Development of a continuously operating process for the enantioselective synthesis of a b-amino acid ester via a solvent-free chemoenzymatic reaction sequence

Article abstract of DOI:10.1002/adsc.201300236

A sequential, chemoenzymatic process for a continuously operating production of the chiral β-amino acid ester ethyl (S)-3-(benzylamino)- butanoate was developed. The reactor set-up combined a plug-flow reactor for the thermal aza-Michael addition of benzy

Full text of DOI:10.1002/adsc.201300236

UPDATES
DOI: 10.1002/adsc.201300236
Development of a Continuously Operating Process for the
Enantioselective Synthesis of a b-Amino Acid Ester via a
Solvent-Free Chemoenzymatic Reaction Sequence
a
b
b,c
a
Simon Strompen, Markus Weiß, Harald Grçger, Lutz Hilterhaus,
a,
and Andreas Liese *
a
Hamburg University of Technology, Institute of Technical Biocatalysis, Denickestr. 15, 21073 Hamburg, Germany
Fax : (+49)-40-42878-2127; phone: (+49)-40-42878-3018; e-mail: liese@tuhh.de
University of Erlangen-Nꢀrnberg, Department of Chemistry and Pharmacy, 91052 Erlangen, Germany
Present address: Bielefeld University, Faculty of Chemistry, Universitꢁtsstr. 25, 33615 Bielefeld, Germany
harald.groeger@uni bielefeld.de
b
c
Received: April 9, 2013; Published online: August 9, 2013
Supporting information for this article is available on the WWW under http://dx.doi.org/10.1002/adsc.201300236.
À3
Abstract: A sequential, chemoenzymatic process
for a continuously operating production of the
chiral b-amino acid ester ethyl (S)-3-(benzylamino)-
butanoate was developed. The reactor set-up com-
bined a plug-flow reactor for the thermal aza-Mi-
chael addition of benzylamine to trans-ethyl croto-
nate coupled to a subsequent packed-bed reactor
for the lipase (Novozym 435)-catalyzed kinetic res-
olution of the racemic intermediate product, which
was formed in the initial step. The coupled reactors
were operated continuously for a time period of 4
days without significant loss of enzyme activity. The
target b-amino acid ester was obtained with 92%
conversion in the plug-flow reactor and 59% con-
version in the packed bed reactor at high enantio-
density [kg m ]; 1 =density of carrier material
p
À3
À1
À1
[kg m ]; Q=productivity [kgkg
nel radius of curvature [cm]; ttn=total turnover
number [molmol ]; T=temperature [8C]; t=resi-
dence time [h]; u =fluid velocity [cm min ]; n˙ =
volumetric flow rate [mL min ]; n =flow rate of
substance [mmol min ]; V=volume [mL]; c =
mole fraction of compound i; C =conversion of
compound i; STY=space-time yield [kgL d ].
d ]; R=chan-
N435
À1
À1
f
À1
i
À1
i
i
À1 À1
Keywords: b-amino acids; biocatalysis; chemoenzy-
matic reaction sequence; continuously operating
process; kinetic resolution; solvent-free conditions
meric excess of >98%. A space-time yield of
À1 À1
0
.4 kgL d was calculated for the total reactor
À1 À1
system and 1.8 kgL d
based solely on the
Introduction
volume of the packed bed reactor. A total turnover
number of 158,000 was calculated for the biocata-
lyst assuming the same deactivation rate as ob-
served in batch experiments. The continuously oper-
ating, solvent-free process thus represents an effi-
cient method for the enantioselective production of
a value added (S)-b-amino acid ester starting from
cheap substrates.
In order to improve the industrial applicability of
(bio)catalytic processes it is generally desirable to op-
erate at high substrate concentrations. The use of sol-
vent-free reactions as the extreme is therefore espe-
cially interesting. Higher yields and reaction rates can
often be achieved, while energy costs and waste for-
mation are reduced. Additionally, the necessity for
smaller reactor sizes leads to decreased capital invest-
ments. A number of solvent-free biocatalytic reactions
including, e.g., polymerizations and esterifications
[
1]
Abbreviations: CALB=Candida antarctica lipase
B; CSTR=continuously stirred tank reactor; d =
i
inner diameter [mm]; ee=enantiomeric excess; e=
À1
[2]
porosity; kcat,obs =apparent turnover number [h ];
have been described in the literature up to date.
À1
kdeact =enzyme deactivation constant [h ]; L=
However, only rather few examples in solvent-free
asymmetric catalysis exist using either chemo- or bio-
length of reactor [m]; m¯ =average mass of a single
p
[
3]
particle [g]; N435=Novozym 435 (Candida antarcti-
ca lipase B immobilized on acrylic resin); PBR=
packed-bed reactor; PFR=plug-flow reactor;
catalysts.
With regard to reactor design, continuously operat-
ing processes represent an attractive approach applied
in the chemical and pharmaceutical industry. As
PTFE=polytetrafluoroethylene; 1 =bulk particle
bp
Adv. Synth. Catal. 2013, 355, 2391 – 2399
ꢂ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
2391

Simon Strompen et al.
UPDATES
Scheme 1. Solvent-free chemoenzymatic reaction sequence for the synthesis of b-amino acid ester (S)-3.
major advantages one can see an increased level of Results and Discussion
control that facilitates scale-up, the possibility to
achieve high space-time yields by making efficient use Theoretical Aspects
of substrates and catalysts, improved safety aspects
and a significantly reduced physical space require- The aza-Michael addition of 1 and 2 has been found
[
4]
ment as opposed to conventional batch reactors.
previously to yield rac-3 as the main product
[7]
Additionally, a decreased labor input of skilled work- (Scheme 1). However, in a slow successive side reac-
ers is required in case of stable processes with long tion, the aminolysis of rac-3 to form rac-4 was ob-
operation times. The stability of the process, however, served. While a continuous stirred tank reactor
is often limited by catalyst deactivation or wash-out, (CSTR) would favor the formation of the final prod-
contamination in the case of bioprocesses, or mechan- uct in racemic form (rac-4) and thus lead to decreased
ical failure of pumps, valves, membranes or pipes (de- yields of rac-3, maximal formation of the intermediate
pending on the specific process, e.g., caused by pre- product rac-3 can be achieved using a plug-flow reac-
cipitate formation, membrane fouling etc.).
tor (PFR). Similarly, the lipase (Novozym 435)-cata-
In an attempt to classify (continuously operating) lyzed aminolysis of rac-3 may be carried out continu-
catalytic reactions in the field of applied biocatalysis, ously in a CSTR or a PFR in the form of a packed-
the consideration of the type of biological principles bed reactor (PBR). For kinetic reasons, the latter re-
[
5]
of cell metabolism has been suggested as a concept.
actor type is favored in order to achieve maximal
Hereafter, coupled reactions carried out in continuous yields of the desired chiral product (S)-3 with high
flow are termed fourth generation processes. Such
processes mimic the concept of cell metabolism in
which a constant flux of nutrients in a continuous
multireaction network is required for the cell to stay
alive. The compatibility of all (chemo)enzymatic steps
involved is the major challenge in the development of
new cascade processes, which is most probably why
relatively few examples of fourth generation process-
es can be found in the literature.
This article describes the development of such
a multistep process for the production of ethyl (S)-3-
(
benzylamino)butanoate [(S)-3] that combines the
above-mentioned advantages of a continuously oper-
ating reactor set-up and solvent-free reactions. The
reaction sequence presented initially by Weiß and
[6]
Grçger comprises a thermal aza-Michael addition of
benzylamine (1) and trans-ethyl crotonate (2), and
a subsequent Candida antarctica lipase B (CALB)-
catalyzed kinetic resolution of the racemic intermedi-
ate b-amino acid ester via aminolysis (Scheme 1). A
kinetic investigation of both individual steps has been
published previously. An environmental assessment
of the batch process including downstream processing
to produce the free b-amino acid revealed the effi-
ciency of the process with regard to raw material con-
Figure 1. Simulation of enantiomeric excess as a function of
conversion in the CALB (Novozym 435)-catalyzed kinetic
resolution of rac-3 with 1 in a PBR and CSTR reactor. The
simulation was based on the mass balance describing each
[
7]
[10]
[7]
reactor and the kinetic model. Initial concentrations for
each compound were defined as obtained from an aza-Mi-
chael addition carried out without prior purification of inter-
À1
À1
mediates. [1] =2.15 mmolg , [(R)-3] =1.50 mmolg , [(S)-
0
0
À1
À1
sumption, as well as environmental and safety 3] =1.50 mmolg ,
[(R)-4] =0.69 mmolg ,
[(S)-4] =
0
0
0
[8]
À1
issues.
0.069 mmolg .
2392
asc.wiley-vch.de
ꢂ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2013, 355, 2391 – 2399

Development of a Continuously Operating Process for the Enantioselective Synthesis of a b-Amino Acid Ester
Consequently, in a PFR the concentration changes
with the length of the reactor instead of time as in
a batch reactor. In order to verify if the assumption of
an ideal reactor is true for envisioned reactor geome-
tries, the conversion in an ideally mixed batch reactor
was compared to the conversion obtained experimen-
tally in a tube reactor (Figure 2a). A significantly de-
creased conversion was observed for the tube reactor
as compared to the batch reactor indicating non-ideal
flow characteristics.
This effect may be attributed to the very low flow
velocities used here in order to allow for long reaction
times and thereby sufficiently high degrees of conver-
Scheme 2. Coupled reactor set-up for the continuously oper-
ating chemoenzymatic production of ethyl (S)-3-(benzylami-
no)butanoate [(S)-3]. C1, C2: substrate containers. C3:
product container. M1: mixing T-piece. R1: tube reactor.
R2: packed-bed reactor. Optimized parameters for tempera-
ture T, reactor volume V and residence time t of the cou-
pled reactor are specified in Table 2.
[
9]
enantiomeric excess. In a CSTR, the continuous op-
eration at high conversion of the favored enantiomer
(
R)-3 necessarily leads to operation at a high resulting
concentration of the unfavored enantiomer (S)-3, thus
kinetically favoring its conversion and causing de-
creased yields of (S)-3 as remaining substrate
(
Scheme 1). The effect becomes obvious when plot-
ting the enantiomeric excess as a function of conver-
sion for both PBR and CSTR as shown in Figure 1
using the respective equations describing the mass
[
10]
balance for each reactor type
rameters determined previously (see the Supporting
Information and Ref. ).
and the kinetic pa-
[
7]
Taking these aspects into account, a coupled reac-
tor design comprising a tube reactor for the aza-Mi-
chael addition and a packed bed for the subsequent
lipase (Novozym 435)-catalyzed aminolysis was de-
vised as shown in Scheme 2.
Continuously Operating Process in a Tube Reactor
for Aza-Michael Addition
Figure 2. Aza-Michael addition in a plug flow reactor car-
ried out with a 1.7:1 initial molar ratio of 1 and 2. (a) Pre-
diction of conversion in ideal reactor based on kinetic
With the kinetic and thermodynamic data of the aza-
Michael addition from batch experiments in hand,
a transfer of the system to a continuously operated
plug-flow reactor can be realized. In an ideal PFR
back-mixing or turbulent flow profiles do not occur.
The reaction time in a batch process and the resi-
dence time t in a PFR are used analogously. In an
[7]
À2
À1 À1
model
(–) with k
=14.1Æ0.6·10 gmmol
h
and
1
,808C
À3
À1 À1
k2,808C =2.1Æ0.3·10 gmmol
h
and experimental results
(
~). Visual aid (–). Tube reactor (PTFE), V=41.6 mL, L=
8
2.7 m, d =0.8 mm, T=808C. (b) Conversion as a function
i
of flow rate at constant residence time t=4 min using plug
flow reactors of constant diameter but increasing length.
otherwise identical mathematical description of the Tube reactor (stainless steel), V=0.5–3 mL, L=1–6 m, d =
two reactors, these two dimensions are exchanged.
i
[
10]
0.75 mm, R=2.5 cm, T=1408C.
Adv. Synth. Catal. 2013, 355, 2391 – 2399
ꢂ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
2393

Simon Strompen et al.
UPDATES
sion. The Reynolds number (Re) is commonly used to same time drastically reducing the reaction time or in-
characterize the flow regime in continuous reactors creasing the length of the tube. It was therefore decid-
and can be calculated for tube reactors according to ed to elongate the reaction time for the tube reactor
Eq. (1):
compared to batch experiments in order to partly
compensate for the decreased reaction rate.
Packed-Bed Reactor for Biocatalytic Aminolysis
While no data on the dynamic viscosity of rac-3 or In order to transfer previous results from batch ex-
reaction mixtures of varying substrate/product compo- periments to a continuous packed-bed reactor set-up,
sition are available, Reynolds numbers for the pure geometrical parameters of reactor and catalyst prepa-
substrates were calculated for the reactor with an ration must be considered. A borosilicate glass reac-
inner tube diameter d =0.8 mm and a flow velocity of tor of d =1 cm was chosen as a suitable PBR due to
i
i
À1
u =6.3 mh . Thermophysical properties of the com- its chemical resistance and flexibility with regard to
f
pounds benzylamine (m1
=0.715 MPas; 11,808C
=
=
adjustment of reactor length. The heterogeneous bio-
catalyst Novozym 435, according to the manufacturer,
,808C
À1
0.933 gmL ) and trans-ethyl crotonate (m2,708C
À1
0.439 MPas; 12,708C =0.870 gmL ) were taken from has a bulk particle density, i.e., an average density of
[11]
the Infotherm database.
Accordingly, Reynolds a large volume of particles, of
numbers of Re=1.8 for benzylamine (1) and Re=2.8
for trans-ethyl crotonate (2) were calculated. Given
the very low Reynolds numbers, the flow regime can
be considered laminar also for substrate/product mix-
tures. However, for tube or pipe reactors small vorti-
ces causing a local, lateral mixing of the liquid phase
are usually desired. In the absence of vortices in an
With a Sauter mean diameter of the particles of
entirely laminar flow regime, a non-ideal flow profile 498 mm and an average mass of a single particle of
À9
within the tube is caused by friction of the fluid and m¯ =6.095·10 g, the density of the carrier material is
p
the tube wall. Consequently, the fluid velocity is calculated to
slowed down towards the tube wall while higher ve-
locities are found in the center of the tube. Effects
such as channeling or stagnant regions, however,
reduce the performance of the reactor and should be
[12]
avoided. In fact, this is a common problem encoun-
tered for slow reactions in continuous flow necessitat-
ing several hours reaction time. To circumvent such
The porosity e describing the void volume in the re-
a problem, a nested-pipe reactor has been presented actor may be calculated as
by Minnich and co-workers with reactor volumes
[
13]
ranging from 0.25–8 L.
Dean vortices can be used to circumvent the prob-
[
13]
lem of channel wall effects. Such vortices are gener-
ated in a curved pipe by a pressure driven flow of the
fluid when the higher velocity stream in the center ex-
periences a greater centripetal force and is hence di-
rected outward. In Figure 2b, conversion of the aza- pending on the amount of catalyst and the volumetric
Michael addition at 1408C in a coiled steel capillary flow rate. However, deviations were observed on
reactor of varying length is plotted as a function of comparing calculated and experimentally determined
the flow rate at a constant residence time of t= bulk particle density and porosity which, in turn,
The data were used to estimate residence times de-
[14]
4
min. With increasing flow rate an increased conver- impact on parameters such as void and total volume
sion is observed. The improved mixing resulting from and the residence time (see Table 1 for observed pa-
Dean vortices is likely to be responsible for this rameters of PBR used in coupled reactor set-up). All
effect. In a microreactor setup, Howell et al. observed calculations and simulations were based on experi-
Dean vortex formation starting at Reynolds numbers mentally determined parameters.
between 1–10 that became stronger with increasing
Long reaction times of the aza-Michael addition de-
[15]
flow velocity.
Unfortunately, the necessary high scribed in the previous section necessitate low flow
flow velocities to improve conversion cannot be real- rates. However, in reactions carried out with enzymes
ized easily in a tube reactor set-up without at the immobilized on porous carriers, external mass transfer
2394
asc.wiley-vch.de
ꢂ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2013, 355, 2391 – 2399

Development of a Continuously Operating Process for the Enantioselective Synthesis of a b-Amino Acid Ester
Table 1. Data characterizing two packed-bed reactors in series containing immobilized C. antarctica lipase B (Novozym 435)
used in a coupled, continuous flow reactor set-up.
À1
À1
À1
PBR di [cm] mN435 [g] Vtotal [mL] Vvoid [mL] L [cm] n˙ [mL min ] u [cm min ] t [h] 1bp [g mL ] e [À]
f
1
2
1
1
4.3
2.0
11.9
5.1
7.8
3.4
15.2
6.5
0.052
0.049
0.10
0.09
2.5
1.2
0.36
0.39
0.65
0.67
PBR experiments. Negligible differences are visible
comparing batch reactor and PBRs.
Coupled Reactor Set-Up for Continuous Production
of (S)-3
Optimization of the overall reactor performance
mainly depends on the enzymatic reaction carried out
in the PBR. A high operational stability of the biocat-
alyst and a good turnover frequency are crucial in
order to develop any economically attractive process.
Important aspects to consider with respect to the sub-
strate ratio effects in the biocatalytic aminolysis are
therefore summarized graphically in Scheme 3.
A high enantiomeric excess (ee) of ideally >99%
of the product rac-3 is a prerequisite in order to ach-
Figure 3. Diffusion limitation in packed bed reactor for ieve a value-added product, for which approximately
CALB (Novozym 435)-catalyzed aminolysis. Comparison of
6
0% conversion of rac-3 are necessary given the in-
conversion of benzylamine 1 as a function of residence time
in batch reactor and two packed bed reactors. T=608C,
c0,rac-3 =62%. Batch reactor (*); PBR , V =3.5 mL void
trinsic enantioselectivity of CALB (Novozym 435) for
this substrate. Consequently, benzylamine 1 as a re-
maining component from stage 1 must be available as
a substrate in a sufficient amount in stage 2. At the
same time, the excess of 1 should be kept as low as
possible for kinetic, stability and atom efficiency rea-
sons. The requirements can be fulfilled best using an
1
1
volume, m =2 g Novozym 435 (&); PBR , V =7 mL void
1
2
2
volume, m =4 g Novozym 435 (~); simulation based on ki-
1
netic model (- -).
or internal diffusion limitations are a common prob- initial molar ratio of approximately 1.7:1 of substrates
lem. For batch reactions, such limitations are not sig- 1 and 2. Despite non-ideal flow conditions in the PFR
[7]
nificant. The potential presence of diffusion limita- (stage 1), 92% conversion can be achieved in a reason-
tion can be determined experimentally for packed- able time frame (t=13.3 h, T=808C) when applying
bed reactors by comparing the conversion of two this starting molar ratio.
PBRs with different amounts of catalyst. Equal resi-
dence times in both reactors of different catalyst load- and ref. ) was used to simulate ee
The kinetic model (see the Supporting Information
[7]
and conversion
S)-3
(
ing can be achieved by varying the flow rate. Similar as a function of residence time with a fixed amount of
conversion levels at equal residence times are expect- 6.3 g CALB (Novozym 435) assuming no impact of
ed in case no diffusion limitation is present. A de- diffusional limitation. As starting conditions, the con-
creased degree of conversion in the reactor that is op-
erated at lower flow rates (lower catalyst load) on the
other hand indicates the presence of diffusional limi-
tations. Figure 3 shows the conversion as a function of
residence time for two PBRs of 3.5 mL (PBR ) and
1
7
mL (PBR ) void volume. Additionally, the conver-
2
sion for a batch reaction in terms of reaction time
known to be not limited by mass transfer phenomena
is shown. Since equally high catalyst concentrations
cannot be realized in batch experiments, correspond-
ing values were calculated from reactions carried out Scheme 3. Summary of substrate ratio effects as obtained
[7]
with a lower catalyst concentration as compared to from batch experiments carried out at 608C.
Adv. Synth. Catal. 2013, 355, 2391 – 2399
ꢂ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
2395

Simon Strompen et al.
UPDATES
Figure 4. Simulation of conversion (À) and enantiomeric
excess (- -) as a function of residence time in a packed bed
[7]
reactor (mN435 =6.3 g, 608C) based on kinetic model. Initial
substrate concentrations used for simulation were those ob-
tained in previous experiments from aza-Michael addition
carried out in a PFR with an initial substrate ratio of 1.7:1
for 1 and 2 at t=13.3 h, T=808C and 92% conversion of 2.
centrations of the substrates 1, rac-3 and product rac-
4
were used as observed at the tube reactor outlet.
Figure 4 shows that a residence time t of approxi-
mately 2.7 h is necessary to obtain 99% ee at 58%
conversion.
Experiments employing the coupled reactor set-up
(
Scheme 2) were carried out starting from the sub-
Figure 5. Conversion and ee as a function of reaction time in
coupled reactor for the continuous aza-Michael addition of
strates benzylamine 1 and trans-ethyl crotonate 2.
Table 1 summarizes the geometric reactor data and
operational parameters characterizing the PBR used
in this continuous reactor set-up. In order to be able
to collect conversion/ee data at two different degrees
of conversion within a single experiment, two PBRs
were coupled in series. Slightly lower flow rates were obtained from a simulation based on the kinetic model) see
observed for the second reactor due to a density in- Supporting Infoirmation and ref. PTFE tube reactor:
crease from approximately 1.00 gmL at ambient 808C, V=41.6 mL, L=82.7 m, d
temperature at the inlet of the first PBR to about
1
second PBR caused by ongoing conversion in the sol-
vent-free system.
The reactor was continuously operated for 88 h
without significant loss of activity. With a residence
1
and 2 and subsequent CALB (Novozym 435)-catalyzed
aminolysis. Samples were taken using sampling valves at (a)
residence time tPBR =2.5 h and (b) tPBR =3.5 h at the outlet
of the packed bed reactor. Symbols represent experimental
data: Crac-3 (^); ee(S)-3 (&). Lines represent predicted data as
[7]
À1
=0.8 mm. A constant con-
i
version of 92% was measured at the outlet of the tube reac-
tor at t=13.3 h (see Figure 2a).Packed-bed reactor: 608C,
À1
.06 gmL at ambient temperature at the inlet of the
Vtotal =17 mL, Ltotal =21.7 cm, d =1 cm, 6.3 g Novozym 435.
i
Flow rate for substrates
1
and 2, respectively: n1 =
À1
À1
0
.303 mmolmin , n =0.180 mmolmin .
2
time t=2.5 h a conversion of 55% at 97% ee(S)-3 was (Figure 3). The operation of the process was aborted
achieved (Figure 5a). A further elongation of the resi- after 88 h due to blocked capillaries and sampling
dence time to t=3.7 h led to an increased conversion valves caused by a colorless precipitate. The precipi-
of 59% and an average ee(S)-3 of 98% (Figure 5b). tate may have been formed from benzylamine, which
With regard to conversion, the results for both resi- upon exposure to carbon dioxide forms the corre-
dence times are in good agreement with the simulated sponding carbamic acid salt or the corresponding acid
data based on the kinetic model. The enantioselectivi- of rac-3 after enzyme-catalyzed hydrolysis with resid-
[
16]
ty is slightly lower as expected. Enzyme deactivation ual water contained in the substrate solutions.
was not observed to a significant degree in the time Parameters characterizing the performance of the
frame of operation. Diffusional limitations were not sequential process are summarized in Table 2 for the
regarded in the kinetic model due to observations in- aza-Michael addition and for the CALB (Novozym
dicating the presence of no or only weak limitations 435)-catalyzed aminolysis. For the PBR, a space-time
2396
asc.wiley-vch.de
ꢂ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2013, 355, 2391 – 2399

Development of a Continuously Operating Process for the Enantioselective Synthesis of a b-Amino Acid Ester
Table 2. Basic reaction engineering parameters characterizing the performance of the coupled-reactor (Scheme 2) for the
continuous production of (S)-3.
[e]
À1 À1
À1
-
1
À1
f
-
spec. ttn [mol ]
N435
1
c [À] T [8C] t [h] C [%] STY [kg L d ] Q [kg kg
d ] ttn [À]
ee(S)-3 [%]
0
N435
[a]
[c]
TR
0.63[
80
60
13.3
3.67
92
59
1.2
1.8
b]
[d]
PBR 0.53
4.9
158.000 239
98
[
[
[
[
[
[
a]
b]
c]
d]
e]
f]
Initial molar ratio of benzylamine (1).
Initial molar ratio of rac-3.
Conversion of trans-ethyl crotonate (2).
Conversion of rac-3.
Based solely on the volume of the respective reactor unit.
Expected according to stability determined in batch experiments [see Eq. (5)].
À1 À1
À1 À1
yield (STY) of 1.8 kgL d was obtained at about between 0.1 and 130 gL h . For comparison, a-
À1 À1 [18]
5
9% conversion, 98% ee and a catalyst productivity Q amino acids are produced at 30–130 gL h . The
of 4.9 kgkg h . The total turnover number ttn was productivity of the process presented here is thus
À1
À1
N435
calculated according to the method proposed by comparable to established industrial processes and
[
17]
Rogers and Bommarius from kcat,obs and the deacti- shows the high potential of solvent-free processes
vation constant kdeact Eq. (5):
even in the case of only moderate reaction rates.
Experimental Section
General Remarks
Benzylamine (99.5%, ꢀ0.3% water K.F.) and trans-ethyl
crotonate (96%) were obtained from Acros Organics (Bel-
gium). Commercial Novozym 435 (Candida antarctica lipase
B immobilized on acrylic resin) was obtained from Novo-
À1
With a production rate of 10.4 Ug Novozym 435
under process conditions, a molecular mass of CALB
À1
of 33000 gmol and an estimated protein load of 5%
w/w) CALB on the carrier material, an apparent zymes (Denmark). Analytical grade solvents used for HPLC
(
À1
kcat,obs =414 h is calculated. A deactivation constant analysis were purchased from Carl Roth (Germany). Refer-
À1
of kdeact =0.0026 h had been determined in batch ex- ence compounds rac-3 and rac-4 were synthesized as de-
[6,7]
scribed previously and in the Supporting Information sup-
plied with this article. A commercial PTFE capillary (Boh-
lender, Germany) was used as a tube reactor and a Super-
periments and was used for an estimation of the ttn in
a continuous process. Even though the process mode
in general may influence enzyme stability, experimen-
tal results indicate a high lipase stability for both
batch reactor and continuously operated PBR.
ꢃ
formance borosilicate glass column (Merck, Germany)
filled with Novozym 435 as the packed-bed reactor.
Continuously Operating Process in a Tube Reactor
for Aza-Michael Addition
Conclusions
The substrates 1 and 2 were mixed prior to the reaction in
order to avoid changes in substrate composition due to
back-pressure changes and pump inaccuracies upon change
of flow rates. The premixed substrates were stored on ice/
The chemoenzymatic reaction sequence towards the
synthesis of short-chain aliphatic b-amino acids devel-
[
6]
oped initially by Weiß and Grçger has been trans-
ferred successfully to a continuously operated, cou- NaCl (À158C) in order to largely prevent an ongoing reac-
pled reactor set-up. Suitable conditions for an in- tion. Substrates were subsequently pumped through a tem-
creased reaction rate and stability of the catalyst had perature controlled tube for the continuous reaction using
[
7]
a Pharmacia 2248 HPLC pump (Sweden). Flow rates were
adjusted in the range from 0.05 mLmin to 0.8 mLmin . A
polytetrafluoroethylene (PTFE) tube (V=41.6 mL, L=
been identified previously in a kinetic investigation.
À1
À1
The developed kinetic model proved useful for the
simulation and optimization of the continuous pro-
cess. The coupled chemo-enzymatic process could be
operated for more than 80 h without significant loss
of activity. An STY of 1.8 kgL d (128 gL h ) for
the desired b-amino acid ester product (S)-3 was ach-
8
2.7 m, d =0.8 mm) coiled around a custom-made steel spin-
i
dle with a diameter of 10.8 cm was inserted into a thermo-
statted cylinder was applied at temperatures up to 808C.
Coiled stainless steel capillaries (V=0.5–3 mL, L=1–6 m,
di =0.075 mm, R=2.5 cm) were applied for experiments on
À1 À1
À1 À1
[
18]
ieved. According to Straathof et al., , industrial pro- flow velocity effects at 1408C. Temperature control was
cesses for the production of fine chemicals range in achieved by immersing the steel capillary reactor into a ther-
Adv. Synth. Catal. 2013, 355, 2391 – 2399
ꢂ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
2397

Simon Strompen et al.
UPDATES
mostat (Ecoline Star Edition E200, Lauda, Germany) filled
with silicon oil. Samples for quantification were collected at
the outlet of the reactor and quenched by dilution in 1,3-di-
oxolane prior to HPLC analysis.
Retention times: 1: 3.9 min, 2: 8.5 min, rac-3: 15.5 min, rac-
4: 18 min.
Determination of Enantiomeric Excess
The enantiomeric excess of 3 was determined using a Chiral-
cel OD-H column (5 mm, 25 cm, 0.46 cm; Daicel, Japan)
cooled down to 108C and UV detection at 223 nm. iso-
Hexane/2-propanol 95/5 (v/v), 0.2% (v/v) DEA was used as
Continuously Operating Process in a Packed-Bed
Reactor for Biocatalytic Aminolysis
In analogy to continuous aza-Michael addition experiments,
substrate mixtures consisting predominantly of benzylamine
À1
the eluent at a flow rate of 0.75 mLmin . Retention times:
(
R)-3: 10 min, (S)-3: 12.5 min. The ee of the product 4 was
(
1) and rac-3 were prepared and used as a substrate feed for
measured on a Nucleocel a S column (5 mm, 25 cm, 0.46 cm;
Macherey–Nagel, Germany) at 208C and UV detection at
the Novozym 435-catalyzed aminolysis in order to avoid
fluctuation of the substrate composition. The mixtures were
obtained as (crude) product resulting from the solvent-free
aza-Michael addition of benzylamine (1) to trans-ethyl crot-
onate (2). A representative batch reaction for preparative
substrate synthesis was carried out using 0.98 mol of benzyl-
amine and 0.58 mol of trans-ethyl crotonate at 608C for 26 h
in a round-bottom glass flask. At 95% conversion the com-
position of the product representing the substrate feed solu-
2
58 nm. A mixture consisting of n-hexane/2-propanol 95/5
(
v/v), 0.2% (v/v) DEA was used as the eluent at a flow rate
À1
of 0.8 mLmin . Retention times: (S)-4: 29 min, (R)-4:
3
2 min.
Computational Methods
Simulation of reactor performances based on kinetic param-
eters published previously was performed using the inherent
À1
tion for the biocatalytic aminolysis was 2.3 molg benzyl-
À1
A
C
H
T
U
N
G
T
R
E
N
N
U
N
G
amine (1), 0.152 molg
trans-ethyl crotonate (2),
[7]
ode45 algorithm of MATLAB (MathWorks, Natick, USA).
À1
À1
3
.031 molg rac-3 and 0.117 molg rac-4. A borosilicate
glass reactor (Superformance R, Merck, Germany) of d =
i
1
6
cm was filled with Novozym 435 and thermostatted at
08C using an Ecoline Star Edition E100 thermostat
References
(
Lauda, Germany). Substrates were pumped through the
packed-bed reactor using a Pharmacia 2248 HPLC pump at
flow rates of 0.02–0.4 mLmin . Pressure was monitored at
the reactor inlet using an analogous pressure gauge. Samples
were collected at the outlet of the reactor and quenched in
[
1] Selected examples on the concept, benefits and chal-
lenges of solvent-free approaches a) P. T. Anastas, J.
Warner, in: Green Chemistry: Theory and Practice;
Oxford University Press, New York, 1998; b) H. R.
Hobbs, N. R. Thomas, Chem. Rev. 2007, 107, 2786–
À1
1,3-dioxolane or iso-hexane:2-propanol 90:10 (v/v) for quan-
tification and chiral analysis via HPLC.
2820; c) P. J. Walsh, H. Li, C. A. de Parrodi, Chem. Rev.
2007, 107, 2503–2545; d) J. M. DeSimone, Science 2002,
297, 799.
Coupled Reactor Set-Up for Continuous Production
of (S)-3
[
2] For efficient examples see a) L. Hilterhaus, O. Thum,
A. Liese, Org. Process Res. Dev. 2008, 12, 618–625;
b) C. Korupp, R. Weberskirch, J. J. Mꢀller, A. Liese, L.
Hilterhaus, Org. Process Res. Dev. 2010, 14, 1118–1124;
c) A. Mahapatro, A. Kumar, R. A. Gross, Biomacro-
molecules 2004, 5, 62–68.
The PTFE tube reactor (V=41.6 mL, L=82.7 m, d =
i
0
2
.8 mm) and packed-bed reactor (Vtotal =17 mL, Ltotal
=
1.7 cm, d =1 cm, 6.3 g Novozym 435) were coupled for the
i
continuous production of (S)-ethyl 3-(benzylamino)buta-
noate [(S)-3]. Two Pharmacia 2248 HPLC pumps were used
[
3] See for example a) J. von Langermann, A. Mell, E.
À1
for the continuous supply of substrates 1 (0.303 mmolmin )
Paetzold, T. Daußmann, U. Kragl, Adv. Synth. Catal.
À1
and 2 (0.180 mmolmin ). PTFE pressure retention valves
2
007, 349, 1418–1424; b) K.-N. Uhm, S.-J. Lee, H.-k.
Kim, H.-Y. Kang, Y. Lee, J. Mol. Catal. B: Enzym.
007, 45, 34–38; c) J. Xiong, J. Wu, G. Xu, L. Yang,
(
Bohlender, Germany) set to 2 bar were installed at the
pump outlet in order to improve the performance of the
pumps and maintain constant flow velocities. Mixing was
achieved using a simple mixing T-piece prior to the reactor
inlet. The tube reactor for the aza-Michael addition was
thermostatted at 808C using a water bath. The temperature
of the packed bed reactor was maintained at 608C. Pressure
gauges and sampling valves were installed before and after
both reactor units.
2
Chem. Eng J. 2008, 138, 258–263; d) H.-H. Li, Y.-H.
He, Y. Yuan, Z. Guan, Green Chem. 2011, 13, 185–189.
4] Selected reviews on continuous (bio)catalytic process-
es; a) C. Wiles, P. Watts, Eur. J. Org. Chem. 2008, 1655–
[
1
671; b) N. N. Rao, S. Lꢀtz, K. Wꢀrges, D. Minçr, Org.
Process Res. Dev. 2009, 13, 607–616; c) P. Lozano, E.
Garcia-Verdugo, S. V. Luis, M. Pucheault, M. Vaultier,
Curr. Org. Synth. 2011, 8, 810–823.
[
5] R. Yuryev, S. Strompen, A. Liese, Beilstein J. Org.
Determination of Conversion
Chem. 2011, 7, 1449–1467.
Conversions of all reactions were analyzed by HPLC on an
Agilent system equipped with a diode array detector
[6] M. Weiß, H. Grçger, Synlett 2009, 1251–1254.
[7] S. Strompen, M. Weiß, T. Ingram, I. Smirnova, H.
Grçger, L. Hilterhaus, A. Liese, Biotechnol. Bioeng.
2012, 109, 1479–1489.
(
215 nm) and a Nucleodur C8 ec column (5 mm, 25 cm,
.46 cm; Macherey–Nagel, Germany). A mixture consisting
0
of 47.5/52.5 (v/v) MeOH/sodium phosphate buffer (50 mM,
pH 6.5) was used as the eluent at a flow rate of 1 mLmin .
[8] a) M. Weiß, T. Brinkmann, H. Grçger, Green Chem.
2010, 12, 1580–1588; b) M. Eissen, M. Weiß, T. Brink-
À1
2398
asc.wiley-vch.de
ꢂ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2013, 355, 2391 – 2399

Development of a Continuously Operating Process for the Enantioselective Synthesis of a b-Amino Acid Ester
mann, S. Steinigeweg, Chem. Eng. Technol. 2010, 33,
29–637.
9] J. L. L. Rakels, H. T. Paffen, A. J. J. Straathof, J. J. Heij-
[14] D. Kaufhold, F. Kopf, C. Wolff, S. Beutel, L. Hilterhaus,
M. Hoffmann, T. Scheper, M. Schlꢀter, A. Liese, J.
Membrane Sci. 2012, 423–424, 342–347.
6
[
nen, Enz. Microb. Technol. 1994, 16, 791–794.
10] H. Chmiel, Bioprozesstechnik, Spektrum, Heidelberg,
[15] P. B. Howell Jr, D. R. Mott, J. P. Golden, F. S. Ligler,
[
Lab Chip 2004, 4, 663–669.
2
006.
[16] L. Heuer, Benzylamine, in: Ullmannꢀs Encyclopedia of
Industrial Chemistry, Wiley-VCH, Weinheim, 2006.
[17] T. A. Rogers, A. S. Bommarius, Chem. Eng. Sci. 2010,
65, 2118–2124.
[
[
11] Website: www.fiz-chemie.de/infotherm/.
12] O. Levenspiel, Chemical Reaction Engineering, John
Wiley & Sons, New York, 1999.
[
13] C. B. Minnich, L. Greiner, C. Reimers, M. Uerdingen,
[18] A. J. Straathof, S. Panke, A. Schmid, Curr. Opin. Bio-
technol. 2002, 13, 548–556.
M. A. Liauw, Chem. Eng. J. 2011, 168, 759–764.
Adv. Synth. Catal. 2013, 355, 2391 – 2399
ꢂ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
2399