Modelling the reaction course of a dynamic kinetic resolution of amino acid derivatives: Identifying and overcoming bottlenecks

Article abstract of DOI:10.1021/op0502349

Theoretically, dynamic kinetic resolution gives stereochemically pure compounds in 100% yield. Practically, this target is seldom reached. In this work, we identified limiting factors for a given reaction. The dynamic kinetic resolution of α-amino acid esters at 25 °C in a water/acetonitrile mixture gives optically active α-amino acids in good yields and optical purity. The Alcalase-catalysed hydrolysis of the ester is combined with in situ racemisation catalysed by 3,5-dinitrosalicylaldehyde. Kinetic analysis and modelling of the system made optimisation possible. Determination of reaction kinetics showed that catalyst deactivation is the limiting factor, and modelling and optimisation of the reaction system lead to an improved yield and shorter reaction time.

Full text of DOI:10.1021/op0502349

Organic Process Research & Development 2006, 10, 622−627
Modelling the Reaction Course of a Dynamic Kinetic Resolution of Amino Acid
Derivatives: Identifying and Overcoming Bottlenecks
Vera Zimmermann,† Matthias Beller,‡ and Udo Kragl*,†
UniVersita¨t Rostock, Abteilung fu¨r Analytische, Technische und Umweltchemie, Albert-Einstein-Strasse 3a,
18059 Rostock, Germany, and Leibniz-Institut fu¨r Katalyse an der UniVersita¨t Rostock e.V., Albert-Einstein-Strasse 29a,
18059 Rostock, Germany
Abstract:
resolution offers a tool to prepare optically active compounds
with 100% yield and 100% ee.12 Experience shows that this
aim is practically seldom reached. In this work we investi-
gated the kinetics of a known DKR and modelled the course
of reaction to examine the factors that are limiting for yield
and optical purity.
Theoretically, dynamic kinetic resolution gives stereochemically
pure compounds in 100% yield. Practically, this target is seldom
reached. In this work, we identified limiting factors for a given
reaction. The dynamic kinetic resolution of r-amino acid esters
at 25 °C in a water/acetonitrile mixture gives optically active
r-amino acids in good yields and optical purity. The Alcalase-
catalysed hydrolysis of the ester is combined with in situ
racemisation catalysed by 3,5-dinitrosalicylaldehyde. Kinetic
analysis and modelling of the system made optimisation possible.
Determination of reaction kinetics showed that catalyst deac-
tivation is the limiting factor, and modelling and optimisation
of the reaction system lead to an improved yield and shorter
reaction time.
The reaction examined was the dynamic kinetic resolution
of DL-phenylalanine ethyl ester shown in Scheme 1.13 Several
examples for the DKR of amino acid derivatives using
protease-catalysed hydrolysis combined with substrate ra-
cemisation catalysed by an aldehyde have been reported,14-17
but all of them suffer from disadvantages such as low yield
or optical purity or high catalyst concentration or the use of
very expensive catalysts. In our case the Alcalase-catalysed
hydrolysis of the ester is coupled with in situ racemisation
catalysed by 3,5-dinitrosalicylaldehyde 5. The major enzyme
component in Alcalase is subtilisin (E.C. 3.4.21.62). The
reaction is carried out at 25 °C in a 1:1 mixture of water
and acetonitrile and an apparent pH of 7.5 and gives
L-phenylalanine 3 in 80% yield and 96% ee after 30 h of
reaction time. These were the conditions giving the best
results during development of the reaction system in previous
studies.18
1. Introduction
The synthesis of chiral, enantiomerically pure compounds
still attracts growing attention in chemical synthesis due to
the increasing need for such compounds in the pharmaceuti-
cal and agricultural industries.1,2 Although asymmetric
catalysis has developed quickly during the past decades,
chiral compounds are often obtained via resolution of racemic
mixtures.3,4 The disadvantages of this procedure are well-
known: The maximum product yield cannot exceed 50%,
and the remaining substrate has to be discarded or recycled
by more or less intricate racemisation. The dynamic kinetic
resolution (DKR) is an improvement of the kinetic resolution
and avoids these drawbacks. During DKR the kinetic
resolution of a racemic mixture is coupled with in situ
racemisation of the substrate so that all of the substrate can
be converted into the desired, enantiomerically pure product.5
There are several excellent reviews summarising the systems
developed so far.6-11 Theoretically, the dynamic kinetic
To examine why the yield is limited to 80%, the kinetics
of the reaction were determined, and the course of the
reaction was modelled. We show in the following that a
deactivation of the catalysts is the limiting factor in this
reaction and how the obtained yield can be raised while
shortening the reaction time.
(8) Noyori, R.; Tokunaga, M.; Kitamura, M. Bull. Chem. Soc. Jpn. 1995, 68,
36-56.
(9) Pamies, O.; Backvall, J. E. Chem. ReV. 2003, 103(8), 3247-3261.
(10) Ward, R. S. Tetrahedron: Asymmetry 1995, 6(7), 1475-1490.
(11) Gihani, M. T. E.; Williams, J. M. J. Curr. Opin. Chem. Biol. 1999, 3, 11-
15.
* To whom correspondence should be addressed. Telephone: +49 381/498-
6450. Fax: +49 381/498-6452. E-mail: udo.kragl@uni-rostock.de.
† Universita¨t Rostock.
(12) Strauss, U. T.; Felfer, U.; Faber, K. Tetrahedron: Asymmetry 1999, 10,
107-117.
(13) Schichl, D. A. Racemisierung und dynamische kinetische Racematspaltung
von Aminosa¨urederivaten und sekunda¨ren Alkoholen (Racemization and
dynamic kinetic resolution of amino acid derivatives and secondary
alcohols). Ph.D. Thesis, TU Mu¨nchen, Mu¨nchen, 2001.
(14) Parmar, V. S.; Singh, A.; Bisht, K. S.; Kumar, N.; Belokon, Y. N.;
Kochetkov, K. A.; Ikonnikov, N. S.; Orlova, S. A.; Tararov, V. I.; Saveleva,
T. F. J. Org. Chem. 1996, 61, 1223-1227.
‡ Leibniz-Institut fu¨r Katalyse an der Universita¨t Rostock.
(1) Breuer, M.; Ditrich, K.; Habicher, T.; Hauer, B.; Kesseler, M.; Sturmer,
R.; Zelinski, T. Angew. Chem., Int. Ed. 2004, 43(7), 788-824.
(2) Schoemaker, H. E.; Boesten, W. H. J.; Broxterman, Q. B.; Roos, E. C.;
Kaptein, B.; van denTweel, W. J. J.; Kamphuis, J.; Meijer, E. M.; Rutjes,
F. P. J. T. Chimia 1997, 51(6), 308-310.
(3) Schmid, A.; Dordick, J. S.; Hauer, B.; Kieners, A.; Wubbolts, M.; Witholt,
B. Nature 2001, 409, 258-268.
(15) Chen, S.-T.; Huang, W.-H.; Wang, K.-T. J. Org. Chem. 1994, 59, 7580-
7581.
(4) Pesti, J. A.; DiCosimo, R. Curr. Opin. Drug DiscoVery DeV. 2000, 3(6),
764-782.
(5) Sheldon, R. A. Chirotechnology; Marcel Dekker: New York, 1993.
(6) Pellissier, H. Tetrahedron 2003, 59, 8291-8327.
(7) Huerta, F. F.; Minidis, A. B. E.; Ba¨ckvall, J.-E. Chem. Soc. ReV. 2001,
30(6), 321-331.
(16) Hacking, M. A. P. J.; Wegman, M. A.; Rops, J.; van Rantwijk, F.; Sheldon,
R. A. J. Mol. Catal. B: Enzym. 1998, 5, 155-157.
(17) Liljeblad, A.; Kiviniemi, A.; Kanerva, L. T. Tetrahedron 2004, 60(3), 671-
677.
(18) Beller, M.; Schichl, D. A.; Riermeier, T. Dynamic kinetic resolution of amino
acid esters. Submitted.
622
•
Vol. 10, No. 3, 2006 / Organic Process Research & Development
10.1021/op0502349 CCC: $33.50 © 2006 American Chemical Society
Published on Web 03/24/2006

Scheme 1. Dynamic kinetic resolution of DL-phenylalanine ethyl estera
a Desired reaction (inside frame) and side reactions.
Scheme 2. Mechanism of the racemisation of phenylalanine ethyl ester
2. Experimental Section
mixture of perchloric acid (pH ) 2) and methanol 9:1 at 1.5
mL/min, and monitored at 265 nm.
Chemicals. L-Phenylalanine and L-phenylalanine ethyl
ester were obtained from Fluka (Buchs, Switzerland). L-
phenylalanine ethyl ester was always applied as the hydro-
chloride. 3,5-Dinitrosalicylaldehyde and perchloric acid were
obtained from Sigma-Aldrich (Steinheim, Germany). Alca-
lase was purchased as a brown liquid from Merck (Darm-
stadt, Germany) with a specific activity of 0.6 Anson-U/g.
Sodium hydroxide was obtained from Lachema (Prague,
Czech Republic), acetonitrile and methanol from J. T. Baker
(Deventer, Netherlands).
Kinetic Constants. The kinetic constants were obtained
by determination of initial reaction rates for various substrate
concentrations at a constant temperature of 25 °C in a solvent
mixture of acetonitrile and water (1:1). Substrate concentra-
tions were varied between 0.03 and 1.2 mol/L while the
catalyst concentrations remained constant (0.005 mol/L
dinitrosalicylaldehyde or 1 vol % Alcalase). Periodically,
aliquots of 100 µL were withdrawn and diluted 1:10 in
hydrochloric acid (pH 1). The solution was filtered with
MICROCON YM 30 centrifugal filters and analysed by
HPLC.
3. Results and Discussion
Kinetic Model. To model the reaction course of the
dynamic kinetic resolution, the kinetics of every reaction step
were determined. The reactions taking place simultaneously
during the dynamic kinetic resolution can be distinguished
in Scheme 1. The enzymatic hydrolysis of L-phenylalanine
ethyl ester 1 is coupled with the aldehyde-catalysed race-
misation of the substrate. Possible side reactions are the
cleavage of D-phenylalanine ethyl ester 2 by the enzyme,
racemisation of the product L-phenylalanine and the base-
catalyzed chemical saponification of both D- and L-pheny-
lalanine ethyl ester. The reaction kinetics for the main
reaction steps were determined by initial rate measurements.
Racemisation. It has formerly been assumed that race-
misation of amino acid esters in the presence of aldehydes
takes place via Schiff base formation and is well described
in the literature19,20 (Scheme 2). The mechanism of the
racemisation of amino acid derivatives via Schiff base
intermediates is described by Wegman et al.20 After forma-
tion of the Schiff base the acidic proton located at the
stereogenic center is removed under alkaline conditions to
form an enolate species which has lost every optical
information.
DKR. DL-Phenylalanine ethyl ester was dissolved in 1:1
mixture of water/acetonitrile to give a 0.2 mol/L solution.
3,5-Dinitrosalicylaldehyde was dissolved in this solution to
give a concentration of 0.005 mol/L. The pH of the solution
was adjusted to 7.5 with concentrated sodium hydroxide and
controlled during the reaction. To start the reaction, 1 vol %
of Alcalase was added. Aliquots were treated as described
above. The reaction was carried out at 25 °C.
The Michaelis-Menten-like saturation kinetics reported
by Wegman et al.20 for this type of racemisation could not
be observed in our case. This is likely due to the different
reaction conditions, substrates, and catalysts. The reaction
HPLC. The concentrations of D- and L-phenylalanine
ethyl ester and D- and L-phenylalanine were determined by
HPLC-measurement using a DAICEL Crownpak CR(-)
analytical column. Samples were eluted at 10 °C with a
(19) Pugnie´re, M.; Commeyras, A.; Previero, A. Biotechnol. Lett. 1983, 5(7),
447-452.
(20) Wegman, M. A.; Hacking, M. A. P. J.; Rops, J.; Pereira, P.; van Rantwijk,
F.; Sheldon, R. A. Tetrahedron: Asymmetry 1999, 10, 1739-1750.
Vol. 10, No. 3, 2006 / Organic Process Research & Development
•
623

kinetics reported by Wegman et al. corresponded to phe-
nylglycine methyl ester and phenylglycine amide racemised
by pyridoxal and salicylaldehyde in tert-butyl alcohol. For
the racemisation of phenylalanine ethyl ester the initial
racemisation rate was linearly proportional to the initial
concentration in a concentration range from 0.03 to 0.3 mol/
L. The reaction kinetics can be described by first-order
kinetics according to:
d[LE]
dt
ν )
) k′rac[LE]
with
k′rac ) krac[cat]
(1)
krac is replaced by k′rac, corresponding to krac multiplied by
the concentration of 3,5-dinitrosalicylaldehyde, because the
catalyst concentration is assumed to be constant during the
reaction. k′rac was determined to be (3.7 ( 0.5) × 10-3 min-1
at a catalyst concentration of 0.005 mol/L. Hence krac has a
value of 0.74 min-1. The kinetics found are probably still
concordant with the mechanism and kinetics proposed by
Wegman et al. because saturation kinetics can be ap-
proximated as first-order kinetics if the concentration range
determined is below the value of the dissociation constant
Figure 1. Optical purity of the product depends on the
enantiomeric ratio E of the enzymatic ester cleavage and the
ratio kr/kS.
by first-order kinetics:
d[LE]
ν )
) k′sap[LE]
dt
with
K
diss of the pre-equlibrium. This constant is probably higher
than 0.3 mol/L in this case.
k′sap ) ksap[H2O] ) (9.0 ( 2.4) × 10-6 min-1
(3)
Enzymatic Hydrolysis. The enzymatic saponification of
L-phenylalanine ethyl ester obeyed Michaelis-Menten kinet-
ics according to:
Because the water concentration is constant, ksap is
replaced by k′sap, corresponding to ksap multiplied by the water
concentration. A comparison of the reaction rates shows that
the enzymatic cleavage is about 350-fold faster than the
chemical cleavage.
V′max[LE]
ν )
(2)
KM ) [LE]
Observation of the racemisation of L-phenylalanine in the
presence of 3,5-dinitrosalicylaldehyde showed very little
product formation after 3 days, but the HPLC peak was to
small to be analysed. Compared with the other reaction rates
determined, the rate of product racemisation was too slow
to be taken into account.
with KM ) 0.74 ( 0.2 mol/L and V′max ) (3.5 ( 0.5) ×
10-3 mol/(L‚min). No data for this or similar solvent systems
can be found in the literature although L-phenylalanine ethyl
ester is a well-known substrate for subtilisin. In aqueous
ethanol (7% ethanol, 30 °C, pH 7.5) a KM value of 0.48
mol/L was determined.21 Both KM values are in the same
order of magnitude.
The initial substrate concentration was varied between
0.03 and 1.2 mol/L. To investigate possible product inhibi-
tion, the initial reaction rate was determined at fixed substrate
concentrations (0.05, 0.07, and 0.1 mol/L) with varying
amounts of product added (0-0.12 mol/L). No substrate or
product inhibition was observed.
Side Reactions. Kinetic rate equations for all possible
side reactions were determined as well. The investigations
showed that the chemical saponification of DL-phenylalanine
ethyl ester had to be taken into account, while the reaction
rate of the product racemisation was negligible. The saponi-
fication was examined with L-phenylalanine ethyl ester as
substrate. The reaction proceeded very slowly so that the
kinetics could not be determined by initial rate measurements.
The course of the reaction was observed over 3 days via the
decrease of substrate concentration and could be described
The knowledge of the kinetic equations for every reaction
step enabled us to determine the enantiomeric ratio E, which
describes the ratio of the enzymatic conversion of both
substrate enantiomers,22 and the ratio of racemisation and
enzymatic cleavage velocity krac/kS. The knowledge of these
parameters allows an estimation of the efficiency of the
DKR.23 For an efficient dynamic kinetic resolution, the
parameters E and krac/kS have to be carefully tuned.12 To
achieve good optical purity of the product (more than 95%
ee), E has to be greater than 50 while the ratio of the kinetic
constants krac and kS should exceed at least 1 to avoid a
depletion of the reacting substrate enantiomer which is, in
this case, L-phenylalanine ethyl ester. The optical purity of
the product depends on these parameters as shown in Figure
1.
(22) Chen, C. S.; Fujimoto, Y.; Girdaukas, G.; Sih, C. J. J. Am. Chem. Soc.
1982, 104(25), 7294-7299.
(23) Kitamura, M.; Tokunaga, M.; Noyori, R. Tetrahedron 1993, 49(9), 1853-
(21) Philipp, M.; Bender, M. L. Mol. Cell. Biochem. 1983, 51, 5-32.
1860.
624
•
Vol. 10, No. 3, 2006 / Organic Process Research & Development

Table 1. Kinetic constants
The E values found for the Alcalase-catalysed cleavage
varied between 120 and 630. The E value was calculated
from the ee values of product and substrate:24
constant
value
k′rac
KM
Vmax
k′S
kDrac
kDe
(3.7 ( 0.5) × 10-3 min-1
0.74 ( 0.20 mol/L
1 - ees
ln
(3.5 ( 0.5) × 10-3 mol/(L ‚ min)
(9.0 ( 2.4) × 10-6 min-1
(1.1 ( 0.3) × 10-3 min-1
(1.2 ( 0.1) × 10-3 min-1
1 + (ees/eep)
E )
1 + ees
ln
1 + (ees/eep)
This big variance is probably due to the difficult determi-
nation of ee values higher than 98%. Anyway, the observed
E value is by far high enough for an efficient kinetic
resolution.
For determination of krac/kS, the Michaelis-Menten-
kinetics found for the enzymatic saponification of 1 was
approximated by first-order kinetics for concentrations
smaller than KM, using kS as kinetic constant. The value found
for kS was 3.2 × 10-3 min-1 which resulted in a ratio krac/kS
) 1.2 under the chosen reaction conditions. This simplifica-
tion is possible as most of the reaction has to be performed
at substrate concentrations less than 0.3 mol/L due to the
limited substrate solubility.
Kinetic Model. We next combined the kinetic equations
for all reaction steps to model the reaction course. The system
of four differential equations describing the concentration
of D- and L-phenylalanine ethyl ester and D- and L-
phenylalanine during reaction was solved using MATLAB.25
Because the model was derived for a constant pH, reactive
forms of the components and hence dissociation constants
for acidic groups are not considered. The HPLC analysis
applied allowed the determination of components 1-4
concurrently. First, experiments showed that the developed
model was not able to describe the reaction course completely
over 30 h. The reaction proceeded more slowly than predicted
by the model and terminated at about 70% conversion.
Because there was no product inhibition found, a deactivation
of one or both catalysts could be an explanation. Further
investigations showed that both catalysts lose activity when
exposed to the reaction system (data not shown). The model
was adjusted to this by introducing a new factor into the
kinetic equation for racemisation as well as enzymatic
hydrolysis. As the catalyst concentrations were assumed to
be constant, they were included in k′rac and V′max. This is
now used to describe the deactivation of the catalysts.
Assuming first-order kinetics, the deactivation can be
described as a depletion of the catalyst concentration for both
catalysts:
Figure 2. Modelled reaction course and measured points for
a reaction carried out with 0.2 mol/L substrate (0.2 mol/L, DL-
phenylalanine ethyl ester; 2.5 mol %, 3,5-dinitrosalicylaldehyde;
1 vol %, Alcalase, pH 7.5; 25 °C; acetonitrile/water, 1:1).
standard conditions was used. The constants are shown in
Table 1. The enzyme loses about 7% of its initial activity
within the first hour of reaction time, the salicylaldehyde,
about 5%. With this improved model the reaction course of
the DKR could be described completely as shown in Figure
2. The model has been verified using catalyst concentrations
within a range from 0.005 to 0.05 mol/L and substrate
concentrations within a range from 0.04 to 0.2 mol/L. The
Alcalase concentration was varied from 1 to 5 vol % (data
not shown).
The catalyst deactivation has no influence on the optical
purity of the product in this case because the enzyme loses
its activity faster than the racemisation catalyst. Hence, the
ratio krac/kS is even increasing in the course of the reaction.
Optimisation. Having developed a model that describes
the complete reaction course of the dynamic kinetic resolu-
tion, we considered an optimisation of the reaction. The
determined values for E and krac/kS show that an enhancement
of the optical purity of the product is rarely possible because
these parameters are already well adjusted (compare Figure
1). Because of the catalyst deactivation, a yield of 100% is
as well difficult to reach, but variation of the reaction
conditions (substrate and catalyst concentrations) showed that
an increase of one of the catalyst concentrations by the factor
of 5 increases the yield from 80% to 90%. In case of an
increased Alcalase concentration the reaction time is as well
shortened from 30 to 24 h. The shorter reaction time and
higher yield increase the volumetric productivity by a factor
of 1.4. For economic reasons further reactions were carried
k′rac ) krac[cat] exp(-kDract)
V′max ) Vmax[E] exp(-kDet)
(4)
(5)
The values for the deactivation constants kDrac and kDe
were calculated and optimised by least-squares fitting using
the software SCIENTIST.26 For this fitting one dataset of
measured concentrations of the reaction performed under
(24) Rakels, J. L. L.; Straathof, A. J. J.; Heijnen, J. J. Enzyme Microb. Technol.
1993, 15(12), 1051-1056.
(25) MATLAB, 6.5 R13; MathWorks: Natick, Massachusetts.
(26) SCIENTIST, 2.0.1; Micro Math Scientific Software: Salt Lake City, 1995.
Vol. 10, No. 3, 2006 / Organic Process Research & Development
•
625

Figure 3. Yield (triangles) and optical purity (squares) of the
product as function of Alcalase concentration: modelled (open
symbols) and experimental values (filled symbols).
Figure 5. Modelling of the reaction course with only one
catalyst deactivated at a time shows that the obtainable
conversion depends on the deactivation of catalysts (conditions
as in Figure 2).
out with 5-fold Alcalase-concentration (based on catalogue
prices, under standard conditions dinitrosalicylaldehyde
contributes to the catalyst costs with 97%, while Alcalase is
only about 3%27). The depletion of optical purity due to a
decrease of krac/kS by change of the Alcalase concentration
is about 6%, which corresponds very well with the theoretical
value (compare Figures 1 and 3). However, there is another
factor limiting the increase in Alcalase concentration. Besides
the loss of optical purity of the product a further increase
would not result in an appreciable increase of yield (Figure
3). Figure 3 also demonstrates a good agreement of experi-
mental and calculated values.
Investigation of the reaction rates indicated that the
increased conversion and shorter reaction time are due to a
much higher effective formation rate of L-phenylalanine ethyl
ester in the beginning of the reaction when the Alcalase
concentration is increased (Figure 4). The effective formation
rate is defined as
ethyl ester by racemisation Vr-. Because of the higher
effective formation rate with increased Alcalase concentration
the rate of enzymatic hydrolysis Venz of 1 is much faster as
well. Most of the reaction is completed after a few hours so
that the effect of the catalyst deactivation is declining.
To establish further optimisation strategies, the role of
catalyst deactivation was investigated. By simulation of the
reaction with only one of the catalysts deactivatedswhich
can easily be done having established the modelsit can be
investigated easily which one is more importantsthe deac-
tivation of the chemocatalyst or of the enzyme. The data
shown in Figure 5 prove that the deactivation of the enzyme
is in fact the limiting factor in this reaction. Without any
deactivation the theoretically reachable conversion of 100%
is reached after 48 h. With deactivation of both catalysts
only about 80% are predicted, which is reached by the
experiments described earlier. Considering which catalyst
deactivation affects the system most and therefore has to be
improved in the first place, we simulated the reaction course
with only one catalyst deactivated at a time. The data are
displayed in Figure 3 as well. When the enzyme deactivation
Veff ) Vr+ - nr-
(6)
with the formation rate of L-phenylalanine ethyl ester by
racemisation Vr+ and the formation rate of D-phenylalanine
Figure 4. Development of reactions rates of racemisation and cleavage and effective formations rate in the course of the reaction
(left: conditions as in Figure 2; right: 5-fold Alcalase concentration); note different scaling of y-axis.
626
•
Vol. 10, No. 3, 2006 / Organic Process Research & Development

calculated with the model is not shown because small
deviations between calculated and real concentrations of the
enantiomers lead to noteworthy variations in the ee value.
In this calculation recovery of the catalysts was not taken
into account.
4. Conclusions
A model describing the complete reaction course of a
dynamic kinetic resolution was developed. Kinetic investiga-
tions and modelling showed that the deactivation of catalysts
was the prominent constraint in reaching 100% conversion
with high optical purity of the product, as can be achieved
with DKR theoretically. Nevertheless, optimisation leads to
an improvement in yield of 10% and a shortening of reaction
time by a factor of 2.
Simulation of the reaction with only one catalyst deac-
tivated at a time revealed that with respect to yield and
productivity the enzyme should be the first target for further
optimisation. Further work will therefore focus on a screening
of commercially available proteases catalysing the same
reaction. If this will not be successful, development of an
enzyme with improved stability by directed evolution could
be a target. However, it should be kept in mind that for a
final judgment also the process economiessin this case the
product specific catalyst consumption of both catalystsshave
to be considered.
Modelling the reaction course in continuously operated
reactors showed that batch and PFR respectively are the best
reactor types to obtain high yield and optical purity in
dynamic kinetic resolutions.
Due to the progress in fermentation methods and genetic
engineering the method of choice for production of L-
phenylalanine is certainly a fermentation process using
microorganisms. However, for non-natural amino acids or
amino acids having the other stereochemistry a process as
described here might be the better choice. Our study
demonstrates what are the limiting factors and how kinetic
studies can be helpful for designing a better process.
Figure 6. Calculated conversion in different continuously
operated reactors (feed/initial concentrations: 0.2 mol/L, DL-
phenylalanine ethyl ester; 2.5 mol %, 3,5-dinitrosalicylaldehyde;
5 vol % Alcalase. Conditions: pH 7.5, 25 °C, 1:1 acetonitrile/
water).
is taken into account, the achievable conversion is insignifi-
cantly higher than with both catalysts deactivated, whereas
a conversion of 95% is obtainable when only the salicyla-
ldehyde is losing activity during reaction.
As expected, the optical purity of the product decreases
slightly (1% ee) if only the racemisation catalyst is subject
to catalyst deactivation because krac/kS decreases. If the
enzyme is subject to deactivation as well, the effect on the
optical product purity is negligible (<0.5% ee). As a
consequence, a continuous or batch-wise dosing of the
Alcalase could be considered. This will reduce the amount
of enzyme by a factor of 2, but as the enzyme is relatively
cheap, as described before, this was not investigated further.
Continuously Operated Reactors. With the determined
kinetic constants it was possible to simulate the reaction
course of the DKR in continuously operated reactors as well.
The concentrations calculated for steady state in an ideal
working CSTR, a three-step cascade and plug flow reactor
(PFR) allowed the determination of accomplishable yield and
optical purity. The results are shown in Figure 6. The highest
yield can be obtained in a PFR or batch reactor. In the CSTR
and the cascade the course of the graph shows a maximum.
First the yields rise with increasing retention time, but after
about 18 h the catalyst deactivation becomes an important
factor, and the yields decline. The enantiomeric excess
Acknowledgment
We thank Dr. W. Ruth and Dr. C. Fischer for their help
and useful comments concerning analytical questions. This
work was supported by Fonds der Chemischen Industrie.
Received for review December 2, 2005.
OP0502349
(27) Alcalase (Merck, Darmstadt) 57 Euros/500 mL; 3,5-dinitrosalicylaldehyde
(Aldrich, Steinheim), CAS number 2460-59-5, 150, 50 Euros/5 g.
Vol. 10, No. 3, 2006 / Organic Process Research & Development
•
627