently has this membrane process proven to be a
complished include a variety of consecutive-parallel viable separation technique.
reactions where an intermediate is the desired product. One example of condensation reactions is esterifica-
Several studies have been conducted on membrane tion of carboxylic acids and alcohols. Esterification
reactors applied to gas-phase reactions such as catalytic reactions are characterized by thermodynamic limita-
dehydrogenation, hydrogenation, and decomposition tions on conversion. To achieve a high ester yield, it is
reactions.1-8 However, much less work has been done typical to use a large excess of one of the reactants,
on liquid-phase reactions because of a lack of suitable usually alcohol, or follow the reaction by distillation to
membranes with good permselectivity and solvent re- remove in situ product(s) to drive the equilibrium to the
sistance. ester side.15 While employing a large excess of one
Pervaporation, a membrane process specially suited reactant leads to a higher cost of subsequent separations
for organic-water and organic-organic separations,9-13 to recover the unused reactant, reactive distillation has
is an ideal candidate for enhancing conversion in several pitfalls, which have been outlined by Feng and
reversible condensation reactions, generating water as Huang.13 Pervaporation-aided reactors are attractive in
this regard because, being a rate-controlled separation
* To whom correspondence should be addressed. Tel.: (312) process, the separation efficiency in pervaporation is not
567-3044. Fax: (312) 567-8874. E-mail: parulekar@iit.edu. limited by relative volatility as in distillation, the energy
†
Illinois Institute of Technology. consumption in pervaporation is low, and it can be
‡
Argonne National Laboratory. carried out at a temperature that is optimal for the
10.1021/ie020850i CCC: $25.00 © 2003 American Chemical Society
Published on Web 05/01/2003
 Ind. Eng. Chem. Res., Vol. 42, No. 11, 2003 2283

Figure 1. Schematic of the reaction-pervaporation system employed for homogeneous/heterogeneous esterification with/without
pervaporation.

reaction. This last feature is particularly important for (SS) screen], each capable of holding up to 40 g of
enzymatic esterification, where enzyme stability con- catalyst. The 5 L SS stirred contained solids reactor
siderations impose severe restrictions on the operating (SCSR) contained four spinning baskets (diameter ) 1.5
temperature. Pervaporation-aided reactors have been in., length ) 6.5 in., 400 mesh SS screen), each capable
used for esterification of carboxylic acids, such as acetic, of holding up to 100 g of catalyst. Each reactor was
erucic, oleic, propionic, tartaric, and valeric acids, and heated with two resistance heaters clamped around the
alcohols, such as benzyl alcohol, butanol, cetyl alcohol, reactor. The process streams were well insulated with
ethanol, methanol, and propanol, with various acids or fiberglass insulation tape and foam pipe wrap. A
enzymes as catalysts.16-27 Substantial acceleration of magnetically coupled, explosion-proof centrifugal pump
these reactions can be achieved using appropriate was employed to introduce the reactor feed. The reactor
commercially available solid catalysts. The increasing temperature was monitored by a T-type thermocouple
interest in pervaporation-aided esterification is also and controlled by a proportional-integral-derivative
revealed by theoretical studies in the past few controller. The reactor pressure was maintained with
years.13,24-29 a silicon-braided Teflon-packed bearing mounted on the
With this in mind, esterification of lactic acid (C3H6O3) reactor head and was monitored by glycerin-filled pres-
and ethanol (C2H5OH) is studied in well-mixed batch sure gauges. A bag filter (10 µm) was installed between
reactors with/without a solid catalyst (Amberlyst XN- the reactor and the pervaporation test cell, and a cold
1010) in this paper. Lactic acid is an important inter- trap was placed between the test cell and the vacuum
mediate in glucose metabolism in most living systems. pump. The pervaporation membrane employed in the
Rate expressions for homogeneous and heterogeneous present study, a GFT-1005 membrane (Deutsche Car-
esterification are obtained from the experimental data bone AG), is an organic acid-compatible, poly(vinyl
using differential and integral methods. The equilibrium alcohol)-based dehydration membrane. The pervapora-
conversion for this mildly exothermic reaction decreases tion of water was accomplished by applying a vacuum
with increasing temperature. Experiments with a closed- on the permeate side of the pervaporation module by a
loop system of a “batch” catalytic reactor and a per- 10-3 Torr vacuum pump. The permeate pressure was
vaporation unit reveal that fractional conversions of the monitored by a 10-4 Torr vacuum transducer, which was
two reactants and a yield of ethyl lactate exceeding the connected to a digital vacuum indicator/controller.
corresponding maximum values in a reaction-only op- Permeate was condensed and collected in two 20 mL
eration are obtained by stripping of the byproduct impingers at cryogenic temperatures with liquid nitro-
(water). The efficacy of pervaporation-aided esterifica- gen (-185 °C), with the permeate stream being well
tion is illustrated by the substantial gains in fractional insulated between the pervaporation module and the
conversion of each reactant. A protocol for the recovery impingers. During reaction-only (homogeneous or het-
of ethyl lactate from pervaporation retentate is pro- erogeneous reaction) operation, the pervaporation mod-
posed. Simulations based on empirical correlations for ule was bypassed. For conduct of the homogeneous
kinetics of esterification and pervaporation reveal the reaction, the catalyst baskets were removed from the
trends observed in experiments. reactor.
2.2. Materials. Concentrated lactic acid (in water)
2. Experimental Section
is miscible in ethanol and hence suitable for esterifica-
2.1. Reaction-Pervaporation System. The sche- tion. The starting materials used in this study, there-
matic of the reaction-pervaporation system is presented fore, were 88% (w/w; in water) lactic acid and anhydrous
in Figure 1. Two reactors with volumes of the reaction ethanol. Lactic acid and ethyl lactate (used as the
mixture of 2 and 5 L were employed in all experiments. product standard) were acquired from Aldrich Chemical,
The 2 L reactor contained four spinning baskets [diam- while ethanol was obtained from AAPER Chemical.
eter ) 1.5 in., length ) 2.5 in., 400 mesh stainless steel Amberlyst XN-1010 (Rohm and Haas), a polymeric
2284 Ind. Eng. Chem. Res., Vol. 42, No. 11, 2003

cation exchanger, was employed as the catalyst. Am-

berlyst XN-1010 is poly(styrenesulfonic acid) (PSSA)
cross-linked with divinylbenzene. The key characteris-
tics of the catalyst were density ) 5.29 g/mL, surface
area ) 540 m2/g, particle size ) 0.2-1.2 mm, and
maximum operating temperature ) 120 °C.
2.3. Analytical Procedures. Two 5 mL samples
were withdrawn at each sampling instant. One sample
was immediately cryofrozen (-185 °C) to stop the
reaction in the sample and then transferred to a
laboratory freezer (-80 °C). The frozen sample was later
analyzed to estimate the concentrations of ethanol, ethyl
lactate, and water using a gas chromatograph (GC). The
second 5 mL sample was cooled to ambient temperature
in a water bath and immediately analyzed to determine
the density of the reaction mixture and lactic acid
concentration. The former was determined by accurate
and direct measurement of the mass and volume of each
sample. The latter was determined by titration with 0.1
M NaOH after a 10-fold dilution of the sample with
deionized water. All titrations were performed in trip-
licate.
A Shimadzu GC-14A GC equipped with an AOC-1400
autosampler and an AOC-17 autoinjector was utilized
to estimate concentrations of ethanol, ethyl lactate, and
water in the reaction mixture. The GC had a thermal
conductivity detector (TCD) in series with a flame
ionization detector (FID) and was connected to a CR 501
Chromtopac integrator. The TCD and FID were con-
nected to a Supelco 30 m capillary column. The auto-
injector was fitted with a 0.5 µL SGE syringe, with the
injection volume being 0.4 µL. Both FID and TCD
utilized 99.999% pure helium as the carrier and refer-
ence gas, with 99.999% pure hydrogen and dry air being
the flame sources for the FID. A carrier gas scrubber
Figure 2. Results from a batch homogeneous esterification
and carrier gas indicator were installed to eliminate the experiment conducted at 95 °C. The concentrations of lactic acid
possibility of oxygen or water entering the chromatog- (A), ethanol (B), ethyl lactate (C), and water (D) in the initial
raphy column and causing oxidation of the gold-coated reaction mixture were 5.95, 6.37, 0.0, and 5.31 mol/L, respectively,
tungsten TCD filaments. The chromatography column and the initial reactor volume was 2 L. (a) Concentration profiles
(length ) 30 m, o.d. ) 0.32 mm, and film thickness ) of lactic acid (A, [), ethanol (B, 9), ethyl lactate (C, 2), and water
(D, b). (b) Phase portraits of CA and CB ([), CA and CC (9), and
0.25 µm) was a water-tolerant Supelcowax 10, bonded-
CA and CD (2). The straight lines denote the stoichiometric
phase medium-polarity poly(ethylene glycol) capillary relations between CA and CJ (J ) B-D), viz., CJ ) CJ0 + νJ(CA0 -
column. A 1 m guard column was installed at the inlet CA), νJ ) -1 for J ) B and 1 for J ) C and D.
to the capillary column to trap nonvolatile species such
as lactic acid and protect the column from contamina- Table 1. Additional Results from the Homogeneous
tion. All reaction samples were analyzed by GC in Batch Esterification Experiment Conducted at 95 °C
quadruplicate by splitting each sample into two vials Large-Time Data
and setting the autoinjector for 2 injections per sample K ) 2.810 789 ∆G ) -3.16 kJ/mol XAe ) 0.522
vial. Acetone was the internal standard (IS), and all Estimations of k1 and k2 (L/mol‚s)
calibration standards and all reaction samples were differential method k1 ) 8.361 × 10-6 k2 ) 2.975 × 10-6
diluted 1:1 with IS. Acetone vials were placed between integral method k1 ) 8.527 × 10-6 k2 ) 3.034 × 10-6
all sample vials to ensure that no residuals remained
in the injector port or the column and that all samples measurements of the concentrations of the four species
were analyzed at the same conditions. participating in the esterification reaction essentially
follow the appropriate stoichiometric relations. The
3. Results and Discussion various indicators for the chemical equilibrium for this
reaction at this temperature, viz., equilibrium conver-
3.1. Homogeneous Esterification. The reaction sion of the stoichiometrically limiting reactant (lactic
under consideration is esterification of lactic acid (A) acid in the present case), equilibrium coefficient, and
and ethanol (B) to generate ethyl lactate (C) and water the Gibbs free energy change for the reaction, were
(D). obtained from the large-time data (Table 1). A candidate
rate expression for this process is
A+BaC+D (1)
1 k1
The results from a batch homogeneous esterification
experiment conducted at 95 °C in a pressurized closed-
(
r ) k1 CACB -
K C D )
C C , K)
k2
(2)

loop system are presented in Figure 2. The phase

portraits in Figure 2b reveal that the independent Both differential and integral methods of analysis of the
 Ind. Eng. Chem. Res., Vol. 42, No. 11, 2003 2285

experimental data were employed to estimate the

kinetic coefficients k1 and k2 (Table 1). For the integral
method, one needs to use the integrated version of the
mass balances, which for the case of a single reaction
carried out in a batch reactor under consideration here
has the form

dXA CA
) r, XA ) 1 - (3)
dt CA0

[ ]
and the following integral form30

2RXA
1+ qk1
â-q
y ) mt, y ) ln , m) ,
2RXA K
1+
â+q
R ) CA0(K - 1),
Figure 3. Concentration profiles of ethyl lactate in batch (het-
erogeneous) esterification experiments conducted at 75, 85, and
â ) -[K(CA0 + CB0) + CC0 + CD0], 95 °C. The mass of the 5% (w/w) acid Amberlyst XN-1010 was

( )
156.8 g, and the reaction volume and agitation speed for the
CC0CD0 catalyst assembly were 5 L and 98 rpm, respectively. The initial
γ ) KCB0 - , q2 ) â2 - 4Rγ g 0 (4) concentrations of lactic acid (A), ethanol (B), ethyl lactate (C), and
CA0
water (D) were (mol/L): CA0 ) 4.5, CB0 ) 5.688, CC0 ) 0.0, and
CD0 ) 5.14 at 75 °C, CA0 ) 5.95, CB0 ) 6.794, CC0 ) 0.0, and CD0
In the differential method, the concentrations of the four ) 5.947 at 85 °C, and CA0 ) 5.8, CB0 ) 6.668, CC0 ) 0.0, and CD0
species were expressed as polynomials in time and ) 5.502 at 95 °C. The symbols 2, 9, and [ refer to data from
linear regression was employed to identify optimal k1 experiments conducted at 75, 85, and 95 °C, respectively.
and k2.31 The consistency of the estimations is indicated
by the less than 2% difference in the estimation of k1 Table 2. Results from the Heterogeneous Batch
and k2 by the two methods and the closeness of K values Esterification Experiments Conducted at 75, 85, and
estimated from the differential method and the large- 95 °Ca
time behavior of the reaction process (Table 1). The k
validity of the rate expression in eq 2 was confirmed by T ∆G [L/(kg of KA a
the use of the integral method (eq 4). (°C) XAe K (kJ/mol) catalyst)‚min] (L/mol) (L/mol)
3.2. Heterogeneous Esterification. The results 75 0.557 3.023 366 -3.201 0.3381 1.1515 0.380 861
from heterogeneous batch esterification experiments 85 0.534 2.884 329 -3.153 0.4137 0.8031 0.278 427
conducted at 75, 85, and 95 °C are presented in Figure 95 0.535 2.781 769 -3.130 0.6014 0.5603 0.201 429
3. For the sake of brevity, profiles of only the desired a The initial reactor compositions for these experiments are

product, ethyl lactate, are presented. The concentration listed in the legend for Figure 3. K ) exp(-∆G/RT).
profiles of the four species nearly followed the stoichio-
metric relations CJ ) CJ0 + νJ(CA0 - CA) with νB ) -1 and may be disguised as the reaction kinetics at the
and νC ) νD ) 1. The profiles of A and B thus were catalyst surface. Only experiments where the transport
nearly parallel to each other, and the same was true processes are not rate limiting must be considered for
for profiles of C and D (data not shown). A comparison estimation of the true kinetics of catalytic reactions.
of the profiles of CC in Figures 2 and 3 for T ) 95 °C Whether or not the transport processes are rate limiting
reveals that the catalyst promotes esterification of can be determined using a variety of criteria already
ethanol and lactic acid as anticipated. This comparison reported and widely used.30 The Weisz-Prater crite-
also reveals a slower initial increase in CC in experi- rion32 is especially suitable at the stage of estimation
ments with solid catalyst. The slower increase is at- of the kinetics of a catalytic reaction when the exact
tributable to the lags associated with heating of the form of the rate expression is unknown. For example,
reaction mixture and the catalyst to reaction tempera- the significance of external mass transport can be
ture and wetting of the catalyst. An increase in the examined by comparing the concentration of a species
temperature leads to accelerated esterification as an- in the bulk fluid (CJ, measured) to that at the external
ticipated (Figure 3). The various indicators for the surface of the catalyst (CJs, estimated). For the reaction
chemical equilibrium for the reaction process at these under consideration, mass balances around a catalyst
temperatures, obtained from the large-time data, are particle lead to the following relations:
provided in Table 2.
In a heterogeneous catalytic reaction process employ- CJs Lrj
ing porous catalysts, the rate of an individual reaction ψJ ) ) 1 + νJ , J ) A-D,
CJ kmJCJ
is influenced by the following processes, which occur in
series and parallel: external transport (from the bulk
liquid to the external surface of the catalyst), intrapar-
rj ) ∫
1 L
L 0
r′Fc dz, νA ) νB ) -1, νC ) νD ) 1,
ticle transport (porous diffusion), and the kinetics of R
L) for a spherical catalyst (5)
reactions at the catalyst surface (adsorption/desorption 3
and surface reaction steps). When either or both of the
transport processes are the slower steps, the observed with rj being the experimentally observed reaction rate.
rate of reaction reflects the rate of a transport process The significance of the intraparticle mass transport can
2286 Ind. Eng. Chem. Res., Vol. 42, No. 11, 2003

be estimated by evaluating the Weisz modulus (an

observable)

jrL2
ΦJ ) , J ) A-D (6)
DeJCJ

The closeness of ψJ to unity implies that the external

mass-transfer resistance is negligible, while ΦJ , 1
implies that the intraparticle mass-transfer resistance
is negligible. For the results presented in Figure 3,
throughout each experiment, ψJ’s (J ) A-D) were found
to be near unity and ΦJ’s much less than unity. For
example, for lactic acid (J ) A), at the three tempera-
tures studied, the maximum value of ΦA was found to
be 0.0201 (T ) 95 °C) and the minimum value of ψA
was 0.9859 (T ) 85 °C).33 It is, therefore, evident that
extraparticle and intraparticle transport resistances
were minimal in these experiments and the observed Figure 4. Determination of the activation energy and the heat
kinetics represented the true kinetics of the catalytic of reaction for the catalytic reaction and the enthalpy of adsorption
reaction. of lactic acid based on the data reported in Table 2. The symbols
Candidate rate expressions for the reaction in eq 1 9, [, and 2 denote the values of k, K, and KA, respectively, at the
three temperatures.
based on single-site mechanisms34 were tested next to
identify expression(s) that adequately represent the
experimental data. Using differential method in con-
From Figure 4, it can be deduced that, in the narrow
junction with the linear regression method to analyze
temperature range under investigation, the heat of
the data gathered in batch experiments, it was deduced
esterification reaction and the enthalpy of adsorption
that the rate expression that satisfactorily describes the
of lactic acid are invariant, with the values of these and
experimental data has the following form (a, k, and K
other pertinent parameters for the catalytic reaction
being functions of temperature):
being reported in Table 3. The esterification of lactic
acid and ethanol, therefore, is a mildly exothermic
1
r′ )
(
k CACB - C C
K C D ) (7)
reaction, while adsorption of lactic acid is substantially
more exothermic.
CB + aCCCD 3.3. Heterogeneous Esterification with Pervapo-
ration. The results from an experiment conducted at
The values of the kinetic parameters in eq 7 at the 95 °C, with the initial reaction mixture (5 L in volume)
three temperatures are reported in Table 2. From the containing lactic acid, ethanol, and water in the molar
denominator on the right-hand side of eq 7, it should ratios 1.00:2.348:1.1044 and being devoid of ethyl
be evident that adsorption of A (lactic acid) must be the lactate, are presented in Figure 5. The agitation rate
rate-determining step in the single-site mechanism and for this experiment was set at 48 rpm, the amount of
the reaction products, ethyl lactate and water, are not catalyst used was 200 g, and the withdrawal rate from
adsorbed (or adsorbed insignificantly) onto the catalyti- the reactor in the recirculation loop was kept at 2 gal/
cally active sites. The reaction mechanism can thus be min. With both reactants and both products remaining
described as (r′ ) rA ) rsr) in the reaction phase, the fractional conversion of the
stoichiometrically limiting reactant (lactic acid for the
A + S a AS, rA ) kA CACS -( CAS
KA ) (8a)
initial composition under consideration) at equilibrium
can be deduced to be 0.714. The equilibrium conversion
of lactic acid is obtained as per solution of the following:
AS + B a C + D + S,

( )
1 NJ0
rsr ) ksr CASCB - C C C (8b) K(1 - XA)(θB - XA) ) (θC + XA)(θD + XA), θJ ) ,
Ksr C D S NA0
J ) B-D (10)
For ksr . kA, the driving force for the surface reaction
(8b) can be considered to be negligible, which upon using with θB ) 2.348, θC ) 0, and θD ) 1.1044 in the present
the active sites balance (Ct ) CS + CAS) leads to case. In the absence of stripping of one or both products
expression (7) with a ) KA/K and k ) kACt. The from the reaction mixture, the equilibrium conversion
variations in the equilibrium coefficients for adsorption represents the upper limit on the fractional conversion
of lactic acid (KA) and the overall esterification reaction of the stoichiometrically limiting reactant.
(K) and the kinetic coefficient k for the catalytic reaction Throughout this experiment, variation in the density
with variation in temperature can be described by the of the reaction mixture was insignificant. One can
van’t Hoff and Arrhenius relations, viz., deduce that the reactor volume decreased with time
because of the removal of water from the reaction
d ∆HR d ∆HA mixture by pervaporation. Because the flux of water
{ln K} ) , {ln KA} ) , through the pervaporation membrane was not moni-
dT RT 2 dT RT2 tored in this experiment, estimation of the variation in
d E
{ln k} ) (9) the reactor volume with time is not possible. Neverthe-
dT RT2 less, a conservative estimate of the fractional conversion
 Ind. Eng. Chem. Res., Vol. 42, No. 11, 2003 2287

Table 3. Thermokinetic Parametersa for Esterification of Lactic Acid and Ethanol Using Amberlyst XN-1010 Catalyst
E ) 30.54 kJ/mol k0 ) 1.257 × 104 L/(kg of catalyst)‚min ∆HR ) -4.441 kJ/mol
∆HA ) -38.37 kJ/mol KA0 ) 2.02 × 10-6 L/mol
a k ) k0 exp(-E/RT), and KA ) KA0 exp(-∆HA/RT).

Figure 5. Concentration profiles of lactic acid (A, ]), ethanol (B,

0), ethyl lactate (C, 4), and water (D, b) in a heterogeneous
esterification with pervaporation experiment conducted at 95 °C. Figure 6. Concentration profiles of lactic acid (A, [), ethanol
The mass of the 10% (w/w) acid Amberlyst XN-1010 was 200 g, (B, 9), ethyl lactate (C, 2), and water (D, b) in a heterogeneous
and the reaction volume and agitation speed for the catalyst esterification with pervaporation experiment conducted at 95 °C.
assembly were 5 L and 48 rpm, respectively. The concentrations The mass of the 5% (w/w) acid Amberlyst XN-1010 was 62.7 g,
of lactic acid, ethanol, ethyl lactate, and water in the initial and the reaction volume and agitation speed for the catalyst
reaction mixture were 4.31, 10.12, 0.0, and 4.76 mol/L, respec- assembly were 2 L and 98 rpm, respectively. The concentrations
tively. of lactic acid, ethanol, ethyl lactate, and water in the initial
reaction mixture were 5.76, 6.93, 0.0, and 7.0 mol/L, respectively.
of the stoichiometrically limiting reactant (XAL) can be
obtained as conservative estimate of the fractional conversion of
lactic acid is provided by XAL (defined in eq 11), which

( )
NA CAV CA at the end of the experiment under consideration was
XA ) 1 - ) 1- g XAL, XAL ) 1 - 0.712. If the reaction is carried out sufficiently longer,
NA0 CA0V0 CA0 near total utilization of lactic acid can be accomplished.
(11) The efficacy of pervaporation-aided esterification in
increasing the equilibrium conversion of lactic acid is
On the basis of the last sample (at 81.65 h), XAL can be once again evident from these results.
estimated to be 0.9855 (i.e., XA > 0.9855 at 81.65 h). It The phase portrait of the flux of water through the
is evident that the fractional conversion in excess of the pervaporation membrane and the concentration of water
equilibrium conversion attainable in a reactor without on the feed side of the pervaporation unit (i.e., in the
product separation can be attained when one of the reactor) for this experiment is shown in Figure 7. The
products (water) is selectively removed from the reaction flux of water was correlated to the concentration of
mixture by pervaporation. As anticipated, the stripping water on the feed side of the pervaporation membrane
of water pushes the equilibrium conversion very close (i.e., in the reaction mixture) as
to unity, demonstrating the efficacy of pervaporation-
aided esterification. JD ) R1CDâ1, R1 ) 0.508, â1 ) 1.1242
The results from another experiment conducted at 95
°C, with ethanol being in excess in the initial reaction (R1 and â1 at 95 °C) (12)
mixture, are presented in Figure 6. The composition of
the initial reaction mixture was (concentrations in mol/ with JD in kg/m2‚h and CD in mol/L. The high water
L) as follows: lactic acid, 5.76; ethanol, 6.93; water, 7.00; flux in this experiment results from the high flow rate
ethyl lactate, 0.0. A total of 5% (w/w) acid Amberlyst of the reaction mixture (increased turbulence) on the
XN-1010 catalyst (amount ) 62.7 g) was added in the feed side of the pervaporation unit, higher operating
catalyst baskets, which were rotated at 98 rpm. The temperature (95 °C) for the same (better adsorption of
initial reaction mixture was 2 L in volume with a water by the membrane), and low permeate pressure
density of 1.036 kg/L, and the withdrawal rate from the (3.3 Torr; increased driving force for water transport
reactor in the recirculation loop was 3.3 gal/min. across the membrane). For each time interval between
Throughout this experiment, variation in the density consecutive samples, the flux of water was obtained as
of the reaction mixture was insignificant. the ratio of the mass of water collected on the permeate
At 95 °C and for the initial composition under side to the product of the membrane area and time
consideration, with both reactants and products remain- interval. The value thus obtained is the average flux in
ing in the reaction phase, the fractional conversion of that time interval.
lactic acid at equilibrium is 0.522 (eq 10). As discussed The flux of water through a pervaporation membrane
earlier, the reactor volume decreased with time owing is anticipated to depend on the operating temperature,
to the selective removal of water by pervaporation. A with pervaporation being promoted with increasing
2288 Ind. Eng. Chem. Res., Vol. 42, No. 11, 2003

Figure 8. VLE diagram of ethanol (1) and ethyl lactate (2) at

Figure 7. Pervaporation flux profile in a semibatch heterogeneous constant pressure (760 Torr) demonstrating no azeotrope forma-
esterification with pervaporation experiment conducted at 95 °C. tion. X1 and Y1 represent mole fractions of ethanol in the liquid
The mass of the 5% (w/w) acid Amberlyst XN-1010 was 62.7 g, and vapor, respectively. The temperatures at which VLE were
and the reaction volume and agitation speed for the catalyst established for ethanol mole fractions (X1) of 0.0, 0.022, 0.026,
assembly were 2 L and 98 rpm, respectively. The withdrawal rate 0.082, 0.2, 0.331, and 1.0 were 154, 150, 148, 138, 121, 99, and 78
from the reactor in the recirculation loop was 3.3 gal/min. The °C, respectively.
concentrations of lactic acid, ethanol, ethyl lactate, and water in
the initial reaction mixture were 5.76, 6.93, 0.0, and 7.0 mol/L, Table 4. Distillation Parameters for Recovery of Ethyl
respectively. Lactate (EtLac) from the Pervaporation Retentate (P,
Pressure; Tb, Boiling Temperature; Tv, Vapor
temperature. Pervaporation experiments with deionized Temperature)
water were conducted at different temperatures using first fraction Tb ) 40-90 °C, P ) 25 Torr, Tv ) 65 °C
the GFT-1005 membrane to estimate the apparent (90% EtOH, 10% EtLac)
activation energy for pervaporation (Ep). From a plot of second fraction Tb ) 96 °C, P ) 25 Torr, Tv ) 87 °C
ln(JD) versus 1/T, Ep was estimated to be 53.652 kJ/ (75% EtLac, 25% EtOH)
mol (data not shown).33 third fraction Tb ) 98-115 °C, P ) 25 Torr, Tv ) 89 °C
(100% EtLac)
3.4. Recovery of Ethyl Lactate from the Per-
vaporation Retentate. Ethyl lactate can be removed fication reactor coupled with a pervaporation system.
from the pervaporation retentate by distillation. The The conservation equations for A-D and the total
transfer of ethanol accompanies the transport of ester reaction mixture can be stated as
from the liquid phase to the vapor phase. Mixtures of
alcohols and esters are commonly known to form azeo- dNJ
tropes. Some examples are methanol-methyl acetate, ) r′JW - FJ, NJ(0) ) NJ0, J ) A-D; FJ ) 0,
methanol-ethyl acetate, ethanol-ethyl acetate, and dt
1-butanol-butyl acetate. Azeotrope formation is not d(FV)
J ) A-C; ) -FDMD, FD ) AmJD/MD;
desirable because breaking an azeotrope increases dt
considerably the downstream processing cost. The vapor- -r′A ) -r′B ) r′C ) r′D ) r′ (13)
liquid equilibrium (VLE) of mixtures of ethanol and
ethyl lactate was, therefore, investigated at atmospheric In view of the stoichiometric relations in eq 13, the
pressure to examine if this binary system results in following relations can be obtained among the NJ’s
azeotrope formation. The results of the VLE study of (J ) A-D) and V.
ethanol-ethyl lactate are presented in Figure 8 and do
not indicate azeotrope formation. Conventional distil- NB ) NB0 - NA0 + NA, NC ) NA0 - NA,
lation is, therefore, adequate to separate and recover
FV - F0V0
ethyl lactate from the pervaporation retentate. ND ) ND0 + NC + (14)
On the basis of the results of the VLE study, a three- MD
step vacuum distillation is utilized to separate ethanol
and ethyl lactate from the pervaporation retentate. In Because the degrees of freedom for pervaporation-aided
the first step, the majority of ethanol (more volatile esterification is 2 (two independent rate processes), the
component) and some ethyl lactate (less volatile com- number of linearly independent conservation equations
ponent) is recovered as distillate. In the second step, in eq 13 is two. To simulate the performance of this
the remainder of ethanol and a large portion of ethyl system, it is, therefore, sufficient to solve an appropriate
lactate are obtained as distillate. The distillate from the pair of conservation equations (one such pair being NA
third step is exclusively ethyl lactate. The information and the total mass of the reaction mixture, FV) in
on the suggested operating parameters and the result- conjunction with relations (14).
ing distillate composition in the three steps is provided The results of simulation for pervaporation-aided
in Table 4. esterification at conditions listed in Figure 6 (F ) F0 )
3.5. Simulation for Pervaporation-Aided Esteri- 1.036 kg/L and Am ) 0.0182 m2) are presented in Figure
fication. The information on the kinetics of catalytic 9. The experimentally observed trends for concentra-
esterification (eq 7) and pervaporation of water (eq 12) tions of the four species (Figure 6) are revealed by the
is used next to simulate the performance of an esteri- simulation. Any discrepancies between simulation re-
 Ind. Eng. Chem. Res., Vol. 42, No. 11, 2003 2289

the economic benefits of this process can be understood.

Such an analysis, however, is beyond the scope of this
paper.

4. Conclusions
Esterification of lactic acid and ethanol with/without
solid catalyst and with/without pervaporation was
studied in this paper. While the kinetics of homogeneous
esterification resembles the kinetics for an elementary
reversible reaction, for heterogeneous esterification,
adsorption of lactic acid was deduced to be the rate-
determining step in the single-site mechanism, with
ethyl lactate and water being adsorbed insignificantly
onto the catalytically active sites. Fractional conversions
in excess of the equilibrium conversion attainable in a
reactor without product separation were attained by
selective removal of water from the reaction mixture by
pervaporation. Stripping of water pushes the equilib-
rium conversion very close to unity, demonstrating the
Figure 9. Simulation results for heterogeneous esterification at
95 °C with pervaporation (solid curves) and without pervaporation efficacy of pervaporation-aided esterification. High wa-
(dashed curves) for conditions specified in the legend for Figure ter flux through the pervaporation membrane was
6. (a) Profiles of XA and y ()V/V0). (b) Profiles of CA (lower solid obtained by maintaining a high recirculation rate for
and dashed curves) and CB (upper solid and dashed curves). the reactor and a low permeate pressure. Pervaporation
(c and d) Profiles of CC and CD, respectively. was promoted with increasing temperature. Conven-
tional multistage distillation was adequate to separate
sults and experimental data are attributable in large and recover ethyl lactate from the pervaporation reten-
part to an inability to precisely measure the flux of tate because mixtures of ethanol and ethyl lactate are
water through the pervaporation membrane. As men- not prone to azeotrope formation. Simulations based on
tioned earlier, the profile of flux of water was obtained expressions for kinetics of esterification and pervapo-
as a piecewise constant function. This is an additional ration reproduced the trends observed in experiments.
source for discrepancies between simulation results and
experimental data. As was observed in the experiments Acknowledgment
(Figure 6), the simulation also predicts a maximum in
the concentration of water. The results of simulation for This work was supported by the U.S. Department of
esterification without pervaporation (FD ) JD ) 0) are Energy, Assistant Secretary for Energy Efficiency and
also presented in Figure 9. In a reaction-only batch Renewable Energy, under Contract W-31-109-Eng-38.
operation, the concentration of water (product) will D.J.B. was a recipient of a Laboratory Graduate As-
increase with time (Figures 2a and 9d). When pervapo- sistantship at Argonne National Laboratory.
ration is employed, the concentration of water in the
reactor (operating in a semibatch mode) undergoes a Nomenclature
maximum as a result of the effect of serial processes of
production by reaction and removal by pervaporation. A ) lactic acid
The rearranged form of the mass balance in eq 13 for AS, S ) active sites occupied by A and vacant active sites,
water (D) shown in the following indeed reflects the two respectively
serial processes. Am ) membrane area (m2)
a ) kinetic coefficient in eq 7 (L/mol)

( )
dCD CDMD B ) ethanol
V ) r′W - FD 1 - (15) C ) ethyl lactate
dt F CAS, CS ) concentrations of active sites occupied by A and
vacant active sites, respectively
Water on the retentate side (reactor) in this case CJ ) concentration of J (mol/L)
exhibits behavior typical of an intermediate in consecu- CJs ) concentration of J at the external surface of a catalyst
tive reactions. The simulations reveal that the esteri- particle (mol/L)
fication reaction can be driven to completion in a reactor Ct ) total concentration of active sites
coupled with a pervaporation unit. The maximum D ) water
fractional reduction in the reactor volume, (V0 - V)/V0, DeJ ) effective diffusivity of J in the porous catalyst (cm2/
corresponding to near complete conversion (XA ≈ 1) is s)
0.222. E ) activation energy for the kinetic group in a heteroge-
The technical feasibility of pervaporation-aided es- neous esterification reaction (kJ/mol)
terification for near total conversion of lactic acid to Ep ) activation energy for pervaporation (kJ/mol)
ethyl lactate has been demonstrated in this paper. It is FD ) molar flow rate of water through pervaporation
anticipated that the benefits of increased production of membrane (mol/s)
the desired product (ester) in the separation-aided J ) species J
reaction process considered here will outweigh the costs JD ) flux of water through pervaporation membrane (kg/
associated with the additional equipment required, such m2‚h)
as the pervaporation module and vacuum pump. A K ) equilibrium coefficient for an esterification reaction
detailed cost-benefit analysis will be required before KA ) equilibrium coefficient for adsorption of A (L/mol)
2290 Ind. Eng. Chem. Res., Vol. 42, No. 11, 2003

k1, k2 ) kinetic coefficients for forward and reverse steps, (7) Gao, H.; Xu, Y.; Liao, S.; Liu, R.; Yu, D. Catalytic Hydro-
respectively, in homogeneous esterification (L/mol‚s) genation and Gas Permeation Properties of Metal-Containing Poly-
k ) kinetic group in the rate expression for catalytic (phenylene oxide) and Polysulfone. J. Appl. Polym. Sci. 1993, 50,
esterification [L/(kg of catalyst)‚min] 1035.
(8) Gao, H.; Xu, Y.; Liao, S.; Liu, R.; Liu, L.; Li, D.; Yu, D.; Zhao,
kA, ksr ) kinetic coefficients for adsorption of lactic acid Y.; Fan, Y. Catalytic Polymeric Hollow Fiber Reactors for Selective
and surface reaction steps, respectively Hydrogenation of Conjugated Dienes. J. Membr. Sci. 1995, 106,
kmJ ) external mass-transfer coefficient for species J 213.
(cm/s) (9) Huang, R. Y. M.; Rhim, J. W. Separation Characteristics of
k0, KA0 ) provided in Table 3 [L/(kg of catalyst)‚min and Pervaporation Membrane Separation Processes. In Pervaporation
L/mol, respectively] Membrane Separation Processes; Huang, R. Y. M., Ed.; Elsevier:
MD ) molecular weight of water (g/gmol) Amsterdam, The Netherlands, 1991; pp 111-180.
m, q, y ) defined in eq 4 (10) Neel, J. Introduction to Pervaporation. In Pervaporation
Membrane Separation Processes; Huang, R. Y. M., Ed.; Elsevier:
NJ ) number of moles of J in the reactor (mol)
Amsterdam, The Netherlands, 1991; pp 1-109.
R ) radius of the spherical catalyst particle (cm) (11) Neel, J. Pervaporation. In Membrane Separations Technol-
R ) universal gas law constant (atm‚L/mol‚K) ogy, Principles and Applications; Noble, R. D., Stern, S. A., Eds.;
r ) rate of a homogeneous esterification reaction (mol/L‚s) Elsevier: Amsterdam, The Netherlands, 1995; pp 143-211.
rj ) volume-average reaction rate (experimentally observed) (12) Fleming, H. L.; Slater, C. S. In Membrane Handbook; Ho,
in the catalyst particle, defined in eq 5 (mol/L‚min) W. S. W., Sirkar, K. K., Eds.; Van Nostrand Reinhold: New York,
r′ ) rate of a heterogeneous catalytic reaction [mol/(kg of 1992; Chapters 7-10, pp 105-159.
catalyst)‚min] (13) Feng, X.; Huang, R. Y. M. Studies of a Membrane Reac-
r′J ) rate of formation of J by the catalytic reaction [mol/ tor: Esterification Facilitated by Pervaporation. Chem. Eng. Sci.
1996, 51, 4673.
(kg of catalyst)‚min]
(14) Jennings, J. F.; Binning, R. C. Removal of Water Generated
rA, rsr ) rates of adsorption of lactic acid and surface in Organic Chemical Reactions. U.S. Patent 2,956,070, 1960.
reaction, respectively [mol/(kg of catalyst)‚min] (15) Reid, E. E. Esterification. In Unit Processes in Organic
T ) absolute temperature (K) Synthesis, 4th ed.; Groggins, P. H., Ed.; McGraw-Hill: New York,
t ) time (h or s) 1952; pp 596-642.
V ) reactor volume (L) (16) Kita, H.; Sasaki, S.; Tanaka, K.; Okamoto, K.-I.; Yama-
XA ) fractional conversion of A, defined in eq 3 moto, M. Esterification of Carboxylic Acid with Ethanol Ac-
XAL ) conservative estimate of XA, defined in eq 11 companied by Pervaporation. Chem. Lett. 1988, 2025.
(17) David, M.-O.; Gref, R.; Nguyen, T. Q.; Neel, J. Pervapo-
Greek Symbols ration-Esterification Coupling. I. Basic Kinetic Model. Trans. Inst.
Chem. Eng. 1991, 69, 335.
R, â, γ ) defined in eq 4 (18) David, M.-O.; Gref, R.; Nguyen, T. Q.; Neel, J. Pervapo-
R1, â1 ) parameters in eq 12 ration-Esterification Coupling. II. Modeling of the Influence of
∆G ) Gibbs free energy change for an esterification Different Operating Parameters. Trans. Inst. Chem. Eng. 1991,
reaction (kJ/mol) 69, 341.
∆HA, ∆HR ) heats of adsorption of lactic acid and an (19) Nijhuis, H. H.; Kemperman, A.; Derksen, J. T. P.; Cuperus,
esterification reaction, respectively (kJ/mol) F. P. Pervaporation Controlled Bio-Catalytic Esterification Reac-
∆SR ) entropy change associated with an esterification tions. In Proceedings of Sixth International Conference on Per-
vaporation Processes in the Chemical Industry; Bakish, R., Ed.;
reaction (J/mol‚K)
Bakish Materials Corp.: Englewood Cliffs, NJ, 1992; pp 368-379.
νJ ) stoichiometric coefficient of J, -1 for A and B and 1 (20) Okamoto, K.-L.; Yamamoto, M.; Otoshi, Y.; Semoto, T.;
for C and D Yano, M.; Tanaka, K.; Kita, H. Pervaporation-Aided Esterification
ψJ ) defined in eq 5 of Oleic Acid. J. Chem. Eng. Jpn. 1993, 26, 475.
φJ ) Weisz modulus, defined in eq 6 (21) Keurentjes, J. T. F.; Janssen, G. H. R.; Gorissen, J. J. The
F ) density of the reaction mixture (kg/L) Esterification of Tartaric Acid with Ethanol: Kinetics and Shifting
Fc ) density of the catalyst particle (kg of catalyst/L) the Equilibrium by Means of Pervaporation. Chem. Eng. Sci. 1994,
θJ ) defined in eq 10 for J ) B-D 49, 4681.
(22) Kwon, S. J.; Song, K. M.; Hong, W. H.; Rhee, J. S. Removal
Subscripts of Water Product from Lipase-Catalyzed Esterification in Organic
Solvent by Pervaporation. Biotechnol. Bioeng. 1995, 46, 393.
e ) chemical equilibrium (23) Ni, X.; Xu, Z.; Shi, Y.; Hu, Y. Modified Aromatic Polyimide
J ) species J Membrane Preparation and Pervaporation Results for Esterifica-
0 ) initial condition (t ) 0) tion System. Water Treat. 1995, 10, 115.
(24) Ronnback, R.; Salmi, T.; Vuori, A.; Haario, H.; Lehtonen,
Literature Cited J.; Sundqvist, A.; Tirronen, E. Development of a Kinetic Model
for the Esterification of Acetic Acid with Methanol in the Presence
(1) Mohan, K.; Govind, R. Analysis of Equilibrium Shift in of a Homogeneous Acid Catalyst. Chem. Eng. Sci. 1997, 52, 3369.
Isothermal Reactors With a Permselective Wall. AIChE J. 1988, (25) Domingues, L.; Recasens, F.; Larrayoz, M. Studies of a
34, 1493. Pervaporation Reactor: Kinetics and Equilibrium Shift in Benzyl
(2) Shu, J.; Grandjean, B. P. A.; Neste, A. V.; Kaliaguire, S. Alcohol Acetylation. Chem. Eng. Sci. 1999, 54, 1461.
Catalytic Palladium-Based Membrane Reactors: A Review. Can. (26) Krupiczka, R.; Koszorz, Z. Activity-Based Model of the
J. Chem. Eng. 1991, 69, 1036. Hybrid Process of an Esterification Reaction Coupled with Per-
(3) Ionnides, T.; Gavalas, G. R. Catalytic Isobutene Dehydro- vaporation. Sep. Purif. Technol. 1999, 16, 55.
genation in a Dense Silica Membrane Reactor. J. Membr. Sci. (27) Liu, Q. L.; Chen, H. F. Modeling of Esterification of Acetic
1993, 77, 207. Acid with n-Butanol in the Presence of Zr(SO4)2‚4H2O Coupled
(4) Itoh, N.; Xu, W.-C. Selective Hydrogenation of Phenol to Pervaporation. J. Membr. Sci. 2002, 196, 171.
Cyclohexanone Using Palladium-Based Membranes as Catalysts. (28) Xuehui, L.; Lefu, W. Kinetic Model for an Esterification
Appl. Catal. A 1993, 107, 83. Process Coupled by Pervaporation. J. Membr. Sci. 2001, 186, 19.
(5) Ziaka, Z. D.; Minet, R. G.; Tsotsis, T. T. Propane Dehydro- (29) Eliceche, A. M.; Davioua, A. M. C.; Hocha, P. M.; Uribeb,
genation in a Packed Bed Membrane Reactor. AIChE J. 1993, 39, I. O. Optimization of Azeotropic Distillation Columns Combined
526. with Pervaporation Membranes. Comput. Chem. Eng. 2002, 26,
(6) Ziaka, Z. D.; Minet, R. G.; Tsotsis, T. T. A High-Temperature 563.
Catalytic Membrane Reactor for Propane Dehydrogenation. J. (30) Froment, G. F.; Bischoff, K. B. Chemical Reactor Analysis
Membr. Sci. 1993, 77, 221. and Design, 2nd ed.; Wiley: New York, 1990.
 Ind. Eng. Chem. Res., Vol. 42, No. 11, 2003 2291

(31) Ogunnaike, B. A.; Ray, W. H. Process Dynamics, Modeling, (34) Yang, K. H.; Hougen, O. A. Determination of Mechanism
and Control; Oxford University Press: New York, 1994. of Catalyzed Gaseous Reactions. Chem. Eng. Prog. 1950, 46, 146.
(32) Weisz, P. B.; Prater, C. D. Interpretation of Measurements
in Experimental Catalysis. Adv. Catal. 1954, 6, 143. Received for review October 28, 2002
(33) Benedict, D. J. Heterogeneous Reaction Kinetics and Revised manuscript received March 20, 2003
Pervaporation-Aided Esterification of Organic Acids Derived from Accepted March 20, 2003
Alternative Feedstocks. M.S. Thesis, Illinois Institute of Technol-
ogy, Chicago, IL, 1998. IE020850I