Enzyme and Microbial Technology 27 (2000) 157–166 www.elsevier.

com/locate/enzmictec

Highly enantioselective esterification of racemic ibuprofen in a packed

bed reactor using immobilised Rhizomucor miehei lipase夞
Antoni Sánchez, Francisco Valero*, Javier Lafuente, Carles Solà
Departament d’Enginyeria Quı́mica, Escola Tècnica Superior d’Enginyeria, Universitat Autònoma de Barcelona, 08193, Bellaterra, Barcelona, Spain

Received 10 June 1999; received in revised form 8 February 2000; accepted 15 March 2000

Abstract

A systematic study of the enantioselective resolution of ibuprofen by commercial Rhizomucor miehei lipase (Lipozyme威 IM20) has been
carried out using isooctane as solvent and butanol as esterificating agent. The main variables controlling the process (temperature, ibuprofen
concentration, ratio butanol:ibuprofen) have been studied using an orthogonal full factorial experimental design, in which the selected
objective function was enantioselectivity. This strategy has resulted in a polynomial function that describes the process. By optimizing this
function, optimal conditions for carrying out the esterification of racemic ibuprofen have been determined. Under these conditions,
enantiomeric excess and total conversion values were 93.8% and 49.9%, respectively, and the enantioselectivity was 113 after 112 h of
reaction. These conditions have been considered in the design of a continuous reactor to scale up the process. The esterification of ibuprofen
was properly described by pseudo first-order kinetics. Thus, a packed bed reactor operating as a plug-flow reactor (PFR) is the most
appropriate in terms of minimizing the residence time compared with a continuous stirred tank reactor (CSTR) to achieve the same final
conversion. This reactor shows a similar behavior in terms of enantioselectivity, enantiomeric excess, and conversion when compared with
batch reactors. A residence-time distribution (RTD) shows that the flow model is essentially a plug flow with a slight nonsymmetrical axial
dispersion (Peclet number ⫽ 43), which was also corroborated by the model of CSTR in series. The stability of the system (up to 100 h)
and the possibility of reutilization of the enzyme (up to four times) lead to consider this reactor as a suitable configuration for scale up of
the process. © 2000 Elsevier Science Inc. All rights reserved.

Keywords: Ibuprofen; Lipozyme; Esterification; Experimental design; Enantiomeric excess; Kinetics; Packed bed reactor; Residence-time distribution

1. Introduction 2-aryl propionic acids (profens) make up an important

Lipases (triacylglycerol hydrolases, EC 3.1.1.3) are able
to catalyze esterification reactions in organic solvents that
are often more enantioselective than the corresponding hy-
drolytic reactions carried out in water [1]. Recently, these
enzymes have been used in the resolution of racemic mix-
tures for the preparation of optically pure compounds, con-
stituting an alternative method to perform chemical asym-
metric synthesis.
夞 This work has been supported by the European Program on Biotech-
nology (project BIO-4-96-0005), and the Spanish Program on Chemical
Processes Technologies (CICYT-QUI97-0506-C03). The Department of
Chemical Engineering (UAB) is member as unit of Biochemical Engineer-
ing of the Centre de Referència en Biotecnologia de Catalunya (Generalitat
de Catalunya).
* Corresponding author. Tel.: ⫹34-93-5811809; fax: ⫹34-93-
5812013.
E-mail address: valero@uab-eq.uab.es (F. Valero).
group of non-steroidal anti-inflammatory drugs (NSAIDs).
The pharmacological activity of these compounds is mainly
shown by their corresponding (S)-enantiomer [2]. For ex-
ample, it has been reported that (S)-enantiomer of ibuprofen
is 160-fold more active than its antipode in the in vitro
synthesis of prostaglandin [3]. Therefore, important efforts
are being made to synthesize pure enantiomers of 2-aryl
propionic acids, since pharmacological studies have indi-
cated that gastrointestinal problems are the most frequent
side effects associated with profens consumption [4].
Ibuprofen, 2-(4-isobutyilphenyl) propionic acid, is a ra-
cemic carboxylic acid each enantiomer of which shows a
physiologically different behavior, being the (S)-enantiomer
the form that exhibits an anti-inflammatory property [5]. In
the last years, several works dealing with the resolution of
ibuprofen have been published. Therefore, it was pointed
out that only Candida rugosa and Rhizomucor miehei
lipases were able to catalyze the esterification reaction, and

0141-0229/00/$ – see front matter © 2000 Elsevier Science Inc. All rights reserved.
PII: S 0 1 4 1 - 0 2 2 9 ( 0 0 ) 0 0 2 0 7 - 6
158 A. Sánchez et al. / Enzyme and Microbial Technology 27 (2000) 157–166

ties used to describe the solvents (log P, dielectric

Nomenclature constant and normalized electron acceptance index)
is a clear and predictive tool in the esterification of
CSTR Continuous stirred rank reactor ibuprofen [10].
PFR Plug-flow reactor 2. Water content is reported as one of the most im-
HPLC High-performance liquid chromatography portant factors affecting the enantioselectivity of
Yobj Objective function lipases [11]. By using C. rugosa lipase, it has been
Y Modeling function obtained from experimental shown that an increase of water activity increases
design the enzyme enantioselectivity [12]. The same be-
bi Coefficients of Y (B, coefficient vector) havior has been reported for C. antarctica lipase
xi Independent variables of Y [8].
M Matrix of experiments 3. Other operational parameters like temperature, con-
[S] (S)-ibuprofen concentration (mM) centration of reagents, enzyme concentration, reac-
[R] (R)-ibuprofen concentration (mM) tion time, stirring rate, acid and alcohol moiety, etc.
EE Enantiomeric excess (%) referred to ibuprofen have been also investigated. Usually, all these fac-
E Enantioselectivity tors are taken into account separately and some
X Conversion (%) interactions between factors may be omitted [5,9].
t Time (h or min) Thus, in some works the use of experimental design
r2 Coefficient of correlation techniques to study the influence of the variables on
r Rate of esterification (mM/h) the esterification reaction has been proposed [6,8].
k Apparent first order constant of esterification
rate (h⫺1 or min⫺1) Although there are numerous references about the use of
CA (S)-ibuprofen concentration (mM) immobilized lipases in bioreactors [13,14], there are few
CA0 Initial (S)-ibuprofen concentration (mM) publications dealing with the resolution of chiral com-
␶ Residence time (h or min) pounds in continuous systems working with organic sol-
V Reactor volume (ml) vents. For instance, several configurations of bioreactor
QL Flow rate (ml min⫺1) have been developed for hydrolysis reactions involving
␪ Dimensionless time chiral centers: a packed bed column with immobilized C.
E␪ Dimensionless concentration of tracer in pulse cylindracea lipase has been used for the resolution of
Mt Mass of tracer (g tracer) naproxen by hydrolysis [15], and there are a wide range of
C Tracer concentration (g ml⫺1) publications using hollow fiber reactors in hydrolysis reac-
N Number of stirred tanks. tions [16,17].
A major operational problem in a continuous reactor
working with organic solvents is usually the control of
water activity. To control water activity in organic media,
several methods have been used. Most of these methods are
the influence of different parameters on the esterification based on the use of saturated salt solutions, which give a
conversion, like temperature, solvent, water content, and defined water activity [18]. Other techniques for water re-
enzyme and substrate concentrations were studied [5]. moval are evaporation/pervaporation [19] and addition of
More recently, studies including resolution of ibuprofen molecular sieves [20,21]. However, these methods are of
and other 2-aryl propionic acids like naproxen or ketoprofen limited use in continuous systems. Recently, some systems
have been reported [6,7,8,9]. From these works, a set of have been proposed to have an accurate control of water
conclusions can be extracted: activity in continuous process involving packed bed reac-
1. Solvents, which are preferably highly hydrophobic, tors, including a new configuration with two reactors sepa-
have an important effect in the esterification con- rated by a water trap [22]. Other works have proposed
version. Solvent hydrophobicity is often correlated systems where the use of reduced pressure combined with
with the value of log P, logarithm of the partition molecular sieves is developed for the elimination of water
coefficient (P) of the solvent between 1-octanol and [23,24].
water. Generally, the catalytic activity is carried out Nevertheless, there is little information about resolution
in organic solvents with log P ⬎ 2. Among the of 2-aryl propionic acids in organic solvents using easily
organic solvents used, n-hexane, cyclohexane and scaleable systems. To date, only batch processes have been
isooctane are the most commonly used [8]. Re- implemented for the resolution of (R,S)-ibuprofen to pro-
cently, it has been concluded that although a higher duce gram quantities of enantio-enriched ibuprofen esters
enantioselectivity is obtained in hydrophilic sol- [25].
vents, a higher reaction rate is achieved in hydro- The aims of this work are: first, to optimize the condi-
phobic solvents. Nevertheless, none of the proper- tions of the enantioselective esterification of ibuprofen in a
 A. Sánchez et al. / Enzyme and Microbial Technology 27 (2000) 157–166
batch system using an orthogonal full factorial experimental
design. Second, to design a continuous and easily scaleable
reactor where the resolution may be carried out converting
gram quantities, and third, to study the stability and behav-
ior of both the reactor and the enzyme in this continuous
process.
2. Materials and methods
2.1. Materials
Lipozyme威 IM20 (lipase from R. miehei immobilized on
Duolite A568) was a generous gift from Novo–Nordisk
(Spain, Madrid). Butanol, isooctane and racemic ibuprofen
were obtained from Sigma Chemicals Co. (St. Louis, MO,
USA). Solvents and other chemicals used in this work were
of analytical grade.
2.2. Batch experiments
Batch experiments for the orthogonal experimental de-
sign were performed in 25 ml reactors. The reaction volume
was 10 ml. Isooctane (log P ⫽ 4.50) saturated with water
was used as solvent and butanol as esterificating agent.
Enzyme loading was 10 mg Lipozyme per millilitre of
reaction mixture. Samples were extracted from reaction
media and filtered through 0.45 ␮m before HPLC analysis.
Stirring was kept to 250 rev./min in an orbital stirrer with
temperature control. Other conditions were variable accord-
ing to the proposed experimental design. The nomenclature
used to label each experiment corresponds to the following
key:
Temperature (°C)/concentration of ibuprofen (mM)/ratio
butanol:ibuprofen.
2.3. Continuous packed bed reactor
An empty stainless steel HPLC column (Aminex威 HPX-
87H from Bio–Rad, Richmond, CA, USA) was used as a
tubular packed bed reactor. Dimensions of the reactor were
30 cm (length) ⫻ 7.8 mm (internal diameter), resulting a
total volume of 14.3 ml. 5.8 g of Lipozyme (density ⫽ 1.12
g/ml⫺1) were packed into the reactor, leaving a working
volume of 9.1 ml. The reactor was submerged into a water
bath set at the desired temperature.
Reaction mixture (isooctane, ibuprofen, and butanol)
was pumped into the reactor by means of a microburette
(model Crimson MicroBu 2013), which had its flow rate
programmed externally by a personal computer with a soft-
ware control developed in Borland威 Turbo C⫹⫹ version
3.0.
159
2.4. Numerical procedures
2.4.1. Statistical model
Optimization of esterification conditions was performed
by means of an orthogonal full factorial experimental design
[26 –28]. Briefly, orthogonal full factorial experimental de-
sign is based on the evaluation of the coefficients fitting a
polynomial function, which is proposed to describe the
system under study. This polynomial function (Y) is an
algebraic expression that combines the different factors that
have been taken into account. Coefficients vector (B) of the
function Y is calculated according to Eq. (1):
B ⫽ (MTM)⫺1 MTYobj (1)
Where M is the matrix of experiments that includes the
normalized levels of the considered factors, and Yobj is the
proposed objective function. A number of experiments are
replicated to validate statistically the consistence of the
proposed function representing the system.
Calculation of Eq. (1) was carried out by a self-made
software developed with Microsoft Fortran Powerstation威
4.0 (1994 –1995).
2.4.2. Optimization of objective function
Optimization of objective function was carried out by a
quasi-Newton method using IMSL威 libraries included in
Microsoft Fortran Powerstation威 4.0 (1994 –1995).
2.5. Analytical procedures
2.5.1. Analysis of ibuprofen
Chiral analysis of both enantiomers of ibuprofen were
conducted by chiral HPLC analysis using a Chiralcel OD
column (Mallinckrodt Chemical, Inc., Deventer, Holland)
with the mobile phase, hexane:2-propanol:trifluoroacetic
acid (98:2:0.1 v/v) at a flow of 1 ml min⫺1.
Enantiomeric excess (EE) referred to the form (R) of the
remaining ibuprofen and enantiomeric ratio (E) were calcu-
lated according to Eqs. (2) and (3)29
关R兴 ⫺ 关S兴
EE ⫽ 䡠 100 (2)
关S兴 ⫹ 关R兴
ln关共1 ⫺ X兲共1 ⫺ EE兲兴
E⫽ (3)
ln关共1 ⫺ X兲共1 ⫹ EE兲兴
2.5.2. Analysis of n-butyl propionate
Analysis of n-butyl propionate (tracer for the RTD) was
carried out by Gas Chromatography by using a Hewlett–
Packard 5890 Gas Chromatograph equipped with a HP
Ultra 1 (Crosslinked Polyethylene Glycol) column (30 m ⫻
0.25 mm ⫻ 0.25 ␮m film) under the following conditions:
Carrier gas: He, Injector Temp.: 250°C, Detector Temp.:
275°C, Oven Temp.: 40°C, Injection volume: 0.5 ␮l, Gas
Flow: 52 ml min⫺1, Split: 1:50.
160 A. Sánchez et al. / Enzyme and Microbial Technology 27 (2000) 157–166

Table 1 Table 2
Levels of the factors considered in the experimental design Batch experiments for each normalized level of the three considered
factors
Variable Left point Centre point Right point
(⫺) (0) (⫹) x1 x2 x3 EE (%) X (%) Yobj (E)

Temperature (x1) (°C) 35 40 45 ⫺ ⫺ ⫺ 88.1 70.1 6

Ibuprofen concentration (x2) (mM) 10 25 50 ⫺ ⫺ 0 96.2 69.1 9
Molar ratio butanol:ibuprofen (x3) 1:1 2:1 4:1 ⫺ ⫺ ⫹ 91.8 67.9 7
⫺ 0 ⫺ 93.9 59.7 15
⫺ 0 0 97.5 58.0 24
⫺ 0 ⫹ 94.1 56.9 21
2.5.3. Analysis of water content ⫺ ⫹ ⫺ 88.7 49.0 74
Water concentration (% w/w) in reaction media was ⫺ ⫹ 0 92.3 50.0 83
⫺ ⫹ ⫹ 93.6 51.3 59
determined by the Karl–Fisher method using a Karl–Fisher 0 ⫺ ⫺ 86.6 61.2 9
Titrator MKC-210 from Kyoto Electronics, equipped with a 0 ⫺ 0 89.9 61.2 11
Moisture Evaporator Model ADP-351. 0 ⫺ ⫹ 89.2 58.6 13
0 0 ⫺ 90.1 54.0 23
0 0 0 94.1 53.3 37
0 0 ⫹ 97.2 58.6 22
3. Results and discussion 0 ⫹ ⫺ 93.6 49.9 113
0 ⫹ 0 94.1 50.0 118
0 ⫹ ⫹ 93.3 50.3 85
3.1. Batch experiments: orthogonal experimental design ⫹ ⫺ ⫺ 92.3 67.7 8
⫹ ⫺ 0 91.3 62.2 11
The study of the factors affecting the enantioselective ⫹ ⫺ ⫹ 94.0 78.2 5
esterification of racemic ibuprofen, using butanol as alco- ⫹ 0 ⫺ 90.1 59.8 12
⫹ 0 0 94.1 59.0 16
hol, in batch process was carried out by means of an or-
⫹ 0 ⫹ 94.5 60.7 14
thogonal full experimental design [26,27,28]. Temperature, ⫹ ⫹ ⫺ 90.3 49.5 76
concentration of ibuprofen and ratio butanol:ibuprofen were ⫹ ⫹ 0 92.6 49.8 98
selected as the main experimental factors that control the ⫹ ⫹ ⫹ 89.3 50.2 50
enantioselectivity of the esterification reaction, therefore, x1, temperature (⫺ ⫽ 35°C; 0 ⫽ 40°C; ⫹ ⫽ 45°C)’ x2, concentration of
they were considered in the optimization of the process. ibuprofen (⫺ ⫽ 10 mM; 0 ⫽ 25 mM; ⫹ ⫽ 50 mM); x3, ratio butanol:
Other factors affecting the process such as stirring rate (250 ibuprofen (⫺ ⫽ 1:1; 0 ⫽ 1:2; ⫹ ⫽ 1:4).
rev./min in an orbital shaking flash), enzyme load (10 mg
ml⫺1) and reaction volume (10 ml) had been determined in of ibuprofen, different polynomial functions were tested.
previous studies [30]. Generally, the selected solvent for The simplest function (linear function) proposed is repre-
lipases in organic media has a log P, the hydrophobicity- sented by Eq. (4):
hydrophilicity ratio of organic solvents, higher than 2. It is

well known that the catalytic efficiency of enzymes de-
creases as the hydrophilicity of the solvent increases [31].
Thus, isooctane saturated with water (log P ⫽ 4.5) was
selected as solvent. Although the enantioselectivity with
1-octanol is higher in the esterification of ibuprofen using
cyclohexane as solvent, the highest ester conversion was
obtained with the shorter primary alcohol (1-butanol),
whereas the conversion using secondary (2-propanol) and
cyclic (cyclohexanol) alcohols was nil [8]. Thus, 1-butanol
was selected as the alcohol in the esterification reaction.
The levels of the factors considered in the experimental
design are presented in Table 1.
The election of these factors combined with the chosen
冘b x
3
Y ⫽ b0 ⫹ i i (4)
i⫽1
Different new terms were included in the linear function,
Eq. (4), to take into account possible interactions between
the considered factors. Coefficients (bi) and the accuracy of
the fitting represented by the coefficient of multiple corre-
lation (r2) were determined for each proposed function us-
ing Eq. (1). The values of r2 showed that the best modeling
function was a full second-order function taking into ac-
count quadratic terms and interactions for x1, x2, and x3 (Eq.
5), where the value of r2 was 0.89:

冘b x ⫹ 冘b x
levels implied the need for 27 batch experiments, which 3 3
were carried out following in each case both enantiomeric Y ⫽ b0 ⫹ i i i i
2
⫹ b 7x 1x 2 ⫹ b 8x 1x 3 ⫹ b 9x 2x 3
excess and total conversion with reaction time. Maximum i⫽1 i⫽1
enantioselectivity (E) was chosen as objective function;
⫹ b 10x 1x 2x 3 (5)
obviously this maximum value was obtained at different
reaction times for each experiment. The results obtained for The values of coefficients bi (corresponding to normal-
each experiment are presented in Table 2. ized values of x1, x2, and x3) are presented in Table 3.
To check the influence of the factors on the esterification This result was statistically validated by means of a
 A. Sánchez et al. / Enzyme and Microbial Technology 27 (2000) 157–166
Table 3
Values of bi coefficients for Eq. (5)
bi Coefficient of
b0 — 50.22
b1 x1
b2 x2
b3 x3
b4 x21
b5 x22
b6 x23
b7 x1 䡠 x2
b8 x1 䡠 x3
b9 x2 䡠 x3
b10 x1 䡠 x2 䡠 x3
F-test. Thus, a set of three experiments corresponding to the
conditions 40/10/2, 40/25/2, and 40/50/2 were randomly
chosen and carried out as replications to determine the
purely experimental uncertainty variance. Experimental
value of F-test was 44.8 higher than the tabulated value for
a confidence of 99%, which was 2.4. That meant that the
significance of factors effects was significant with a confi-
dence equal to or greater than 99%, which reflected a very
low variance due to lack of fit.
The high values of b2 (37.84) indicated that the most
important factor affecting the esterification of ibuprofen was
the concentration of ibuprofen, whereas temperature (b1 ⫽
⫺0.72) and ratio butanol:ibuprofen (b3 ⫽ ⫺3.75) had a
lower influence on the function Y. These results are in
accordance with the esterification of ibuprofen with butanol
in cyclohexane [8]. Again, the alcohol:acid ratio (x3) was
not significant, which suggested that there are no diffusional
restrictions of both the acid and alcohol. However, not only
linear factors had influence on the esterification. For in-
stance, quadratic terms of the three independent variables,
and interactions between ibuprofen concentration and ratio
butanol:ibuprofen (b9 ⫽ ⫺6.97) had a significant weight in
the final value of the function Y.
3.2. Optimization of the function Y
Function Y was used to determine the optimal conditions
for the esterification of ibuprofen. This function was numer-
ically maximized using a quasi-Newton method restricted
within the ranges of x1 (35– 45°C), x2 (10 –50 mM), x3
(1:1– 4:1) and the value of the optimal variables was:
Y Temperature (x1): 40°C.
Y Concentration of ibuprofen (x2): 50 mM.
Y Ratio butanol:ibuprofen (x3): 1.9.
Y Y value for these optimal values: 110.
The optimal ibuprofen concentration was 50 mM, which
was the maximum value tested. When experiments with
higher values of ibuprofen concentration were carried out,
161
Value
⫺0.72
37.84
⫺3.75
⫺18.22
18.80
⫺13.12
0.45
⫺1.99
⫺6.97
⫺1.57
Fig. 1. Optimal conditions obtained from orthogonal experimental design.
Conversion (X) and enantiomeric excess (EE) in batch reaction under the
optimal conditions predicted by the orthogonal experimental design.
the enantioselectivity maintained values over 100, but the
conversion did not reach values over 25% in any case. So,
these optimal conditions (40°C, 50 mM and 1.9) were cho-
sen for the esterification of ibuprofen using Lipozyme as
biocatalyst.
3.3. Characterization of the optimal conditions
One single experiment at the optimal conditions was
carried out. It is showed in Fig. 1. After 112 h of esterifi-
cation, enantiomeric excess, and total conversion values
were 93.8% and 49.9%, respectively, and the enantioselec-
tivity was 113, which was considerably higher than those
previously reported using Rhizomucor miehei lipase [8], in
which the best enantioselectivity value was 14.6. However,
the conditions of the reaction were different, cyclohexane
was used as solvent, the temperature was 37°C, concentra-
tion of both, acid and alcohol, were 125 mM, and the
amount of biocatalyst was at least 15⫻ higher than in this
experiment.
Nevertheless, the combination of experimental design
procedure and numerical optimization technique provided a
tool to improve the enantioselectivity of the reaction under
study.
These conditions were chosen for the design of a con-
tinuous system for the production of gram quantities of
highly enriched (S)-esters of ibuprofen.
3.4. Characterization of a pseudo first-order kinetics for
the optimal conditions
Although several approximations to the rigorous study of
lipase kinetics have been carried out [32–35], the most
common procedure is the use of simple pseudokinetics for
162 A. Sánchez et al. / Enzyme and Microbial Technology 27 (2000) 157–166
Fig. 2. Determination of pseudo first-order kinetics for the optimal condi-
tions. Linear fit for a pseudo first order kinetics in batch reaction under the
optimal conditions predicted by the orthogonal experimental design.
the description of lipase-catalyzed kinetics. Among pseudo-
kinetic models, first-order and Michaelis–Menten models
are the most commonly used [36,37]. Due to the high
enantioselectivity, it could be assumed that only (S)-ibupro-
fen was esterified by 1-butanol, whereas (R)-ibuprofen was
considered as an inert compound. Under this hypothesis,
different kinetics were tested, and the esterification of ibu-
profen under the optimal conditions was fitted to first-order
kinetics. The fitting of experimental data to first-order ki-
netics is presented in Fig. 2. Esterification kinetics could be
represented by Eq. (6):
r ⫽ kCA (6)
Where the value of apparent first-order constant (k) was
0.044 h⫺1.
The esterification of ibuprofen was properly described by
a first-order kinetics (r2 ⫽ 0.998). Thus, analyzing this
results it is evident that a plug-flow reactor is more appro-
priate in terms of minimizing the residence time (volume)
than a stirred tank reactor to achieve the same final conver-
sion in the esterification of ibuprofen in a continuous sys-
tem. Consequently, assuming that only (S)-ibuprofen syn-
thesis took place, the residence time of a plug-flow reactor
to achieve a final conversion of 95%, referred to (S)-enan-
tiomer, was 6⫻ lower than a stirred tank reactor.
3.5. Esterification of racemic ibuprofen in a packed bed
reactor
As it was described in Section 2, the tubular packed reactor
has a total working volume of 14.3 ml. It was completely
packed with 5.8 g of Lipozyme, resulting a working volume of
9.1 ml (average density of Lipozyme: 1.12 mg ml⫺1, manu-
facturer’s data). Considering that esterification rate apparent
constant (k) was proportional to the enzyme load, its value in
the tubular packed reactor would be 63.7⫻ higher than the
optimal batch experiment. Thus, the new value of the kinetic
constant would be 0.047 min⫺1. For example, assuming that
the objective is to reach a conversion of 92% of (S)-ester
(46% of total conversion), from a mass balance on the plug
flow reactor, the residence time would be 53.5 min, requer-
ing a flow rate of 0.17 ml min⫺1.
In Fig. 3, the behavior of the tubular packed reactor
working at different flow rates when steady state was
reached is presented. Interestingly, when the reactor worked
at the predicted flow rate, 0.17 ml min⫺1 (residence time ⫽
53.5 min), the results obtained were an enantiomeric excess
of 93.1% and a total conversion of 49.7%. These results are
only slightly higher than those obtained in a batch reactor
and demonstrate that the hypotheses considered in the de-
sign of the tubular packed reactor were reasonably correct.
As expected, when residence time decreased below this
value, both conversion and enantiomeric excess decreased.
However, when residence time increased, conversion ex-
ceeded 50%. Thus (R)-ibuprofen is being esterified and,
consequently, enantiomeric excess decreased. This behavior
had not been observed in batch reactions, in which (R)-
ibuprofen was not esterified (Fig. 1).
This fact provoked that an accurate control of the flow
rate of the reactor was required to ensure high enantiomeric
excess of the ibuprofen ester obtained.
Hence, a flow rate of 0.17 ml min⫺1 was chosen as the
flow rate that produced the best results in terms of enantio-
meric excess and enantioselectivity for the tubular packed
reactor. At this flow rate, enantioselectivity reached a value
of 112, which was also in accordance with batch results.
3.6. RTD of the packed bed reactor
Residence-time distribution (RTD) technique is a pow-
erful tool to study the behavior of continuous reactors and to
predict possible deviation from ideality [38]. RTD was
carried out in the tubular packed reactor by pumping into
the reactor a solution of N-butyl propionate, an inert com-
pound (tracer), in pulse (Mt ⫽ 0.1 g). This tracer was
selected because the physical properties are similar to those
of the reacting mixture and it is easy to analyze. Flow rate
selected was 0.17 ml min⫺1.
Concentration of tracer was followed with time and is
represented in Fig. 4 as dimensionless concentration [cal-
culated according to Eq. (8)] versus dimensionless time
[calculated according to Eq. (7)]:
t
␪⫽ (7)
␶
V
E␪ ⫽ 䡠C (8)
Mt
A consistence mass balance of the tracer was closed with
an error inferior to 5%.
 A. Sánchez et al. / Enzyme and Microbial Technology 27 (2000) 157–166
Fig. 3. Effect of residence time on the esterification of ibuprofen in the packed bed reactor. Conversion (X), enantiomeric excess (EE) and enantioselectivity
(E) in the packed bed reactor at steady state for different values of the residence time.
As shown in Fig. 4, flux model was essentially a plug
flow with a slight nonsymmetrical axial dispersion. The
dispersed plug flow model or simply the dispersion model
describes the plug flow of a fluid [38]. Peclet number can be
regarded as the ratio between the rate of transport by con-
vection and the rate of transport by diffusion or dispersion.
For this system Peclet number was 43, which corresponded
Fig. 4. Dimensionless RTD for the packed reactor. A pulse of tracer was
pumped into the reactor. Continuous line shows theoretical RTD for ideal
plug flow reactor.
163
to an intermediate amount of dispersion. Also, experimental
residence time was calculated from RTD, and a value of
60.1 min was obtained (theoretical residence time was 53.5
min). Thus, a slight deviation from the ideal plug flow
model was detected.
Series stirred tanks model is another possibility to de-
scribe the flow model in the packed reactor [38]. From mass
balance, the expression, which describes the evolution of
dimensionless tracer concentration when a pulse of tracer is
applied to the reactor, is given by Eq. (9):
N共N ␪ 兲 N⫺1
E␪ ⫽ exp共⫺N ␪ 兲 (9)
共N ⫺ 1兲!
Where N is the number of stirred tanks.
Plots of dimensionless concentration of tracer for differ-
ent values of N, as well as experimental data, are presented
in Fig. 5. The number of stirred tanks that fitted more
accurately the experimental data were 20.
Therefore, both models proposed for the study of RTD
showed that the tubular packed reactor could be essentially
considered as an ideal plug flow reactor with slight devia-
tions. This fact constitutes another validation of the pro-
posed system for the esterification of ibuprofen.
3.7. Stability of the reactor
The suitability of the designed tubular reactor was
checked in relation to its behavior over long time continu-
ous operation. The evolution of enantiomeric excess and
total conversion for the tubular packed reactor working up
to 100 h (close to 100 residence times) is presented in Fig. 6.
164 A. Sánchez et al. / Enzyme and Microbial Technology 27 (2000) 157–166
Fig. 5. Stirred tanks in series model. Representation for different values of
the number of stirred tanks (N). Thick continuous line corresponds to
experimental data.
As shown in Fig. 6, both parameters did not show ap-
preciable variation or deviation of their values. Thus, aver-
age enantiomeric excess was 93.9% (standard deviation ⫽
1.6%) and average conversion was 45.5% (standard devia-
tion ⫽ 0.7%) without considering the first three hours of
operation that corresponded to the start-up of the system.
Therefore, with the tubular packed reactor 6 g of opti-
cally enriched (S)-ibuprofen ester were obtained with an
easily scaleable system.
Fig. 6. Stability of the packed bed reactor. Conversion (X) and enantio-
meric excess (EE) with time in the packed bed reactor under the optimal
conditions of residence time.
Table 4
Evolution of enantiomeric excess (EE) and conversion (X) for different
utilizations of Lipozyme in the packed bed reactor
Utilization EE (%) X (%)
1 92.9 46.2
2 92.7 45.5
3 92.6 47.0
4 93.1 48.1
Average 92.8 46.7
SD 0.2 1.1
3.8. Packed bed reactor water balance and reutilization
of the catalyst
Distribution of the water generated in the packed reactor
was studied due to the influence that this parameter could
have on the conversion and enantiomeric excess [5,6,8,39,
40]. In the tubular reactor there was a continuous generation
of water, which could be estimated from the total average
conversion (45.5%), flow rate (0.17 ml min⫺1) and initial
ibuprofen concentration, at 4.2 mg water h⫺1. Average water
content at the input of the reactor was 0.044% (which corre-
sponded to the saturation level of water in isooctane at 40°C),
whereas at the output of the reactor an average value of
0.067% (standard deviation ⫽ 0.005%) was reached. This
corresponded to an elimination of water of 1.62 mg water h⫺1.
This meant that although there was a percentage of water
eliminated at the output (approximately 40%), an important
amount of water remained in the reactor. Measures of the
water content present in Lipozyme after long operation time
showed that water content increased with time (data not
showed). However, Duolite A568 (support on which lipase
from Rhizomucor miehei is immobilized in Lipozyme) is
delivered with a humidity of 8 to 10%, but it can rise to 64%
(manufacturer’s data), which implied that water produced in
the esterification mainly remained in the support. Assuming
this hypothesis, the humidity of the support should increase
from 10% to 14% after 100 h of operation.
Nevertheless, a set of experiments, in which Lipozyme
was removed from the reactor, washed with fresh solvent
and dried overnight at room temperature in a silica gel drier,
were carried out. The results are presented in Table 4. As
can be seen, Lipozyme could be reutilized up to four times
without appreciable loss of catalytic activity.
The stability of both operational system and biocatalyst
confirmed the suitability and applicability of the proposed
tubular packed reactor in the resolution of racemic ibupro-
fen by means of esterification reaction, converting gram
quantities of the desired product.
4. Conclusions
The results here presented constitute a new approach to
the resolution of chiral compounds in organic media. On the
one hand, experimental design has provided a powerful tool
 A. Sánchez et al. / Enzyme and Microbial Technology 27 (2000) 157–166
not only to study, but also to optimize the esterification con-
ditions that permit an important improvement of crucial as-
pects such as conversion or enantioselectivity in this process.
On the other hand, experimental design has permitted to
predict the performance of a continuous system designed for
the achievement of gram quantities of optically enriched
esters of ibuprofen. By using this continuous reactor, a
highly enantioselective resolution of ibuprofen (E ⫽ 113)
has been reached, which is considerably higher than the
results previously reported [6,8].
The stability of the reactor and the enzyme (up to 100 h),
as well as the low cost and simple scaling up of the system
make this process an alternative option to perform highly
stereoselective esterifications in organic media. Although it
is well described that water is adsorbed by the catalyst
resulting in lower reaction rates, and an excessive accumu-
lation of water in the reactor leads to a rapid irreversible
inactivation of the enzyme [39], in our system this problem
is not detected during the reaction time tested (up to 100 h).
Nevertheless, it had been shown that the activity of Li-
pozyme was not significantly altered by operations of reuti-
lization and drying, as well as by long operation times.
Moreover, a methodology composed of numerical tech-
niques, biochemical engineering and biocatalysis studies
has been developed to design systems that may be applica-
ble to pilot plant or industrial level. This system is presently
being tested in the highly enantioselective resolution of
other racemic compounds with pharmacological interest.
References
[1] Chen CS, Sih CJ. General aspects and optimization of enantioselec-
tive biocatalysis in organic solvent. Angew Chem Int Ed Engl 1989;
28:695–707.
[2] Hutt AJ, Caldwell J. The importance of stereochemistry in the clinical
pharmacokinetics of the 2-arylpropionic acid non-steroidal anti-in-
flammatory drugs. Clin Pharmacokin 1984;9:371–3.
[3] Adams SS, Bresloff P, Mason GC. Pharmacological difference be-
tween the optical isomers of ibuprofen: evidence for metabolic in-
version of the (⫺)-isomer. J Pharm Pharmacol 1976;28:156 –7.
[4] Shanbhag VR, Crider AM, Gokhale R, Harpalani A, Dick RM. Ester
and amide prodrugs of ibuprofen and naproxen: synthesis, anti-in-
flammatory activity and gastrointestinal toxicity. J Pharm Sci 1992;
81:149 –54.
[5] Mustranta A. Use of lipases in the resolution of racemic ibuprofen.
Appl Microbiol Biotechnol 1992;38:61– 6.
[6] Arroyo M, Sinisterra JV. High enantioselective esterification of
2-arylpropionic acids catalyzed by immobilized lipase from Candida
antarctica: a mechanistic approach. J Org Chem 1994;59:4410 –7.
[7] Kim MG, Lee SB. Enzymatic resolution of racemic ibuprofen by
lipase-catalyzed esterification reaction: effects of water content and
solid supports. J Ferment Bioeng 1996;81:269 –71.
[8] López–Belmonte MT, Alcántara AR, Sinisterra JV. Enantioselective
esterification of 2-aryl propionic acids catalyzed by immobilized
Rhizomucor miehei lipase. J Org Chem 1997;62:1831– 40.
[9] Tsai SW, Lin JJ, Chang CS, Chen JP. Enzymatic synthesis of (S)-
ibuprofen ester prodrug from racemic ibuprofen by lipase in organic
solvents. Biotechnol Prog 1997;13:82– 8.
165
[10] Ducret A, Trani M, Lortie R. Lipase-catalyzed enantioselective es-
terification of ibuprofen in organic solvents under controlled activity.
Enzyme Microb Technol 1998;22:212– 6.
[11] Tawaki SH, Klibanov AM. Chemoselectivity of enzymes in an-
hydrous media is strongly solvent dependent. Biocatalysis 1993;
8:3–19.
[12] Högberg HE, Edlund H, Berglund P, Hedeström E. Water activity
influences enantioselectivity in a lipase catalyzed resolution by ester-
ification in an organic solvent. Tetrahedron: Asymmetry 1993;4:
2123– 6.
[13] Balcao VM, Paiva AL, Malcata FX. Bioreactors with immobilized
lipases: state of the art. Enzyme Microb Technol 1996;18:392– 416.
[14] Rosell CM, Vaidya AM, Halling PJ. Continuous in situ water activity
control for organic phase biocatalysis in a packed bed hollow fiber
reactor. Biotechnol Bioeng 1996;49:284 –9.
[15] Battistel E, Bianchi D, Cesti P, Pina C. Enzymatic resolution of
(S)-(⫹)-Naproxen in a continuous reactor. Biotechnol Bioeng 1991;
38:659 – 64.
[16] Matsumae H, Furui M, Shibatani T. Production of optically active
3-phenylglycidic acid ester by the lipase from Serratia marcescens on
a hollow fiber membrane reactor. J Ferment Bioeng 1994;78:59 – 63.
[17] Balcao VM, Malcata FX. Interesterification and acidolysis of butter-
fat with oleic acid by Mucor javanicus lipase: changes in the pool of
fatty acid residues—influence on the free fatty acid profile. Enzyme
Microb Technol 1998;22:511–9.
[18] Goderis HL, Ampe G, Feyten MP, et al. Lipase-catalyzed ester
exchange reactions in organic media with controlled humidity. Bio-
technol Bioeng 1987;30:258 – 66.
[19] Kwon SJ, Song KM, Hong WH, Rhee JS. Removal of water produced
from lipase-catalyzed esterification in organic solvent by pervapora-
tion. Biotechnol Bioeng 1995;46:1–12.
[20] Vázquez–Lima F, Pyle DL, Asenjo JA. Factors affecting the esteri-
fication of lauric acid using an immobilized biocatalyst: enzyme
characterization and studies in a well-mixed reactor. Biotechnol Bio-
eng 1995;46:69 –79.
[21] Han JJ, Rhee JS. Effect of salt hydrate pairs for water activity control
on lipase-catalyzed synthesis of lysophospholipids in a solvent-free
system. Enzyme Microb Technol 1998;22:158 – 64.
[22] Wehtje E, Kaur J, Adelercreutz P, Chand S, Mattiason B. Water
activity control in enzymatic esterification processes. Enzyme Microb
Technol 1997;21:502–10.
[23] Ergan F, Trani M, Lortie R. Selective esterification of racemic ibu-
profen. Ann NY Acad Sci 1995;750:228 –31.
[24] Roure F, Ducret A, Trani M, Lortie R. Enantioselective esterification
of racemic ibuprofen under reduced pressure. J Chem Tech Biotech-
nol 1997;69:266 –70.
[25] Trani M, Ducret A, Pepin P, Lortie R. Scale-up of the enantioselec-
tive reaction for the enzymatic resolution of (R,S)-ibuprofen. Bio-
technol Lett 1995;17:1095– 8.
[26] Deming SN, Morgan SL. Experimental design: a chemometric ap-
proach, 1st Edition. Amsterdam: Elsevier, 1987.
[27] Montgomery DC. Design and analysis of experiments, 3rd Edition.
New York: John Wiley & Sons, 1991.
[28] Morgan E. Chemometrics. Experimental design, 1st Edition. New
York: John Wiley & Sons, 1991.
[29] Chen CS, Fujimoto Y, Girdaukas G, Sih CJ. Quantitative analyses of
biochemical kinetic resolutions of enantiomers. J Am Chem Soc
1982;104:7294 –9.
[30] Sánchez A, Valero F, Lafuente J, Solà C. Enantioselective esterifi-
cation of 2-arylpropionic acids and trans-2-phenyl-1-cyclohexanol:
comparison between immobilized lipases from Candida rugosa and
Rhizomucor miehei. Biotechnol Lett 1998;20:1145– 8.
[31] Carrea G, Ottolina G, Riva S. Role of solvents in the control of
enzyme selectivity in organic media. Trends Biotechnol 1995;13:
63–70.
[32] Botta M, Cernia E, Corelli F, Manetti F, Soro S. Probing the substrate
specificity for lipases. II: Kinetic and modeling studies on the mo-
166 A. Sánchez et al. / Enzyme and Microbial Technology 27 (2000) 157–166
lecular recognition of 2-arylpropionic esters by Candida rugosa and
Rhizomucor miehei lipases. Biochim Biophys Acta 1997;1337:
302–10.
[33] Parve O, Vallikivi I, Metsala A, et al. Lipase-catalyzed enantioselec-
tive hydrolysis: interpretation of the kinetic results in terms of frontier
orbital localization. Tetrahedron 1997;53:4889 –900.
[34] Chen J, Wang J. Wax ester synthesis by lipase-catalyzed esterification
with fungal cells immobilized on cellulose biomass support particles.
Enzyme Microb Technol 1997;20:615–22.
[35] Chulalaksananukul W, Condoret JS, Delorme P, Willemot RM. Ki-
netic study of esterification by immobilized lipase in hexane. FEBS
Lett 1990;276:181– 4.
[36] Gargouri Y, Chahinian H, Moreau H, Ransac S, Verger, R. Inactiva-
tion of pancreatic and gastric lipases by THL and C12:0-TNB: a
kinetic study with emulsified tributyrin. Biochim Biophys Acta 1991;
1085:322– 8.
[37] Arroyo R, Sánchez–Muñiz FJ, Cuesta C, Burguillo FJ, Sánchez–
Montero JM. Hydrolysis of used frying palm olein and sunflower oil
catalyzed by porcine pancreatic lipase. Lipids 1996;31:1133–9.
[38] Levenspiel O. Chemical Reactor Omnibook, 1st Edition. Corvallis:
OSU Book Stores Inc., 1979.
[39] Selmi B, Gontier E, Ergan F, Barbotin JN, Thomas D. Lipase-
catalyzed synthesis of tricaprylin in a medium solely composed of
substrates. Water production and elimination. Enzyme Microb Tech-
nol 1997;20:322–5.
[40] Mensah P, Gainer JL, Carta G. Adsorptive control of water in ester-
ification with immobilized enzymes: II. Fixed-bed reactor behavior.
Biotechnol Bioeng 1998;60:445–53.