Scalable enantioseparation of amino acid derivatives using continuous liquid-liquid extraction in a cascade of centrifugal contactor separators

Article abstract of DOI:10.1021/op900152e

Using a cascade of six centrifugal contactor separators in a countercurrent liquid-liquid extraction mode allowed the separation of one of the enantiomers of 3,5-dinitrobenzoyl-leucine in 55% yield and 98% ee using a catalytic amount of a chiral host comp

Full text of DOI:10.1021/op900152e

Organic Process Research & Development 2009, 13, 911–914
Scalable Enantioseparation of Amino Acid Derivatives Using Continuous
Liquid-Liquid Extraction in a Cascade of Centrifugal Contactor Separators
Boelo Schuur,† Andrew J. Hallett,‡ Jozef G. M. Winkelman,† Johannes G. de Vries,*,‡,§ and Hero J. Heeres*,†
UniVersity of Groningen, Department of Chemical Engineering, Nijenborgh 4, 9747 AG Groningen, The Netherlands, DSM
Pharmaceutical Products, InnoVatiVe Synthesis and Catalysis, P.O. Box 16, 6160 MD Geleen, The Netherlands, and UniVersity
of Groningen, Stratingh Institute for Chemistry, Nijenborgh 4, 9747 AG Groningen, The Netherlands
Abstract:
alcohol19-22 separation en route to single enantiomer drugs.23-25
An enantiopure host in catalytic amounts is used to preferentially
transport one of the two enantiomers across a phase boundary.
The potential of enantioselective liquid-liquid extraction lies
in its versatility and ease of scale up.26,27 However, the reported
moderate selectivities for most systems would require the use
of multistage processing to achieve high product ee’s.12,14-20,22-28
Remarkably, experimental studies on multistage chiral liquid-
liquid extraction are scarce. We are aware of only two reports
from Maier in which full separation of a racemate was achieved,
one describing the use of a number of membranes in series
and another one in which an analytical centrifugal partition
chromatograph (CPC) was used.7,10 Although these results are
quite remarkable, the methods are not scalable. The productivity
of the membrane systems is much too low, and the CPC only
exists as an analytical instrument, although recently the con-
struction of a scaled-up version with a volume of 18 L was
reported.29 Other than Chiral SMB, no scalable method exists
for liquid phase enantioseparation.
Using a cascade of six centrifugal contactor separators in a
countercurrent liquid-liquid extraction mode allowed the separa-
tion of one of the enantiomers of 3,5-dinitrobenzoyl-leucine in 55%
yield and 98% ee using a catalytic amount of a chiral host
compound based on a cinchona alkaloid. This method allows the
preparation of kilograms of enantiopure material in table-top-sized
equipment that can be operated in a fume cupboard. The
methodology can be easily scaled up to ton amounts as large-
volume centrifugal contactor separators are commercially available
as well.
Introduction
A growing demand for enantiopure compounds has stimu-
lated research and development on efficient manufacture and
separation technologies.1 A well-known technique to obtain
enantiopure compounds on industrial scale is resolution of the
racemate through crystallization.2-4 This technique is not always
applicable, and alternatives have been explored. Examples are
enantioseparation using simulated moving bed5,6 related chro-
matographic techniques,7 and the use of (liquid) membrane
technology.8-10 Liquid-liquid extraction is also considered a
promising technology for chiral separation and examples may
be found in the area of amino acid,11-18 amine, and amino
In this paper we describe the use of centrifugal contactor
separators (CCS) for multistage enantioselective liquid-liquid
(12) Koska, J.; Haynes, C. A. Chem. Eng. Sci. 2001, 56, 5853.
(13) Lindner, W.; La¨mmerhofer, M. Cinchona-based carbamates as chiral
selectors of stereodiscrimination. EP 0912563, 1997.
(14) Pickering, P. J.; Chaudhuri, J. B. Chem. Eng. Sci. 1997, 52, 377.
(15) (a) Reeve, T. B.; Cros, J. P.; Gennari, C.; Piarulli, U.; de Vries, J. G.
Angew. Chem., Int. Ed. 2006, 45, 2449. (b) Dzygiel, P.; Monti, C.;
Piarulli, U.; Gennari, C. Org. Biomol. Chem. 2007, 5, 3464. (c)
Dzygiel, P.; Reeve, T. B.; Piarulli, U.; Krupicka, M.; Tvaroska, I.;
Gennari, C. Eur. J. Org. Chem. 2008, 11, 1253.
* Authors for correspondence. E-mail: Hans-JG.Vries-de@dsm.com;
H.J.Heeres@rug.nl.
† University of Groningen, Deptartment of Chemical Engineering.
‡ DSM Pharmaceutical Products.
(16) Takeuchi, T.; Horikawa, R.; Tanimura, T. Anal. Chem. 1984, 56, 1152.
(17) Tan, B.; Luo, G. S.; Wang, H. D. Tetrahedron: Asymmetry 2006, 17,
883.
§ University of Groningen, Stratingh Institute for Chemistry.
(1) Rouhi, A. M. Chem. Eng. News 2003, 81, 45.
(2) Bruggink, A. In Chirality in Industry II; Collins, A. N., Sheldrake,
G. N. , Crosby, J. , Eds.; John Wiley & Sons, Ltd.: Chichester, 1997,
81.
(18) Tan, B.; Luo, G. S.; Wang, J. D. Sep. Purif. Technol. 2007, 53, 330.
(19) Abe, Y.; Shoji, T.; Kobayashi, M.; Qing, W.; Asai, N.; Nishizawa,
H. Chem. Pharm. Bull. 1995, 43, 262.
(20) Steensma, M.; Kuipers, N. J. M.; de Haan, A. B.; Kwant, G. Chirality
2006, 18, 314.
(3) Faigl, F.; Fogassy, E.; Nogradi, M.; Palovics, E.; Schindler, J.
Tetrahedron: Asymmetry 2008, 19, 519.
(21) Tang, L.; Choi, S.; Nandhakumar, R.; Park, H.; Chung, H.; Chin, J.;
Kim, K. J. Org. Chem. 2008, 73, 5996.
(4) Fogassy, E.; Nogradi, M.; Kozma, D.; Egri, G.; Palovics, E.; Kiss, V.
Org. Biomol. Chem. 2006, 4, 3011.
(22) Viegas, R. M. C.; Afonso, C. A. M.; Crespo, J. G.; Coelhoso, I. M.
Sep. Purif. Technol. 2007, 53, 224.
(5) Francotte, E.; Leutert, T.; La Vecchia, L.; Ossola, F.; Richert, P.;
Schmidt, A. Chirality 2002, 14, 313.
(23) Colera, M.; Costero, A. M.; Gavina, P.; Gil, S. Tetrahedron:
Asymmetry 2005, 16, 2673.
(6) Zenoni, G.; Quattrini, F.; Mazzotti, M.; Fuganti, C.; Morbidelli, M.
FlaVour Fragrance J. 2002, 17, 195.
(24) Ohki, A.; Miyashita, R.; Naka, K.; Maeda, S. Bull. Chem. Soc. Jpn.
1991, 64, 2714.
(7) Gavioli, E.; Maier, N. M.; Minguillon, C.; Lindner, W. Anal. Chem.
2004, 76, 5837.
(25) Tang, K.; Chen, Y.; Huang, K.; Liu, J. Tetrahedron: Asymmetry 2007,
18, 2399.
(8) Baragana, B.; Blackburn, A. G.; Breccia, P.; Davis, A. P.; de Mendoza,
J.; Padron-Carrillo, J. M.; Prados, P.; Riedner, J.; de Vries, J. G.
Chem.sEur. J. 2002, 8, 2931.
(26) Steensma, M.; Kuipers, N. J. M.; de Haan, A. B.; Kwant, G. Chem.
Eng. Proc. 2007, 46, 996.
(9) Keurentjes, J. T. F.; Nabuurs, L. J. W. M.; Vegter, E. A. J. Membr.
Sci. 1996, 113, 351.
(27) Steensma, M.; Kuipers, N. J. M.; de Haan, A. B.; Kwant, G. Chem.
Eng. Sci. 2007, 62, 1395.
(10) Maximini, A.; Chmiel, H.; Holdik, H.; Maier, N. W. J. Membr. Sci.
2006, 276, 221.
(28) Steensma, M.; Kuipers, N. J.; de Haan, A. B.; Kwant, G. J. Chem.
Technol. Biotechnol. 2006, 81, 588.
(11) Kellner, K. H.; Blasch, A.; Chmiel, H.; La¨mmerhofer, M.; Lindner,
W. Chirality 1997, 9, 268.
(29) Sutherland, I.; Hewitson, P.; Ignatova, S. J. Chromatogr., A 2009,
1216, 4201.
10.1021/op900152e CCC: $40.75  2009 American Chemical Society
Published on Web 08/19/2009
Vol. 13, No. 5, 2009 / Organic Process Research & Development
•
911

Scheme 1. Structures of the guest (R,S)-DNB-Leu (l) and
the host CA (r)
extraction. These integrated devices combine the functions of
fast mixing and separation and are thus an ideal instrument to
achieve process intensification. We expected to achieve full
enantioseparation with high productivities using a number of
CCS’s in series.
Recently, we demonstrated the potential of these CCS
devices for combined reaction and separation.30 We have
also demonstrated the single-stage chiral separation of the
amino acid 3,5-dinitrobenzoyl-(R),(S)-leucine [DNB-
(R),(S)-Leu, Scheme 1] using the cinchona alkaloid O-(1-
tert-butylcarbamoyl)-11-octadecylsulfinyl-10,11-dihydro-
quinine (CA, see Scheme 1)31,32 a host-guest system
developed by La¨mmerhofer and Lindner.11,13 Optimized
single stage organic phase ee values of 34% were obtained
for the (S)-enantiomer at a yield of 61%. From the
previous work it had become clear that a cascade of CCS
equipment operated in a countercurrent mode would be
required to achieve full enantioseparation.
Figure 1. Centrifugal contactor separator used in this study,
real size from top to bottom is 65 cm.
Results and Discussion
Figure 2. Schematic representation of the cascade with six CCS
devices in series with full recovery of the host in a single back-
extraction stage (F ) feed; BE ) back extraction).
Here, we report the first experimental study on multistage
chiral separation by liquid-liquid extraction. Similar to our
previous studies with single staged extractions, DNB-(R),(S)-
Leu is the model guest and CA the model host. A cascade of
six CCS devices in series was used.
enantiomer, the model predicts that the feed should be posi-
tioned at stage 5 (see Figure 2).
The cascade of six CCS devices of the type CINC V-0233
(Figure 1, maximum throughput, 1.9 L/min) was applied in
combination with a back-extraction step in a single CCS for
recovery and recycling of the host. The optimum configuration
of the cascade (e.g., feed input, concentrations, flow rates) was
determined by mathematical modeling based on a previously
developed equilibrium model. This model uses as input the mass
balances for all components (DNB-(R)-Leu, DNB-(S)-Leu and
CA) and the equilibrium relations for each stage.34 The model
predicts that 12 stages are required to obtain both enantiomers
in high enantiomeric excess (>99% ee). With the experimental
limitation of only six CCS devices available for the cascade
and one for the back extraction unit, the model predicts that it
is possible to obtain one of the enantiomers with an ee exceeding
99% in a yield up to 44%. To obtain the enantiopure (S)-
In an experiment aimed to obtain enantiopure DNB-(S)-Leu,
1.35 L of host solution (4.3 mM) was recycled continuously
through the seven CCSs with 100 mL/min, see Figure 2. The
aqueous DNB-(R),(S)-Leu (1.5 mM, pH 5.7, I ) 0.07 mol/L)35
was fed with 42 mL/min at the fifth stage. At stage 1 an aqueous
phosphate buffer (pH 7.2, I ) 0.08 mol/L) entered the cascade
with 60 mL/min to wash the undesired DNB-(R)-Leu out of
the organic phase. The organic exit from stage 1, containing
DNB-(S)-Leu complexed to the host in high purity was then
back-extracted using 33 mL/min pH 9 phosphate buffer (I )
0.12 mol/L) and the host was recycled to the cascade in stage
6. The DNB-(R),(S)-Leu concentrations were monitored in both
aqueous exit streams (Fout,1 and Fout,2). Figure 3 shows the
aqueous concentrations of the two enantiomers in the back-
extraction outlet, Fout,2, Figure 4 shows the concentrations in
Fout,1. The experiment was run for 10 h. The average concentra-
tions in Fout,2 are 0.57 mM and 8.47 µM for DNB-(S)-Leu and
DNB-(R)-Leu, respectively, corresponding with an average ee
of 98% and a yield of 55% for the S-enantiomer in Fout,2. The
average steady-state concentrations in Fout,1 were 0.11 and 0.28
(30) Kraai, G. N.; van Zwol, F.; Schuur, B.; Heeres, H. J.; de Vries, J. G.
Angew. Chem., Int. Ed. 2008, 47, 3905.
(31) Hallett, A. J.; Kwant, G. J.; de Vries, J. G. Chem.sEur. J. 2009, 15,
2111.
(32) Schuur, B.; Floure, J.; Hallett, A. J.; Winkelman, J. G. M.; de Vries,
J. G.; Heeres, H. J. Org. Process Res. DeV. 2008, 12, 950.
(33) Meikrantz, D. H.; Macaluso, L. L.; Sams, H. W.; Schardin, C. H.;
Federici, A. G. Centrifugal separator. U.S. Patent 5,762,800, 1998.
(34) Schuur, B.; Winkelman, J. G. M.; Heeres, H. J. Ind. Eng. Chem. Res.
2008, 47, 10027.
(35) We use ionic strength, I, instead of concentrations since there are ions
with different valencies present in the aqueous solution. For further
details see ref 34.
912
•
Vol. 13, No. 5, 2009 / Organic Process Research & Development

Figure 3. Back-extraction aqueous outlet (Fout,2) DNB-Leu
concentrations. 0: DNB-(S)-Leu, O: DNB-(R)-Leu.
Figure 5. Single CCS extraction yields versus total liquid
throughput. 0: DNB-(S)-Leu, O: DNB-(R)-Leu. Lines: Single-
stage model prediction.
The experiments were carried out with relatively low reagent
concentrations and flow rates, and this reduces the volumetric
production rate considerably. To gain insights in the highest
attainable production rates in the cascade, the solubility of the
amino acid and also the maximum flow rate in the cascade were
determined. The maximum solubility of DNB-(R),(S)-Leu-Na+
in water at ambient temperature was determined experimentally
to be 29 ( 1 mM. Taking into account the margin needed to
avoid precipitation, a maximum DNB-(R),(S)-Leu concentration
of 15 mM may be fed to the cascade.
The maximum flow rate was determined by experiments in
a single CCS. At otherwise constant conditions (1 mM aqueous
DNB-(R),(S)-Leu, 0.27 mM organic CA, 50 Hz rotational
frequency, and 294 K), the total flow rate was varied, and the
effect on the outlet concentrations was studied. It was found
that the outlet concentrations and hence the extraction yields
were not a function of the flow rates over the entire operating
range (Figure 5). This indicates that equilibrium is established
even at the operating limit of the CCS of 1.8 L/min. With this
information available, the maximum production rate in the
cascade of six CCS devices may be calculated. The model
predicts that, when applying a maximum total flow rate of 1.8
L/min for each CCS, the feed flow should be below 0.36 L/min.
The maximum capacity when using this flow rate combined
with the reagent concentrations set by the solubility limits is
5.4 mmol racemate/min, corresponding to 17.7 kg/week. For
this production level only 60 g of the extractant is required.
Using the largest CCS commercially available a production of
5-10 tons per week is feasible. In addition, it should be possible
to achieve turnover numbers on host of 400-700 per week,
which leads to perfectly acceptable economics. The host is a
very stable compound and did not show any decomposition in
our experiments. We believe that the method described in this
paper has similar potential as chiral SMB for the ton-scale
separation of racemates. We expect the cost of our method to
be much lower as the CCSs are commercially available
equipment that is used routinely in industry and hence are
available at relatively low cost. In addition, the cost of the chiral
Figure 4. Aqueous cascade outlet (Fout,1) DNB-Leu concentra-
tions. 0: DNB-(S)-Leu, O: DNB-(R)-Leu.
mM for DNB-(S)-Leu and DNB-(R)-Leu, respectively, corre-
sponding with an ee of 42%. Mass balance calculation showed
that the balance is closed with an experimental error of 8%.
The model used to determine the optimum settings predicted
99% ee for DNB-(S)-Leu with a yield of 45%. The experimental
values for the ee’s are slightly lower than the model prediction.
This is likely due to the solubility of DCE in water, leading to
slightly higher host concentrations than intended. During the
10 h run, small portions of DCE (65 mL/h) were added to the
system to compensate for this effect, but apparently this was
not sufficient. The fluctuations in the DCE hold-up and
concomitant in the host concentration are also likely the cause
of the small fluctuations in concentrations in the steady-state
profiles (Figures 3 and 4).
Erosion of the ee during the 10 h run time was not observed
(Figures 3 and 4), and the host was recycled 50 times. Thus,
the chemistry appears to be very robust, and the back-extraction
step is very efficient. Dichloroethane (DCE) was used as solvent,
as previous work had shown that use of this solvent leads to
the highest selectivities.31 It will be possible to use other solvents
or mixtures of solvents, although this may necessitate the use
of more CCSs to obtain full separation. No reaction between
DCE and the host CA was observed over long periods of time.
Vol. 13, No. 5, 2009 / Organic Process Research & Development
•
913

host compound is much lower than commercially available
chiral stationary phases that are used in chiral SMB.
from 6 to 1 and finally the back-extraction CCS. After the
recycle of the heavy organic phase was established, the flows
of the light aqueous back-extraction (phosphate buffer pH 9.0,
I ) 0.12 mol/L, 33 mL/min) and wash streams pH 7.2, I )
0.08 mol/L, 60 mL/min) were started, and as soon as the wash
flow had reached the feed stage, the feed flow was started (1.5
mM DNB-(R),(S)-Leu, pH 5.7, I ) 0.07 mol/L, 42 mL/min).
When the aqueous outlet started running, samples were taken
every 15 min during the first 2 h. and afterward every 30 min
from both the aqueous cascade outlet and the aqueous back-
extraction outlet to determine the concentrations of DNB-
(R),(S)-Leu using chiral HPLC (accuracy 3%). Every hour about
65 mL of DCE was added in small portions to the extract
storage vessel to compensate for the losses due to solubility of
DCE in water (0.8% vol). Every 2 h samples were taken from
the aqueous flows in between the stages. The run time of the
experiment was 10 h.
The concentrations of the DNB-Leu enantiomers in the
aqueous phase were determined by HPLC using an Agilent LC
1100 series apparatus, equipped with an Astec Chirobiotic T
column (now Supelco, Sigma-Aldrich). Detection was done
using 270 nm UV light. The eluent was a 3:1 (v/v) mixture of
acetonitrile and methanol, to which 0.25% (vol) triethylamine
and 0.25% (vol) acetic acid were added. The flow rate was set
at 1 mL/min. Before injecting the aqueous phase samples to
the column, 0.10 mL of the samples was diluted with 1.0 mL
eluens and filtered over a syringe filter with pore size 0.45 µm
(Waters Chrom). Quantitative analysis was enabled using
calibration curves.
Experimental Section
O-tert-Butylcarbamoylquinine.13 Quinine (6.255 g, 19.3
mmol), tert-butylisocyanate (2.2 mL, 30.8 mmol), and dibutyl-
tindilaurate (1 drop) were heated at 130 °C in toluene (30 mL)
for 4 h. The solvent was then removed in Vacuo and the product
purified by column chromatography (silica/acetone) to give a
light-yellow waxy powder. Yield 6.941 g (16.4 mmol) 85%.
Spectral data were in accord with the literature.13
O-tert-Butylcarbamoyl-11-octadecylthio-10,11-dihydro-
quinine.13 O-tert-Butylcarbamoylquinine (6.941 g, 16.4 mmol),
octadecan-1-thiol (8.755 g, 30.6 mmol), and AIBN (0.275 g,
1.7 mmol) in CHCl3 (35 mL) were heated at reflux for 24 h.
The solution was concentrated in vacuo and the product purified
by column chromatography (silica/CHCl3). The product was
eluted as the second fraction with CHCl3/MeOH (8:2) as the
eluent. Yield 11.542 g (16.3 mmol) 99%. Spectral data were
in accord with the literature.13
O-tert-Butylcarbamoyl-11-octadecylsulfinyl-10,11-dihy-
droquinine.13 O-tert-Butylcarbamoyl-11-octadecylthio-10,11-
dihydroquinine (11.542 g, 16.3 mmol) and NaIO4 (3.510 g, 16.4
mmol) were stirred at room temp in a mixture of MeOH and
H2O (400 mL, 9:1) for 3 h. H2O (300 mL) was then added and
the pH adjusted to 8.5 with a NaHCO3/Na2CO3 mixture. The
product was extracted into CHCl3 (3 × 150 mL) and then
purified by column chromatography (silica/EtOAc). The product
was eluted with EtOAc/MeOH (9:1). Yield 9.562 g (13.2 mmol)
81%. Spectral data were in accord with the literature.13
Separation of DNB-Leu. Six centrifugal contactor separa-
tors of CINC, CINC V-02,32 were used. The size of the weir
depends on the solvent mixture used. For the dichloroethane/
water mixture used here the size we used was 0.975 in. The
six CCSs were connected with tubing as shown in Figure 2.
Both liquids were transferred to the reactor using Verder
VL1000 Control peristaltic tube pumps equipped with double
pump heads (1.6 × 1.6 × 8R).
Acknowledgment
We thank the Dutch Science Foundation (NWO) for
funding through the Separation Technology program.
Gerard Kwant, Karla Danen (both DSM) and Claudia
Kronenburg (Schering-Plough) are thanked for helpful
discussions. We thank DSM for a loan of two CCS devices.
A.J.H. thanks the European commission for financial support
for his research, which was part of a Research Training Network
“Enantioselective Recognition: Towards the Separation of
Racemates” (HPRN-CT-2001-00182).
The experiment was initiated by switching on the rotors of
all CCSs (40 Hz), opening the tap of the cooling water flow,
cooling the CCSs 2, 4, and 6 in the cascade to maintain a
constant temperature of 294 K, and starting the organic phase
pump. After starting the organic phase flow (4.3 mM host CA
in DCE, 100 mL/min), the CCSs were filled up in the order
Received for review June 16, 2009.
OP900152E
914
•
Vol. 13, No. 5, 2009 / Organic Process Research & Development