Maximise equilibrium conversion in biphasic catalysed reactions: How to obtain reliable data for equilibrium constants?

Article abstract of DOI:10.1002/adsc.200606145

For the prediction and optimisation of the equilibrium conversion in biphasic catalysed reactions, the equilibrium constant of the desired reaction and the partition coefficients of all reactants have to be known. Within this contribution we have examined the alcohol dehydrogenase-catalysed reduction of several linear and aromatic ketones in biphasic reaction media with respect to equilibrium conversion. In this example, the equilibrium constant can be expressed in terms of differences in oxidation-reduction potentials ΔE⁰. However, for a large variety of organic compounds, these data are quite rare in the literature. To overcome this lack of data, we have utilised methods of computational chemistry to calculate data for the Gibbs free energy ΔG_R leading to the equilibrium constants of a homologous series of linear ketones. To obtain comparable data for the reduction of substituted acetophenone derivatives, the Hammett relation leads to the necessary equilibrium constants. Furthermore, we compare the equilibrium conversions of a set of cofactor regeneration methods for the alcohol dehydrogenase-catalysed reductions. These results lead to a time-saving experimental design for the enantioselective reduction of 2-octanone to (R)-2-octanol on a preparative scale utilising biphasic reaction conditions.

Full text of DOI:10.1002/adsc.200606145

FULL PAPERS
DOI: 10.1002/adsc.200606145
Maximise Equilibrium Conversion in Biphasic Catalysed Reac-
tions: How to Obtain Reliable Data for Equilibrium Constants?
a, c,
a
a
b
b
M. F. Eckstein, * J. Lembrecht, J. Schumacher, W. Eberhard, A. C. Spiess,
c
c
c
c
a,
M. Peters, C. Roosen, L. Greiner, W. Leitner, and U. Kragl *
Lehrstuhl für Technische Chemie, Rostock University, A.-Einstein-Str. 3a, 18059 Rostock, Germany
a
Fax : (+49)-381-498-6452; e-mail: udo.kragl@uni-rostock.de
Lehrstuhl für Bioverfahrenstechnik, RWTH Aachen University, Worringerweg 1, 52074 Aachen, Germany
Institut für Technische und Makromolekulare Chemie, RWTH Aachen University, Worringerweg 1, 52074 Aachen,
b
c
Germany
Fax : (+49)-241-80-22177; e-mail: eckstein@itmc.rwth-aachen.de
Received: March 25, 2006; Accepted: June 13, 2006
Supporting information for this article is available on the WWW under http://asc.wiley-vch.de/home/.
Abstract: For the prediction and optimisation of the leading to the equilibrium constants of a homologous
equilibrium conversion in biphasic catalysed reac- series of linear ketones. To obtain comparable data
tions, the equilibrium constant of the desired reac- for the reduction of substituted acetophenone deriv-
tion and the partition coefficients of all reactants atives, the Hammett relation leads to the necessary
have to be known. Within this contribution we have equilibrium constants. Furthermore, we compare the
examined the alcohol dehydrogenase-catalysed re- equilibrium conversions of a set of cofactor regener-
duction of several linear and aromatic ketones in bi- ation methods for the alcohol dehydrogenase-cata-
phasic reaction media with respect to equilibrium lysed reductions. These results lead to a time-saving
conversion. In this example, the equilibrium constant experimental design for the enantioselective reduc-
can be expressed in terms of differences in oxida- tion of 2-octanone to (R)-2-octanol on a preparative
0
tion-reduction potentials DE . However, for a large scale utilising biphasic reaction conditions.
variety of organic compounds, these data are quite
rare in the literature. To overcome this lack of data,
we have utilised methods of computational chemistry Keywords: biotransformation; biphasic catalysis;
to calculate data for the Gibbs free energy DGR equilibria; oxidoreductases; prediction
Introduction
actants given for a set of solvents, while free parame-
ters of optimisation are given by the phase volume
[
10]
Biphasic reaction media are a promising tool to opti- ratio and the initial substrate ratio.
To obtain relia-
mise process conditions of homogeneously catalysed ble data for a suitable set of biphasic systems, these
reactions. So far, biphasic media are used for a broad parameters were often examined experimentally for a
[
1]
[11]
variety of reactions and solvents. Next to water and large number of solvents.
This strategy is a time-
fluorinated solvents , ionic liq- consuming and cost-intensive step of process develop-
and supercritical solvents like supercritical ment, and may quite often be an unattractive perspec-
[
2–4]
[5]
organic solvents,
[
6–8]
uids
[
9]
carbon dioxide are applied in multiphase catalysis tive for the utilisation of biphasic reaction media in
with homogeneous catalysts as well as with biocata- industry.
lysts.
To overcome the effort of numerous experiments
From a reaction engineering point of view, the max- and to offer a time-saving and effective method of ex-
imisation of equilibrium conversion and product yield perimental design, we have recently derived a mathe-
are main targets for optimisation, next to catalyst ac- matical expression that enables the chemist to predict
tivity, selectivity and stability. These factors are influ- the equilibrium conversion and yield of catalysed bi-
[
10]
enced by the thermodynamic driving force of the re- phasic reactions.
This method could successfully be
action, e.g., the difference in oxidation-reduction po- applied on the alcohol dehydrogenase-catalysed re-
0
tentials DE and by the partition behaviour of all re- duction of acetophenone in several biphasic reaction
Adv. Synth. Catal. 2006, 348, 1597 – 1604
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1597

M. F. Eckstein et al.
FULL PAPERS
media, using organic solvents as well as ionic liquids
combined with an aqueous buffer phase.
Here, we use the previously derived equation to
predict the equilibrium conversion of alcohol dehy-
drogenase-catalysed reductions of several linear and
aromatic ketones under biphasic reaction conditions.
The mathematical expression requires reliable data
for the equilibrium constant K as given by the oxida-
tion-reduction potentials of the reactants. For a varie-
ty of ketones and corresponding alcohols, the oxida- Scheme 1. Reduction of prochiral ketones (1) to chiral alco-
0
tion-reduction potentials DE are available in the lit- hols (2) catalysed by the (R)-selective alcohol dehydrogen-
[
12,13]
erature.
However, these data are limited, espe- ase from Lactobacillus brevis (LB-ADH); substrate-coupled
cofactor regeneration with 2-propanol (3) yielding acetone
cially regarding ketones with low water solubilities,
restricting the practical applicability of the developed
mathematical expression. To overcome this lack of
(
4).
data, we investigated alternative ways to obtain the second phase is defined as a non-reactive phase, rep-
equilibrium constants K. To allow a concrete compari- resenting a reservoir for all reactants. This phase may,
son of the different calculation methods, monophasic for example, be represented by non-miscible organic
[
7]
as well as biphasic reaction conditions are compared solvents or ionic liquids. The reactants show a cer-
with respect to equilibrium conversions. tain affinity to each phase, expressed by their specific
Finally, a comparison of cofactor regeneration partition coefficients. It is important to mention that
methods leads to an application of practical relevance. if the catalyst is soluble in both phases, the equilibri-
Therein, the alcohol dehydrogenase-catalysed reduc- um conversion in biphasic systems does not differ
tion of 2-octanone to (R)-2-octanol is performed with from the monophasic case.
high chemical purity and high enantioselectivity Recently, we have shown that the equilibrium con-
under biphasic reaction conditions on a preparative version X of a catalysed bimolecular reaction under
scale.
biphasic reaction conditions, can be calculated using
the following expression [Eq. (1)]:
[
10,18,]
ꢀ
ꢁ
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2
ꢀ1
Results and Discussion
m
Kð þS 1Þ ꢀ ð1 ꢀ S Þ þ 4S Á ðmKÞ
½
19ꢁ
X ¼
:
ð1Þ
2
ðmK ꢀ 1Þ
Alcohol Dehydrogenase-Catalysed Reductions
Thus, X depends on the equilibrium constant K of
The enantioselective reduction of prochiral ketones is the reaction and on the initial substrate ratio S. Fur-
a reaction of synthetic relevance leading to important thermore, X depends on the factor m that includes
intermediates for pharmaceutical compounds and the partition coefficients of all reactants a, b, g, d and
[
14]
food additives.
The application of a large pool of the phase volume ratio V to relate molar amounts
biocatalytic reactions allows the synthesis of these with effective concentrations in the biphasic system
chiral intermediates with high selectivities.[15,16] Here, [Eq. (2)]:[20]
we investigated the alcohol dehydrogenase-catalysed
reduction of several linear and aromatic ketones. All
m ¼
ðgV þ 1ÞðdV þ 1Þ :
ð2Þ
ðaV þ 1ÞðbV þ 1Þ
of these substrates have low water solubilities in
ꢀ
1
common (<0.5 molL ). Thus, the application of a bi-
phasic reaction medium is a necessity to obtain prepa-
Similar expressions were obtained by Martinek
rative yields and sufficiently high space-time yields. et al.[ If the catalyst is soluble in both phases or
The NADPH-dependent alcohol dehydrogenase from monophasic reaction conditions are applied, m equals
Lactobacillus brevis (LB-ADH) is a good biocatalyst 1.
21]
with high selectivity and stability under these reaction
To use Eq. (1) to predict the equilibrium conver-
[
3,7]
conditions.
Cofactor regeneration was carried out sion X for the investigated reductions, we have mea-
in a substrate-coupled approach, by oxidation of sured the partition coefficients of all reactants in the
excess 2-propanol as co-substrate (Scheme 1). presence of methyl-tert-butyl ether (MTBE) as non-
A general description of any biphasic catalysed re- reactive phase (Table 1). In previous works, MTBE
[
17]
action system
includes a reactive phase, in which turned out to be the non-reactive phase of choice for
the catalyst is dissolved and in which the reaction the LB-ADH reduction of acetophenone with respect
takes place. In biocatalytic applications, this phase to maximum equilibrium conversion and yield.[ Ad-
usually is an aqueous buffer phase, containing the dis- ditionally, the utilised LB-ADH has shown high s3t,a7-]
solved biocatalyst and the cofactor, if required. The bilities in the presence of MTBE in earlier studies.[
10]
1598
www.asc.wiley-vch.de
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2006, 348, 1597 – 1604

Maximise Equilibrium Conversion in Biphasic Catalysed Reactions: How to…
FULL PAPERS
0
Table 1. Measured partition coefficients of reactants be- Table 2. Calculated equilibrium conversion X
A
H
R
U
G
tween MTBE and aqueous buffer solution (308C).
to experimental data Xexp obtained under monophasic (aqueous
buffer; xacetonitrile =0.05) and biphasic (aqueous buffer/MTBE;
V=1) reaction conditions (S=20; 308C). X is given in percent.
Substrate
MTBE
Ketone
Alcohol
0
0
Substrate
X (DE )
Xexp mono- X (DE )
Xexp bi-
phasic
Acetone
1.1ꢂ0.2
1.1ꢂ0.2
monophasic phasic
biphasic
2-Pentanone
2-Hexanone
2-Heptanone
2-Octanone
2-Nonanone
10.47ꢂ0.41
33.6ꢂ11.2
170ꢂ28
10.54ꢂ0.52
41.9ꢂ1.7
198ꢂ28
2-Pentanone
2-Hexanone
2-Heptanone 91.8
2-Octanone
2-Nonanone
Acetophenone 90.0
92.3
91.8
89.7
85.2
90.0
90.0
89.2
90.0
92.3
93.6
92.8
-
98.7
84.6
87.9
87.6
77.8
83.6
457ꢂ66
600ꢂ122
[a]
[a]
-
-
-
-
Acetophenone
66.9ꢂ1.5
107ꢂ4.8
345ꢂ52
63.3ꢂ4.0
79.4ꢂ9.7
424ꢂ56
32.1ꢂ2.4
73.1ꢂ4.7
356ꢂ77
-
p-Nitroacetophenone
p-Chloroacetophenone
p-Methoxyacetophenone
m-Methoxyacetophenone
o-Methoxyacetophenone
82.2
64.8ꢂ3.7
5.26ꢂ0.12
9.40ꢂ1.2
difference may be caused by the neglected activity co-
efficients during the measurement of the partition co-
efficients. However, in the case of acetophenone, X
can be predicted with very high accuracy for the
monophasic as well as for the biphasic case. As a gen-
[
a]
For 2-nonanone and 2-nonanol, concentrations in the
aqueous phase were too low to obtain reliable data.
As substrates, we have chosen linear ketones in the eral result we can conclude that the addition of
range from 2-pentanone to 2-nonanone and acetophe- MTBE as non-reactive phase does not have a high
none with some of its derivatives.
impact on the calculated equilibrium conversion
0
For the linear reactants, the partition coefficients X(DE ) of the reduction of lower linear ketones com-
increase with chain length. The linear alcohols show pared to monophasic conditions, due to similar parti-
higher partition coefficients compared to their corre- tion coefficients of substrates and products.
sponding ketones. In the case of 2-nonanone and 2-
nonanol, concentrations in the aqueous phase were could be calculated using DE from the litera-
extremely low and could not be determined reliably. ture.
For most of the results shown so far (Table 2), K
0
[
12,13]
However, investigating the reduction of 2-
For acetophenone derivatives, the position of the sub- octanone and several aromatic ketones like p-chloro-
stituent strongly influences the partition behaviour, acetophenone or p-nitroacetophenone, no comparable
e.g., in the case of the different methoxyacetophe- data were available. To realise the prediction of the
nones the partition coefficient of the ketone differs equilibrium conversion X without available data for
0
from 63.3 up to 424. In contrast, the alcohols corre- DE , we tried to fill this lack of data by calculating K
sponding to meta- and ortho-methoxyacetophenone via computational chemistry or by using the Hammett
show a much lower affinity to the organic solvent sigma constant for the reduction of acetophenone de-
compared to the ketones (Table 1).
rivatives.
Obtain K via DE0
Obtain K via Computational Chemistry
As we consider an oxidation-reduction reaction, the One possibility to obtain K is the calculation of the
0
difference in oxidation-reduction potentials DE can Gibbs free energy of reaction DGa,R in aqueous solu-
0
be used to calculate the equilibrium constant K
A
H
R
U
G
and thus to predict the equilibrium conversion Gibbs free energies of formation of all reactants in
0
X(DE ). The results were compared to experimental aqueous solution DGa,f,i. The latter can also be ex-
data using a monophasic approach (m=1) as well as pressed in terms of the standard Gibbs free energy of
0
a;f;i
biphasic reaction conditions (Table 2). In the mono- formation in aqueous solution DG
or the standard
phasic approach, acetonitrile was added as co-solvent Gibbs free energy of reaction in aqueous solution
0
a;R
with a mole fraction of xacetonitrile =0.05 [~13% (v/v) DG [Eq. (3)]:
acetonitrile], to realise a substrate solubility of at
least 10 mM in the aqueous solution.
X
DGa;R
¼
ui Á DG
a;f;i
¼
i
Except for the reduction of n-hexanone, the experi-
mental results for the monophasic reaction are in
good agreement with the calculated data, while the
experimental results obtained under biphasic reaction
conditions are higher than the calculated values. This
X
0
a;f;i
0
a;R
ui Á DG þ RT ln K ¼ DG þ RT ln K ð3Þ
i
Adv. Synth. Catal. 2006, 348, 1597 – 1604
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.asc.wiley-vch.de
1599

M. F. Eckstein et al.
FULL PAPERS
The reaction equilibrium in aqueous solution is Table 3. Equilibrium constants K calculated using oxidation-
[12]
reduction potentials
(DG) and the predicted conversions X
a monophasic system (S=20; 258C). X is given in percent.
ACHTREUNG
(DE ) and the COSMO-RS model
0
0
given if DGa,R =0. In this case the equilibrium con-
0
A
H
R
U
G
stant K can be calculated from DG with x being
a;R
i
the mole fraction of reactant i in the solution and gi
being its activity coefficient [Eq. (4)]:
0
0
Substrate
K (DE )
K (DG)
X (DE )
X (DG)
0
P
1
A
!
0
0
a;R
ꢀ
u
i
ÁDGa;f;i
2-Butanone
2-Pentanone
0.627
0.580
0.536
0.536
-
2.87
2.48
2.03
1.61
1.59
92.8
92.3
91.8
91.8
-
98.2
98.0
97.5
96.9
96.9
Y
ꢀ
DG
RT
eq eq
u
i
i
@
K¼ ðx g Þ ¼exp
¼exp
ð4Þ
i
i
RT
2
2
2
-Hexanone
-Heptanone
-Octanone
i
0
are tabulated.[22] Howev-
Several values for DG
a;f;i
er, regarding the large variety of molecules treated in
bio-)catalysis, only sparse data are available. In this
(
case, predictive methods, like group contribution carbon atoms is well returned in comparison to
0
methods or continuum solvation models can be ap- K(DE ).
plied.
On the one hand, these results show that, where no
Mavrovouniotis et al. introduced a group contribu- comparable data are available from the literature,
0
a;f;i
[23,24]
tion method to obtain DG in aqueous solution.
COSMO-RS can be used to predict trends in K for
However, this method could not be applied for the in- different reactions quite well. On the other hand, the
vestigated set of reactions, since the group contribu- higher the complexity of the molecules the more
tions on the substrate and product side are the same, effort has to be done with respect to conformer
0
a;R
yielding a standard DG ¼ 0, which is physically not choice and geometry optimisation.
reasonable.
A more sophisticated way to describe solute solvent
systems is the COSMO-RS method based on the Obtain K via the Hammett Relation
COSMO continuum solvation model.[
25,26]
COSMO-
RS combines quantum-chemical methods on the DFT For the conversion of substituted aromatic com-
level with those of statistical thermodynamics and pounds, the Hammett sigma constant sX can be used
[
20]
allows a genuine prediction of VLE and LLE data.
to calculate the equilibrium constant K(s ), if an
X
Besides the prediction of interaction effects between equilibrium constant Kunsubstituted is available for the
solute and solvents, COSMO-RS also gives the total conversion of the unsubstituted compound [Eq.
[
27]
Gibbs free energy of a molecule G . The Gibbs free (6)]:
a,i
energy of formation DGa,f,i is defined by the sum of
sX
the energies of molecules minus the sum of atom en- Kðs Þ ¼ K
Á 10
ð6Þ
X
unsubstituted
ergies. For a reaction, where amount and type of
atoms equal one another on the substrate and product
Acetophenone with s =0.0 is the origin of the
H
side, the atom information cancels out in the Gibbs Hammett relation. Thus, using the equilibrium con-
0
free energy of formation DGa,f,i. Consequently, in this stant K
A
H
R
U
G
for acetophenone, Eq. (6)
unsubstituted
case the Gibbs free energy of reaction DGa,R equals gives K(s ) for acetophenone derivatives substituted
X
the sum of the total Gibbs free energies of molecules in the para and meta positions. Due to steric effects,
Ga,i or the sum of the Gibbs free energies of forma- no Hammett sigma constants s are available for de-
X
[
27]
tion DGa,f,I [Eq. (5):
rivatives substituted in the ortho position.
K(sX)
can be used to predict the equilibrium conversion
X
X
X(s ). To show the applicability of this concept, we
X
Ga;R
¼
ui Á Ga;i
¼
ui Á DGa;i ¼ DGa;f;i
ð5Þ
have chosen a series of substituted acetophenone de-
i
i
rivatives. Where available, we compared X(s ) to
X
0
COSMO-RS does not only calculate the total data calculated for X
A
H
R
U
G
Gibbs free energies in aqueous solution G , but gives sults Xexp obtained for the monophasic and biphasic
a,i
the same in any real solvent. In our work, calculations system (Table 4). In most cases, the calculated equilib-
0
were done for a homologous series of reductions, rium conversions X
A
H
R
U
G
X
keeping one substrate (2-propanol) and one product agreement. However, in the case of p-methoxyaceto-
acetone) constant, while raising the number of phenone, the calculated equilibrium conversions
carbon atoms in the second substrate and product differ up to 14%. One reason may be the accuracy of
Table 3). Although the results obtained for K(DG) the literature data. Unfortunately, using the LB-ADH
differ by a factor of 2 to 4.6 from the values calculat- as catalyst, no reliable experimental results could be
(
(
0
ed via the oxidation-reduction potentials K
A
H
R
U
G
decreasing trend in K with increasing number of is far from equilibrium, which is probably due to a co-
1600
www.asc.wiley-vch.de
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2006, 348, 1597 – 1604

Maximise Equilibrium Conversion in Biphasic Catalysed Reactions: How to…
FULL PAPERS
Table 4. Calculated equilibrium conversions X using oxida-
Table 5. Comparison of cofactor regeneration methods with
[12]
tion-reduction potentials
constant (sX) compared to experimental data Xexp for the
monophasic (buffer; xacetonitrile =0.05) and biphasic (buffer/
A
H
R
U
G
respect to calculated data of the thermodynamic equilibrium
0
[27]
0
0
constant K
A
H
R
U
G
A
H
R
U
G
0
of 2-heptanone (E =121 mV) in a monophasic system
(buffer, xacetonitrile =0.05; S=1;). E is given in mV; X(DE ) is
0
MTBE; V=1) system (S=20; 308C). X is given in per-
A
C
H
T
R
E
U
N
G
0
cent.
given in percent.
0
0
0
Substrates
Monophasic
system
Biphasic system
Co-substrate
E
K
A
H
R
U
G
X(DE )
A
C
H
T
R
E
U
N
G
Substrate-coupled:
X
X
Xexp
X
X
Xexp
[12]
0
0
2-Propanol
+ 29
0.536
42.3
(
DE ) (s )
(DE ) (s )
X
X
Enzyme-coupled:
Glucose (GDH)
Phosphite (PTDH)
[
31]
Acetophenone
p-Chloroacetophe-
none
p-Bromoacetophe-
none
p-Nitroacetophenone
m-Nitroacetophenone 99.1 97.7
p-Methoxyacetophe-
none
m-Methoxyacetophe- 91.2 92.1
none
o-Methoxyacetophe-
none
90.0
-
-
90.0 83.2
-
83.6
+54
ꢀ650
ꢀ199
ꢀ413
125
91.8
[32]
26
10
18
93.6 94.1 -
89.8 83.4
1.2·10
6.7·10
1.5·10
>99.9
>99.9
>99.9
[33]
Formate (FDH)
[
34,35]
95.2 93.6
-
-
-
-
Hydrogen (hydrogenase)
Electrochemical:
[36]
18
-
98.0 99.4 -
97.2 99.5
-
RhMedOx
ꢀ430
4.4·10
>99.9
-
-
-
[
[
[
a]
a]
a]
72.1 83.9
74.2 88.2
-
-
-
proach, where one enzyme catalyses the desired re-
duction as well as the regeneration reaction of
(DE )=42.3% with 2-propanol as
co-substrate in a monophasic system (S=1). To
obtain X>99.0% under these conditions, the sustrate
ratio has to be increased to at least S>180.
trast, utilising different enzyme-coupled regeneration
-
-
59.0 57.0
0
NADPH, leads to X
A
H
R
U
G
98.0
-
65.1
-
[
a]
Irreversible binding of substrate; no reliable data avail-
able.
In con-
methods, where a second enzyme is used to catalyse
0
valent binding of the substrates to the catalytic centre the regeneration reaction, leads to X
A
H
R
U
G
of the enzyme that leads to a deactivation of the the case of glucose as co-substrate and even reaches
0
enzyme. Currently other catalysts are under investiga-
tion to obtain experimental results that allow a fur- are used as co-substrates. Similar results could be cal-
ther comparison and discussion of these data. culated for an electrochemical regeneration method.
The experimental results Xexp obtained for p- In comparison, results obtained for the biphasic
X(DE )>99.9%, if phosphite, formate or hydrogen
A
H
R
U
G
0
chloroacetophenone and p-nitroacetophenone were system (MTBE/buffer; V=1; S=1), X(DE ) values
A
H
R
U
G
predicted with high accuracy for the monophasic differ only slightly for the co-substrates 2-propanol
0
system, closing the gap of DE availability. A differ- (44.1%) and glucose (92.3%), due to the similar parti-
ence between X(s ) and X of about 6% for the re- tion behaviour of 2-heptanone and 2-heptanol
X
exp
duction of p-chloroacetophenone in the biphasic (Table 1).
system may be due to large standard deviations for These calculations underline that the driving force
the measurement of the partition coefficients of the reaction DE can be increased by the right
0
(
Table 1).
choice of co-substrate and regeneration method.
However, next to the regeneration method and sol-
vent combination for biphasic reaction media, enzyme
stability, activity and inhibitory effects influence the
choice of reaction conditions applied on preparative
Choice of Cofactor Regeneration Method
In contrast to whole cell processes, most enzyme-cata- scale.
lysed oxidation-reduction reactions require a regener-
We investigated the LB-ADH-catalysed reduction
ation of the cofactor within the catalytic process. Dif- of 2-octanone to (R)-2-octanol under biphasic reac-
ferent methods can be applied to achive an efficient tion conditions. The main advantage of this approach
cofactor regeneration, even in organic sovents or bi- is an easy work-up of the product (R)-2-octanol. Due
[
29,30]
phasic systems.
Using Eq. (1), it is now possible to a satisfactory half-life time of 60 h under reaction
to predict and compare equilibrium conversions X- conditions (MTBE/buffer), cofactor regeneration was
0
A
C
H
T
R
E
U
N
G
(DE ) of processes with different regeneration meth- realised with a glucose dehydrogenase from Pseudo-
ods and thus to choose a suitable regeneration monas species (PS-GDH) with glucose as co-sub-
0
method for the desired process to maximise X
ACHTREUNG
0
We have calculated X
tion of 2-heptanone, comparing different regeneration with Xexp >99% [X
methods (Table 5). Utilising a substrate-coupled ap- heptanone] and an ee>99.5%. After 2 days of reac-
ACHTREUNG
0
A
H
R
U
G
Adv. Synth. Catal. 2006, 348, 1597 – 1604
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.asc.wiley-vch.de
1601

M. F. Eckstein et al.
FULL PAPERS
ꢀ
1
ꢀ1
tion, this lead to an STY=20 gL
(
d
and a ttn buffer to pH 7 containing 1 mM magnesium chloride if not
NADPH)=7920.[20] While the LB-ADH, NADP(H),
mentioned otherwise.
glucose and gluconate stay in the aqueous buffer
phase, the product stays in the organic phase and can
easily be removed by phase separation (isolated yield Analysis
after distillation 71%). In contrast, utilising 2-propa-
nol as co-substrate with a substrate-coupled cofactor Ketones and alcohols were analysed by gas chromatography,
using a Varian CP 3800 equipped with DB-1701 capillary
regeneration, the co-substrate would be needed in
high excess (S>460) to obtain similar results. Addi-
tionally, this large amount of co-substrate would influ-
ence the phase separation and partition behaviour
and thus, work-up to obtain product with high chemi-
cal purities is more difficult.
column Alltech/Germany (dodecane as internal standard,
ꢀ1
flow rate He 2.0 mLmin ). Samples were diluted with ethyl
acetate prior to measurement. Enantiomeric excess was de-
termined by GC-FID with a FS-Cyclodex-beta-I/P column
ꢀ1
Macherey&Nagel/Germany (He 2.0 mLmin ). To measure
concentrations of acetone and 2-propanol an HPLC method
was used. Therefore, the HPLC was equipped with an RI
detector and a BIORAD Aminex HDK-87 H Ion Exclusion
ꢀ1
Conclusions
column (300”7.8) with sulfuric acid (6 mM; 0.8 mLmin ).
Samples were diluted with water as appropriate.
In conclusion, the lack of data for the prediction of
equilibrium conversion in catalysed biphasic and
monophasic reactions can partly be overcome by ab Partition Coefficients
initio calculations and by empirical calculations.
While ab initio methods like COSMO-RS are valua- Partition coefficients were measured adopting methods re-
ble to predict relative trends in equilibrium constants ported previously.[
37,38]
All measurements were carried out
at least 7-fold. Standard deviations are given as errors. Sub-
strate solutions containing the ketone and alcohol were pre-
pared in MTBE. In screw capped vials (8 mL), the aqueous
phase (4 mL) was covered with the organic phase (4 mL).
The samples were shaken (308C; 400 rpm). The mixtures
were stored in a water bath (308C; 3 days). Samples were
taken from each phase and analysed. Partition coefficients
of acetone and 2-propanol were determined following the
decline of concentration in the aqueous phase before and
and conversions, the Hammett relation is a suitable
method to calculate equilibrium constants for the con-
version of substituted aromatic substrates. Next to ox-
idation-reduction reactions, this method is also appli-
cable for other reactions types. However, due to rea-
sons of steric hindrance, no Hammett sigma constants
are available for derivatives substituted in the ortho
[
27]
position.
Furthermore, it was possible to predict and com- after the addition of the organic phase. Samples were taken
pare equilibrium conversions for 6 different cofactor from the aqueous phase containing acetone and 2-propanol
prior to contacting with solvents. Following the same proce-
regeneration methods. If enzyme availability and sta-
dure as above, samples were analysed by HPLC. The parti-
bility allow their use under the reaction conditions,
tion coefficient e[
20]
is then calculated according to e=
enzyme-coupled cofactor regeneration should be em-
ployed for the preparative production by enzymatic
means.
[
(starting concentration)/(residual concentration)] ꢀ 1. In
this method, activity coefficients describing the deviations
from dilute behaviour were neglected. Measurements at dif-
ferent substrate concentrations assured that partition coeffi-
cients were not concentration depended in this concentra-
tion range.
Experimental Section
Materials
Equilibrium Conversion (LB-ADH), Monophasic
Acetophenone, 1-phenylethanol, 4-chloroacetophenone, 1-
(
4-chlorophenyl)-ethanol, 4-methoxyacetophenone and, all
Acetonitrile was added to the buffered aqueous solution in
+
linear ketones and alcohols, n-dodecane and MTBE were
purchased from Merck/Germany. 3-Methoxyacetophenone
and 1-(2-methoxyphenyl)-ethanol were supplied by Acros
Organics/Germany, 2-methoxyacetophenone was obtained
from Fluka/Germany. 2-Propanol and acetone were pur-
chased from J.T. Baker/Germany. All chemicals were ob-
tained in p.a. quality.
a mole fraction x=0.05 (13% v/v). LB-ADH, NADP
ꢀ1
(0.1 mmolL ), 2-propanol and the ketone were dissolved in
this mixture (2.5 mL) and stored at 308C. Samples were
withdrawn, diluted with ethyl acetate as appropriate and an-
alysed by GC.
The alcohol dehydrogenase from Lactobacillus brevis (EC
1
.1.1.2), the glucose dehydrogenase from Pseudomonas spe- Equilibrium Conversion (LB-ADH), Biphasic
+
cies (E.C. 1.1.1.47) and the NADP (disodium salt) were
supplied by Jülich Chiral Solutions/Germany. Aqueous solu-
tions were buffered using 50 mM potassium phosphate
+
An aqueous solution (2.5 mL) of LB-ADH, NADP
(0.1 mmolL ) and 2-propanol was covered with MTBE
ꢀ1
1602
www.asc.wiley-vch.de
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2006, 348, 1597 – 1604

Maximise Equilibrium Conversion in Biphasic Catalysed Reactions: How to…
FULL PAPERS
(
2.5 mL) containing the ketone (10 to 80 mM) and the mix-
[2] H. Grçger, H. Hummel, C. Rollmann, F. Chamouleau,
H. Hüsken, H. Werner, C. Wunderlich, K. Abokitse, K.
Drauz, S. Buchholz, Tetrahedron 2004, 60, 633–640.
[3] M. Villela Filho, T. Stillger, M. Müller, A. Liese, C.
Wandrey, Angew. Chem. Int. Ed. 2003, 42, 2993–2996.
[4] H. Klein, R. Jackstell, M. Beller, Chem. Commun.
ture was shaken in a vertical shaker (308C; 200 rpm). Sam-
ples were withdrawn from the non-reactive phase and ana-
lysed by GC. Mass balance of the ketone and alcohol were
corrected by their partition coefficients:
total concentration of E=[(concentration of E in non-re-
active phase)/e] + concentration of E in non-reactive phase.
Conversion was calculated according to:
X=(total product concentration)/[(total substrate concen-
tration) + (total product concentration)].
The same reaction conditions were applied to obtain 1-(4-
methoxyphenyl)-ethanol, 1-(3-methoxyphenyl)-ethanol and
2
005, 2283–2285.
[
5] D. J. Adams, D. J. Cole-Hamilton, E. G. Hope, P. J. Po-
gorzelec, A. M. Stuart, J. Org. Chem. 2004, 68, 1413–
1
417.
[
6] P. J. Dyson, Chimia 2005, 59, 66–71.
[
7] M. Eckstein, M. Villela Filho, A. Liese, U. Kragl,
1
-(4-nitrophenyl)-ethanol for the investigation of partition
Chem. Commun. 2004, 1084–1085.
coefficients.
[
8] H. Pfründer, M. Amidjojo, U. Kragl, D. Weuster-Botz,
Angew. Chem. Int. Ed. 2004, 43, 4529–4531.
9] W. Leitner, Pure Appl. Chem. 2004, 76, 635–644.
For the reduction of 2-octanone to 2-octanol on a prepa+-
rative scale, LB-ADH (300 mg), PS-GDH (18 mg), NADP
[
(
40 mg) and glucose (90 g) were dissolved in the aqueous
phase and covered by MTBE (200 mL) and 2-octanone
51 g). Samples were withdrawn from the organic phase and
[
10] M. Eckstein, M. Peters, J. Lembrecht, A. Spieß, L.
Greiner, Adv. Synth. Catal. 2006, 348, 1591–1596.
11] W. Keim, Green Chem. 2003, 5, 105–111.
(
[
[
analysed by GC-FID. After the reaction, the organic phase
was separated and dried over sodium sulfate. MTBE was
evaporated and the remaining (R)-2-octanol was distilled for
further purification (isolated yield: 71.2%).
12] H. Adkins, R. M. Elofson, A. G. Rossow, C. C. Robin-
son, J. Am. Chem. Soc. 1949, 71, 3622–3629.
13] S. Gladiali, G. Mestroni, in: Transition metals for or-
ganic synthesis – Building blocks and fine chemicals,
[
(
1
Eds.: M. Beller, C. Bolm), Wiley VCH, Weinheim,
998, Vol. 2.
Calculations with COSMO-RS
[
[
[
14] A. J. J. Straathof, S. Panke, A. Schmid, Curr. Opinion
Biotechnol. 2002, 13, 548–556.
15] W. Kroutil, H. Mang, K. Edegger, K. Faber, Curr.
Opinion Biotechnol. 2004, 8, 120–126.
The DFT/COSMO calculations with the B-P density func-
tional and the TZVP basis set combination were performed
using Turbomole 5.7.1. For the following statistical thermo-
dynamics, COSMOtherm-C2.1–01.054 was used. The free
energy of a molecule in the solvent provided from COSMO-
therm was calculated for 258C and corrected by adding
16] M. Breuer, K. Ditrich, T. Habicher, B. Hauer, M.
Keßler, R. Stürmer, T. Zelinski, Angew. Chem. Int. Ed.
2
004, 43, 788–824.
[
[
17] For a general scheme of biphasic reactions see Support-
ing Information.
18] L. Greiner, S. Laue, A. Liese, C. Wandrey, Chem. Eur.
RTLn(x). The mixture composition was iterated until the
i
free energy of reaction Gs,R was zero [Eq. (5)]. With this
mixture composition and the referring activity coefficients,
also provided from COSMOtherm, the equilibrium constant
K(DG) was calculated using Eq. (4).
J. 2006, 12, 1818–1823.
19] For mK=1, it can be shown that X=S (S + 1)
ꢀ1
[
[
.
20] For a comprehensive lists of all mathematical defini-
tions, abbreviations and symbols used, see the Support-
ing Information.
Acknowledgements
[21] K. Martinek, A. M. Klibanov, G. P. Samokhin, A. N. Se-
menov, I. V. Berezin, Bioorg. Khimiya 1977, 3, 696–
702.
We thank Dr. T. Daußmann from Julich Chiral Solutions
(
www.julich.com) for providing the alcohol dehydrogenase
[22] D. D. Wagmann, W. H. Evans, V. B. Parker, R. H.
Schumm, I. Halow, S. M. Bailey, K. L. Churney, R. L.
Nuttall, J. Phys. Chem. Ref. Data 1982, 11, suppl.2.
[23] M. L. Mavrovouniotis, Biotech. Bioeng. 1990, 36, 1070–
1082.
from Lactobacillus brevis. M.F. Eckstein thanks the Ministry
of Innovation, Science, Research and Technology of the State
of North Rhine-Westphalia, Germany (MIWFT) for a Lise-
Meitner-Scholarship. Parts of this work were financially sup-
ported by the DFG-Graduiertenkolleg 1213: Neue Methoden
für Nachhaltigkeit in Katalyse und Technik and the DFG-
Graduiertenkolleg 1166: Biocatalysis in non-conventional
media (www.bionoco.org). Furthermore, we want to thank
Prof. J. Büchs and Prof. M. Liauw (RWTH Aachen Universi-
ty) for their support and fruitful discussions.
[
24] R. G. Forsythe, P. D. Karp, M. L. Mavrovouniotis,
CABIOS – Computer applications in the biosciences
1
997, 13, 537–543.
[
[
25] A. Klamt, J. Phys. Chem. 1995, 99, 2224–2235.
26] W. Arlt, O. Spuhl, A. Klamt, Chem. Eng. Process. 2004,
4
3, 221–238.
[
27] P. Sykes, Reaktionsmechanismen der organischen
Chemie, 9th edn., VCH, Weinheim, 1988.
0
[
[
28] For values of DE and sX see Supporting Information.
References
29] P. Adlercreutz, Biocatal. Biotrans. 1996, 14, 1–30.
[
1] B. Cornils, W. A. Herrmann, I. T. Hovat, W. Leitner, S.
Mecking, H. Olivier-Bourbigou, D. Vogt, Multiphase
homogeneous catalysis, Wiley-VCH, Weinheim, 2005.
[30] M. Eckstein, T. Daußmann, U. Kragl, Biocatal. Bio-
trans. 2004, 22, 89–96.
Adv. Synth. Catal. 2006, 348, 1597 – 1604
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.asc.wiley-vch.de
1603

M. F. Eckstein et al.
FULL PAPERS
[
31] W. M. Clark, Oxidation-reduction potentials of organic
systems, R. E. Krieger Publishing Company, Hunting-
ton, N. Y., 1972.
[35] R. Mertens, L. Greiner, E. C. D. van den Ban,
H. B. C. M. Haaker, A. Liese, J. Mol. Catal. B 2003,
24–25, 39–52.
[
[
[
32] J. M. Vrtis, A. K. White, W. M. Metcalf, W. A. van der
Donk, Angew. Chem. Int. Ed. 2002, 114, 3391–3393.
33] R. P. S. Chaplin, A. A. Wragg, J. Appl. Electrochemistry
[36] K. Vuorilehto, S. Lütz, C. Wandrey, Bioelectrochemistry
2004, 65, 1–7.
[37] K. Fan, P. Ouyang, X. Wu, Z. Lu, J. Chem. Technol.
Biotechnol. 2001, 76, 851–856.
2
003, 33, 1107–1123.
34] L. Greiner, D. H. Müller, E. C. D. van den Ban, J. Wçl-
[
38] J. Bruijn de, F. Busser, W. Seinen, J. Hermens, Environ-
tinger, C. Wandrey, A. Liese, Adv. Synth. Catal. 2003,
mental Toxicology Chemistry 1989, 8, 499–512.
3
45, 679–683.
1604
www.asc.wiley-vch.de
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2006, 348, 1597 – 1604