Continuous separation of racemic 3,5-dinitrobenzoyl-amino acids in a centrifugal contact separator with the aid of cinchona-based chiral host compounds

Article abstract of DOI:10.1002/chem.200800797

The resolution of racemates is mostly performed by crystallisation of diastereomeric salts. Direct physical separation could be much more efficient, but so far most concepts, with the exception of SMB, have proven to be non-scaleable. Here we report the f

Full text of DOI:10.1002/chem.200800797

FULL PAPER
DOI: 10.1002/chem.200800797
Continuous Separation of Racemic 3,5-Dinitrobenzoyl-Amino Acids in a
Centrifugal Contact Separator with the Aid of Cinchona-Based Chiral Host
Compounds
Andrew J. Hallett, Gerard J. Kwant, and Johannes G. de Vries*[a]
Abstract: The resolution of racemates
is mostly performed by crystallisation
of diastereomeric salts. Direct physical
separation could be much more effi-
cient, but so far most concepts, with
the exception of SMB, have proven to
be non-scaleable. Here we report the
first scalable process for the resolution
of N-protected amino acid derivatives
through selective extraction, with the
aid of a catalytic amount of a chiral
host compound based on Cinchona al-
kaloids. The method hinges on the use
of centrifugal contact separators
(CCSs) for fast mixing and separation.
Although the highest ee obtained was
only 80%, the concept can be extended
through the use of a series of CCSs in
countercurrent mode to effect full sep-
aration.
Keywords: amino acids · host–guest
systems · liquid-phase separation ·
membranes
· process intensifica-
tion · transport
Introduction
for the preparation of enantiopure compounds, including
resolution of racemates, fermentation, chiral pool synthesis
and asymmetric synthesis.[1] Currently, most industrial pro-
cesses are still based on the use of enantiopure materials ob-
tained by classical resolution through crystallisation of dia-
stereomeric salts. Nevertheless, resolutions with enzymes[2]
and asymmetric catalysis[3] seem to be rapidly catching up.
Relatively few examples of processes for the separation of
racemic or non-enantiomerically pure compounds by meth-
ods other than resolution of diastereomeric salts or enzy-
matic resolution can be found in the literature.[4] Certain
synthetic receptors are capable of binding their substrates
with sufficient power to draw them across phase boundaries:
for example, from an aqueous medium into an organic sol-
vent. If the receptor is enantioselective, one enantiomer of a
racemic substrate will be prefer-
Chiral molecules are of great importance in biological pro-
cesses. A wide range of biological functions arise through
molecular recognition, which often requires a strict match-
ing of chirality. The two enantiomers will show different in-
teractions with biological receptors or enzymes and hence
give rise to different biological effects. The body, being ex-
tremely enantioselective, will interact with the components
of a racemic drug differently and metabolize each enantio-
mer by a separate pathway to produce different pharmaco-
logical activity. One enantiomer may thus produce desired
therapeutic activities, while the other may be inactive or, in
the worst cases, produce unwanted effects. The efficient and
economic production of enantiomerically pure compounds,
particularly from racemic mixtures, is thus one of the major
challenges facing the modern chemical industry. At present,
the majority of commercially available drugs are both syn-
thetic and chiral. However, a number of chiral drugs are still
marketed as racemic mixtures. There are several methods
entially transported, and some
measure of resolution will be
achieved. Chiral separation is
thus achieved by the active
transport of one enantiomer in
a racemic solution from a feed-
ing phase, through a transport
phase containing a chiral host
[a] Dr. A. J. Hallett, Dr. G. J. Kwant, Prof. Dr. J. G. de Vries
DSM Pharmaceutical Products
Innovative Synthesis & Catalysis
P.O. Box 18, 6160 MD Geleen (The Netherlands)
Fax : (+31)46-4767604
compound, into
a
receiving
phase (Figure 1). The transport
phase differs in polarity from
the feeding and receiving
E-mail: Hans-JG.Vries-de@dsm.com
Figure 1. Chiral
through enantioselective trans-
separation
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/chem.200800797.
phases (water and organic sol- port across phase boundaries.
Chem. Eur. J. 2009, 15, 2111 – 2120
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
2111

vent or vice versa). The chiral host compound selectively ex-
tracts one of the enantiomers of the racemic mixture from
the feed phase and delivers it into the receiving phase. Usu-
ally the transport is driven by the concentration gradient of
the substrate between the feeding and the receiving phase,
although other incentives such as pH or temperature shifts
can be applied if full transport is desired.
There are several possible forms of interaction between
the host and the desired enantiomer (guest). These include
ion-pair formation, hydrogen-bonding, p–p interactions and
van der Waals interactions. These interactions result in one
enantiomer binding preferentially to the host compound.
Since the binding is reversible, the guest can also be re-
leased into the receiving phase, although this can be aided
by altering, for example, the pH or temperature of the re-
ceiving phase.
A great variety of chiral host compounds for the separa-
tion of racemates have been reported in the literature.
Chiral macrocyclic hosts, for example, have been used to
separate amino acids, amino acid esters[5] and amines,[6] cal-
ixarenes to separate the enantiomers of amines,[7] guanidini-
um compounds for the separation of carboxylic acids and
amino acids,[8] carbamoylated cinchona derivatives for the
separation of carboxylic acids and N-protected amino
acids,[9] tartaric borate derivatives for the separation of b-
amino alcohols,[10] metal complexes for the separation of
amino acid derivatives,[11] steroidal receptors for the separa-
tion of carboxylic acids,[12] and binaphthol hosts for the sepa-
ration of amines and amino acid esters.[13]
Historically, U-tubes or hollow fibre membrane systems
have been used to demonstrate the ability of a chiral host to
extract an enantiomer selectively from the feed phase and
to deliver it to a receiving phase.[14,15] U-tubes cannot be
scaled up, whereas the membrane systems have several asso-
ciated problems when the process is scaled up for industrial
use. The rate of mass transfer in such a device is too low for
an industrial-scale process. Hollow fibre membranes are lim-
ited by diffusion rate, slowing down transport, and can only
be used with water/nonpolar solvent combinations such as
water/hexane.[15] Use of membranes that had been soaked
with a solution of a chiral cinchona host allowed somewhat
higher rates, but here crystallisation on the membrane sur-
face formed an additional problem, which could be solved
by the addition of an immobilised Cinchona alkaloid.[16]
Maier has reported the use of a centrifugal partition chro-
matograph containing an MTBE solution of bis-1,4-(dihy-
droquinidinyl)phthalazine as the stationary chiral host solu-
tion and was able fully to separate the herbicide 2-(2,4-di-
chlorophenoxy)propionic acid (dichlorprop), which was fed
dissolved in aqueous buffer as the mobile phase.[17] The only
drawback was that full separation could be obtained only
when a host–guest ratio of 1.0 was used.
extraction-based separation process in a fine-chemical
(multi-product) environment it is important to have equip-
ment with high extraction efficiency and a small liquid hold-
up that is compatible with a wide range of solvents. The
small liquid hold-up facilitates fast start-up and reduces the
required amount of the—often chemically complex—expen-
sive host. Although existing mixer devices can be used for
the extraction, phase separation may be very slow and is
likely to become the capacity-limiting step.
Ideally we would like to develop a continuous process by
using a device that combines fast mixing with fast separation
in a small volume. Centrifugal contact separators (CCSs)
such as those produced by CINC have the right characteris-
tics.[18] The CINC Model V2 separator uses centrifugal force
to separate two immiscible liquids of different densities. It
operates both as a mixer and a separator. The immiscible
liquids are rapidly mixed at speeds of up to 6000 rpm, giving
excellent transport across phases. The centrifuge has an ad-
justable weir that can be adjusted to separate the aqueous
and a variety of organic phases. We have previously report-
ed on the use of this device for performing continuous two-
phase catalytic reactions.[19] In this paper we report on the
use of centrifugal contact separators for the separation of
racemates with the aid of a chiral host compound.
Results and Discussion
Synthesis of Cinchona alkaloid host compounds: The Cin-
chona alkaloids (CAs) represent a class of compounds likely
to provide enantioselective receptors because they have
multiple functional groups and rigid skeletons with distinct
conformational preferences. There are three positions in
which the chemistry of the Cinchona alkaloid can be altered.
The hydroxy group can be allowed to react with, for exam-
ple, an isocyanate to give a carbamoylated Cinchona alka-
loid. The double bond can be employed to immobilise the
CA to a silica support to create a chiral stationary phase or
to add a long-chain lipophilic substituent. Finally, the me-
thoxy group on the quinoline moiety can be altered to a
range of alkoxy groups.
So far, the only physical separation of racemates that is
used on production scale is the chiral simulated moving bed
(SMB) process. This method is rather costly because the
chiral host is immobilised on a solid phase and the equip-
ment is expensive because of the higher pressures. For an
It has been shown by Lindner et al. that O-(1-adamantyl-
carbamoyl)-11-octadecylsulfinyl-10,11-dihydroquinine
can
selectively extract one enantiomer of an N-protected amino
acid derivative, such as 3,5-dinitrobenzoylleucine (DNB-
Leu), from an aqueous buffered solution into an organic
2112
www.chemeurj.org
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 2111 – 2120

Continuous Separation
FULL PAPER
phase containing the chiral selector. These were stoichio-
metric experiments, in which two equivalents of racemate
(with respect to the chiral host compound) were used. The
variables explored were organic solvent, host/guest ratio, pH
and buffer concentration of the aqueous phase and the
structures of different N-protected leucine derivatives.
Yields were 70–80% and ee values of 90% were observed in
some cases.[20] Maier et al. have reported the use of these
hosts for the separation of DNB-Leu in a membrane device.
They achieved complete separation by using a number of
membranes in series.[16] Here we report the results of experi-
ments in which these host compounds are used in catalytic
amounts for the extractive separation of racemates.
Since the long-chain lipophilic CAs had performed so
well in the reported research we decided to use these host
compounds to explore the use of CCSs for continuous chiral
separations. We prepared a range of derivatised CAs as
shown in Scheme 1. These CAs were examined for their
Scheme 1. Preparation of lipophilic carbamoylated CAs.
are only suited to give data on the equilibrium distributions
of enantiomers between the phases and the binding to the
host. Under the right mixing conditions (fast mixing), how-
ever, they give information on the binding rate (intrinsic) as
well. The model is based on a “simple” mass balance for the
two enantiomers. Binding of an enantiomer to the host in
the organic phase is described by a Langmuir isotherm (il-
lustrated for enantiomer 1):
K1C1
1þK1C1þK2C2
ð1Þ
CH,1 ¼ CH,0
where C1 and C2 are the concentrations of the two enantio-
mers in the aqueous phase and CH,0 is the initial (or total)
host concentration. The total concentration of an enantio-
mer in the organic phase is assumed to be composed of the
amount bound to the host and the physically extracted
matter, described by the distribution coefficient m1:
K1C1
ð2Þ
COrg,1 ¼ m1C1þC
H,0 1þK1C1þK2C2
The model results are calculated from the mass balance for
every enantiomer over the equilibrium cell:
Va,1ðC1,oꢀC1Þ ¼ VOrgðCOrg,1,oꢀCOrg,1
Þ
ð3Þ
where the intrinsic selectivity is defined as the ratio of ad-
abilities selectively to extract DNB-Leu, 3,5-dinitrobenzoyl-
alanine (DNB-Ala), 3,5-dinitrobenzoylvaline (DNB-Val),
3,5-dinitrobenzoylphenylalanine (DNB-Phe) and 3,5-dinitro-
benzoylphenylglycine (DNB-PhG) from the aqueous phase
into an organic phase. All of the synthesised CAs were pre-
pared from either quinine or cinchonidine and so have the
(8S,9R) configuration.
sorption constants (K1/K2) and the operational selectivity as
the ratio of concentrations for the organic phase (COrg,1
/
COrg,2) and for the aqueous phase (C1/C2). From the analysis
we may obtain the values for the initial concentration of
each enantiomer in the aqueous phase, the concentration of
each enantiomer in the aqueous phase at equilibrium and
thus the total yield, the yield of each enantiomer and the en-
antiomeric excess.
Two-phase extraction experiments: Two-phase extraction ex-
periments were carried out in beakers. In principle, these
Two-phase extraction experiments can be used to deter-
mine the ideal conditions for selective chiral extraction.
Chem. Eur. J. 2009, 15, 2111 – 2120
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
2113

J. G. de Vries et al.
There are a number of variables to consider, including the
pH of the aqueous phase, the relative concentrations of host
and guest, the organic solvent and the temperature.
with tert-butyl methyl ether and no host complex was per-
formed; this resulted in a similar high yield, showing that in
this case direct extraction had taken place, bypassing the in-
termediacy of the chiral host. The highest yields and selec-
tivities obtained in solvents other than heptane were gener-
ally obtained with hosts containing the adamantyl carbamoyl
group and the neopentoxy substituent on the quinoline.
Chlorinated solvents gave similar results in analogous ex-
periments: CA D, for example, gave yields of 27 and 26%
and selectivities of 54 and 57% in dichloromethane and 1,2-
Enantioselective extraction of N-(3,5-dinitrobenzoyl)-d,l-
leucine: The CAs were initially used in the two-phase ex-
traction of d,l-DNB-Leu (Table 1). In these experiments,
d,l-DNB-Leu in a buffered aqueous phase (2 mL, 1 mmol
solution) was mixed with a CA in an organic solvent (2 mL,
0.5 mmol solution). There are therefore two molecules of
amino acid derivative (or one of each enantiomer) for each
molecule of CA. The theoretical maximum yield is 50%. In
all cases it was ascertained that the true equilibrium had
been reached by following the extraction over time until
constant values were obtained. The highest yields and selec-
tivities were observed when the aqueous feed phase was at
pH 6. The lowest yields, and generally selectivities, were ob-
served at pH 9.
dichloroethane, respectively (Experiments 2 and
Table 3).
4
in
Enantioselective extraction of other DNB-amino acids:
Two-phase extraction experiments were also performed with
other DNB-protected amino acids, including d,l-DNB-Phe,
d,l-DNB-Ala, d,l-DNB-Val and d,l-DNB-PhG (Table 2).
For a full list of two-phase extraction experiments with all
amino acid derivatives see the Supporting Information.
Table 1. Yields and selectivities for extraction of N-(3,5-dinitrobenzoyl)-
d,l-leucine.
Table 2. Yields and selectivities for extraction of DNB-Leu.
Exp.
Host
Organic solvent
pH
Yield
[%]
ee [%]
Exp.
Host
Organic solvent
pH
Yield [%]
ee [%]
28
29
30
31
32
33
34
35
36
37
38
39
40
41
CA A
CA B
CA D
CA F
CA G
CA H
CA D
CA F
CA G
CA H
CA D
CA D
CA D
CA D
dichloroethane
dichloroethane
dichloroethane
dichloroethane
dichloroethane
dichloroethane
dichloroethane
dichloroethane
dichloroethane
dichloroethane
dichloroethane
heptane
6
6
6
6
6
6
7
7
7
7
9
6
6
6
58
46
46
37
52
22
13
10
20
6
5
1
2
3
4
5
6
7
8
CA D
CA D
CA D
CA D
CA D
CA D
CA D
CA D
CA D
CA D
CA D
CA D
CA D
CA D
CA D
CA E
CA F
CA F
CA F
CA G
CA G
CA G
CA H
CA H
CA H
CA A
CA B
heptane
6
6
6
6
6
7
7
7
7
7
9
9
9
9
9
6
6
6
6
6
6
6
6
6
6
6
6
38
27
71
26
20
21
14
27
18
7
3
7
8
7
4
7
8
25
83
24
36
86
17
16
83
49
32
85[a]
54
14
57
1
19
19
30
19
19
38
55
30
2
dichloromethane
tert-butyl methyl ether
dichloroethane
toluene
heptane
96[a]
3
dichloromethane
tert-butyl methyl ether
dichloroethane
toluene
7
5
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
40
29
15
20
12
6
64
2
67
6
20
59
6
23
42
6
0
0
heptane
50
33
14
43
15
6
dichloromethane
tert-butyl methyl ether
dichloroethane
toluene
dichloromethane
heptane
dichloromethane
tert-butyl methyl ether
heptane
dichloromethane
tert-butyl methyl ether
heptane
chloroform
toluene
Use of a pH 6 aqueous phase results in the highest yields,
as in the case of the DNB-Leu substrate. With this racemate,
however, pH 7 had given the highest selectivities, whereas at
pH 9 no extraction was observed. As in the case of N-(3,5-
dinitrobenzoyl)leucine, extractions in heptane gave the best
result (50% yield, ee=43%, Exp. 39, Table 2). Use of tolu-
ene again resulted in the lowest yield and selectivity (14%
yield, ee=6%, Exp. 41).
dichloromethane
tert-butyl methyl ether
dichloromethane
dichloroethane
13
55
The yields and selectivities obtained with CA D in hexane
(Figure 2) and in 1,2-dichloroethane (Figure 3) are shown
for each of the amino acid derivatives used. In this case the
yield is represented as the percentage of the maximum that
can be extracted. With CA D in heptane the highest selec-
tivities are observed in the separation of DNB-Leu and
DNB-Val. Selectivities for DNB-Phe and DNB-PG are
moderate. However, the yield in the extraction of DNB-Phe
is excellent. Both the yield and selectivity for DNB-Ala are
low. With CA D in 1,2-dichloroethane the yields are similar
to those above but the selectivities are generally lower, with
the exception of DNB-Ala, for which the ee has greatly in-
[a] The high enantioselectivities obtained in these experiments are proba-
bly due to classical resolution by crystallisation of a diastereomeric salt.
Vide infra.
The use of heptane as the organic solvent seemed to
induce the highest selectivities. Chlorinated and aromatic
solvents can disrupt the interactions—such as the hydrogen
bonding and the p–p stacking—between the host and the
guest. Use of tert-butyl methyl ether resulted in unexpected-
ly high yields of 70–80% for all host complexes. Since the
theoretical maximum yield is 50%, a control experiment
2114
www.chemeurj.org
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 2111 – 2120

Continuous Separation
FULL PAPER
Figure 2. Extraction of DNB-protected amino acids by CA D in heptane;
&
&
yield: , ee:
.
Figure 4. Schematic cross-section of
a centrifugal contact-separator
(Courtesy of CINC-Solutions, the Netherlands).
Figure 3. Extraction of DNB-protected amino acids by CA D in 1,2-di-
&
&
chloroethane; yield: , ee:
.
creased. Again, the leucine and valine derivatives are ex-
tracted with higher selectivities than the phenylalanine and
phenylglycine derivatives.
Enantioselective enrichment by centrifugal separation: A
centrifugal separator (CCS) is in essence a centrifuge
(Figure 4). The immiscible liquid phases are introduced by
pump in the narrow annular mixing zone between the out-
side of the rotor and the inside of the outer housing. Here,
very efficient and fast mixing between the two phases
occurs, which is highly conducive towards high mass trans-
fer. The dispersion is then sucked inside the centrifuge,
where the two phases are gradually but very efficiently sepa-
rated whilst moving upwards, after which they leave the
device through separate exits. In addition to the fast mixing
and separating, another major advantage of the CCS over
hollow fibre membranes is the ability to use a variety of
polar and nonpolar solvents. There are two different designs
of transport experiments: with and without recycling. In our
experiments we have employed the recycling setup
(Figure 5), in which the aqueous phases are fed into the
CCSs and, after separation from the organic phase, are re-
turned to the original container. Two CCS units and three
peristaltic pumps are employed in the setup for extraction
experiments (Figure 5).
Figure 5. Setup of the CINC model V2 separators for extraction experi-
ments.
ferent densities. An aqueous phase can therefore be mixed
with a range of organic solvents such as alkanes, cycloal-
kanes, halogenated solvents (chloroform, dichloromethane,
1,2-dichloroethane etc.), higher alcohols and aromatic sol-
vents. The aqueous phase can include buffer solutions to
maintain constant pH values, or salts such as KBr to provide
counter-transport of—in this case—anions.
A variety of CA hosts were tested in the separation of
d,l-DNB-protected amino acids under different conditions.
Samples both from the feed phase and from the receiving
phase were taken over time. The first transport experiment
performed in the CCS made use of the CA host O-(1-tert-
butylcarbamoyl)-11-octadecylsulfinyl-10,11-dihydroquinine
(CA D) as the host compound and heptane as the solvent.
This is the combination that gave the highest yield (38%)
The only criteria for the use of solvents in the CCS are
that the two chosen solvents must be immiscible and of dif-
Chem. Eur. J. 2009, 15, 2111 – 2120
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
2115

J. G. de Vries et al.
and selectivity (85%) in the two-phase extraction experi-
ments (Exp. 1). However, there were difficulties in dissolv-
ing DNB-Leu at pH 6 at higher concentrations, so the feed
phase was prepared at pH 6.5. The receiving phase was pre-
pared at pH 9 in order to “strip” the amino acid derivative
from the host. After three hours of operation, there was no
DNB-Leu observed in the receiving phase. The concentra-
tion of DNB-Leu in the feed phase had decreased by 10%
(equal to the concentration of host compound) and a white
solid was present in the feed CCS at the completion of the
experiment. It appeared that DNB-Leu was forming a salt
with the host compound (as the concentration in the feed
phase was reduced) but was not being delivered to the re-
ceiving phase. KBr was added to the receiving phase to aid
the removal of the substrate from the host molecule. Again,
after three hours, no DNB-Leu was observed in the receiv-
ing phase and there was precipitation of the host/guest on
complexation. In retrospect, this experiment suggests that
the high ee obtained in the 2:1 screening experiments was
due to classical resolution through preferential crystallisa-
tion of one of the diastereomeric salts. To overcome this
problem, a more polar solvent—1,2-dichloroethane—was
employed (Table 3).
It can be seen that the change of solvent results in com-
plete transport of DNB-Leu from the feed phase to the re-
ceiving phase (Figure 6). It can also be seen that enantiose-
lective transport takes place. The concentration of the R en-
Figure 6. a) Enantiomeric separation of N-(3,5-dinitrobenzoyl)-d,l-leu-
~
&
^
cine with use of CA D and two CCSs ( =(S) (feed), =(R) (feed),
=
&
*
^
(R) (receiving), =(S) (receiving)). b) ee of feed ( ), ee of receiving ( ).
Table 3. Conditions for enantioselective transport in two CCSs.
20%. The conditions used in
this experiment will from now
on be defined as the standard
conditions, and in all other ex-
periments performed their devi-
ations from these conditions
composition of feeding phase and flow rate
0.1m phosphate buffer (pH 6.5), 5 mm substrate, 500 mL
H2O, 625 mLminꢀ1
composition of receiving phase and flow rate
0.1m phosphate buffer (pH 9), 0.1m KBr, 1000 mL H2O,
625 mLminꢀ1
transport phase: host compound, concentration, sol- CA D, 0.5 mm, 500 mL 1,2-dichloroethane, 625 mLminꢀ1
vent and flow rate
rotational speed
feed CCS 3500 rpm
will be highlighted. The results
stripping CCS 3500 rpm
of other experiments will also
be compared with the results
from these standard conditions.
^
antiomer ( ) in the feed phase is reducing at a faster rate
The CA host O-(1-tert-butylcarbamoyl)-11-octadecylsul-
finyl-10,11-dihydroquinine (CA D) was used for this experi-
ment because it had given the highest yield and selectivity
in two-phase extraction experiments. However, this had
been the case when the organic solvent used was heptane.
The highest yield and selectivity observed in 1,2-dichloro-
ethane had been achieved with the CA host O-(1-adaman-
tylcarbamoyl)-11-octadecylsulfinyl-10,11-dihydroquinine
(CA G). This host led to a yield of 36 and 59% enantiose-
lectivity in 1,2-dichloroethane (Exp. 21), in comparison with
a yield of 27% and a selectivity of 54% for CA D in the
same solvent (Exp. 2). The CCS extraction experiment was
repeated under identical conditions with the CA G host
(Figure 7).
&
than that of the S enantiomer ( ). This is also observed in
the receiving phase, where the concentration of the R enan-
tiomer ( ) is increasing at a faster rate than that of the S en-
~
*
antiomer ( ). Eventually both enantiomers are transported
completely to the receiving phase. The transport is relatively
fast with the CCS (ca. 90 min for 5 mmol of amino acid) in
comparison with enantiomeric separations using hollow
fibre membranes.[15] The course of ee change in feeding and
receiving phases in this experiment clearly shows that initial-
ly (S)-DNB-Leu is transported with an ee of 58%. (Fig-
ure 6b) This number quickly erodes over time as increasing-
ly also the R enantiomer is being transported, due by the
fact that the rate of transport is proportional to the concen-
tration in the feeding phase. Interestingly, it would also be
possible to recover the R enantiomer from the feeding
phase, where after 90 min an ee of 83% has been reached,
although the yield of DNB-Leu has by now declined to
Consistently with the two-phase extraction experiments,
the transport is faster in this experiment than in the stan-
dard one. Also, the selectivity has slightly increased. The
highest selectivity in the two-phase experiments had been
2116
www.chemeurj.org
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 2111 – 2120

Continuous Separation
FULL PAPER
Figure 7. Enantiomeric separation of DNB-d,l-leucine with CA G with
the setup of Figure 5 (for legend see Figure 6).
Figure 8. Enantiomeric separation of 3,5-dinitrobenzoyl-d,l-leucine with
CA F and CCS extractors (for legend see Figure 6).
obtained with the CA host O-(1-tert-butylcarbamoyl)-6-neo-
pentoxy-11-octadecylsulfinyl-10,11-dihydrocinchonidine (CA
F). This host had induced a selectivity of 67%, but the yield
had been relatively low at 25% (Exp. 18). Again, the CCS
extraction experiment was performed under the standard
conditions with CA F host (Figure 8). Host CA F gave a
greater selectivity but a lower rate of transport, in accord-
ance with the two-phase extraction experiments.
and less efficient extraction.[19] Conversely, lowering the
speed of mixing to 2800 rpm resulted in slower transport but
higher selectivity. Having the feed CCS set at 2800 rpm and
the stripping CCS set at 4800 rpm resulted in a higher selec-
tivity but slower transport. The results are almost identical
to those obtained when both CCS extractors were set at
2800 rpm. This suggests that in the current setup the rota-
tional speed of the feed CCS is more important for control-
ling yield and selectivity than the rotational speed of the
stripping CCS.
In the experiments described thus far, the concentration
of host has been equal to 10% of the concentration of sub-
strate. Increasing the concentration of host results, as ex-
pected, in faster transport but lower selectivity. On lowering
the flow rate of the transport phase by a third and, at the
same time, increasing the concentration of host by a third,
the same number of molecules passed through the CCS ex-
tractors per minute but at a higher concentration, which re-
sulted in faster transport but lower selectivity.
Many extraction experiments using the centrifugal separa-
tors were performed with different hosts and substrates and
different flow rates, rotational speeds and concentrations.
For a complete listing and graphical representation of each
experiment see the Supporting Information. Use of CA B
resulted in faster transport but slightly lower selectivity.
Without KBr in the receiving phase, the transport was
much slower. A number of explanations for this effect are
possible. It is assumed that at the pH used during transport
the host is present mostly in its protonated form. This
means that when the DNB-LEU is stripped into the receiv-
ing phase an exchange of counter-ions occurs. If no bromide
is present, more polar anions such as hydroxide or phos-
phate would lead to a poorly soluble host compound. In ad-
dition, the thermodynamics for anion exchange may become
unfavourable. For fast transport the host should not all be in
the host–guest form, because this would lead to the extrac-
tion becoming the rate-determining step. A subtle balance
between counter-ions is thus necessary, and this so far seems
to be optimal with bromide.
Changing the substrate to DNB-Phe resulted in generally
fast transport but with lower selectivity than observed with
DNB-Leu (For CAs B, D and F). Use of CA D led to
slower transport of DNB-Val than of the other two sub-
strates.
Outlook
Increasing the speed of rotation in each CCS from 3500 to
4800 rpm resulted in lower selectivity. This is due to the
characteristics of the equipment: at higher speed the volume
of the mixed phase reduces, leading to shorter contact time
In order to achieve full separation of both enantiomers a
number of CCSs in series has to be used, with the aqueous
and organic streams running in countercurrent directions.
One enantiomer will be obtained in pure form from the
Chem. Eur. J. 2009, 15, 2111 – 2120
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
2117

J. G. de Vries et al.
HPLC analyses of the 3,5-dinitrobenzoyl-protected amino acids were car-
ried out with a Chirobiotic T column (250ꢁ4.6 mm ID) from Astec, Inc.
CAT 12024. A mixture of acetonitrile (75%), methanol (25%), acetic
acid (0.25%) and triethylamine (0.25%) (polar ionic mode: pim mode)
was used as eluent. The analysed samples contain feeding (or receiving)
phase (0.1 mL) and eluent (0.9 mL). The flow rate was 1.2 mLminꢀ1. The
injection volume was 20 mL, and detection was performed with a spectro-
photometer at UV 270 nm. All analyses were performed at ambient tem-
perature.
aqueous stream (exit of CCS no. 1); the other enantiomer
will be continuously extracted from the organic phase in
CCS no. 5. We are currently studying this system in more
detail, allowing us to calculate the necessary number of
CCSs and the other process parameters.[21]
We would expect it to be possible to separate an entire
class of compounds, such as DNB-amino acids, with a single
host compound. Although there will be differences in selec-
tivity for each substrate, these can be compensated for by
using more CCSs in series. Early calculations show it is pos-
sible to produce 10–20 kg of pure enantiomer per week in
the setup shown in Figure 9. With the largest commercially
Retention times for the first-eluting enantiomer l-DNB-Leu and the
second-eluting enantiomer d-DNB-Leu were 3.92 and 7.00 min, respec-
tively. Retention times for the first-eluting enantiomer l-DNB-Phe and
the second-eluting enantiomer d-DNB-Phe were 3.74 and 6.13 min, re-
spectively. Retention times for the first-eluting enantiomer l-DNB-Val
and the second-eluting enantiomer d-DNB-Val were 3.68 and 5.41 min,
respectively. Retention times for the first-eluting enantiomer l-DNB-Ala
and the second-eluting enantiomer d-DNB-Ala were 4.26 and 5.41 min,
respectively. Retention times for the first-eluting enantiomer l-DNB-Phg
and the second-eluting enantiomer d-DNB-Phg were 3.39 and 10.66 min,
respectively.
Conditions for two-phase extraction experiments: Two-phase extraction
experiments were carried out with a mixture of the CA host compound
in an organic solvent (2 mL, 0.5 mmol solution) and the substrate in a
buffered aqueous solution (2 mL, 1 mmol solution). The phases were
stirred in a sample vial (20 mL) overnight and the mixture was then al-
lowed to stand for 1 h in order to allow phase separation. Concentrations
of each enantiomer in the aqueous phase were determined by chiral
HPLC analysis. Extraction yields and selectivity were calculated by com-
parison with the analysis of the original starting aqueous phase.
Figure 9. Full separation of racemates with a series of CCSs in a counter-
current setup.
available CCS a production of 5–10 tons per week should be
feasible. If the host compound is sufficiently stable—and so
far we have not detected any decomposition products—it
should be possible to achieve turnover numbers on host of
400–700 per week, which leads to perfectly acceptable eco-
nomics. We believe that the method described in this paper
has potential similar to that of chiral SMB for the ton-scale
separation of racemates, but at much reduced cost.
Conditions for CCS extraction experiments: For the CCS extraction ex-
periments, two CINC model V2 separators were employed with use of
the recycle setup. The feed and transport phases were thus pumped into
the first CCS with peristaltic pumps (Watson Marlow 101U/R) set at
70 rpm, giving a flow rate of 625 mLminꢀ1 for each phase. After mixing
and separation of the phases, the aqueous phase was returned to the orig-
inal feed solution, whereas the organic phase was transferred to a second
CINC model V2 separator. The receiving phase was pumped into the
second CINC model V2 separator with
a peristaltic pump (Watson
Marlow 101U/R) also set at 70 rpm, giving a flow rate of 625 mLminꢀ1
.
Other host compounds will need to be developed for the
separation of other classes of racemates. A patent for this
continuous flow system has been filed.[22]
After mixing and separation of the phases, the aqueous phase was re-
turned to the original receiving solution and the organic phase was re-
turned to the original transport solution. There are therefore three closed
loops in the system: two aqueous and one organic.
The tubing used for the experiments depends on the solvents. For the
aqueous phases, Marprene (Watson Marlow, 8 mm 5/16’’, serial number
902.0080.016) was used in the peristaltic pumps to drive the solutions to
the CCSs. For the organic phase (chlorinated solvents), Viton (Watson
Marlow, 8 mm 5/16’’, serial number 970.0080.016) was used in the peri-
staltic pump. All of the outlets of the CCSs were fitted with chemically
resistant T-tubes, allowing air to enter the CCS in order to prevent a
pressure build-up.
Conclusion
In conclusion, we have shown that it is possible to separate
a racemic mixture of an N-acylated amino acid by extraction
across a liquid membrane containing a chiral host compound
based on a Cinchona alkaloid. We have now shown for the
first time that it is possible to perform this separation in
continuous mode by use of a catalytic amount of chiral host
compound in a centrifugal contact separator. Further work
directed towards full separation of racemates by the same
principles is in progress.
The tubing from the outlets of the CCSs also depends on the solvents.
For the aqueous phase, Sta-Pure Silicone (Watson Marlow, 16 mm/
22 mm, serial number 910.0159.032) was used to return the phases to the
original solutions, although any tubing can be employed for this. For the
organic phase (chlorinated solvents), ꢂso-Versinic, Verneret (Merck Eu-
rolab, 15 mm/21 mm, serial number VERN770460-01) was used to trans-
port the phase between the two CCSs and to return the phase to the orig-
inal solution.
The experiment was started by setting the CCSs to the required rpm and
pumping the organic phase (the heavier phase) into the first CCS. After
5–10 seconds, the feed phase was also pumped into the first CCS. Once
the aqueous solution could be seen returning to the feed phase, the re-
ceiving phase was pumped into the second CCS. When only one phase is
within the CCSs, the solution will exit by the heavy-phase outlet.
Experimental Section
General: All starting materials were purchased from commercial chemi-
cal suppliers and were used as received. Organic solvents were purchased
from Fluka, were of dried reagent grade and were used without further
purification.
The solutions were pumped continuously through the system for 3 h,
during which samples were taken both from the feed and from the re-
2118
www.chemeurj.org
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 2111 – 2120

Continuous Separation
FULL PAPER
ceiving phases. Samples were taken at various times between 0 and 180
minutes and were analysed by chiral HPLC.
Collins, G. N. Sheldrake, J. Crosby), Wiley, New York, 1992; d) Chir-
ality in Industry II: Developments in the Commercial Manufacture
and Applications of Optically Active Compounds (Eds.: A. N. Col-
lins, G. N. Sheldrake, J. Crosby), Wiley, New York, 1997; e) R. A.
Sheldon, Chirotechnology, Industrial Synthesis of Optically Active
Compounds, Marcel Dekker, New York, 1993; f) S. Kotha, Tetrahe-
dron 1994, 50, 3639–3662; g) Handbook of Chiral Chemicals (Ed.:
D. J. Ager), Marcel Dekker, New York, 1999; h) Handbook of
Chiral Chemicals, 2nd ed. (Ed.: D. J. Ager), CRC, Boca Raton,
2005.
For a typical experiment (Exp. 3), the feed phase consisted of 3,5-dinitro-
benzoyl-d,l-leucine (0.82 g, 5 mmol) in distilled water (500 mL). The so-
lution was set at pH 6.5 with a phosphate buffer (0.1m). The transport
phase was O-(1-tert-butylcarbamoyl)-11-octadecylsulfinyl-10,11-dihydro-
quinine (CA D) (0.182 g, 0.5 mmol) in 1,2-dichloroethane (500 mL). The
receiving phase consisted of KBr (11.9 g, 0.1m) in distilled water
(1000 mL). The solution was set at pH 9 with a phosphate buffer (0.1m).
All solutions were pumped into the CCSs at 625 mLminꢀ1, and the CCSs
were set at 3500 rpm.
[2] H. Schoemaker, D. Mink, M. Wubbolts, Science 2003, 299, 1694–
1697.
Synthesis of chiral hosts: The chiral hosts based on Cinchona alkaloids
were synthesised as described by Lindner and Lꢃmmerhofer.[23] Quinine,
cinchonidine or 6-neopentoxycinchonidine were first carbomoylated with
1-adamantyl, 1-cyclohexyl or 1-tert-butyl isocyanate. The resulting carba-
mate was then treated with the long-chain thiol to give the thioether,
which could in turn be oxidised to the desired product.
[3] a) E. N. Jacobsen, Comprehensive Asymmetric Catalysis (Eds.: A.
Pfaltz, H. Yamamoto), Springer, Berlin, 1999; b) Asymmetric Cataly-
sis on Industrial Scale: Challenges, Approaches and Solutions (Eds.:
H. U. Blaser, E. Schmidt), Wiley-VCH, Weinheim, 2004; c) J. G. de
Vries in Encyclopedia of Catalysis, Vol. 3 (Ed.: I. T. Horvath), Wiley,
New York, 2003, p295–347; d) H. Kumobayashi, Recl. Trav. Chim.
Pays-Bas 1996, 115, 201–210.
[4] a) Chiral Separations-Application and Technology (Ed.: S. Ahuja),
ACS Professional Reference books, ACS, Washington, 1997;
b) Chiral Separation Techniques-A Practical Approach (Ed.: G. Sub-
ramanian), Wiley-VCH, Weinheim, 2001; c) Enantiomer Separation-
Fundamentals and Practical Methods (Ed.: F. Toda), Kluwer Aca-
demic, Dordrecht, 2004.
[5] a) S. C. Peacock, L. A. Domeier, F. C. A. Gaeta, R. C. Helgeson,
J. M. Timko, D. J. Cram, J. Am. Chem. Soc. 1978, 100, 8190–8202;
b) D. S. Lingenfelter, R. C. Helgeson, D. J. Cram, J. Org. Chem.
1981, 46, 393–406; c) M. Sawada, Y. Takai, H. Yamada, J. Nishida,
T. Kaneda, R. Arakawa, M. Okamoto, K. Hirose, T. Tanaka, K.
Naemura, Perkin Trans. 2 1998, 701–710.
[6] M. Steensma, N. J. M. Kuipers, A. B. de Haan, G. J. Kwant, Chirality
2006, 18, 314–328.
[7] I. S. Antipin, I. I. Stoikov, E. M. Pinkhassik, N. A. Fitseva, I. Stibor,
A. I. Konovalov, Tetrahedron Lett. 1997, 38, 5865–5868.
[8] P. Blondeau, M. Segura, R. Perez-Fernandez, J. de Mendoza, Chem.
Soc. Rev. 2007, 36, 198–210.
[9] a) M. Lꢃmmerhofer, W. Lindner, J. Chromatogr. A 1996, 741, 33–
48; b) P. Franco, J. Blanc, W. R. Oberleitner, N. M. Maire, W. Lind-
ner, C. Minguillon, Anal. Chem. 2002, 74, 4175–4183.
[10] a) Y. Abe, T. Shoji, M. Kobayashi, W. Qing, N. Asai, H. Nishizawa,
Chem. Pharm. Bull. 1995, 43, 262–265; b) Y. Abe, T. Shoji, S. Fukui,
M. Sasamoto, H. Nishizawa, Chem. Pharm. Bull. 1996, 44, 1521–
1524.
[11] a) T. Takeuchi, S. Yamazaki T. Tanimura, Anal. Chim. Acta 1991,
242, 291–294; b) T. B. Reeve, J.-P. Cros, C. Gennari, U. Piarulli, J. G.
de Vries, Angew. Chem. 2006, 118, 2509; Angew. Chem. Int. Ed.
2006, 45, 2449–2453.
6-Hydroxycinchonidine: 6-Hydroxycinchonidine was prepared by the lit-
erature procedure;[24] BBr3 (80 mL, 1m solution in CH2Cl2, 80.0 mmol)
was thus slowly added at ꢀ788C to
a solution of quinine (6.51 g,
20.1 mmol) in CH2Cl2 (500 mL). The mixture was allowed to reach ambi-
ent temperature and was then heated to 408C for 1 h. After the mixture
had cooled to 58C, NaOH (150 mL, 10% soln.) was added. The aqueous
phase was separated, washed with CH2Cl2 (150 mL) and acidified with
HCl (50 mL, 37% soln.). The acidic solution was in turn brought to
pH 9.5 with NH4OH (58%), and the product was extracted with butan-1-
ol (2ꢁ200 mL). After drying over Na2SO4 and filtering, the solvent was
removed under reduced pressure. Yield 5.04 g, 16.2 mmol (81%).
6-Neopentoxy-cinchonidine: 6-Neopentoxy-cinchonidine was prepared by
the literature procedure;[25] finely powdered caesium carbonate (7.56 g,
23.2 mmol, dried at 1208C in vacuo for 16 h) and neopentyl bromide
(3 mL, 23.8 mmol) were thus added to a solution of 6-hydroxycinchoni-
dine (5.04 g, 16.2 mmol, dried at 1208C in vacuo for 1 h) in N-methylpyr-
rolidone (50 mL). The mixture was heated to 1308C with stirring for
24 h, after which it was allowed to reach ambient temperature. Water
(300 mL) was added, and the product was extracted with toluene (5ꢁ
100 mL). The combined organic phases were washed with water (3ꢁ
200 mL) and brine (3ꢁ200 mL) and were then dried over MgSO4. The
mixture was filtered, and the solvent was removed under reduced pres-
sure. Yield 2.93 g, 7.5 mmol (45%).
(3,5-Dinitrobenzoyl)-protected amino acids: The N-(3,5-dinitrobenzoyl)-
protected amino acids were prepared in quantitative yields by the litera-
ture procedure.[26] Thus, to obtain N-(3,5-dinitrobenzoyl)-d,l-leucine, 3,5-
dinitrobenzoyl chloride (9.35 g, 40.5 mmol) was added to a suspension of
d,l-leucine (5.32 g, 40.5 mmol) in dry THF (100 mL). The mixture was
cooled in an ice bath, and propylene oxide (2.9 mL, 41.4 mmol) was
added dropwise. After stirring for 2 h at ambient temperature, the solu-
tion was filtered through celite, and the solvent was removed under re-
duced pressure to give the desired product. N-(3,5-Dinitrobenzoyl)-d,l-
valine, N-(3,5-dinitrobenzoyl)-d,l-alanine, N-(3,5-dinitrobenzoyl)-d,l-
phenylalanine and N-(3,5-dinitrobenzoyl)-d,l-phenylglycine were pre-
pared in a similar manner.
[12] A. P. Davis, Coord. Chem. Rev. 2006, 250, 2939–2951.
[13] a) D. Wang, T. J. Liu, W. C. Zhang, W. T. Slaven, C. J. Li, Chem.
Commun. 1998, 1747–1748; b) T. J. Liu, Y. J. Chen, K. S. Zhang, D.
Wang, D. W. Guo, X. Z. Yang, Chirality 2001, 13, 595–600.
[14] W. H. Pirkle, W. E. Bowen, Tetrahedron: Asymmetry 1994, 5, 773–
776.
[15] B. BaragaÇa, A. G. Blackburn, P. Breccia, A. P. Davis, J. de Mendo-
za, J. M. Padron-Carrillo, P. Prados, J. Riedner, J. G. de Vries, Chem.
Eur. J. 2002, 8, 2931–2936.
[16] A. Maximini, H. Chmiel, H. Holdik, N. W. Maier, J. Membr. Sci.
2006, 276, 221–231.
Acknowledgements
We thank Josꢄ Padron, Andrꢄ de Haan and Amrish B. Bahara for helpful
discussions, calculations and help with the initial experiments. We thank
the European commission for financial support for this project, which
was part of a Research Training Network “Enantioselective Recognition:
Towards the Separation of Racemates” (HPRN-CT-2001-00182)
[17] E. Gavioli, N. W. Maier, C. Minguillꢅn, W. Lindner, Anal. Chem.
2004, 76, 5837–5848.
[18] D. H. Meikrantz, L. L. Macaluso, H. W. Sams, III, C. H. Chardin, Jr.,
A. G. Federici, US 5762800, 1998, to Costner Industries Nevada,
Inc. See also: http://www.auxill.nl/uk/cinc.separators.php
[19] G. N. Kraai, F. van Zwol, B. Schuur, H. J. Heeres, J. G. de Vries,
Angew. Chem. 2008, 120, 3969–3972; Angew. Chem. Int. Ed. 2008,
47, 3905–3908.
[1] a) J. Jacques, A. Collet, S. H. Wilen, Enantiomers, Racemates and
Resolutions, Wiley, New York, 1981; b) J. Crosby, Tetrahedron 1991,
47, 4789–4846; c) Chirality in Industry, The Commercial Manufac-
ture and Applications of Optically Active Compounds (Eds.: A. N.
[20] K. H. Kellner, A. Blasch, H. Chmiel, M. Lꢃmmerhofer, W. Lindner,
Chirality 1997, 9, 268–273.
Chem. Eur. J. 2009, 15, 2111 – 2120
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
2119

J. G. de Vries et al.
[21] a) B. Schuur, W. J. Jansma, J. G. M. Winkelman, H. J. Heeres, Chem.
Eng. Process. 2008, 47, 1484–1491; b) B. Schuur, J. Floure, A. J. Hal-
lett, J. G. M. Winkelman, J. G. de Vries, H. J. Heeres, Org. Process
Res. Dev. 2008, 12, 950–955.
[22] J. G. de Vries, G. Kwant, A. J. Hallett, Continuous process for the
enantioselective enrichment and separation of enantiomers using
centrifugal separation, EP 1676614, 2006 to DSM IP Assets BV.
[23] W. Lindner, M. Lꢃmmerhofer, N. Maier, Cinchonan based chiral se-
lectors for separation of stereoisomers, WO97/046557.
[24] L. D. Small, H. Rosenberg, P. U. Nwangwu, T. L. Holcslaw, S. J.
Stohs, J. Med. Chem. 1979, 22, 1014–1016.
[25] N. M. Maier, S. Schefzick, G. M. Lombardo, M. Feliz, K. Rissanen,
W. Lindner, K. B. Lipkowitz, J. Am. Chem. Soc. 2002, 124, 8611–
8629.
[26] W. H. Pirkle, T. C. Pachapsky, J. Am. Chem. Soc. 1987, 109, 5975–
5982.
Received: April 25, 2008
Revised: October 1, 2008
Published online: January 13, 2009
2120
www.chemeurj.org
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 2111 – 2120