Enantiomer separation by countercurrent chromatography using cinchona alkaloid derivatives as chiral selectors

Article abstract of DOI:10.1021/ac020209q

Cinchona-derived anion-exchange type chiral selectors were used in countercurrent chromatography (CCC) for the separation of enantiomers of N-derivatized amino acids and 2-aryloxypropionicacids. The accurate optimization of the enantioseparation in terms of solvent system composition, pH values, ionic strength, and CCC operating conditions was carried out. Successful resolutions were obtained in systems such as ammonium acetate buffer/tert-amyl alcohol/methanol/heptane and ammonium acetate buffer/methyl isobutyl ketone (MIBK) or diisopropyl ether (DIPE). Different α values were obtained in CCC and HPLC for a given pair of selector/racemate. For n-(3,5-dinitrobenzoyl)-(±)-leucine and N-(3,5-dinitrobenzyloxycarbonyl)-(±)-neopentylglycine, enantioselectivity factors were lower in CCC, while for N-(3,5-dinitrobenzyloxycarbonyl)-(±)-β-phenylalan ine, α values slightly increased. No significant separation was observed for any of the of the aryloxypropionic acid derivatives tested in system containing (MIBK). In contrast, some discriminations for the enantiomers of these compounds was detected when DIPE was used in the binary system. The results exhibited the high potential of CCC as a separative separation technique.

Full text of DOI:10.1021/ac020209q

Anal. Chem. 2002, 74, 4175-4183
Enantiomer Separation by Countercurrent
Chromatography Using Cinchona Alkaloid
Derivatives as Chiral Selectors
Pilar Franco,† Javier Blanc,† Wolfgang R. Oberleitner,‡ Norbert M. Maier,‡ Wolfgang Lindner,*,‡ and
Cristina Minguillo´n*,†
Laboratori de Qu´ımica Farmace`utica, Facultat de Farma`cia, Universitat de Barcelona, Avda. Diagonal s/n, E-08028
Barcelona, Spain, and Institute of Analytical Chemistry, University of Vienna, Wa¨hringerstrasse 38, A-1090 Vienna, Austria
context, the applications of alkaloids as CSs have also been
extensively reported in the literature.4-6 Concretely, Lindner and
co-workers described the ability of cinchona derivatives, especially
quinine (QN) and quinidine (QD), as efficient CSs in anion-
exchange mode.7 In a first approach, this type of selectors was
tested in HPLC,8,9 although these selectors high potential to form
very stable complexes with chiral acidic compounds has allowed
their efficient use as CSs also in capillary electrophoresis (CE),10,11
capillary electrochromatography (CEC),12 supercritical fluid chro-
matography (SFC), and extraction experiments,13,14 among others.
Exhaustive investigations have been performed in order to
understand the influence of every substituent and functionality
to enhance the chiral recognition abilities of this type of
selectors8,15-17 (See general structure in Figure 1.). On one hand,
the possibility of protonating the quinuclidine nitrogen of their
molecule makes them especially suitable for the resolution of
acidic compounds, such as amino acid derivatives and other chiral
acidic substances due to ion pair formation. On the other hand,
the derivatization of the hydroxy group to a carbamate function
in the O9-position, especially bearing bulky substituents, such as
tert-butyl or adamantyl, was leading to the highest enantioselec-
tivity values. Selectivity factors of ∼30 were obtained at 25 °C in
HPLC for the resolution of the enantiomers of N-(3,5-dinitroben-
Cinchona-derived anion-exchange-type chiral selectors
have been adapted and employed in countercurrent
chromatography (CCC) for the separation of enantiomers
of N-derivatized amino acids and 2 -aryloxypropionic
acids. The accurate optimization of the enantioseparation
in terms of solvent system composition, pH values, ionic
strength, and CCC operating conditions was performed.
A wide range of solvent mixtures was evaluated. Success-
ful resolutions were achieved in systems such as am-
monium acetate buffer/ ter t-amyl alcohol/ methanol/ hep-
tane and especially ammonium acetate buffer/ methyl
isobutyl ketone or diisopropyl ether. Up to 3 0 0 mg (0 .9 2
mmol) of N-(3 ,5 -dinitrobenzoyl)-(()-leucine was totally
resolved in a single run using a 1 0 mM concentration of
chiral selector in 122 mL of stationary phase. This amount
could be increased up to 9 0 0 mg (2 .7 7 mmol) when pH-
zone-refining mode was applied. The results here pre-
sented account for the high potential of CCC as a pre-
parative enantiomer separation technique.
The search for novel either broadly or dedicatedly applicable
chiral selectors (CSs) and the development of separation tech-
niques for the resolution of enantiomers is an open field still
attracting the creativity and efforts of researchers all over the
world.1 The large number of publications and books about the
subject is a consequence of the well-known interest in chirality
and its scientific and economic impact on pharmaceutical and
biological sciences. Chiral selectors were primarily found in
nature, although with time, sophisticated modifications and fully
synthetic compounds have demonstrated to be very efficient in
the recognition of enantiomers.
(4) Sinibaldi, M.; Vinci, M.; Federici, F.; Flieger, M. Biomed. Chromatogr. 19 97 ,
11, 307-310.
(5) Rosini, C.; Bertucci, C.; Pini, D.; Altemura, P.; Salvadori, P. Chromatographia
1 9 8 7 , 24, 671-676.
(6) Stalcup, A. M.; Gahm, K. H. J. Microcolumn Sep. 1 9 9 6 , 8, 145-150.
(7) Lindner, W., L¨ammerhofer, M.; Maier, N. M. PCT/ EP97/ 02888, 1997.
(8) La¨mmerhofer, M.; Maier, N. M.; Lindner, W. Am. Lab. 1 9 9 8 , 30, 71-78.
(9) La¨mmerhofer, M.; Lindner, W. J. Chromatogr., A 1 9 9 6 , 741, 33-48.
(10) La¨mmerhofer, M.; Zarbl, E.; Lindner, W. J. Chromatogr., A 2 0 0 0 , 892, 509-
521.
(11) Zarbl, E.; L¨ammerhofer, M.; Franco, P.; Petracs, M.; Lindner, W. Electro-
phoresis 2 0 0 1 , 22, 3297-3307.
(12) Tobler, E.; La¨mmerhofer, M.; Lindner, W. J. Chromatogr., A 2 0 0 0 , 875,
Among the natural building blocks, oligo- and polysaccharides,
proteins, peptides, and amino acids, alkaloids, and antibiotics are
the most commonly used scaffolds for chiral selectors.2,3 In this
341-352.
* Corresponding authors. W.L.: (fax) +43-1-3151826; (e-mail) wolfgang.lindner@
univie.ac.at. C.M.: (fax) +34-93-4035941; (e-mail) minguill@farmacia.far.ub.es.
† Universitat de Barcelona.
(13) Kellner, K.-H.; Blasch, A.; Chmiel, H.; L¨ammerhofer, M.; Lindner, W.
Chirality 1 9 9 7 , 9, 268-273.
(14) Tobler, E.; La¨mmerhofer, M.; Oberleitner, W. R.;, Maier, N. M.; Lindner,
‡ University of Vienna.
W. Chromatographia 2 0 0 0 , 51, 65-70.
(1) Maier, N. M.; Franco, P.; Lindner, W. J. Chromatogr., A 2 0 0 1 , 906, 3-33
and references therein.
(15) Maier, N. M.; Nicoletti, L.; L¨ammerhofer, M.; Lindner, W. Chirality 1 9 9 9 ,
11, 522-528.
(2) Subramanian, G.. Ed. A Practical Approach to Chiral Separations by Liquid
Chromatography; VCH: Weinheim, 1994.
(16) Franco, P.; L¨ammerhofer, M.; Klaus, P. M.; Lindner, W. J. Chromatogr., A
2 0 0 0 , 869, 111-127.
(3) Subramanian, G., Ed. Chiral Separation Techniques. A Practical Approach,
2nd ed.; Wiley-VCH: Weinheim, 2001.
(17) Franco, P.; La¨mmerhofer, M.; Klaus, P. M.; Lindner, W. Chromatographia
2 0 0 0 , 51, 139-146.
10.1021/ac020209q CCC: $22.00 © 2002 American Chemical Society
Published on Web 07/12/2002
Analytical Chemistry, Vol. 74, No. 16, August 15, 2002 4175

Figure 1. Chemical structure of the quinine (QN)- and quinidine (QD)-derived chiral selectors.
zoyl)-(()-leucine (DNB-Leu) with some of these carbamate
derivatives.15,18
compositions. Solvents can be easily changed and removed and
no irreversible adsorption of the sample in the stationary phase
is possible; hence, a wide range of solvent mixtures can be chosen
to perform the fine-tunable chromatographic process.
Furthermore, this remarkable stereoselectivity observed when
the chiral selector was covalently bonded to the chromatographic
matrix was also displayed in solution: carbamoyl-type QN and
QD derivatives were able to produce stereoselective enrichments
of acidic analytes in liquid-liquid extraction experiments.13 These
experimental data, obtained from a single equilibration stage
between the two immiscible phases, served as promising signs
for an interesting application of multiple liquid-liquid extraction
equilibria as revealed in the countercurrent chromatography
(CCC) concept.
However, as mentioned before, there are not many examples
described of enantiomer separations using this technique.23-34 The
difficulty of finding a suitable CS (being highly selective for the
given racemic compounds) and the appropriate system of solvents,
which might not suppress this selectivity and keep the capacity
to elute the chiral solutes of interest, can presumably be the main
reasons for the scarcity of applications in this field. The CSs
employed until present were in general those directly used also
in liquid chromatography or in other separation techniques. Very
long retention times and rather poor resolutions were common
features of the first attempts. However, this has been in general
overcome in the most recent applications thanks to the improve-
ments in the instrumentation and to further optimizations of the
different parameters involved in the overall separation pro-
cess.23,29,30,32,34 As a consequence, we conceptually designed
modified cinchona-derived selectors in such a way to bear an
adamantylcarbamate function to enhance its chiral recognition
ability and to increase also the overall lipophilicity by introducing
an octadecylthia moiety to avoid the selector distribution into the
mobile phase (see Figure 1). In the present article, its use in the
context of a consequent search of suitable immiscible liquid-
liquid solvent system(s) and the optimization of the resolutions
of several enantiomeric pairs (Figure 2) were brought about in
CCC and its particular modalities, centrifugal partition chro-
matography (CPC) among them, are techniques with a clear
preparative potential and focus. Their successful applications in
the purification of natural products have been sufficiently
reported.19-21 In contrast, the examples devoted to the separation
of enantiomers are still limited, although they have shown that
the general principle of these techniques, the partition phenom-
enon between two immiscible liquid phases, which controls the
retention of the analytes,19,22,23 is perfectly applicable also to chiral
resolutions by addition of an appropriate chiral selector to the
system. The ideal solvent system, one of the key points in the
success of the separation, should meet certain requirements. Thus,
the leakage of CS from one phase to the other (e.g., from the
stationary to the mobile phase) should be avoided while promoting
the desired partition of the analyte enantiomers between the two
liquid phases. Moreover, the solubility of the CS in the stationary
phase has to be sufficient to bring about the separation and the
elution of the analytes in a reasonable amount and time for
preparative purposes.
(24) Oliveros, L.; Minguillo´n, C.; Franco, P.; Foucault, A. P. Enantioseparations
in Counter-current Chromatography (CCC) and Centrifugal Partition Chro-
matography (CPC). In Countercurrent Chromatography, the support-free liquid
stationary phase; Berthod, A., Ed.; Comprehensive Analytical Chemistry
Series; Elsevier: New York, 2002; Chapter 11.
(25) Domon, B.; Hostettmann, K.; Kovacevic, K.; Prelog, V. J. Chromatogr. 1 9 8 2 ,
250, 149-151.
(26) Oya, S.; Snyder, J. K. J. Chromatogr. 1 9 8 6 , 370, 333-338.
(27) Takeuchi, T.; Horikawa, R.; Tanimura, T. J. Chromatogr. 1 9 8 4 , 284, 285-
288.
In these aspects lies the main appeal of the CCC techniques:
solvent mixtures with low cost constitute the stationary and the
mobile phases, and their properties have the particularity of being
mainly modulated thanks to changes in the solvent system
(28) Armstrong, D. W.; Menges, R.; Wainer, I. W. J. Liq. Chromatogr. 1 9 9 0 , 13,
(18) Maier, N. M.; Oberleitner, W. R.; Lindner, W. Improved enantioseparation
of N-acylated amino acid derivatives with a novel generation of O6-modified
cinchona alkaloid carbamate type selectors. Poster presented at the 23rd
Symposium on High Performance Liquid-Phase Separation and related
techniques (HPLC‘1999), Granada, Spain, May-June 1999.
3571-3581.
(29) Oliveros, L.; Franco Pue´rtolas, P.; Minguillo´n, C.; Camacho-Frias, E.;
Foucault, A.; Le Goffic, F. J. Liq. Chromatogr. 1 9 9 4 , 17, 2301-2318.
(30) Ma, Y.; Ito, Y. Anal. Chem. 1 9 9 6 , 68, 1207-1211.
(31) Santano, C.; Minguillo´n, C. Separacio´n de enantiomeros por cromatograf´ıa
de reparto centr´ıfugo (CPC). Ventajas e inconvenientes. Presented at the
XXX Scientific Meeting of the Group of Chromatography and related
Techniques, Valencia, Spain, April 2001.
(32) Breinholt, J.; Lehmann, S. V.; Varming, A. R. Chirality 1 9 9 9 , 11, 768-771.
(33) Armstrong, D. W., Tang, Y.; Chen, S.; Zhou, Y.; Bagwill, C.; Chen, J. R.
Anal. Chem. 1 9 9 4 , 66, 1473-1484.
(19) Foucault, A. P., Ed. Centrifugal Partition Chromatography; Chromatographic
Science Series 68; Marcel Dekker: New York, 1995.
(20) Foucault, A. P.; Chevolot, L. J. Chromatogr., A 1 9 9 8 , 808, 3-22.
(21) Menet, J.-M., Thiebaut, D., Eds. Countercurrent Chromatography; Chromato-
graphic Science Series 82; Marcel Dekker: New York, 1999.
(22) Conway, W. D., Ed. Countercurrent Chromatography: apparatus, theory and
applications; VCH: New York, 1990.
(34) Shinomiya, K.; Kabasawa, K.; Ito, Y. J. Liq. Chromatogr. 1 9 9 8 , 21, 135-
141.
(23) Foucault, A. P. J. Chromatogr., A 2 0 0 1 , 906, 365-378.
4176 Analytical Chemistry, Vol. 74, No. 16, August 15, 2002

collected and analyzed individually during the whole elution of
the analytes in order to get a profile of the enantiomeric excesses
of the peaks. This control of the CCC-collected fractions and the
extraction experiments was performed on the same HPLC system,
changing the CPC device for the appropriate HPLC chiral column
for the resolution of the considered enantiomers. The homemade
HPLC column (150 × 4.6 mm i.d.) contained a tert-butylcarbam-
oylated-QN chiral selector (these columns are now commercially
available from Bischoff, Leonfelder, Germany or Iris Technologies,
Lawrence, KS). The mobile phases were mixtures of methanol
and 1.0 M ammonium acetate adjusted to an apparent pH (pHa)
of 6.0 by adding glacial acetic acid. The ratio methanol/ buffer
depended on the racemic solutes analyzed; it was: 95:5 for DNB-
Leu, 80:20 for N-(3,5-dinitrobenzyloxycarbonyl)-(()-neopentylg-
lycine (DNZ-NPG) and N-(3,5-dinitrobenzyloxycarbonyl)-(()-â-
phenylalanine (DNZ-â-Phe), and 30:70 for the 2-aryloxypropionic
acids. Flow rate was always 1 mL/ min, and the UV detection was
at 254 nm for the amino acid derivatives, 230 nm for dichlorprop,
and 276 nm for DMP and mecoprop. Temperature was kept
constant at 25 °C during the CCC experiments and the HPLC
runs.
Figure 2. Chemical structures of the racemic compounds tested.
CCC and will be discussed in detail. This also includes the
necessary preliminary screening experiments.
Stereoselective Extraction Screening Conditions. In a small
test tube with a screw cap, 100 µL of a 100 mM solution of racemic
DNB-Leu in methanol (10 µmol) were pipetted. Then, 100 µL of
a 100 mM solution of the CS in dichloromethane (10 µmol) was
added, and the solvents were removed under vacuum at 30 °C.
The solvent systems to be tested were prepared in different
flasks and equilibrated before use by stirring vigorously during 1
min. An aliquot of 2 mL of the upper phase and an equal volume
from the lower phase were added to the test tube. The vial was
closed and vortexed vigorously in a water bath at 25 °C for 5 min.
The phases were allowed to separate, and 1 mL of the aqueous
lower phase was transferred to another vial. Most of the solvent
was removed by gentle heating, and the sample was dried under
vacuum at 60 °C. The residue was dissolved in 2 mL of methanol
and directly injected into the HPLC system, where the enantio-
meric excess (ee) was evaluated using an appropriate cinchona-
derived column.15
EXPERIMENTAL SECTION
Materials. 2,4-Dimethylphenol, 2-bromopropionic acid, and
glacial acetic acid were purchased from Aldrich (Steinheim,
Germany). Racemic leucine (Leu) and neopentylglycine (NPG)
were supplied by Fluka (Buchs, Switzerland). Dichlorprop was a
gift from Rhoˆne-Poulenc (Lyon, France), and mecoprop was
purchased from Lancaster (Eastgate, England).
The synthesis of the CSs was described elsewhere.35 Racemic
â-phenylalanine (â-Phe) was prepared following a literature
procedure.36 N-Acylated amino acids were available from prior
studies9,37 or were prepared following well-established protocols.
2-(2,4-Dimethylphenoxy)propionic acid (DMP) was synthesized
from 2,4-dimethylphenol by alkylation with 2-bromopropionic acid
in aqueous basic media.
Mobile phases for chromatography were prepared from
analytical reagent-grade ammonium acetate from Merck and
HPLC-grade water, together with HPLC solvents. Trifluoroacetic
acid (TFA) and ammonia used for the pH-zone-refining mode were
supplied by Panreac Qu´ımica S.A. (Montcada i Reixac, Spain).
The HPLC column used to control the CCC fractions contained a
tert-butylcarbamoylated quinine as chiral selector.15
To check the possible selector leaching into the mobile phase,
the aqueous component was investigated by thin-layer chroma-
tography (TLC) developing with chloroform/ methanol mixtures
(10:1).
CCC Experimental Conditions. The CS was dissolved in the
liquid phase with less density (stationary phase), and therefore,
in all the cases the mobile phase was pumped through in
descending mode. The CCC column was kept in methanol when
not in use after exhaustive cleaning with the suitable solvents.
Thus, before starting any experiment, water was pumped through
to eliminate methanol traces. As schematized in Figure 3, several
preparation steps were performed before the injection of the
samples and they were different depending on the operational
mode. In classical CCC mode, water from the column was
completely displaced with the stationary phase in descending
mode. Centrifugal rotation was then applied, and the mobile phase
was pumped until no bleeding of stationary phase was observed
in the eluate. The displaced volume of stationary phase cor-
responded to the void volume and allowed the calculation by
Apparatus. An HPCPC model LLB-M (EverSeiko, Tokyo,
Japan) was used. It is a bench CPC (30 × 45 × 45 cm) with a
stacked circular partition disk rotor (2136 channels, 220 mL of
internal volume). The CPC was connected to a conventional HPLC
system HP 1100 (pump, autosampler, UV detector, and chroma-
tography data station software) from Hewlett-Packard (Figure 3).
A manual Rheodyne injector provided with a 2.4-mL loop was used.
Commonly, 2.4 mL of racemate solution was injected, but when
larger amounts of sample were analyzed, they were introduced
directly into the system through the pump.
The pH of the mobile phases was measured with a Crison pH
meter, model Basic 20 (Alella, Spain). Fractions of 4.5 mL were
(35) Oberleitner, W. R. Ph.D. Thesis, Vienna, Austria, 2000.
(36) Johnson, T. B.; Livak, J. E. J. Am. Chem. Soc. 1 9 3 6 , 58, 299-303.
(37) Piette, V.; La¨mmerhofer, M.; Bischoff, K.; Lindner, W. Chirality 1 9 9 7 , 9,
157-161.
difference of the retained stationary phase (V ) and, therefore,
st
Analytical Chemistry, Vol. 74, No. 16, August 15, 2002 4177

Figure 3. Separation system and operational scheme.
the amount of CS present in the CCC column. Injection of the
enantiomeric pairs to be separated, dissolved in stationary phase
in order not to disturb the equilibrium conditions, was then
performed. In pH-zone-refining CCC mode, the column was filled
analogously with the acidified stationary phase. However, injection
of the analytes was performed before centrifugal rotation was
applied and the basic mobile phase started to be pumped through.
In the eluate, the first part of the displaced stationary phase
appeared and afterward the collection of fractions was undertaken.
The apparent pH and the enantiomer content were determined
for every fraction. The volume of nondisplaced chiral stationary
it with a 5% (w/ v) NaHCO3 solution (× 3), 0.25 M ammonium
acetate solution (× 1), and water (× 2). The remaining organic
layer was then dried with anhydrous sodium sulfate and filtered,
and the solvent was evaporated under vacuum. When an alcohol
was present in the stationary phase, a partial evaporation of this
component was performed previous to the above-described treat-
ment. The selector was dried overnight at 60 °C under reduced
pressure, resulting in ∼95% recovery.
RESULTS AND DISCUSSION
Screening of Conditions. Several stages were planned in the
development of the complete application of CCC to the separation
of enantiomers. First, an initial procedure was designed to assess
the potential enantioselectivity of the CSs in different solvent
systems by means of extraction experiments. Moreover, important
features of the solvent systems, such as ready separation of the
two immiscible liquid phases, were critical to evaluate. In a second
stage, blank and control CCC runs without the selector dissolved
in the stationary phase helped to establish the operating final CCC
conditions, which were later applied also in the presence of the
CSs. Some of the best resolutions were consequently scaled up.
Extraction Experiments. Liquid-liquid extraction experi-
ments were necessary to find suitable systems of solvents, which
had to meet the following requirements. First, the CS should be
phase or V was also used to calculate the amount of selector
st
present in the CCC column during the separation process.
Evaluation of the Recovery of Solutes and Selectors after
the CCC Runs. The collected mobile-phase fractions (standard
4.5 mL) containing the separated enantiomers were acidified with
a small volume of concentrated hydrochloric acid and extracted
three times with ethyl acetate. The combined organic layers were
dried with anhydrous sodium sulfate and filtered, and the solvent
was evaporated under vacuum. The residue was weighed and
found to be 90-95% of the injected amount of racemic solute by
summing up the yield of the individual fractions.
The QN and QD derivatives dissolved in the stationary phase
were recovered from the organic phase by successively washing
4178 Analytical Chemistry, Vol. 74, No. 16, August 15, 2002

soluble in only one of the two phases (in our case, in the stationary
phase). For a two-phase aqueous/ organic system, the selector
should be either highly lipophilic to remain entirely in the organic
phase or highly hydrophilic (ionic), if the aqueous phase is to be
used as the stationary phase. The CSs considered in the present
study were designed to be lipophilic even in the presence of a
positively charged amino group at the working pH. Therefore,
the organic phase will act as stationary phase in all the solvent
systems tested. Second, the racemic mixtures (see Figure 2)
should be readily soluble in both phases, whereby in the present
case the solubility of the ion-pair enantiomer-CS associate in the
lipophilic phase needs to be considered. The optimum distribution
coefficient is ∼1. This is approximately the same solute concentra-
tion (without specifying the individual enantiomers) in both phases
after extraction. High solute distribution in the stationary phase
will lead to long retention times. On the contrary, higher
concentrations of compound to be separated in the mobile phase
will reduce the chromatographic efficiency of enantioseparation
due to the reduced interaction between enantiomers and CS, this
latter being mainly in the stationary phase, due to the ion-pair-
type extraction.
studies revealed a pronounced tendency of the polysiloxane
derivatives to form stable emulsions, whereas the use of the
corresponding nonpolymerized but lipophilized CS assured a rapid
phase separation for the same solvent systems.
CCC Runs. A CCC run without the addition of CS was useful
to estimate nonstereoselective retention by injecting the racemic
solutes that should later be separated into the enantiomers.
Moreover, this control experiment was performed to check how
the two-phase solvent system behaves under real operating CCC
conditions as outlined in Figure 3. Then, the QN- or QD-derived
CS was introduced and the CCC runs were undertaken. Although
selectors 1 and 2 are in fact diastereomers, as is usual for QN
and QD derivatives acting as enantiodiscriminating agents, their
enantioselectivity was complementary and the expected inversion
of the elution order of enantiomers was always observed. As QN-
CS 1 is an oil and QD-CS 2 a solid, both selectors were tested in
order to check their behavior in solution and their slightly different
chromatographic properties.
Three main types of solvent systems were investigated: (A)
quaternary solvent systems; (B) solvent systems containing MIBK;
(C) solvent systems containing DIPE.
A wide range of binary, ternary, and quaternary solvent
systems was evaluated. First, their readiness for forming nonstable
emulsions was taken into account. The solvent systems that
separated rapidly were subsequently tested with the CS, to detect
any possible distribution between both phases. Initial extraction
experiments were performed using the lipophilized CSs, and
afterward, DNB-Leu was added to the same mixtures as a
racemate. The molar CS/ racemate ratio was 1 for all the experi-
ments. The CS concentration in the extraction experiments was
similar to that used later for the CCC stationary phase, because
in some cases, it is actually the CS that causes emulsification of
the system. The solvent systems that still showed ready separation
in the presence of the CS were candidates for further determina-
tions. The enantioselective partition of DNB-Leu between the two
phases was determined by HPLC.
On the basis of previous results performed by liquid-liquid
extraction and liquid membranes,13 it was expected that a ternary
system consisting of an ammonium acetate buffer, an alcohol, and
a hydrocarbon might yield good enantioselectivity and liquid-
liquid (CCC-type) chromatographic behavior. However, the sys-
tems containing 1-decanol and an hydrocarbon, such as isooctane
or pentadecane, were not appropriate, as their high boiling points
were not suitable for the recovery of the CS after CCC runs. Thus,
other alcohols in combination with heptane were tested. By far
the highest enantioselectivities were observed with systems
containing an alcohol of medium lipophilicity (1-butanol or tert-
amyl alcohol) in combination with heptane. There were also some
other binary and ternary systems with methyl isobutyl ketone
(MIBK), diisopropyl ether (DIPE), or methyl tert-butyl ether
(MTBE) that gave rise to good enantioselectivities.
(A) Quaternary Solvent Systems. Based on the above-
discussed extraction results, several CCC runs were performed
with the system ammonium acetate buffer 0.1 M pH 6.0/ tert-amyl
alcohol/ MeOH/ heptane (10/ 5/ 1/ 5) containing 10.6 mM CS. This
system was rather stable at a rotational speed (ω) of 1000 rpm
and a flow rate of 3 mL/ min. Analytical amounts of the racemic
mixtures (1-10 mg) were injected (see Figure 2). While baseline
separations were observed for DNB-Leu and DNZ-NPG, only a
partial resolution was achieved for DNZ-â-Phe (Figure 4). There-
fore, several variations of the ionic strength and pH of the running
buffer, proportions of the different solvents, concentration, and
QN- or QD-type CSs (see Figure 1) were investigated. Although
some improvement in the separation was obtained in a system of
ammonium acetate buffer 0.25 M pH 7.0/ tert-amyl alcohol/
MeOH/ heptane (10/ 3/ 1/ 7) containing 10.6 mM QN selector 1 ,
no complete resolution was possible for DNZ-â-Phe.
Once a successful analytical-scale enantioseparation was
achieved, a first upscaling procedure was performed to determine
the maximum amount of injectable solutes that still can be near-
baseline separated. This is referred in a first attempt as the
maximum throughput of analyte in this solvent system in a single
run. However, it should be taken into account that the detector
reached signal saturation at ∼3000 mAU. Thus, for semiprepara-
tive separations (10-100 mg), peaks would be much higher and
cannot be recorded that way. Nevertheless, separation could still
be observed by HPLC analysis of individual fractions, thus
reconstructing an elution profile. The off-line HPLC results were
in any case conclusive in order to know the exact composition
(ee) of the different collected fractions.
A summary of the results from the first upscaled CCC runs is
listed in Table 1. In general, different R values were obtained in
CCC and HPLC for a given pair of selector/ racemate. However,
it has to be taken into account that conditions are not directly
extrapolable from one technique to the other. For DNB-Leu and
DNZ-NPG, enantioselectivity factors were lower in CCC, whereas
for DNZ-â-Phe, R values slightly increased.
Concerning the CSs, O9-(1-adamantylcarbamoyl)-QN and QD
derivatives had previously demonstrated the ability to recognize
the enantiomers of amino acid derivatives in HPLC with even
higher R values than the corresponding tert-butyl analogues.16 In
the present case, the adamantyl-derived selectors were made more
lipophilic by addition of an octadecylthia group to the vinyl moiety
(1 and 2 , Figure 1). Other approaches, such as immobilization
onto polysiloxane, were also investigated. However, exploratory
Nevertheless, enantioselectivity might be affected either by the
fixation of the CS to a chromatographic support or by the
Analytical Chemistry, Vol. 74, No. 16, August 15, 2002 4179

Figure 4. CCC chromatograms. Flow rate 3 mL/min; ω ) 1000 rpm; descending mode; λ 240 nm. Solvent systems: (a, c) ammonium acetate
buffer 0.1 M pH 6.0/tert-amyl alcohol/MeOH/heptane (10/5/1/5), 10.6 mM QD selector 2, P ) 70 bar; (b, d) ammonium acetate buffer 0.25 M
pH 7.0/tert-amyl alcohol/MeOH/heptane (10/3/1/7), 10.6 mM QN selector 1, P ) 79 bar.
Table 1. CCC Separations Using the Quaternary System Ammonium AcetatBuffer/tert-Amyl Alcohol/MeOH/Heptanea
mrac (mg)
rCS/ rac
t1 (min)
t2 (min)
k1
k2
R
eo
conditionsb
DNZ-NPG
(RHPLC ) 3.0)c
2.0
10
359
66
36
51
49
70
25
27
36
31
34
37
62
93
0.8
1.5
1.4
2.5
0.2
0.3
0.8
0.6
0.7
0.9
2.1
3.6
3.6
6.2
1.8
2.2
6.2
1.4
1.5
2.0
2.69
2.38
2.53
2.48
7.60
6.79
7.71
2.43
2.22
2.28
S
R
R
S
S
S
R
S
R
R
A
B
B
A
A
A
B
A
B
B
50
13
92
100
2.0
7.8
100
1.0
2.0
12.5
6.6
300
75
5.9
720
350
56
143
57
65
145
48
50
59
DNB-Leu
(RHPLC ) 15)c
DNZ-â-Phe
(RHPLC ) 1.6)c
a
mrac, mass of racemate injected; rCS/ rac, molar ratio CS/ racemate; k1 and k2, retention factors of the enantiomers. eo, elution order, configuration
of the first-eluted enantiomer. b Flow rate, 3 mL/ min; ω ) 1000 rpm; descending mode; λ 240 nm. (A) Ammonium acetate buffer 0.1 M pH 6.0/
tert-amyl alcohol/ MeOH/ heptane (10/ 5/ 1/ 5), 10.6 mM QD-CS 2 . (B) Ammonium acetate buffer 0.25 M pH 7.0/ tert-amyl alcohol/ MeOH/ heptane
(10/ 3/ 1/ 7), 10.6 mM QN-CS 1 . c RHPLC values were obtained with a column containing the adamantylcarbamate-QN-CS bonded to silica gel using
mixtures of ammonium acetate buffer/ methanol as mobile phase.35
differences in solvents used in both techniques affecting its
solvatation and conformational equilibria. In HPLC, the CS-
enantiomer association takes place at the interphase between the
mobile and the stationary phases. In CCC, the racemic solute must
undergo a phase transition from the mobile to the stationary phase
to form an ion-pair with the CS. Generally, for the same volume
of stationary phase and CS concentration, resolution decreased
at higher solute amounts injected and loaded on, as is usual in
HPLC, but to a much lesser extent than expected for the CCC
technique. In Figure 5, the chromatogram corresponding to the
enantioseparation of 50 mg of DNZ-NPG is shown. During elution,
peaks were fractionated and subsequently analyzed by HPLC.
There was only a small mixed zone, and essentially pure enanti-
omers were eluted around the maximums of the chromatographic
peaks.
enantioselectivity. However, solubility of the racemic mixtures was
relatively low, retention times were long, and CS/ racemate ratios
were usually high, but in the same order of the ones commonly
found in HPLC. These features implied a limited loading capacity
for the system used.
(B) Solvent Systems Containing MIBK. After these observa-
tions, we decided to switch to other more promising solvent
systems, which originated lower enantioselectivity values in the
first screenings but offered other features. Secondary solvent
systems, constituted of ammonium acetate buffer and MIBK,
MTBE, or DIPE were under investigation. Some of these organic
solvents (MTBE, DIPE) showed the same reduced solubility for
amino acid derivatives, but this problem was overcome by using
MIBK. Exploratory tests in tube with these binary systems
demonstrated that the best enantioselectivity values were achieved
with MIBK for DNB-Leu.
Despite these promising results, some drawbacks were ob-
served. The solvent system was optimized to yield the highest
4180 Analytical Chemistry, Vol. 74, No. 16, August 15, 2002

Table 2. CCC Separations in Buffer/MIBK Solvent Systemsa
QN-CS 1
QD-CS 2
t1
t2
t1
t2
racemate
DNB-Leu
pH
m
rac (mg)
rCS/ rac
(min)
(min)
k1
k2
R
(min)
(min)
k1
k2
R
6.85
8.0
150
150
150
170
170
170
179
179
179
3
3
211
71
83
d
170
191
d
190
188
282
116
157
11.1
3.1
2.7
15.1
5.6
6.0
1.36
1.84
2.20
222
83
297
158
12.4
4.42
16.9
9.33
1.37
2.11
8.0b
6.85
8.0
2.6c
3
DNZ-NPG
d
175
3
200
227
8.7
7.5
10.4
9.1
1.20
1.21
232
255
10.4
12.7
14.1
15.6
1.36
1.23
8.0b
6.85
8.0
2.6c
3
DNZ-â-Phe
d
210
3
230
215
9.9
7.4
12.1
8.6
1.23
1.16
8.0b
2.6c
a
mrac, mass of racemate corresponding to 0.46 mmol; rCS/ rac, molar ratio CS/ racemate; k1 and k2, retention factors of the enantiomers. Flow
rate, 3 mL/ min; ω ) 1200 rpm; descending mode; λ 230 nm. Ammonium acetate buffer 0.1 M/ MIBK (1:1), 10 mM CS (V )138 mL), except as
st
noted. b Ammonium acetate buffer 0.1 M/ acetone/ MIBK (2:1:2), V ) 122 mL. c The molar ratio CS/ racemate is reduced as a result of the smaller
stationary-phase volume retained in the CCC column for this solvenstt system. d The racemates were retained in the CCC column in these conditions.
Regarding the 2-aryloxypropionic acids tested, only a slight
enantiomeric enrichment was observed for mecoprop at pH 6.85
(data not shown).
In the search for improved conditions, a third solvent was
introduced in the system. Thus, using 0.1 M ammonium acetate
buffer, pH 8.0/ acetone/ MIBK (2/ 1/ 2) containing a concentration
of 10 mM QN selector 1 , retention times were slightly longer
than those obtained with the binary system at the same pH, even
though the volume of stationary phase retained in the CCC system
was smaller. This ternary system was also less stable than the
binary one previously described. Some leaching of the stationary
phase was observed during operation. Nevertheless, enantiose-
lectivity increased for DNB-Leu, the only racemic mixture whose
baseline resolution was achieved. Saturation of the CS was attained
when 450 mg (1.38 mmol) of this solute was injected under these
conditions (molar ratio CS/ racemate, 0.88) (Figure 6).
Figure 5. Resolution of 50 mg of DNZ-NPG. Solvent system:
ammonium acetate buffer 0.25 M pH 7.0/tert-amyl alcohol/MeOH/
heptane (10/3/1/7), 10.6 mM QN selector 1. Flow rate 3 mL/min; ω
) 1000 rpm; descending mode; λ 240 nm, P ) 79 bar. Vertical axis,
arbitrary absorbance units. Horizontal axis, time (min).
(C) Solvent Systems Containing DIP E. As already men-
tioned, no remarkable separation was observed for any of the
aryloxypropionic acid derivatives tested in systems containing
MIBK. In contrast, some discrimination for the enantiomers of
these compounds was detected when DIPE was used in a binary
system (Table 3). In this case, an increase of the ionic strength
of the buffer solution was needed to avoid emulsification. As these
compounds were in general terms less retained, the decrease of
the pH of the buffer was leading to better chiral recognition,
though insufficient for practical purposes.
Unfortunately, amino acid derivatives were not sufficiently
soluble under these conditions. This last drawback was partially
overcome by including 2-propanol in the binary mixture. Although
comparable or slightly better R values could be observed for the
amino acid derivatives in these DIPE systems than with the ones
containing MIBK, they turned out to be not suitable to carry out
a preparative separation due to the limited solubility of these
racemates in this ether.
In Table 2, results corresponding to the separation of amino
acid derivatives using MIBK are presented. It should be noted
that the molar ratios of CS/ racemate were highly reduced thanks
to the increased solubility of the solutes. Modulation of the pH
values of the buffer was, as expected, a key point to control the
elution of the analytes. Higher pHs were led to shorter retention
times but lower enantioselectivity.38 However, total resolutions
were possible for all amino acid derivatives at pH 8.0 with QN-CS
1 . At pH 6.85, DNZ-NPG and DNZ-â-Phe remained retained in
the CCC column.
In contrast to the total resolution achieved in the same
conditions using QN-CS 1 , only a partial resolution was attained
for DNZ-NPG and DNZ-â-Phe using QD-CS 2 . Overloading as a
cause for that observation was discarded as comparable results
in terms of resolution were achieved when half the amount of
racemic solute was injected. A slower kinetics observed in all cases
when CS 2 was used may be the cause for this partial resolution.
Ion-Exchange Displacement Chromatography (pH-Zone-
Refining Mode). To improve the preparative potential of CCC
for the separation of enantiomers, and taking into account the
ionizable character of the studied solutes and CSs, pH zone
refining was applied. This chromatographic mode in CCC is a
displacement type chromatography introduced by Ito et al.,39 and
(38) Franco, P.; Oberleitner, W. R.; Maier, N. M.; Lindner, W.; Minguillo´n, C.
Enantiomer separation by centrifugal partition chromatography (CPC) using
cinchona alkaloid derivatives as chiral selectors. Poster presented at the
12th Symposium on Chiral Discrimination (Chirality 2000), Chamonix,
France, September 2000.
(39) Weisz, A.; Seher, A. L.; Shinomiya, K.; Fales, M.; Ito, Y. J. Am. Chem. Soc.
1 9 9 4 , 116, 704-708.
Analytical Chemistry, Vol. 74, No. 16, August 15, 2002 4181

Figure 6. Elution profiles corresponding to the separation of increasing amounts of DNB-Leu injected. Solvent system: ammonium acetate
buffer 0.1 M pH 8.0/acetone/MIBK (2/1/2), 10 mM QN selector 1. Flow rate 3 mL/min; ω ) 1300 rpm; descending mode, P ) 76 bar. Vertical
axis, arbitrary absorbance units. Horizontal axis, time (min).
Table 3. CCC Separations in Buffer/DIPE Solvent Systemsa
QN-CS 1
QD-CS 2
t1
(min)
t2
(min)
t1
(min)
t2
(min)
racemate
pH
m
rac (mg)
rCS/ rac
k1
k2
R
k1
k2
R
DNB-Leu
6.85
6.85
6.85
6.85
6.85b
6.0b
6.85
6.85b
6.0b
50
9
3
10
3
3.5
3
3
75
173
85
124
68
87
114
64
86
142
235
115
133
71
3.8
10.1
4.4
6.9
5.8
3.8
6.3
5.4
3.7
8.1
14.0
6.3
2.15
1.39
1.44
1.08
1.05
1.06
1.00
1.00
1.02
80
175
98
113
59
75
113
66
79
151
244
143
136
66
3.8
9.5
4.9
5.8
2.6
3.1
5.8
2.9
3.3
8.0
13.6
7.6
2.11
1.44
1.55
1.23
1.18
1.16
1.00
1.02
1.05
DNZ-NPG
DNZ-â-Phe
dichlorprop
170
57.5
108.5
108.5
108.5
89.5
89.5
89.5
7.5
7.1
6.1
4.0
3.0
3.6
92
84
DMP
3.5
3
67
82
3.0
3.5
88
3.8
a
mrac, mass of racemate corresponding to 0.46 mmol, except for DNB-Leu and DNZ-â-Phe (0.15 mmol), where solubility was limited. rCS/ rac
,
molar ratio CS/ racemate; k1 and k2, retention factors of the enantiomers. Flow rate, 3 mL/ min; ω ) 1000 rpm; descending mode; λ 230 nm.
Ammonium acetate buffer 0.1 M/ 2-PrOH/ DIPE (60:19:21), 10 mM CS (V ) 143 mL). b Ammonium acetate buffer 0.4 M/ DIPE (1:1), 10 mM CS
st
(V ) 159 mL at pH 6.85 and 135 mL at pH 6.0).
st
Figure 7. Elution profiles corresponding to the separation of increasing amounts of DNB-Leu by pH-zone-refining CCC. Stationary phase,
MIBK containing TFA (10 mM) and QN selector 1 (10 mM); mobile phase, water containing ammonia (20 mM). Flow rate 3 mL/min; ω ) 1200
rpm; descending mode, Vst ) 166-170 mL, P ) 72 bar. Vertical left axis, arbitrary absorbance units. Vertical right axis, pH. Horizontal axis, time
(min).
it has been successful in the resolution of ionizable compounds.40
The products to be separated elute in blocks according to their
pKa values and partition coefficients. In the chiral field, its principal
usefulness has been demonstrated for the separation of DNB-
Leu with a proline-derived CS.23,41
In the present investigation, the system water/ MIBK was
chosen to perform the experiments, using TFA as retainer in the
organic stationary phase and ammonia as displacer in the aqueous
mobile phase. The separation was brought about by filling the
entire CCC column with the acidified organic phase containing
the CS, followed by introduction of the racemic sample dissolved
in the stationary phase. The mobile phase was then pumped
through the column while the apparatus was running at 1200 rpm.
Along the column, the NH3 border determined the elution in zones
of the two enantiomers, as a function of the different pKa values
of their complexes with the CS.
Increasing amounts of DNB-Leu, up to 1200 mg (3.70 mmol),
were injected in the presence of QN-derived CS 1 in the stationary
phase. The resulting enantiomer separation is shown in Figure
7. Saturation of the system was attained at ∼900 mg of racemate
(2.77 mmol, molar ratio CS/ racemate, 0.60), whereas there was
450 mg (1.38 mmol) when the classical CCC mode was applied.
With the aim of enhancing the incipient separation observed
for dichlorprop using the binary system buffer/ DIPE in conven-
tional CCC, the system water/ 2-propanol/ DIPE was also inves-
tigated in the displacement mode. Although the separation was
somehow improved in these conditions, only a partial resolution
was achieved (Figure 8).
(40) Foucault, A. P.; Chevolot, L. J. Chromatogr., A 1 9 9 8 , 808, 3-22.
(41) Ma, Y.; Ito, Y.; Foucault, A. J. Chromatogr., A 1 9 9 5 , 704, 75-81.
4182 Analytical Chemistry, Vol. 74, No. 16, August 15, 2002

retained enantiomer, the amount of sample can be even larger,
as the limiting factor would be the amount of the more interacting
enantiomer. However, when the nonracemic mixture would be
enriched in the more retained counterpart, less sample might be
injected. At this point, the availability of CSs having similar
physicochemical properties but opposite chiral recognition abili-
ties, as is the case for QN and QD derivatives, becomes
advantageous.
The application of pH zone refining has demonstrated an
improvement in the loading capacity of the technique. However,
it is interesting to point out an important difference in operational
terms between this displacement chromatography and the clas-
sical CCC mode. The separations undertaken by pH zone refining
can be only performed by single-run processes. That is, the
stationary phase containing the CS can only operate once, as the
initial conditions change during the run (see Figure 3). Therefore,
the column should be emptied and later refilled with newly
acidified organic phase. In contrast, the classical CCC mode
maintains the stationary phase unmodified into the column and
consecutive injected samples can be resolved, as in HPLC. The
convenience of using one or the other mode must be investigated
for every case, to take the maximum profit of the planned runs.
Figure 8. Elution profile corresponding to the separation of 108.5
mg (0.46 mmol) of dichlorprop by pH-zone-refining CCC. Solvent
system, DIPE/2-PrOH/water (60/19/21). Stationary phase contained
TFA (10 mM) and QN selector 1 (10 mM); mobile phase, contained
ammonia (20 mM). Flow rate 3 mL/min; ω ) 1200 rpm; descending
mode, Vst ) 164 mL, P ) 64 bar. Vertical left axis, arbitrary
absorbance units. Vertical right axis, pH. Horizontal axis, time (min).
Comparison of Operational CCC Modes: Loading Limits.
The limiting loadability in pH zone refining for DNB-Leu (Figure
7), corresponding to a CS/ racemate molar ratio of 0.6, implies
that 2 equiv of CS is able to resolve 3 equiv of racemate. This
apparently incongruous result led us to the necessity of further
understanding the underlying mechanism of this displacement
chromatographic mode. In HPLC or classical CCC, applied to the
separation of enantiomers, one might expect that at the saturation
limit the molar ratio CS/ racemate has to be 1:1. Thus, if we check
the results presented above (see Figure 6), when 300 mg of DNB-
Leu was separated, the CS/ racemate molar ratio was 1.33; that
is, 4 equiv of CS resolved ∼3 equiv of analyte. The attempt of
achieving 450 mg of racemic mixture in a single run (CS/ racemate
molar ratio, 0.88) was unsuccessful as 9 equiv of CS was not
enough to separate 10 equiv of racemate. Saturation was attained.
However, this was not the case in the displacement mode.
In pH zone refining, the partition of the analyte into the mobile
phase has been drastically reduced at the beginning of the
chromatographic process. All the sample remains in the stationary
phase in equilibrium between the free acidic and the associated
form with the CS. When the pH starts changing and reaches the
pKa value of the analyte enantiomers, the less interacting enan-
tiomer starts to distribute to the mobile phase, while the other
one, which is bonded with more affinity to the CS, stays in the
stationary phase. In contrast to what we have in classical
chromatography, in this displacement mode, the more retained
enantiomer would not pass to the mobile phase until all the less
retained counterpart has abandoned the stationary phase. Two
conditions limit this phenomenon. The number of molecules of
the more retained enantiomer has to be lower or equal to the
number of molecules of CS, and the difference in association
constants between each one of the enantiomers and the CS has
to be high enough.
CONCLUSIONS
The combination of QN- or QD-derived CSs and CCC provides
a very efficient tool for the resolution of certain amino acid
derivatives at preparative scale. The appropriate chemical modi-
fication of the CSs, starting from the naturally available alkaloids,
was needed to ensure the correct distribution in the chromato-
graphic solvent systems. The introduction of a bulky adamantyl-
carbamate functionality together with a lipophilic side chain
resulted in a suitable compromise to enhance chiral recognition,
while keeping appropriate physicochemical properties such as the
CS staying stable in the organic phase during the CCC run. For
comparative reasons, all experiments were performed with 10 mM
concentrations of the CSs; however, this value is below the
solubility limit of these compounds in the solvent systems tested,
and therefore, higher loading capacities could be envisaged. The
results thus obtained confirm the high potential of the technique
for preparative enantioseparations, especially in the pH-zone-
refining mode, and open up new possibilities in the search of other
applicable chiral selectors.
ACKNOWLEDGMENT
Financial support from the European Union (BriteEURAM
Project BE 96-3159) and from the Direccio´n General de Ensen˜ anza
Superior e Investigacio´n Cient´ıfica of Spain (Project PB96-0382)
is acknowledged. The authors gratefully thank the Austrian
Academic Exchange Service and the Ministerio de Educacio´n y
Cultura of Spain for granting the collaboration within the Inte-
grated Actions Program (HU 1998-0018). P.F. thanks the European
Commission for a postdoctoral TMR Marie Curie Research
Training Grant and the Generalitat de Catalunya for a postdoctoral
research contract (RED99-08).
Consequently, for a racemic mixture, the limit of loading
capacity in pH zone refining would be a 2:1 CS/ racemate ratio.
The experimental results are in good agreement with this
rationale. As already mentioned, total resolution of 3 equiv of DNB-
Leu was performed in the presence of 2 equiv of CS, whereas
saturation was reached when 1200 mg of racemic solute was
injected (see Figure 7), as the CS/ racemate molar ratio was 5:11.
Furthermore, this rationale can be extended to nonracemic
mixtures. In this case, if the mixture is enriched in the less
Received for review April 1, 2002. Accepted May 21, 2002.
AC020209Q
Analytical Chemistry, Vol. 74, No. 16, August 15, 2002 4183