Chemical Engineering Science 57 (2002) 1577 – 1584

www.elsevier.com/locate/ces

Application of zeolite T membrane to vapor-permeation-aided

esteri'cation of lactic acid with ethanol
Kazuhiro Tanaka, Ryuuhei Yoshikawa, Cui Ying, Hidetoshi Kita, Ken-ichi Okamoto ∗
Department of Advanced Materials Science and Engineering, Faculty of Engineering, Yamaguchi University, Tokiwadai, Ube 755-8611, Japan

Abstract

Zeolite T membranes were applied to vapor-permeation-aided esteri'cation of lactic acid with ethanol. The hybrid process provided
almost complete conversion within a short reaction time by removing water from the reaction mixture. Zeolite T membrane worked
steadily for a long time. The reaction time-courses were described by a model based on the assumptions that the esteri'cation obeyed
second-order kinetics and the permeation 3ux of each component was proportional to its concentration in the reaction mixture. The 'nal
reaction liquid mixtures consisted mostly of ethyl lactate and ethanol with little ester of polylactic acids, although concentrated lactic acid
solution was used as a source. ? 2002 Elsevier Science Ltd. All rights reserved.

Keywords: Kinetics; Membranes; Reaction engineering; Separations; Zeolites

1. Introduction and found it has high pervaporation (PV) performance to

The combination of membrane separation process with
chemical reaction process has been increasingly attracting
much attention. There have been a number of investigations
done for hydrogen-permeable membrane reactors applied
to reversible gas-phase reactions (Itoh, 1987; Umemiya,
Sato, Ando, & Kikuchi, 1991). Some investigations have
focused on water-permeable membrane reactors applied to
liquid-phase reversible reactions such as esteri'cations of
carboxylic acids with alcohols (Kita, Tanaka, Okamoto,
& Yamamoto, 1987; David, Gref, Nguyen, & Neel, 1991;
Okamoto et al., 1993, 1994; Yamamoto et al., 1995; Lip-
nizki, Field, & Ten, 1999). Vapor permeation (VP) using
asymmetric polyimide hollow 'ber membrane improved
the esteri'cations of oleic acid with alcohols (Okamoto
et al., 1994). Further improvement of the esteri'cation was
achieved by VP separation with zeolite NaA membrane
because of its extremely high permeation 3ux (Yamamoto
et al., 1995). Almost complete conversion was reached
within 1:5 h with the initial molar ratio of alcohol to the
acid m0 of 1.5. However, zeolite A membrane is unstable
to acids and not applied to the system where it directly con-
tact with acids. Recently, we prepared zeolite T membrane
∗ Corresponding author. Tel.: +81-836-85-9660;
fax: +81-836-85-9601.
E-mail address: okamotok@po.cc.yamaguchi-u.ac.jp (K.-i. Okamoto).
water-organic liquid mixtures as well as good stability to
acid solutions (Takagi, Kita, & Okamoto, 1999). The PV
separation with zeolite T membrane was applied to the es-
teri'cation of acetic acid with ethanol (Tanaka, Yoshikawa,
Cui, Kita, & Okamoto, 2001). The conversion exceeded the
equilibrium limit and reached almost 100% within 8 h for
m0 = 1:5 and 2.
Lactic acid esters are biodegradable and can be used as
detergents and solvents. Conventional esteri'cation are used
for commercial production of methyl, ethyl, and n-butyl lac-
tates. The methyl and ethyl lactates are soluble in water and
it is not easy to remove water from the reaction mixtures.
Water is more volatile than ethyl lactate and the VP using a
water-permeable membrane is expected to improve the es-
teri'cation of lactic acid with ethanol. Although lactic acid
has low volatility, zeolite T membrane may be suitable for
this system.
In this study, the VP separation with zeolite T membranes
is applied to esteri'cation of lactic acid with ethanol.
2. Experimental
Zeolite T membrane was grown hydrothermally on the
surface of a porous cylindrical mullite support. The alu-
minosilicate gel used in the synthesis was prepared by
mixing precipitated silica, alkaline aluminate, sodium

0009-2509/02/$ - see front matter ? 2002 Elsevier Science Ltd. All rights reserved.
PII: S 0 0 0 9 - 2 5 0 9 ( 0 2 ) 0 0 0 3 3 - 7
1578 K. Tanaka et al. / Chemical Engineering Science 57 (2002) 1577–1584
Fig. 1. Schematic diagram of the experimental apparatus for
VP-aided-esteri'cation.
hydroxide, potassium hydroxide, and water. The molar
composition of the resulting gel was SiO2 : Al2 O3 : (Na2 O +
K2 O) : H2 O = 1 : 0:05 : 0:35 : 14. The seeded support tube
was placed in the gel. After hydrothermal synthesis at
373 K for 28 h, the tube was taken out, washed with water
and air-dried at ambient temperature.
The VP-aided esteri'cation was carried out using the
apparatus shown in Fig. 1. A reaction cell consisted of a
four-neck round 3ask (300 × 10−6 m3 ), an empty distilla-
tion column (3 × 10−2 m inner diameter and 50 × 10−2 m
in length), and a condenser. The membrane was set in
the distillation column. The eLective membrane area was
49 × 10−4 m2 . The inside of the tube was evacuated, and
the permeate was condensed in a trap cooled by liquid ni-
trogen. The performance of the membrane was evaluated
by a permeation 3ux (Q in kg m−2 h−1 ) and a separation
factor ().
The catalyst used was a cation exchange resin (Amberlyst
15, Organo). The catalyst and ethanol were put in the reactor
and the reactor was kept at boiling temperature. Lactic acid
was then poured into the reactor to start the esteri'cation.
The esteri'cation experiments with and without VP were
carried out. A sample (0:2 × 10−3 kg) was withdrawn peri-
odically and analyzed. Acidity of the reaction mixture was
determined by titration with a standard potassium hydrox-
ide solution. The concentrations of ethanol and water were
determined by gas chromatograph using acetone as an in-
ternal standard. The concentrations of monomeric and poly-
meric lactic acids and their ethyl esters were determined by
high-performance liquid chromatograph (JASCO Co.) with
a column packed with ODS gel. An aqueous solution of
10% methanol with 0:5 wt% phosphoric acid was used as
an eluting solvent. Lactic acids and their ethyl esters were
identi'ed by comparing the retention time.
The composition of a dilute aqueous solution, contain-
ing less than approximately 20% lactic acid, corresponds
essentially to monomeric lactic acid and water, whereas
the concentrated solutions are more complex because of
self-esteri'cation to the so-called polylactic acids of vari-
ous chain lengths (Filachione, 1952). A reagent grade of
lactic acid solution was used, of which the titratable acid-
ity was 73%. Its total acidity was determined to be 91%
by the titration after the complete hydrolysis of the reagent
grade solution. The regent grade of lactic acid solution was
identi'ed as a mixture of monomeric, dimeric and trimeric
lactic acids, L1 , L2 , and L3 , respectively, and water. Their
compositions were determined to be 58, 22, 6 and 14 wt%,
respectively. The conversion of esteri'cation was calcu-
lated based on the titratable acidity, which corresponds to
sum of molar concentration of L1 , L2 and L3 . The con-
version obtained from the titration was in good agreement
with the value calculated from the liquid chromatographic
analysis.
3. Results and discussion
Table 1 lists typical PV and VP performance of zeolite T
membrane for a water=ethanol (10=90 wt%) liquid mixture.
Fig. 2 shows variation in the conversion and the compo-
sition of reaction mixture with reaction time for esteri'ca-
tion of lactic acid without VP for the initial molar ratio of
ethanol to sum of lactic acids m0 of 2.4 at boiling points
(363–365 K). The conversion increased to 65% in 1 h and
slightly decreased, while concentration of ethyl lactate L1 ES
reached 2:7 mol kg−1 in 1 h and further increased slowly.
Since several reactions occur simultaneously in the reaction
mixture, the composition of reaction mixture changed even
when the conversion hardly changed. For better understand-
ing of the eLect of VP separation on the esteri'cation of
lactic acid, it is necessary to follow variation in the concen-
tration of each component in the reaction mixture Ci as well
as variation in the conversion.
Table 1
Performance of a zeolite T membrane for a water-ethanol liquid mixture
(10=90 wt%)
Experiments Water content Water content in Q (kg m−2 h−1 ) 
in feed (wt%) permeate (wt%)
PV at 348 K 10.9 98.3 0.88 480
VP at 393 K 7.7 97.0 0.34 400
 K. Tanaka et al. / Chemical Engineering Science 57 (2002) 1577–1584 1579

Fig. 2. Time-course of esteri'cation without VP for m0 of 2.4 at boiling temperature(363–365 K). The solid lines were calculated from Eqs. (6) to (14)
using the values listed in Table 2. : L1 ,
: ethanol, : ethyl lactate(L1 ES), • : water,  : L2 , 4 : L2 ES,
: L3 ,  : L3 ES.

k4
As a rough approximation we considered the following L2 + H2 O  2L1 (4)
've reactions. k4 =Ke4
5 k
k1 L3 + H2 O→L 1 + L2 ; (5)
L1 + EtOH  L1 ES + H2 O; (1)
k1 =Ke1 where L2 ES and L3 ES are ethyl esters of L2 and L3 , re-
k2 spectively. With reaction time, after 1 h, the CL1 ES slowly
L2 + EtOH  L2 ES + H2 O; (2) increased, and the CL1 also slightly increased, as shown in
k2 =Ke2
Fig. 2(b). This was due to the formation of L1 by the hy-
k3
L3 + EtOH  L3 ES + H2 O; (3) drolysis of L2 and L3 , namely, the reactions (4) and (5).
k3 =Ke3 The hydrolysis slightly decreased the conversion calcu-
1580 K. Tanaka et al. / Chemical Engineering Science 57 (2002) 1577–1584

lated based on the titratable acidity. The CL2 ES increased Table 2

to 0:20 mol kg−1 in 4 h, and then decreased down to Parameter values used to calculate the time course of esteri'cation of
lactic acid with ethanol based on a reaction model
0:10 mol kg−1 . The decrease may be caused by the con-
sumption of L2 . Since the CL2 decreased from 0.75 to Parameters Values
0:05 mol kg−1 during this esteri'cation, most of L2 was −1
k1 [kg mol min ] −1
4.6
considered to react with water into L1 . As shown in Fig. k2 [kg mol−1 min−1 ] 1.0
2(c), the CL3 decreased monotonically and became zero k3 [kg mol−1 min−1 ] 0.4
at 22 h. The initial CL3 was 0:14 mol kg−1 , whereas the k4 [kg mol−1 min−1 ] 1.7
k5 [kg mol−1 min−1 ] 0.5
CL3 ES was ¡ 0:04 mol kg−1 , indicating that most of L3
Ea;1 [kJ mol−1 ] 47
was also transformed into L1 . Since the chromatographic Ea;2 [kJ mol−1 ] 72
peak of L3 ES was broad and followed by the tailing, the Ea;3 [kJ mol−1 ] 93
CL3 ES seemed less accurate than the concentrations de- Ea;4 [kJ mol−1 ] 57
termined from other peaks. It was not clear whether the Ea;5 [kJ mol−1 ] 63
CL3 ES might start to decrease after 4 h as in the case of the a [kg mol−1 ] 0.23
Ke1 2.4
CL2 ES . Ke2 1.2
Although small amounts of L2 , L2 ES and L3 ES still re- Ke3 1.0
mained in the 'nal reaction mixture, the equilibrium con- Ke4 5.0
stant of L1 ES formation calculated from the 'nal values of The values of ki are at 363 K and CH2 O = 7:5 mol kg−1 .
CL1 , CEtOH , CL1 ES and CH2 O was 2.4. This was in good agree-
ment with the value reported in the literature (Filachione,

1952). dCH2 O k1
= k1 CL1 CEtOH − CL1 ES CH2 O + k2 CL2 CEtOH
According to the general kinetic expression for second- dt Ke1
order reversible reaction, the reaction rate of each compo-

k2
nent is written as: − CL2 ES CH2 O + k3 CL3 CEtOH

 Ke2
dCL1 k1

= −k1 CL1 CEtOH + CL1 ES CH2 O k3
dt Ke1 − CL3 ES CH2 O − k4 CL2 CH2 O

 Ke3
k4

+ 2k4 CL2 CH2 O − 2 CL21 + k5 CL3 CH2 O (6) k4
Ke4 + CL21 − k5 CL3 CH2 O : (13)

 Ke4
dCL2 k2


= −k2 CL2 CEtOH + CL2 ES CH2 O 1 Eai
dt Ke2 ki = ki;0 exp − (14)
1 + aCH2 O RT


k4 The equilibrium constants Ke1 was 'xed to be 2.4 as men-
− k4 CL2 CH2 O + CL21 + k5 CL3 CH2 O ; (7)
Ke4 tioned above. The Ke2 value was determined from the max-

 imum value of CL2 ES . The forward rate constants k1 and
dCL3 k3
= −k3 CL3 CEtOH + CL3 ES CH2 O k2 were determined from the initial slope of the increases
dt Ke3
in CL1 ES and CL2 ES . Although the determination of Ke3 and
− k5 CL3 CH2 O ; (8) k3 were ambiguous because of inaccurate CL3 ES , those val-
ues did not aLect the other Ci other than the CL3 ES . The


dCL1 ES k1 rate constants k4 and k5 , were determined from another ex-
= k1 CL1 CEtOH − CL1 ES CH2 O ; (9) periment, where the hydrolysis of the concentrated lactic
dt Ke1

 acid solution was carried out under the same condition as
dCL2 ES k2 this esteri'cation using water instead of ethanol. The re-
= k2 CL2 CEtOH − CL2 ES CH2 O ; (10)
dt Ke2 verse reaction of Eq. (5) was negligible. The eLect of water

 content on the reaction rate was taken into account in the
dCL3 ES k3
= k3 CL3 CEtOH − CL3 ES CH2 O ; (11) form of (1+aCw)−1(Okamoto, Yamamoto, et al., 1994).
dt Ke3 The values of a and activation energies Eai were determined

 from the esteri'cation and hydrolysis of lactic acid at dif-
dCEtOH k1
= −k1 CL1 CEtOH + CL1 ES CH2 O − k2 CL2 CEtOH ferent temperatures and water contents. The lines shown in
dt Ke1

 Figs. 2 and 3 were calculated from Eqs. (6) to (14) us-
k2 ing the values of ks and Kes listed in Table 2. The calcu-
+ CL2 ES CH2 O − k3 CL3 CEtOH
Ke2 lated lines were in rough agreement with the experimental

 values.
k3 Fig. 3 shows the results for the esteri'cations of lactic acid
+ CL3 ES CH2 O ; (12)
Ke3 without VP for m0 of 3.6 at boiling points (360 –362 K).
 K. Tanaka et al. / Chemical Engineering Science 57 (2002) 1577–1584 1581

Fig. 3. Time-course of esteri'cation without VP for m0 of 3.6 at boiling temperature (360–362 K). The solid lines were calculated from Eqs. (6) to (14)
using the values listed in Table 2. : L1 ,
: ethanol, : ethyl lactate(L1 ES), • : water,  : L2 , 4 : L2 ES,
: L3 .

These were similar to the results for m0 of 2.4 described
above. The 'nal conversion for m0 of 3.6 was a little higher
than that for m0 of 2.4.
Figs. 4 and 5 shows the results with VP for m0 of 2.4
and 3.6, respectively. The conversion for VP-aided reaction
exceeded the thermodynamic equilibrium. Almost complete
conversion of 100% was reached within 8 h. The CH2 O in-
creased in the early stage ¡ 1 h, and then decreased. The
water content in the reaction mixture was ¡ 10 wt% dur-
ing the reaction. The L2 and L3 disappeared within 8 h. The
CL1 was ¡ 0:1 mol kg−1 . For comparison, the compositions
of the 'nal reaction mixtures for the esteri'cation with and
without VP were listed in Table 3. The concentration of
L2 ES, a by-product, in the 'nal reaction mixture were higher
for the esteri'cation with VP than for that without VP. This
was due to disappearance of water. The residual amount of
1582 K. Tanaka et al. / Chemical Engineering Science 57 (2002) 1577–1584

Fig. 4. Time-course of esteri'cation with VP for m0 of 2.4 at boiling temperature (363–375 K). The solid lines were calculated from Eqs. (6) to (12),
(14) and (15) using the values listed in Table 2. : L1 ,
: ethanol, : ethyl lactate(L1 ES), • : water,  : L2 , 4 : L2 ES,
: L3 ,  : L3 ES.

L2 ES, however, was ¡ 10% of the amount of formed L1 ES.
It is evident that the selective removal of water by VP sep-
aration using zeolite T membrane shifted the equilibrium in
favor of ester formation of lactic acid.
The complete conversion was achieved within a similar
time for m0 of 2.4 and 3.6, although the 'nal conversion
without VP was a little lower for m0 = 2:4 than for m0 = 3:6.
The amount of 'nally obtained ethyl lactate was larger for
m0 = 2:4 than for m0 = 3:6; that is, the lower m0 was rather
advantageous for VP-aided-esteri'cation. This was due to
higher water content in the vapor phase, resulting in a higher
rate of removing water for m0 = 2:4 than for m0 = 3:6. The
water content in the vapor phase 12% and 10% for m0 of 2.4
and 3.6, respectively, at a reaction time of 2 h. The Q values
at 2 h were in 0.5 and 0:3 kg m−2 h−1 for m0 of 2.4 and
3.6, respectively. The composition of vapor phase depended
 K. Tanaka et al. / Chemical Engineering Science 57 (2002) 1577–1584 1583

Fig. 5. Time-course of esteri'cation with VP for m0 of 3.6 at boiling temperature (360–366 K). The solid lines were calculated from Eqs. (6) to (12)
and (14) using the values listed in Table 2. : L1 ,
: ethanol, : ethyl lactate(L1 ES), • : water,  : L2 , 4 : L2 ES,
: L3 .

mainly on the composition of volatile components, namely,
ethanol, water and L1 ES, in the reaction mixture. The content
of L1 ES in the vapor was 12 wt% at 8 h. The contents of
L1 and L2 ES were ¡ 0:5 wt%. The L2 , L3 , and L3 ES were
not present in the vapor. Therefore, the water content in the
vapor phase was higher than that in the reaction mixture. It
was advantageous to apply the water-permeable membrane
to VP-aided-esteri'cation. The permeate was nearly water
with a little ethanol (¡ 1%). The total amount of permeated
ethanol as 0:4 wt% of initial amount of ethanol. The  was
higher than 500.
The membrane used in the VP-aided-esteri'cation of lac-
tic acid directly contacted with the monomeric lactic acid
vapor. The VP performance of the zeolite T membrane
used in a series of the VP-aided-esteri'cation experiments
was measured at 393 K for a water-ethanol liquid mixture
(10=90 wt%) before and after the use in the reaction system.
The results were listed in Table 4. In this VP experiment,
the content of water in the vapor phase was 7:7 wt%. The
use in the VP-aided-esteri'cation did not aLect Q and .
Permeation 3ux of each component is generally consid-
ered to be proportional to its composition. The permeated
1584 K. Tanaka et al. / Chemical Engineering Science 57 (2002) 1577–1584

Table 3 where qw10 was the water 3ux in kg m−2 h−1 at 10 wt%
Compositions of 'nal reaction mixtures for the esteri'cation with and water composition of the reaction mixture, S eLective mem-
without VP
brane area in m2 , and CH 2 O the concentration of water in
Components Concentration [mol kg−1 ] wt%. The amount of reaction mixture V was in kilogram and
m0 = 2:4 m0 = 3:6 calculated by subtracting the permeate and sampling amount
from the initial amount of reaction mixture V0 . The solid
Without VPa With VPb Without VPa With VPb
lines in Figs. 4 and 5 were calculated from these equations
L1 1.6 0.08 0.98 0.06 using the parameter values and the water 3ux measured by
L2 0.05 0.00 0.02 0.00 VP experiment for the water-ethanol liquid mixture (10 wt%
L3 0.00 0.00 0.00 0.00
L1 ES 3.6 5.3 3.0 4.0
water) as qw10 . The solid lines were also in rough agreement
L2 ES 0.12 0.46 0.07 0.23 with the experimental results for VP-aided esteri'cation.
L3 ES (0.04) (0.03) – –
Water 7.4 0.9 6.6 0.8
Ethanol 6.8 3.8 9.1 9.5 4. Conclusion
a Reaction time was 22 h.
b Reaction time was 8 h.
The VP separation with zeolite T membrane was success-
fully applied to the esteri'cation of lactic acid with ethanol.
The conversion exceeded the equilibrium limit. It reached
Table 4 almost 100% within 8 h for m0 = 2:4 and 3.6 for the ester-
VP performance of a zeolite T membrane for a water-ethanol liquid i'cation. The VP-aided esteri'cation processes were well
mixture (10=90 wt%) at 393 K
simulated using a simple model.
Experiments Q (kg m−2 h−1 ) 
VP 0.34 400 References
VP-aided-esteri'cation
VP 0.37 1100
David, M. O., Gref, R., Nguyen, T. Q., & Neel, J. (1991). Pervaporation
VP-aided-esteri'cation
Esteri'cation coupling: Part I. Basic kinetic model. Transactions of
VP 0.31 1200
the Institution of Chemical Engineers., 69, 335–340.
VP-aided-esteri'cation
Filachione, E. M. (1952). Lactic acid. In R. E. Kirk & D. F. Othmer,
VP 0.37 3100
Encyclopedia of chemical technology, vol. 8 (pp. 167–180). New
VP-aided-esteri'cation
York: Interscience Publishers.
VP 0.32 2000
Itoh, N. (1987). A membrane reactor using Palladium. A.I.Ch.E. Journal,
The membrane was used in a series of the VP-aided-esteri'cation 33, 1576–1578.
experiments. Kita, H., Tanaka, K., Okamoto, K., & Yamamoto, M. (1987). The
esteri'cation of oleic acid with ethanol accompanied by membrane
separation. Chemistry Letters, (1987) 2053–2056.
amounts of organic components may be neglected because Lipnizki, F., Field, R. W., & Ten, P. K. (1999). Pervaporation-based
hybrid process: A review of process design, applications and
of higher . On these assumptions, the time course of
economics. Journal of Membrane Science, 153, 183–210.
VP-aided-esteri'cation was calculated from Eqs. (6) to (14) Okamoto, K., Yamamoto, M., Otoshi, Y., Semoto, T., Yano, M., Tanaka,
together with the following equation instead of Eq. (13). K., & Kita, H. (1993). Pervaporation-aided esteri'cation of oleic acids.

 Journal of Chemical Engineering of Japan, 26, 475–481.
dCH2 O k1
= k1 CL1 CEtOH − CL1 ES CH2 O + k2 CL2 CEtOH Okamoto, K., Yamamoto, M., Noda, S., Semoto, T., Otoshi, Y., Tanaka,
dt Ke1 K., & Kita, H. (1994). Vapor-permeation-aided esteri'cation of oleic

 acid. Industrial and Engineering Chemistry Research, 33, 849–853.
k2 Takagi, S., Kita, H., & Okamoto, K. (1999). Kagaku Kogaku Symposium
− CL2 ES CH2 O + k3 CL3 CEtOH
Ke2 Series, 66, 90.

 Tanaka, K., Yoshikawa, R., Cui, Y., Kita, H., & Okamoto, K. (2001).
k3 Application of zeolite membranes to esteri'cation reactions. Catalysis
− CL3 CEtOH − k4 CL2 CH2 O Today, 67, 121–125.
Ke3
Umemiya, S., Sato, N., Ando, H., & Kikuchi, E. (1991). The water gas


k4 shift reaction assisted by a Palladium membrane reactor. Industrial
+ CL21 − k5 CL3 CH2 O and Engineering Chemistry Research, 30, 585–589.
Ke4 Yamamoto, M., Munehisa, N., Kaibara, M., Horii, K., Tanaka, K., Kita,
H., & Okamoto, K. (1995). Vapor-permeation-aided esteri'cation of
qW 10 S(1000) CH 2 O oleic acid. Application of pressurized vapor circulation system and
− ; (15)
V (18)(60) 10 zeolite membrane. Maku (Membrane) , 20, 143–148.