Water Activity Control: A Way To
Improve the Efficiency of Continuous
Lipase Esterification
Sophie Colombié, Russell J. Tweddell, Jean-Stéphane Condoret,
Alain Marty
Laboratoire de Biotechnologie-Bioprocédés, INSA, UMR 5504, L.A. INRA,
Complexe Scientifique de Rangueil, 31077 Toulouse Cedex 04, France;
telephone: +33 5 61 55 94 39; fax: +33 5 61 55 94 03; e-mail:
alain.marty@insa-tlse.fr
Received 26 September 1997; accepted 10 April 1998
Abstract: During continuous lipase-catalyzed oleic acid
esterification by ethanol in n-hexane, the oleic acid con-
version, initially at 95%, decreases to 20% after 2 h. This
decrease is caused by the accumulation of the water pro-
duced in the course of the reaction in the packed-bed
reactor (PBR). In order to improve the PBR efficiency, it is
necessary to evacuate the water produced. In this study,
different approaches have been tested to control the wa-
ter content in the PBR during continuous esterification.
The first approach consisted in improving the water solu-
bility by increasing the reaction medium polarity. The
addition of polar additives to n-hexane, the use of more
polar solvents, and the use of solvent-free reaction me-
dium were tested as a means to favor the water evacu-
ation from the PBR. First of all, the use of n-hexane
supplemented with acetone (3 M ) or 2-methyl-2-
propanol (1 M) enabled the conversion to be maintained
at higher values than those obtained in pure n-hexane.
The replacement of n-hexane by a more polar solvent,
like the 5-methyl-2-hexanone, resulted in the same effect.
In all cases, conversions at steady-state were always less
than 95%, as obtained in pure n-hexane. This is ex-
plained by a decrease in the enzyme activity due to the
increase in the medium polarity. Nevertheless, an in-
crease in enzyme quantity allowed 90% conversion to be
maintained during 1 week using 3 M acetone amended
n-hexane. Good results (a steady-state conversion of
about 80%) were obtained when esterification was car-
ried out in a solvent-free reaction medium containing 2
M 2-methyl-2-propanol as a polar additive. The second
approach consisted in the evaporation of the accumu-
lated water by use of an intermittent airflow. Although
this process did not enable constant esterification rate to
be maintained, it did enable the initial conversion (95%)
to be restored intermittently. © 1998 John Wiley & Sons, Inc.
Biotechnol Bioeng 60: 362–368, 1998.
Keywords: esterification; continuous reaction; water ac-
tivity; lipase; organic solvent; packed-bed reactor
INTRODUCTION
The advantages of using organic solvents or low water ac-
tivity media to operate enzymatic reactions have been fre-
quently pointed out (Zaks and Klibanov, 1985, 1988; Zaks
Correspondence to: A. Marty
© 1998 John Wiley & Sons, Inc.
and Russell, 1988; Dordick, 1989; Laane and Tramper,
1990). Compared to aqueous media, organic media favor
the stability of the enzymes (Zaks and Klibanov, 1984; Chu-
lalaksananukul et al., 1990), increased solubility of hydro-
phobic compounds, and easy recovery of enzymes as they
are generally insoluble in these media. Moreover, in the
case of hydrolytic enzymes, the reaction may be reversed
because thermodynamic equilibrium is shifted due to the
low thermodynamic water activity and due also to changes
in substrate and product activities (Halling, 1994). For in-
stance, the use of an hydrophobic solvent allowed a low
ester activity to maintained during an esterification reaction.
Consequently it is possible to carry out, in organic media,
synthesis reactions catalyzed by hydrolytic enzymes.
The use of lipases (triacylglycerol acylhydrolase, EC
3.1.1.3) as catalysts in organic media has been frequently
described (Yokoseki et al., 1982; Zaks and Klibanov, 1985,
1988; Laane et al., 1986; Brink et al., 1988; Mustranta et al.,
1993). Lipases are enzymes that primarily catalyze hydro-
lysis of acylglycerides. This reaction is reversible, and the
potential of these enzymes for synthesis of esters in organic
solvents has been well documented (Chulalaksananukul et
al., 1993; Perraud and Laboret, 1995; Zaks and Russell,
1988)
Lipase-catalyzed esterification in organic solvents is a
reaction where water plays a crucial role. A minimal amount
of water is necessary for the enzyme to ensure its optimal
conformation and then to become optimally active (Zaks
and Klibanov, 1985; Monot et al., 1991; Valivety et al.,
1993). But, an excess of water decreases the enzyme cata-
lytic activity both from kinetic and thermodynamic points of
view (Chulalaksananukul et al., 1990; Marty et al., 1992).
The optimal level of enzyme hydration for optimal activity
appears to be enzyme dependent (Zaks and Klibanov, 1988;
Dordick, 1989). For instance it was demonstrated that for an
immobilized lipase, the lipozyme, hydration values of the
enzymatic support of 8–10/100 (g/g of dry enzymatic sup-
port, which correspond to a thermodynamic water activity
of 0.5) are optimal to catalyze the esterification of oleic acid
by ethanol in n-hexane and in supercritical carbon dioxide
CCC 0006-3592/98/030362-07
(SCCO2) (Chulalaksananukul et al., 1990, 1993; Marty et
al., 1992; Halling, 1994). Moreover, since water is a product
of the esterification reaction, the amount of water in the
organic solvent, which varies as the reaction progresses,
affects not only the enzyme activity but also the thermody-
namics of the equilibrated reaction (Halling, 1994).
Immobilized biocatalyst packed-bed reactors (PBR) have
traditionally been used for most large-scale catalytic reac-
tions because of their efficiency, low cost, and ease of con-
struction and operation. They were notably used to develop
new industrial applications for continuous ester production
by reversion of hydrolysis (Balcao et al., 1996). In our
recent studies, which dealt with the continuous lipase-
catalyzed esterification of oleic acid by ethanol in n-hexane
using a PBR, we noted the drop in the esterification rate
versus time that appeared during operation of continuous
packed bed of immobilized enzymes (Marty et al., 1997).
The oleic acid conversion, initially at 95%, decreases dras-
tically within 2 h to 20%. Such variation is obviously a great
handicap to the efficient use of such continuous reactions. In
this paper, we suggested that this phenomenon was caused
by the fact that the water, produced in the course of the
reaction, accumulates in the PBR, due to its adsorption on
the support. Indeed, the produced water is partitioned be-
tween the enzymatic support and the solvent according to
specific adsorption isotherms (Condoret et al., 1997; Marty
et al., 1997). The high hydrophobicity of the solvent leads to
an excessive accumulation of water in the vicinity of the
enzyme, which is unfavorable to the reaction. This decrease
in the conversion is assumed to be due to the negative effect
of water activity on both the kinetics of the reaction and the
thermodynamic equilibrium. After this initial drop in the
conversion, a steady value is reached. It is obtained at a
conversion where the quantity of the water produced may be
totally evacuated in the solvent outflow, at a concentration
given by the partition of water between the solvent and the
solid support.
This work showed that the control of the water content,
and consequently of the water activity, in the PBR enables
a high and constant esterification rate to be maintained.
Some authors have already proposed controlling water ac-
tivity in different ways, such as pervaporation (Keurentjes et
al., 1994; Kwon et al., 1995), use of saturated salt solutions
(Valivety et al., 1993; Wehtje et al., 1993; Rosell et al.,
1996), use of solid salt pairs (Halling, 1992), and use of
partial vacuum (Napier et al., 1996).
The objective of this work is therefore to propose other
solutions for controlling the water activity in the PBR, and
in this way, to maintain a high and constant conversion
during esterification of oleic acid with ethanol. Different
ways to evacuate the water produced were tested: (i) addi-
tion to the hydrophobic solvent, n-hexane, of small fractions
of a more polar solvent in order to increase the solubility of
water in the mixture, (ii) use of more polar solvents, (iii) use
of a solvent free reaction medium, taking advantage of the
hydrophilicity of one of the substrates, alcohol, and (iv)
‘‘drying’’ of PBR by intermittent injection of air.
MATERIALS AND METHODS
Materials
For all the experiments, the catalyst was lipozyme (200
U/g). Lipozyme, a lipase (EC 3.1.1.3) from Mucor miehei,
immobilized on macroporous anionic resin beads (Duolite
A568) was kindly provided by Novo Industry (Bagsveard,
Denmark).
Oleic acid (cis-9-octadecanoic acid) of low purity (70%)
was used for economic reasons. The validity of this use has
been previously established (Marty et al., 1994). Oleic acid,
absolute ethanol, and n-hexane were supplied by Prolabo
(France). Acetone and 5-methyl-2-hexanone were pur-
chased from Carlo Erba Reagent (Italy). 2-Methyl-2-
propanol and 2-methyl-2-butanol were supplied by Aldrich.
Experimental Apparatus
The continuous reaction vessel was a tubular packed-bed
reactor, which consisted of a thermostated column (diam-
eter: 9 mm) containing the immobilized enzyme. The reac-
tion mixture was pumped using a peristaltic pump (Gilson
Medical Electronics, France) and percolated the packed bed
of enzyme. Samples were collected manually at the reaction
vessel exit.
Esterification Conditions
In organic solvents (pure n-hexane, 5-methyl-2-hexanone,
or polar additive amended n-hexane), continuous oleic acid
esterification by ethanol was performed with 400 mg of
lipozyme. The reaction medium flow rate was 1 mL/min.
The reaction temperature was 40°C, and substrate concen-
trations were 200 and 300 mM for oleic acid and ethanol,
respectively. In solvent-free medium and in solvent-free
medium supplemented with 2-methyl-2-propanol, the actual
substrate concentrations being different (but in the same
ratio, 2:3, as previously used), operating conditions were
calculated to obtain a conversion of 95% to ensure that the
enzyme was in contact with the same molar quantity of
substrates per unit of time as previously.
Batch reactions were run in 10-mL glass tubes; the solu-
tions were mixed by magnetic stirring, and 30 mg of en-
zyme was used.
Analytical Methods
Oleic acid and ethyl oleate concentrations were determined
by an HPLC system equipped with a Kontron 420 pump, a
Varian R14 refractive index monitor (Varian Associates,
Orsay, France) and a Nucleosil C18, 5-␮m column (250 ×
4.0 mm) (Touzart et Matignon, Vitry sur Seine, France).
Elution was conducted at 50°C with methanol/acetic acid
COLOMBIÉ ET AL.: WATER ACTIVITY CONTROL 363
(99.7:0.3, v/v) at a flow rate of 1 mL/min. The conversion
was calculated as follows:
100.关Ethyloleate兴 t
X= , view, the industrial scale-up must be achieved with a reac-
关Oleic acid兴 t + 关Ethyloleate兴 t
where [ ]t are concentrations at time t.
The water content in the reaction media was systemati-
cally determined by Karl Fischer titration. The Karl Fischer
device consists of a 703KF Trinitro buret automatically con-
trolled by a titrator (Metrohm, Germany). The titrant (Hy-
dranal, composite 2 or 5) was a pyridine free solution (Rie-
del de Haën, Germany). The solvent used was methanol
except when acetone or 5-methyl-2-hexanone was present
in the sample. In these cases, the solvent was Hydranal
auxiliaire K, and the titrant was Hydranal composite 5K,
because ketones lead to a secondary reaction which pro-
duces water.
RESULTS AND DISCUSSION
Whatever the system tested for the esterification of oleic
acid by ethanol in the PBR, conditions such as enzyme
quantity, enzyme hydration level and flow rate were ad-
justed to convert 95% of oleic acid into ethyl oleate. This
was done using mathematical modeling based on the reac-
tion mechanism, on values of kinetic parameters and on the
assumption of plug flow reactor, as described in our previ-
ous works (Chulalaksananukul et al., 1990; Marty et al.,
1994). It was noticed that a very good correlation was ob-
tained between the predicted conversion and initial experi-
mental conversion when pure n-hexane is used, whatever
the conditions (Fig. 1 and data not shown).
As stated in the Introduction, we will now consider the
different solutions for controlling water activity in the PBR.
Addition of a Polar Cosolvent ton-Hexane To
Increase the Medium Polarity
As a first step, we carried out almost the same experiment
as presented in our previous work (Marty et al., 1997),
Figure 1. Oleic acid conversion (%) (closed symbols) and water content
at the outflow of the reactor (g/L) (open symbols) versus time, in n-hexane,
pure (䊉, 䊊) or amended with different acetone concentrations: 1 M (⽧);
2 M (䉲); 3 M (䊏, 䊐); 4 M (䉱). Oleic acid (200 mM) esterification by
ethanol (300 mM) performed in a continuous packed-bed reaction vessel
(flow rate: 1 mL/min) containing 400 mg of lipozyme.
364 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 60, NO. 3, NOVEMBER 5, 1998
concerning the lipase-catalyzed esterification between oleic
acid and ethanol in n-hexane. We used a higher concentra-
tion of oleic acid, because, from an economical point of
tion solution as concentrated as possible. This experiment,
performed in a very conventional organic solvent, n-hexane,
will be our reference (Fig. 1, curve with black circles).
n-Hexane appeared to be a poor solvent for the evacuation
of the undesired water produced (Fig. 1, filled black circles).
It appears that the conversion decreases until n-hexane is
able to evacuate the water produced. In order to improve
this evacuation, one possible method could be an increase in
the solvent polarity by using a mixture of n-hexane and
another more polar solvent. Acetone, 2-methyl-2-propanol
and 2-methyl-2-butanol were tested. They were chosen not
only because of their hydrophilic behavior but also because
of their inert behavior with respect to the reaction (tertiary
alcohols and ketones are not substrates for this lipase).
Firstly, we will consider the addition of acetone. Optimal
acetone concentration to ensure complete evacuation of the
water produced was determined using the following calcu-
lations based on mass balance and water partition equilib-
rium. The water mass balance equation at the apparent
steady state, in a continuous PBR, is:
q[H2O]solvent, inflow + qX[Oleic acid]to ⳱ q[H2O]solvent, outflow
(1)
where [H2O]solvent, inflow is the water concentration at the
reactor entry (mM); [H2O]solvent, outflow is the water concen-
tration at the reactor exit (mM), assumed to be in equilib-
rium with the solid; [Oleic acid]to is the oleic acid concen-
tration in the substrate solution (mM); q is the total flow rate
(mL min−1); X is the oleic acid conversion (%); and qX-
[Oleic acid]to represents the molar flow rate of the water
produced (mmol min−1).
When the water solubility in the reaction medium is low,
the final conversion is limited by the water solubility, and
consequently it can be assumed that the steady state outflow
water concentration is approximately equal to the water
solubility in the medium. That corresponds to the final flat
part of an adsorption curve (water dissolved in the solvent
versus water adsorbed on the solid) (Marty et al., 1997).
Consequently, Eq. (1) can be used to estimate the water
solubility necessary to maintain a given conversion. For
instance, to obtain a conversion of 95%, the water concen-
tration in the outflow has to be 3.5 g/L. Published data
(Ragaini et al., 1974) about water solubility in a ternary
system composed of n-hexane, acetone, and water at 40°C
predict that an approximately 3 M concentration of acetone
in n-hexane allows 3.5 g of water/L to be solubilized.
To check the validity of this approach, we have tested the
effect of 1, 2, 3, and 4 M acetone reaction mixtures. Oleic
acid conversions as a function of time using reaction me-
dium inflow containing either pure n-hexane or n-hexane
amended with different acetone concentrations were deter-
mined. As shown in Fig. 1, when no acetone is added to the
system (previously mentioned curve with black circles), the
conversion decreases and becomes constant at 20% after
120 min. In acetone-amended n-hexane, whatever the ac-
etone concentration, the steady state conversion is higher
(37, 48, 62, and 37% for 1, 2, 3, and 4 M acetone, respec-
tively) than in pure n-hexane. As predicted in our compu-
tations, a 3 M acetone concentration appears optimal. It
must be reminded that it corresponds to a given value of the
initial oleic acid concentration. Acetone concentrations of 1
and 2 M seem too low since a drop in the conversion is
observed before a steady state is reached. At acetone con-
centrations of 3 and 4 M, there is no delay in reaching the
apparent steady state, indicating that acetone allowed all the
water produced to be evacuated. Indeed, the higher the ac-
etone concentration, the higher the concentration of water at
the reactor outflow and the lower the water content on the
enzymatic support (Table I). At 3 M acetone, the water
adsorbed on the enzymatic support at the end of the experi-
ment is optimal. Surprisingly we observe that the conver-
sion obtained in the medium supplemented with 4 M ac-
etone is lower than with 3 M, this latter being lower than the
95% expected. At 3 M concentration of acetone, the solvent
evacuates water at a concentration of 2.5 g/L (Table I),
which allows the previously presented mass balance to be
verified (Table I: at steady state, experimental conversion,
62%; calculated conversion, 64%). This value (2.5 g/L) is
lower than the predicted solubility of water in this medium
(3.5 g/L) because the conversion (62%) is lower than the
one expected (95%). n-Hexane supplemented with 3 M ac-
etone would be able to evacuate a higher quantity of water
than in this particular case. The fact that the initial value is
lower than 95% indicates a new effect not related to water
accumulation (the initial value cannot be affected by water
accumulation). One might have been reminded of the limi-
tation by the thermodynamic equilibrium resulting from the
influence of acetone on the thermodynamic water activity
or, more likely, on the thermodynamic activities of other
compounds (the thermodynamic water activity is lower in
the presence of acetone, which is favorable for the synthe-
Table I. Influence of acetone concentration in n-hexane.
Acetone concentration in n-hexane (M)
Parameters 0 1 2 2.5 3 4
Initial conversiona (%) 95 84 68 64 62 38
Experimental apparent steady
state conversion (%) 20 37 48 56 62 37
Theoretical apparent steady
state conversion (%) from
Eq. (1) 14 31 44 50 64 NDb
Water concentration in the
solvent outflow (g/L) at
the end of the experiment 0.6 1.3 1.8 2 2.5 2.1
Water content adsorbed on
the enzymatic support (g/g)
at the end of the experiment 0.5 0.33 0.16 0.12 0.1 0.06
a
For all acetone concentrations tested, operating conditions were fixed to
have a 95% conversion.
b
ND, not determined.
sis). This hypothesis was discarded after determining that a
decrease in the flow rate or an increase in the amount of
enzyme in the PBR allows the conversion of oleic acid to
increase at both concentrations, 3 and 4 M. For instance, in
the case of n-hexane amended with 3 M acetone, the de-
crease in the flow rate, from 1 to 0.32 mL min–1 (operating
conditions being the same), leads to an increase in the con-
version from 60 to 90% (data not shown). Similarly, the use
of 1 g of enzyme instead of 400 mg results in a 85% con-
version. In addition, it was observed that after 3 h of op-
eration with 3 M acetone amended hexane, the use of pure
n-hexane enables the reactor to recover a 95% conversion.
All these observations correspond, therefore, to a reversible
decrease in the enzyme activity due to the presence of ac-
etone. The most probable hypothesis is a modification of the
kinetic parameters of the enzyme. In support of this hypoth-
esis, additional batch experiments (data not shown), point
out that, in a 3 M amended n-hexane medium, the initial
velocity is 3 times lower than in pure n-hexane, and that no
significant change in the thermodynamic equilibrium posi-
tion is detected.
The stability of the 3 M acetone-amended n-hexane re-
actor was tested using 1 g of immobilized enzyme and the
same operating conditions (200 mM oleic acid; 300 mM
ethanol: flow rate 1 mL min−1). A stable conversion of 85%
is obtained for a 1-week period (data not shown).
In order to investigate the influence of the reaction me-
dium, another type of polar cosolvent, a tertiary alcohol,
was tested. 2-Methyl-2-propanol and a slightly less polar
solvent, 2-methyl-2-butanol, were selected. Due to the lack
of water solubility data for these mixtures, it was not pos-
sible to predict the tertiary alcohol concentration necessary
to ensure the water evacuation, so arbitrary concentrations
were tested. The same phenomenon was observed: in all
cases the steady state was obtained without delay which
indicates that the water produced is totally evacuated, but its
value being lower than the 95% expected (Fig. 2). As in the
case of acetone-amended n-hexane, it appears that tertiary
Figure 2. Oleic acid conversion (%) (closed symbols) and water content
at the outflow of the reactor (g/L) (open symbols) versus time, in n-hexane
amended with different tertiary alcohols: 2-methyl-2-butanol, 1 M (䉱) and
2 M (䉲); or 1 M 2-methyl-2-propanol (䊏, 䊐). Oleic acid (200 mM)
esterification by ethanol (300 mM) performed in a continuous packed-bed
reaction vessel (flow rate: 1 mL/min) containing 400 mg of lipozyme.
COLOMBIÉ ET AL.: WATER ACTIVITY CONTROL 365
alcohols affect enzyme activity. This confirms that an in-
crease in solvent polarity unfavorably modifies kinetic pa-
rameters together with an improvement of the ability to
evacuate the produced water.
The use of 1 M 2-methyl-2-propanol amended n-hexane
allows the conversion to be maintained at a value of 70%,
which is slightly better than the best result obtained with the
addition of acetone.
Use of a More Polar Solvent than n-Hexane
As seen above, one way to increase the polarity of the
reaction medium is to add polar cosolvent to n-hexane but
another way could be to replace n-hexane by a more polar
solvent. 5-Methyl-2-hexanone was then tested as reaction
medium for oleic acid esterification. Before use, this solvent
was hydrated in order to adjust the water activity to the
optimal value of 0.5 at the outflow of the reaction vessel.
The adjusted water concentration at the input of the reactor
was calculated as follows: the water adsorption isotherm,
determined by the method developed by Condoret et al.
(1997), allows the water content of the solvent in equilib-
rium with the optimal 8% g/g of water adsorbed on the
support to be determined. A value of 8 g/L was estimated.
For 95% conversion, the quantity of water produced may be
computed, and the mass balance leads to a value of 4 g/L in
the inflow of the reactor. In this way, throughout the reactor,
a gradient of water content of the enzymatic support from 5
to 10% will be theoretically obtained, which is a good com-
promise for enzyme activity.
In this case, no decrease in the conversion versus time is
observed (data not shown). Nevertheless, the value of the
apparent steady state conversion (40%) shows that 5-
methyl-2-hexanone appears to affect the enzymatic activity,
as already observed for n-hexane amended with 4 M ac-
etone. Also in this case, it can be assumed that kinetic
constants are negatively influenced by the solvent change.
This fact explains also why the outflow water concentration
is only 6 g/L. It was verified at the end of the experiment
that water hydration of the enzymatic support was almost
optimal (6% g/g of support). It appears that this solvent
leads to better results than those observed in n-hexane (40%
against 20%), but it is not as efficient as n-hexane supple-
mented with 3 M acetone.
In one of our previous studies (Marty et al., 1994), an-
other solvent, supercritical carbon dioxide (SCCO2), was
tested. It is slightly more polar than n-hexane (water solu-
bility about 1.3 g/L at 40°C and 150 bar against 0.15 g/L in
n-hexane at 40°C). Operating conditions were fixed in order
to get the oleic acid conversion at a value of 95%. The
experimental conversion was not dependent of the time, and
its value is equal to the predicted value of 95%. At the end
of the experiment (300 min), it was checked by Karl Fischer
titration that the catalyst optimal hydration had been main-
tained, leading to the conclusion that this solvent had the
capacity to evacuate the water produced under these condi-
tions of temperature and pressure (40°C, 200 bar), without
366 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 60, NO. 3, NOVEMBER 5, 1998
affecting enzyme activity. Although enzyme activity was
shown to be lower in SCCO2 than in n-hexane (Marty et al.,
1992), SCCO2 allows the necessary water evacuation. Its
natural character and its capacity to integrate reaction and
fractionation being greatly appreciated (Marty et al., 1994;
Aaltonen and Rantakylä, 1991) confer numerous advan-
tages to this solvent.
Solvent-Free Reaction Medium
Due to the polarity of ethanol, which is one of the two
reactants, one could think about conducting the reaction in
the absence of solvent. Coulon et al. (1995) and Napier et al.
(1996) have already performed lipase-catalyzed esterifica-
tion in solvent-free mixture. In our case, in order to assess
the ability of ethanol to solubilize the water produced, the
esterification was performed in a medium exclusively com-
posed of ethanol and oleic acid (in the same ratio as the one
previously used). Substrate concentrations being different,
to ensure that the enzyme is in contact with the same molar
quantity of substrates per unit of time, the flow rate was
calculated and then fixed to obtain a conversion of 95%
using the values of kinetic constants determined in pure
n-hexane. This value was approximately verified as shown
on Fig. 3 (black circles), but, as in n-hexane, a decay in the
conversion precedes a final steady state. Nevertheless the
steady state conversion value (50%) is higher than the one
observed in n-hexane (20%). This value is in agreement
with the mass balance prediction that gives 45% conversion
calculated with Eq. (1). As expected, ethanol, in addition to
its substrate role, allows the evacuation of a significant
quantity of water (22 g/L) (Fig. 3). At the end of the ex-
periment, the water adsorbed on the enzymatic support was
about 11/100 (g/g of dry enzymatic support).
As a way to point out and to improve the efficiency of
this process, different quantities of 2-methyl-2-propanol
Figure 3. Oleic acid conversion (%) (closed symbols) and water content
at the outflow of the reactor (g/L) (open symbols) versus time, in solvent-
free reaction medium (䊉, 䊊) and in solvent-free reaction medium amended
with 2-methyl-2-propanol: 1 M (䉱); 2 M (䊏, 䊐). Oleic acid (2.5, 2.2, and
2.1 M, respectively) esterification by ethanol (3.7, 3.3, and 3.1 M, respec-
tively) performed in a continuous packed-bed reaction vessel (flow rate:
0.16, 0.17, and 0.19 mL/min ) containing 800 mg of lipozyme.
were added to the reaction medium. It appears that addition
of 2 M 2-methyl-2-propanol allows the steady state to be
obtained almost immediately and increases significantly the
efficiency of the system (Fig. 3). In medium containing 2 M
2-methyl-2-propanol, 30 g/L of water is evacuated com-
pared to 20 g/L for the original solvent-free medium.
Among all the systems tested, these amended solvent-free
reaction media are the most productive and present the ad-
vantages of being less toxic, less dangerous, and cheaper.
Air Drying of the Packed-Bed Enzyme
The previous experiments have demonstrated that it is pos-
sible to continuously evacuate the water produced by in-
creasing the polarity of the reaction medium. This can be
done during the reaction, as done by Napier et al. (1996),
but in their case, water was the only volatile component. In
our case which involves ethanol, this would deplete one of
the substrates. Therefore, after a 2-h period of reaction, air
stripping of the water produced is proposed. The air pres-
sure will be low even at a large scale because the pressure
drop is low in this kind of reactor. As shown on Fig. 4, oleic
acid conversion drops from almost 100% to 20% in 2 h, at
the same time water concentration at the reactor out flow
increases. In our system, at 2-h intervals, air was system-
atically injected through the packed-bed enzyme for 75 min,
after the input flow had been stopped. This intermittent
removal of water by evaporation enabled the initial conver-
sion to be restored during 4 cycles of reaction. Although
these results are preliminary and not optimized, this process
seems promising because air does not affect enzyme activity
and involves low cost and no toxicity. Considering the pos-
sible value of this process, future studies will be undertaken
in order to (i) optimize parameters such as air drying dura-
tion, flow rate, and frequency and (ii) assess the feasibility
of using continuous percolation airflow to avoid water ac-
cumulation on the enzymatic support when no volatile sub-
strate is present.
Figure 4. Effect of intermittent air drying (75 min every 2 h) on oleic
acid conversion (䊉) and water content at the outflow of the reactor (g/L)
(䊊), versus time during oleic acid (200 mM) esterification by ethanol (300
mM), in n-hexane, in a continuous packed-bed reaction vessel (flow rate:
1 mL/min) containing 400 mg of lipozyme.
CONCLUSION
During continuous lipase-catalyzed oleic acid esterification
by ethanol in n-hexane, the oleic acid conversion decreases
as the reaction progresses and reaches a steady state. This
decrease is caused by an accumulation of water in the PBR.
In order to improve the PBR efficiency, it is necessary to
evacuate the water produced during the esterification. In this
study, two approaches have been tested to control the water
content in the PBR during continuous esterification. The
first approach consisted of improving the water solubility in
the reaction medium by increasing its polarity. The latter
must be optimized, taking into account its negative influ-
ence on the enzyme performance, which could be overcome
by an increase in enzyme quantity. The second approach
consisted of evaporating accumulated water by means of an
intermittent airflow.
In this work, it has been shown how the choice of oper-
ating conditions, even as basic as the choice of the solvent,
could be questioned when continuous operation is envis-
aged. The performance of the process is governed by the
control of one parameter, the water activity, which, if al-
ready mentioned as important, was not suspected to influ-
ence so strongly the development of a continuous enzymatic
reaction towards an industrial process.
The authors are very grateful to Valérie Sert and Jean-François
Naudi for their technical assistance and to Mr. Mel Sladdin for
the English style corrections.
References
Aaltonen, O., Rantakylä, M. 1991. Biocatalysis in supercritical CO2.
Chemtech 21: 240–248.
Balcao, V. M., Paiva, A. L., Malcata, F. X. 1996. Bioreactors with immo-
bilized lipases: State of the art. Enzyme Microb. Technol. 18:
392–416.
Brink, L. E. S., Tramper, S., Luyben, K. Ch. A. M., Van’t Riet, K. 1988.
Biocatalysis in organic media. Enzyme Microb. Technol. 10: 736–743.
Chulalaksananukul, W., Condoret, J. S., Delorme, P., Willemot, R. M.
1990. Kinetic study of esterification by immobilized lipase in n-
hexane. FEBS Lett. 276: 181–184.
Chulalaksananukul, W., Condoret, J. S., Combes, D. 1993. Kinetics of the
lipase catalyzed transesterification of geraniol in supercritical CO2.
Enzyme Microb. Technol. 15: 691–698.
Condoret, J. S., Vankan, S., Joulia, X., Marty, A. 1997. Prediction of water
adsorption curves for heterogeneous biocatalysis in organic and super-
critical solvents. Chem. Eng. Sci. 52: 1, 213–220.
Coulon, D., Girardin, M., Rovel, B., Ghoul, M. 1995. Comparison of direct
esterification and transesterification of fructose by Candida antartica
lipase. Biotechnol. Lett. 17: 183–186.
Dordick, J. S. 1989. Enzymatic catalysis in monophasic organic solvents.
Enzyme Microb. Technol. 11: 194–211.
Halling, P. J. 1992. Salt hydrates for water activity control with biocata-
lysts in organic media. Biotechnol. Tech. 6: 271–276.
Halling, P. J. 1994. Thermodynamic predictions for biocatalysis in non-
conventional media: Theory, tests and recommendations for experi-
mental design and analysis. Enzyme Microb. Technol. 16: 178–205.
Keurentjes, J. T. F., Janssen, G. H. R., Gorissen, J. J. 1994. The esterifica-
tion of tartaric acid with ethanol: Kinetics and shifting the equilibrium
by means of pervaporation. Chem. Eng. Sci. 49: 4681–4689.
COLOMBIÉ ET AL.: WATER ACTIVITY CONTROL 367
Kwon, S. J., Song, K. M., Hong, W. H., Rhee, J. S. 1995. Removal of
water produced from lipase catalysed esterification in organic solvent
by pervaporation. Biotechnol. Bioeng. 46: 393–395.
Laane, C., Tramper, J. 1990. Tailoring the medium and reactor for bioca-
talysis. Chemtech 20: 502–506.
Laane, C., Tramper, J., Lilly, M. D. 1986. Optimization of biocatalysis in
organic media, pp. 65–84. In: C. Laane, J. Tramper, and M. D. Lilly
(eds.), Biocatalysis in organic media. Elsevier Science Publishers, Am-
sterdam.
Marty, A., Chulalaksananukul, W., Willemot, R. M., Condoret, J. S. 1992.
Kinetics of lipase catalyzed esterification in supercritical CO2. Bio-
technol. Bioeng. 39: 273–280.
Marty, A., Combes, D., Condoret, J. S. 1994. Continuous reaction separa-
tion process for enzymatic esterification in supercritical carbon diox-
ide. Biotechnol. Bioeng. 43: 497–504.
Marty, A., Dossat, V., Condoret, J. S. 1997. Continuous operation of li-
pase-catalyzed reactions in nonaqueous solvents: Influence of the pro-
duction of hydrophilic compounds. Biotechnol. Bioeng. 56, 2,
232–237.
Monot, F., Borzeix, F., Bardin, M., Vandecasteele, J. P. 1991. Enzymatic
esterification in organic media: Role of water and organic solvent in
kinetics and yield of butyl butyrate synthesis. Appl. Microbiol. Bio-
technol. 35: 759–765.
Mustranta, A., Forssel, P., Poutanen, K. 1993. Applications of immobilized
lipases to transesterification and esterification reactions in non-
aqueous systems. Enzyme Microb. Technol. 15: 133–139.
Napier, E. P., Lacerda, H. M., Rossel, C. M., Valivety, R. H., Vaidya,
A. M., Halling, P. J. 1996. Enhanced organic-phase enzymatic esteri-
368 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 60, NO. 3, NOVEMBER 5, 1998
fication with continuous water removal in a controlled air-bleed evacu-
ated headspace reactor. Biotechnol. Prog. 12: 47–50.
Perraud, R., Laboret, F. 1995. Optimization of methyl propionate produc-
tion catalyzed by Mucor miehei lipase. Appl. Microbiol. Biotechnol.
44: 321–326.
Ragaini, V., Raverdino, V., Santi, S. 1974. Ternary system hexane/2-
propanone/water, at 40°C. Dechema chemistry data series; 1980, Part
2, 491.
Rosel, C. M., Vaidya, A. M., Halling, P. J. 1996. Continuous in situ water
activity control for organic phase biocatalysis in a packed-bed hollow
fiber reactor. Biotechnol. Bioeng. 49: 284–289.
Valivety, R. H., Halling, P. J., Macrae, A. R. 1993. Water as a competitive
inhibitor of lipase catalysed esterification in organic media. Biotech-
nol. Lett. 15: 1133–1138.
Wehtje, E., Svensson, I., Aldercreutz, P., Mattiasson, B. 1993. Continuous
control of water activity during biocatalysis in organic media. Bio-
technol. Tech. 12: 873–878.
Yokozeki, K., Yamanaka, S., Takinami, K., Hirose, Y., Tanaka, A., So-
nomoto, K., Fukui, S. 1982. Eur. J. Appl. Microbiol. Biotechnol. 14:
1–5.
Zaks, A., Klibanov, A. M. 1984. Enzymatic catalysis in organic media at
100°C. Science 224: 1249–1251.
Zaks, A., Klibanov, A. M. 1985. Enzyme-catalyzed processes in organic
solvents. Proc. Natl. Acad. Sci. U.S.A. 82: 3192–3196.
Zaks A., Klibanov A. M. 1988. The effect of water on enzyme action in
organic media. J. Biol. Chem. 263: 8017–8021.
Zaks, A., Russell, A. J. 1988. Minireview. Enzymes in organic solvents:
Properties and applications. J. Biotechnol. 8: 259–270.