Scale-up of a chiral resolution using cross-linked enzyme crystals

Article abstract of DOI:10.1021/op980032v

The scale-up of a biocatalytic process using a cross-linked enzyme crystal (ChiroCLEC-PC, Altus Biologics Inc., Cambridge, MA) for the resolution of a racemic mixture (R/S-sec-phenethyl acetate to R-sec-phenethyl alcohol) is reported. Reaction parameters were initially investigated at the 20-mL and 1-L scale to determine optimum reactor conditions for operation at the 100-L scale. At this scale, reproducibility trials were conducted with a single charge of 50 g of ChiroCLEC-PC for the entire nine trials. Results demonstrated that, although there was a steady decrease in reaction rate over the trials (primarily as a result of sub-optimal CLEC catalyst recovery between each trial) all nine reactions proceeded to >48% conversion, with ee_p's of 99.9% and ee_s's of 91-95% recorded. In total, 270 kg of racemate was converted into R-phenethyl alcohol and S-phenethyl acetate during the reproducibility trials using a standard stirred tank design with pH control. With further work to improve CLEC catalyst recovery between runs, the technology is suitable for industrial exploitation to compete favourably alongside alternative chiral technologies.

Full text of DOI:10.1021/op980032v

Organic Process Research & Development 1998, 2, 400−406
Scale-Up of a Chiral Resolution Using Cross-Linked Enzyme Crystals
Anne M. Collins, Christopher Maslin, and R. Julian Davies*
Industrial Research Limited, P.O. Box 31-310, Lower Hutt, New Zealand
Abstract:
reactions over organic synthetic reactions. Lalonde4 sum-
marised the main advantages in using enzymes: they are
highly regio/stereoselective, they are efficient, they work
under ambient conditions at neutral pH, they are environ-
mentally benign, a broad range of reactions are possible, and
they are ideal “green” catalysts, producing less waste and
consuming less energy.
However, on the downside, enzyme-catalysed reactions
can exhibit slow reaction rate, generally occur in higher
dilution than organic syntheses, are subject to inhibition, and
can be unstable owing to narrow operating conditions, their
downstream processing can be complex, and the bulk supply
is currently limited to hydrolases. These disadvantages,
along with the generally held view by chemists and chemical
engineers that enzymes are delicate, probably account for
the reason enzymes have yet to make ground in the chemical
process industries.
One technology to emerge which addresses the disad-
vantageous properties of enzymes is the use of cross-linked
enzyme crystals (CLEC, a registered trademark of Altus
Biologics Inc.).5 CLEC catalysts are reported to compre-
hensively address all the cited limitations of biocatalysts by
combining the desirable features of a pure catalyst with those
of immobilised biocatalysts in a stable form that is amenable
to large-scale production.6 The basis of CLEC technology
lies in the crystallisation of the protein catalyst and then
cross-linking of these highly pure microcrystals with a
bifunctional reagent such as glutaraldehyde. CLEC catalysts
are reported to maintain their structure and hence remain
active in environments that can be detrimental to enzyme
function, including multiple reaction cycles, prolonged
exposure to high temperatures, near-anhydrous organic
solvents, and aqueous-organic solvent mixtures.
This paper describes the scale-up of a biocatalytic process
using a CLEC catalyst for the resolution of a racemic mixture
and assesses whether such a technology is applicable for use
in the pharmaceutical/chemical process industries to compete
alongside alternative chiral technologies. It is an attempt to
show organic chemists and chemical engineers in the
specialised organics industry that biocatalysis can fit well
into overall chemical processes that convert material inputs
into products.
The scale-up of a biocatalytic process using a cross-linked
enzyme crystal (ChiroCLEC-PC, Altus Biologics Inc., Cam-
bridge, MA) for the resolution of a racemic mixture (R/S-sec-
phenethyl acetate to R-sec-phenethyl alcohol) is reported.
Reaction parameters were initially investigated at the 20-mL
and 1-L scale to determine optimum reactor conditions for
operation at the 100-L scale. At this scale, reproducibility trials
were conducted with a single charge of 50 g of ChiroCLEC-
PC for the entire nine trials. Results demonstrated that,
although there was a steady decrease in reaction rate over the
trials (primarily as a result of sub-optimal CLEC catalyst
recovery between each trial) all nine reactions proceeded to
>48% conversion, with eep’s of 99.9% and ees’s of 91-95%
recorded. In total, 270 kg of racemate was converted into
R-phenethyl alcohol and S-phenethyl acetate during the repro-
ducibility trials using a standard stirred tank design with pH
control. With further work to improve CLEC catalyst recovery
between runs, the technology is suitable for industrial exploita-
tion to compete favourably alongside alternative chiral tech-
nologies.
Introduction
In a recent article, Persidis1 predicted that by the year
2000 total pharmaceutical sales will exceed $US 300 billion.
About $US 100 billion of these sales will be chiral
therapeutics. The annual growth rate for sales of single
enantiomer chiral pharmaceuticals has been steady at over
20% for several years. Recently, the U.S. Food and Drug
Administration (FDA) has increased the necessity for
investigating the chirality of drug candidates by requesting
evaluation of the chiral forms of all potential new drugs
where technically possible.2,3 The reason for the upsurge in
interest in chirality is that the distinct chiral forms of a drug
can have different reactivities and pharmokinetic properties.
These events have contributed to a resurgence of interest in
developing chiral technologies that should prove to be a
lucrative area of biopharmaceutical research and commer-
cialisation.
Stinson,3 in a report on chiral drugs, stated that among
the ways to produce enantiomeric drugs, resolution of
racemic mixtures often is more economical than asymmetric
syntheses. When companies do choose enantioselective
preparations, they have frequently chosen enzyme-catalysed
In this work, we follow a structured approach to biocata-
lytic process design as proposed by Woodley and Lilly7 in
order to select the correct process on the basis of fundamental
(4) Lalonde, J. Chem. Eng. 1997, 9, 108-112.
(5) Margolin, A. L. Trends Biotechnol. 1996, 14, 223-230.
(6) Lalonde, J. Manuf. Chem. 1996, 4, 19-22.
(7) Woodley, J. M.; Lilly, M. D. Chem. Eng. (Rugby, England) 1996, 611,
28-30.
(1) Persidis, A. Nature Biotechnol. 1997, 15, 594-595.
(2) Caldwell, J. Chem. Ind. 1995, 6 March, 176-179.
(3) Stinson, S. C. Chem. Eng. News 1995, 73 (Oct 9), 44-74.
400
•
Vol. 2, No. 6, 1998 / Organic Process Research & Development
10.1021/op980032v CCC: $15.00 © 1998 American Chemical Society and Royal Society of Chemistry
Published on Web 11/04/1998

Table 1. Reactor conditions for 1-L scale runs
agitation
(rpm)
impeller
type
[ChiroCLEC-PC]
phase ratio
(initial)
run
in reactor (g L-1
)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
700
500
700
500
500
500
700
700
700
700
500
700
500
700
700
700
500
500
500
500
700
rushton
rushton
marine
marine
marine
rushton
rushton
rushton
marine
rushton
rushton
rushton
marine
marine
marine
marine
rushton
marine
rushton
marine
rushton
1
1
1
0.5
1
1
0.5
1
0.5
0.5
0.5
1
0.5
0.5
1
1
0.5
1
1
0.5
0.5
0.1
0.4
0.1
0.1
0.1
0.1
0.1
0.4
0.4
0.1
0.1
0.1
0.1
0.1
0.1
0.4
0.4
0.4
0.4
0.4
0.4
Figure 1. Hydrolysis of racemic sec-phenethyl acetate by
ChiroCLEC-PC.
characterisation of the reactants, products, reactions, and
biocatalyst. The biocatalytic process chosen is the chiral
resolution of R-sec-phenethyl alcohol from racemic R/S-sec-
phenethyl acetate using ChiroCLEC-PC (Figure 1).
Materials and Methods
Biocatalyst. ChiroCLEC-PC (a lipase from Pseudomonas
cepacia), sourced from Altus Biologics Inc. (Cambridge,
MA), was used for the work. The CLEC catalyst is supplied
in slurry form in a 10 mM Tris, 10 mM CaCl2, pH 6.0 buffer.
Inherent Activity Assay. CLEC esterase activity (units
per mg of ChiroCLEC-PC) was measured as the rate of
hydrolysis of a 6% (v/v) glycerol triacetate (BDH no. 28456
4H, >99% by GLC) solution, using a Mettler DC40 RC
Memo Titrator. One unit is the activity of ChiroCLEC-PC
that hydrolyses 1 µmol of glycerol triacetate per minute, at
pH 6.0 and 25 °C.
duplicates. The reactor conditions for these experiments are
shown in Table 1.
CLEC Characterisation: Stability Trials. To 1.5-mL
plastic Eppendorf tubes were added 475 µL of a buffer (pH
3.0 (10 mM citric acid), 4.0, 4.5, 5.0, 5.5 (10 mM sodium
acetate), 6.0, 7.0, and 8.0 (10 mM phosphate)) and 25 µL of
ChiroCLEC-PC (90.9 g L-1), and these solutions were
incubated at 30, 40, 50, or 70 °C. Ten-microliter samples
were taken at intervals for inherent activity determination.
Solvent Effects. Two milliliters of solvent (methanol,
ethanol, acetone, isopropyl alcohol, sec-phenethyl acetate,
and water (control)) and 1 mL of ChiroCLEC-PC catalyst
(10 g L-1) were put into a sealed glass test tube for 1 h with
regular vortexing. The solvent/CLEC catalyst mixtures were
filtered (2.5 cm diameter, No. 1 Whatman filter paper) under
vacuum, and the CLEC cake was washed with reverse
osmosis (RO) water (10 mL) followed by 10 mM phosphate
buffer, pH 6.0 (10 mL). Note: Phenethyl acetate-suspended
ChiroCLEC-PC was washed only with 10 mM phosphate
buffer, pH 6.0 (30 mL). The washed CLEC cake was
resuspended (10 mM phosphate buffer, pH 6.0) to 1 mL for
activity determination.
Two-Liquid-Phase Reactions20-mL Scale. Fourteen
milliliters of 0.2 M phosphate buffer, pH 6.0, and 6 mL of
R/S-sec-phenethyl acetate were agitated in a 50-mL glass,
jacketed vessel (attached to a circulating waterbath at 30 °C)
for 1 h, and then 1 g L-1 (aqueous basis) ChiroCLEC-PC
was added. The reaction was allowed to run for 24 h, with
samples being taken periodically to monitor the reaction
progress.
Experimental Designs1-L Scale. A series of experi-
ments were planned using a factorial design to allow us to
determine the effects of all the major parameters while
reducing the number of experiments required. Parameters
varied were stirrer speed, impeller type, and CLEC loading.
A total of 21 experiments were required, including five
Two-Liquid-Phase Reactions1-L Scale. These reac-
tions were carried out in 2-L glass bioreactors (New
Brunswick Multigen System), with a working volume of 1
L. Agitation was achieved by a magnetically coupled
agitator with either a single rushton turbine or a marine
impeller. pH was controlled to 6.0 by the addition of 10 M
NaOH. The agitation rate was set to at least 500 rpm in
order to ensure complete homogeneity of the phases. The
aqueous (RO water) and organic (sec-phenethyl acetate)
phases were added to the reactor and allowed to equilibrate
(under agitation) for 30 min at 30 °C. ChiroCLEC-PC
(between 0.25 and 1 g L-1, reactor basis) was added to the
reactor. The weight of NaOH added was monitored over a
24-h period. One-milliliter samples were taken at 0, 1, 2,
3, 4, 6, 8, and 24 h for GC analysis.
Conditions for all the 1-L experiments were planned, and
results were analysed using statistical software, EChip v6.01
from Echip Inc. The design was a simple linear screen, with
the analysis concentrating on interaction effects.
CLEC Recoverys1-L Scale. Reactor contents were
filtered through No. 42 Whatman filter paper in a Buchner
funnel. The CLEC catalyst was rinsed using the desired
solvent solution (generally 70% IPA in RO water). This
wash was repeated as required and was followed by one or
more washes using a 10 mM phosphate buffer, pH 6.0. The
CLEC catalyst was removed from the filter and resuspended
in 5 mL of 10 mM phosphate buffer, pH 6.0 for storage.
The ChiroCLEC-PC was tested for activity using the inherent
activity assay.
Two-Liquid-Phase Reactions100-L Scale. These reac-
tions were carried out in a 250-L SS316 bioreactor (IRL in-
house design), with a working volume of 100 L. Agitation
was achieved by a bottom entry agitator with a single 45°
pitched six-bladed impeller. pH was controlled to 6.0 by
Vol. 2, No. 6, 1998 / Organic Process Research & Development
•
401

the addition of 18 M NaOH. The agitation rate was 200
rpm in order to ensure complete homogeneity of the phases.
The aqueous (RO water) and organic (sec-phenethyl acetate)
phases were added to the reactor with a phase ratio of 0.3
and allowed to equilibrate (under agitation) for 30 min at
30 °C. ChiroCLEC-PC (initially at 0.5 g L-1, reactor basis)
was added to the reactor. The weight of NaOH added was
monitored over the course of the reaction, and samples were
taken at intervals for GC analysis.
pH Control. The reaction produces acetic acid, which
required neutralisation. This was done using ∼18 M NaOH
dosed in via a peristaltic pump that was controlled using a
software package (FixDMACS, Intellution Software) and a
PLC (Modicon Compact). For the first hour of each run,
pH deadband was set at (1 pH, dropping to (0.2 pH for
the next phase, with a final reduction to (0.1 pH for the
final 30-60 min of each run.
Data Logging. Throughout the 100-L reactions, tem-
perature, pH, and mass of NaOH consumed were monitored
using the data logging capabilities of the reactor control
system (FixDMACS SCADA software). Data were logged
directly to the hard disk of the machine before being
transferred to a personal computer for analysis and manipu-
lation.
Sampling from a Two-Liquid-Phase Reactions1-L
and 100-L Scale. Samples of ∼1.5 mL were placed in 1.5-
mL glass reaction vials with conical bases and sealed with
a PTFE lined lid. These were centrifuged (at room temper-
ature) at 4000 rpm for 5 min. Three phases separated:
organic, aqueous, and solid CLEC catalyst (at low catalyst
concentrations, the crystals were frequently found adhered
to the liquid-liquid interface). A 100-µL sample of the
liquid was taken (with an aqueous:organic ratio equivalent
to the phase ratio in the vessel). This was extracted twice
with 1 mL of butyl acetate, using octanol internal standard,
and was analysed by chiral GC.
Chiral GC. A J&W 30-m × 0.25-mm × 0.25-µm
CyclodexB column was used. Hydrogen carrier was at 85
kPa, split 50 mL/min, purge 2.5 mL/min. Sample was 10
mg/mL in methanol. Manual injection of 0.1 µL was done.
Temperature program was as follows: initial, 80 °C; gradient
rate, 1 °C/min; final temperature, 99 °C.
Monitoring Reaction. Reaction progress was monitored
by viewing the mass of NaOH consumed. As the reaction
neared completion, consumption of NaOH (used to neutralise
the acetic acid produced in the reaction) dropped to very
low levels. The reaction was considered complete after a
period of 60 min with no addition of NaOH. Samples were
taken at intervals to allow us to compare the theoretical
conversion as calculated from NaOH consumption with
actual conversion at a later date.
Catalyst Recoverys100-L Scale. Reactor contents were
filtered through a 27-µm filter cloth (27-µm nylon mono-
mesh, Filter Corp., Auckland, New Zealand) (runs 1-3) or
a 5-µm filter cloth (heavy duty nylon, code BCNY-HD005-
5, NZ Filter Specialists Ltd., Auckland, New Zealand) (runs
4-9). The CLEC catalyst was rinsed by resuspending the
catalyst in 10 L of RO water for 10-30 min, followed by
filtration. This was repeated four times and was followed
by a wash using 1 L of 10 mM phosphate buffer, pH 6.0.
The CLEC catalyst was removed from the filter and
resuspended in 500 mL of 10 mM phosphate buffer, pH 6.0,
for storage. The ChiroCLEC-PC was tested for activity using
the inherent activity assay. The CLEC catalyst recovered
from each run was used as the catalyst in the subsequent
run. A small sample of the recovered CLEC catalyst from
each run was taken and dried for 24 h at 80 °C. The dry
weight of this sample was used to estimate the total dry mass
of ChiroCLEC-PC present in the subsequent run.
Experimental Rationale
Solvent Effects. Initial work concentrated on determining
the effects of solvents and the substrate on the activity of
the ChiroCLEC-PC. Methods for use of CLEC catalysts
recommend that the catalyst be washed in a suitable solvent
to improve the catalyst life. To this end, a number of
solvents selected from those used in similar systems in the
literature were examined for possible detrimental effects on
the catalyst activity.
Reaction Characteristics. ChiroCLEC-PC (Altus Bio-
logics Inc) is used to preferentially hydrolyze the R-phenethyl
acetate to the R-phenethyl alcohol. There is interest in this
fine bulk chemical for synthesis, and it is presently produced
via asymmetric reduction of acetophenone.
At the small scale, CLEC was found to confer the greatest
stability and activity for this reaction at pH 6.0 and 30 °C
(results not shown). These conditions were used for all
experiments referred to here.
Phenethyl acetate is poorly water soluble (approximately
1.5 mM), and the product phenethyl alcohol partitions
between the organic (phenethyl acetate) and aqueous phases
in the ratio 9:1. Interestingly, phenethyl acetate and phen-
ethyl alcohols have specific densities of 1.028 and 1.02,
respectively, and there is high interfacial tension between
phenethyl acetate and the aqueous phase. Therefore, to
achieve a high substrate concentration in the reactor, the
phenethyl acetate will be present as a discrete second liquid
phase.
While the ChiroCLEC-PC appeared to act as a classic
lipase, with the CLEC catalyst adhering to the interface, it
is unlikely that the reaction was occurring at the interface
on a molecular level. Lalonde et al.8 showed that there is
no interfacial activation with ChiroCLEC-CR by studying
systems with and without an interface and observing that
the reaction rate in biphasic systems was not appreciably
different from the reaction rate in single-phase systems. It
is thought that this is because the lipase is held rigidly in
the CLEC structure and is not able to align itself with the
interface.
While the reaction may not be occurring at the interface
on the molecular level, because of the propensity of the
ChiroCLEC-PC to physically locate itself at the interface
and in order to maximize mass transfer between the phases,
experimental conditions were chosen that facilitated high
interfacial area.
(8) Lalonde, J.; et al. J. Am. Chem. Soc. 1995, 117, 6845-6852.
402
•
Vol. 2, No. 6, 1998 / Organic Process Research & Development

Acetic acid is a byproduct of this reaction, and therefore
pH has to be controlled to maintain optimum activity of the
ChiroCLEC-PC. A stirred tank reactor was selected for this
reaction as it provides a high interfacial area between the
liquid phases and also allows easy control of pH. It is worth
noting that the stirred tank reactor is the type of reactor most
commonly found in process plants and a type of equipment
that most engineers are familiar with.
A reaction with a phase ratio (volume fraction organic)
of 0.3 will produce approximately 1 mol L-1 of acetic acid.
Because of this, it was impractical to buffer the system
(aqueous phase), and instead alkali addition was used as a
means to control the pH. This was also an easy and accurate
method to follow the rate of reaction. The disadvantage of
this method is that, on a large scale, large quantities of alkali
are required. Therefore, a high molarity sodium hydroxide
(18 M) was used to minimise the increase in volume of the
reactor contents.
Reactor Operation. By maximizing the liquid-liquid
interfacial area, the reaction rate is unlikely to be affected
by diffusion rate limitations. The agitation rate, the phase
ratio, and the impeller type (type of mixing) will affect the
interfacial area. Increasing the liquid-liquid interface by
increasing the agitation rate and using a higher shear impeller
type will increase the shear, a situation that is detrimental
to traditional biocatalysts (above a certain level).
Figure 2. Effects on ChiroCLEC-PC activity after mixing for
1 h in various solvents.
To operate an efficient process, one needs to maximize
the product produced per unit time and biocatalyst mass. For
an optimal system, the catalyst concentration needs to be
just sufficient to complete the reaction in the required time
frame. Use of more catalyst than the minimum would be
economically undesirable, while the desired time frame will
be highly dependent on the cost structure of the facilities
being used.
These four reaction variablessimpeller type, phase ratio,
biocatalyst concentration, and agitation rateswere identified
as parameters to be further investigated. A full factorial
experimental design was carried out to investigate the effect
of these parameters and their interactions on reaction
performance.
Figure 3. Enantiomeric analysis of a 20-mL scale ChiroCLEC-
PC-catalysed R/S-sec-phenethyl acetate hydrolysis reaction.
(i) CLEC concentration has a positive effect on reaction
rate and a negative effect on specific initial activity (both at
the 5% level).
(ii) Phase ratio has a positive effect on reaction duration
(at the 1% level).
Results
Effect of Solvents on Catalyst Activity. The effect of
mixing the ChiroCLEC-PC in various organic solvents was
quite severe even after 1 h (Figure 2). All the common
solvents, which were originally intended to be used for
washing the CLEC catalyst between runs, decreased the
inherent activity by >50%. However, the drop in activity
in the presence of the reaction substrate, phenethyl acetate,
was within experimental error of the tests.
Two-Liquid-Phase Reaction Trials20-mL Scale. Fig-
ure 3 shows the reaction profile of a ChiroCLEC-PC
resolution of racemic sec-phenethyl acetate on the 20-mL
scale. Chiral GC results showed that S-phenethyl acetate
was unaffected by ChiroCLEC-PC activity and no S-
phenethyl alcohol was produced.
(iii) [rpm × phase ratio] and impeller type both have a
positive effect on the initial activity.
The qualitative effects were determined by multivariate
analysis using a statistical software package. The most
significant effect was the effect of phase ratio on the time
taken for completion of the reaction. This was expected, as
the greater the amount of substrate, the longer it will take to
completely react at any given rate. The ChiroCLEC-PC
concentration has a negative effect on the specific initial
activity (Figure 4) but a positive effect on the overall reaction
rate. The initial activity was significantly affected by the
agitation rate and phase ratio interaction (Figure 5), and
higher reaction rates were also found with the rushton
impeller, a higher shear impeller than the marine impeller.
It appears that the CLEC catalyst is not detrimentally affected
by high shear in the reactor, with the rpm and impeller type
having no statistically significant effects on the inherent
Two-Phase Reactions1-L Scale. The results from the
21 small-scale (1 L) reactions are shown in Table 2, while
the qualitative effects of the reactor parameters are as follows:
Vol. 2, No. 6, 1998 / Organic Process Research & Development
•
403

Table 2. Results from 1-L scale experiments
initial activity
specific activity
(mmol min-1
NaOH basis
)
(mmol min-1 g-1
NaOH basis
)
time for 48%
conversion (h)
E at end
of reaction
rate to 48%
(mmol/h)
run
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
11.7
6.6
10.1
5.5
11.67
6.59
10.12
11.00
7.91
18.00
28.36
12.67
29.00
8.00
21.34
16.00
9.00
15.00
6.55
9.00
8.40
7.67
8.62
10.60
12.46
1.00
24.00
1.15
2.62
4.35
1.23
2.08
5.46
18.33
3.35
1.69
1.46
2.58
2.54
1.65
7.46
23.62
6.81
7.48
7.68
18.69
1256
2001
2227
1821
1072
1635
993
11541
5476
1695
1532
2095
9717
9348
976
1906
2389
4105
4371
2063
3334
621.2
103.5
540.2
237.1
142.8
505.0
298.7
455.1
135.6
185.4
367.6
425.5
240.8
244.6
376.5
333.1
105.2
364.9
332.2
323.5
132.9
7.9
18.0
14.2
12.7
14.5
4.0
10.7
16.0
4.5
7.5
6.5
9.0
4.2
7.7
8.6
5.3
6.2
Figure 5. Initial activity as a function of phase ratio and
impeller speed.
Figure 4. Specific initial activity as a function of CLEC catalyst
concentration and impeller speed.
was chosen in conjunction with the phase ratio to fit the
economic constraints. Agitation at 200 rpm on the 100-L
scale produced a high level of agitation in the reactor,
arguably even greater than that used in the 1-L runs. The
45° pitched blade impeller was the closest in type to the
rushton impeller that was available for the 100-L reactor.
The temperature of 30 °C and pH 6.0 were chosen from the
optima for CLEC activity and stability as determined at the
1-mL scale (results not shown).
Pilot Plant Trials and Reaction Reproducibility. The
identification of the significant reactor parameters allowed
this reaction to be scaled-up for the pilot plant (100-L
working volume). The reaction parameters were chosen to
ensure that the process took approximately 18 h, whilst
activity of the ChiroCLEC-PC after the completion of the
trial. This confirms the results presented by Lalonde et al.9
Although there were no unexpected results, this exercise
highlighted the relevant significance of parameters in this
system and their interactions. The findings from this work
were used to define the reactor conditions for scaling-up to
the 100-L scale.
100-L Experimental Conditions. Phase ratio was de-
cided as 0.3 mainly for economic reasons. Any phase ratio
between 0.1 and 0.4 could be chosen, depending on the
required economics. A catalyst concentration of 0.5 g L-1
(9) Lalonde, J.; et al. Proceedings of Industrial Biocatalysis: InBio 96; Spring
Innovations Ltd.: Bramhall, England, 1996.
404
•
Vol. 2, No. 6, 1998 / Organic Process Research & Development

Figure 8. 100-L reproducibility trialssenantiomeric excess and
conversion.
Figure 6. 100-L reproducibility trialsshydrolysis of sec-
phenethyl acetate using ChiroCLEC-PC.
activities as the runs proceed to nine. It can be clearly
observed that there is a steady decrease in reaction rate over
the nine runs (Figure 7). This decrease in activity by a factor
of approximately 10 appears to be due to a physical loss of
catalyst (from 50 to 12.5 g) and a drop in catalyst activity
(as indicated by the specific initial activity) by a factor of
2.5 (from 0.96 to 0.37 mol h-1 g-1).
The extent of the loss of CLEC catalyst during the
recovery process was not realised until after run 3 was
completed due to delays in obtaining the dry weights of
samples. Thereafter, the original 27-µm filter was used in
conjunction with a similar 5-µm cloth.
While the majority of the catalyst loss occurred during
the first three runs, some loss was still evident after run 4.
It was initially thought that the high agitation rate of 200
rpm in the reactor may have broken up the ChiroCLEC-PC
into smaller particles which were then lost during filtration,
but literature indicates that this is unlikely to have been the
case.10 Subsequent contact with Altus revealed that the batch
of ChiroCLEC-PC used had an average particle size of 20-
µm and a standard deviation of 6.67 µm.11 This indicates
that it is certainly possible that the smaller particles were
lost during the recovery process in the first three runs. The
combination of reducing the pore size of the filter cloth and
the fact that the majority of the CLEC catalyst remaining
would tend to be at the larger end of the size distribution
would lead to the observed trend of sharply decreasing mass
of catalyst over the first few runs, followed by a sustained
but lower loss over the remaining runs. It is apparent that
the pore size of the filter cloth was still too coarse to retain
all the catalyst. In future work, it was decided to use an
inert material added with the CLEC catalyst to improve both
the filtration rate and the overall recovery of the catalyst.
Even after nine runs with only a quarter of the original
CLEC catalyst present (38 g lost), the reaction proceeded to
completion (albeit for a longer duration as the runs stacked
up), with no adverse effect on the quality of the reaction, as
shown in Figure 8.
Figure 7. 100-L reproducibility trialssdrop in initial activity
and CLEC catalyst.
maintaining productivity (maximizing product formed per
unit time and per unit mass of catalyst). The reactor
conditions are outlined in the Materials and Methods Section.
During run 1, the reaction reached 48.8% conversion in
18 h, with enantiomeric excess of the product and substrate
of 99.9% and 94.8%, respectively. The initial activity of
the reaction was 48.5 mol h-1, and the specific initial activity
was 16.2 mmol min-1 g-1.
Run 1 demonstrated that the reaction could run efficiently
on the pilot scale with high enantioselectivity. Runs 2-9
were conducted in a manner simulating a production process
with recovery of the ChiroCLEC-PC for use in the next
identical production run. No fresh top-up of catalyst was
performed, so that the effects of processing (such as catalyst
recovery and effects due to repeated contact with solvent)
on the reaction rate and enantioselectivity could be assessed.
The ChiroCLEC-PC from these pilot-scale runs was
recovered using filtration. After recovery, the CLEC catalyst
was resuspended in 10 mM phosphate buffer, with 1 mM
CaCl2 at pH 6.0, to a total volume of 500 mL. A 1-mL
homogeneous sample was taken to determine the CLEC
catalyst dry weight. This process was repeated for eight
reactor trials, and at no time was any fresh catalyst added.
Figures 6 and 7 respectively are graphical representations
of the reaction profiles and trend in the drop of specific initial
(10) Lalonde, J.; et al. Proceedings of Chiral USA ‘97 USA; Boston, MA, May
1997; Spring Innovations Ltd.: 1997.
(11) Baust-Timpson, M., Altus Biologics. Personal communication, 1998.
Vol. 2, No. 6, 1998 / Organic Process Research & Development
•
405

Since this experimental work was completed, improve-
ments have been made in the recovery and reuse of CLEC
catalysts. With ChiroCLEC-PA, recovery and reuse has been
proven for 500 cycles without any measurable material or
activity loss.10
Enantiomeric excess of the product is 99.9%, and that of
the substrate is between 91 and 95% (for runs 1-8). All
reactions (apart from run 9) were allowed to run to
approximately 48% conversion. Enantioselectivity (E values)
were greater than 104 for all reaction runs.
washing, for the next substrate charge. In this manner, no
handling of small CLEC catalyst quantities is required
between runs, minimising catalyst losses.
Until CLEC catalyst losses are minimised by this (or
other) means, the effect of other processing activities and
conditions on the activity of the CLEC catalyst cannot be
adequately addressed. Figure 2 demonstrates that pure
phenethyl acetate had no effect on the ChiroCLEC-PC
activity after 1 h. This needs to be repeated for a much
longer duration, and it also cannot be discounted that the
phenethyl alcohol product is affecting ChiroCLEC-PC activ-
ity.
Discussion
The key issues to consider with regard to the suitability
of this technology for industrial-scale application are, in our
view, the following:
(1) What is the cause of the decline in activity as runs
proceed through the cycles and can it be overcome?
(2) Is the quality of the reaction affected by the recycling
of the catalyst?
(2) Quality of Reaction. Even though there was a 62%
loss in specific activity and a 75% physical loss in catalyst
mass, the quality of the reaction was unaffected, with
enantiomeric excess for product of >99.9% recorded for all
nine runs. Enantiomeric excess for substrate is lower, at 91-
95%, as runs were stopped before reaction completion owing
to time constraints.
(3) Given the ChiroCLEC-PC charge, is the yield of
enantiomers acceptable and can they be manufactured for
an acceptable cost?
At this juncture, only some of these questions can be
answered.
(3) Yield and Cost. Fifty grams of ChiroCLEC-PC was
used to convert 270 kg of racemate over the nine runs.
Without catalyst loss, it is estimated that this could be further
improved to at least 20 cycles. Undoubtedly, there is a limit
to the number of cycles one could subject ChiroCLEC-PC
to under these operating conditions.
If 20 cycles were possible, each with a 30-kg racemate
charge, then this would produce 223 kg of chiral R-sec-
phenethyl alcohol. Fifty grams of ChiroCLEC-PC costs $US
5000 at $100 g-1 (Altus Biologics, personal communication).
This ChiroCLEC-PC kinetic resolution process is relatively
simple to operate and would not require a large expense in
capital equipment. The other major operating costs would
be the substrate cost (600 kg × $US 15 kg-1 ) $US 9000)
and utility costs during distillation to separate the S-acetate
from the R-alcohol. Depending on the production scale and
market demand, the value of the product should be between
$US 100 and 1000 kg-1, which would indicate that this
process is economically feasible.
(1) Decline in Catalyst Activity. Run 1 demonstrated
48.8% conversion in 18 h with an initial specific activity of
approximately 1 mol h-1 g-1 ChiroCLEC-PC. By run 9,
this had decreased to 0.37 mol h-1 g-1, or 38% residual
activity. Over these runs, it was observed that 75% of the
CLEC catalyst was lost. It is a reasonable assumption that
this loss was mainly of the smaller CLEC particles rather
than the larger particles breaking through the filter. These
smaller particles probably exhibit higher specific activity
owing to a higher surface area-to-volume ratio.
Further repeat runs are required which retain 100%
ChiroCLEC-PC recovery between cycles. A stirred tank
design incorporating a 0.2-µm flat sheet filter at the outlet
would be an ideal CLEC reactor. The inclusion of an inert
filter aid would be necessary to facilitate filtration of the
products from the reactor at the completion of each batch.
The CLEC catalyst would remain in the reactor amongst the
inert filter aid cake caught on the 0.2-µm filter ready, after
Received for review April 6, 1998.
OP980032V
406
•
Vol. 2, No. 6, 1998 / Organic Process Research & Development