High-performance liquid chromatography separation of enantiomers of mandelic acid and its analogs on a chiral stationary phase

Article abstract of DOI:10.1002/chir.20767

The enantiomers of mandelic acid and its analogs have been chromatographically separated on a chiral stationary phase (CSP) derived from 4-(3,5-dinitrobenzamido) tetrahydrophenanthrene. The rationale of separations of these compounds is discussed with respect to the method development for determining enantiomeric purity and possibility of obtaining enantiomerically pure materials by high-pressure liquid chromatography. The relationship of analyte structure to the extent of enantiomeric separation has been examined and separation factors (a) are presented for various groups of structurally related compounds. Chiral recognition models have been suggested to account for the observed separations. These models provide mechanistic insights into the chiral recognition process.

Full text of DOI:10.1002/chir.20767

CHIRALITY 22:479–485 (2010)
High-Performance Liquid Chromatography Separation
of Enantiomers of Mandelic Acid and Its Analogs on a Chiral
Stationary Phase
2
3
RITU ANEJA,1 PRATIBHA MEHTA LUTHRA, AND SATINDER AHUJA
*
*
1Department of Biology, Georgia State University, Atlanta, Georgia
2BR Ambedkar Center for Biomedical Research, University of Delhi, Delhi, India
3Ahuja Consulting, Calabash, North Carolina
ABSTRACT
The enantiomers of mandelic acid and its analogs have been chromato-
graphically separated on a chiral stationary phase (CSP) derived from 4-(3,5-dinitroben-
zamido) tetrahydrophenanthrene. The rationale of separations of these compounds is
discussed with respect to the method development for determining enantiomeric purity
and possibility of obtaining enantiomerically pure materials by high-pressure liquid
chromatography. The relationship of analyte structure to the extent of enantiomeric sep-
aration has been examined and separation factors (a) are presented for various groups
of structurally related compounds. Chiral recognition models have been suggested to
account for the observed separations. These models provide mechanistic insights into
C
VV
the chiral recognition process. Chirality 22:479–485, 2010.
2009 Wiley-Liss, Inc.
KEY WORDS: mandelic acid; chiral stationary phase; chiral recognition
INTRODUCTION
still not fully understood. For this study, we employed the
Whelk-O-1 column (Brush or Pirkle type CSP) that has
4-(3,5-dinitrobenzamido) tetrahydrophenanthrene system
covalently bonded to 3-propyl silica (5 lm) participating in
chiral recognition (Fig. 1A). It was initially designed to
separate naproxen; however, it has been shown to be use-
ful for the separation of underivatized enantiomers of vari-
ous classes of compounds such as amides, epoxides,
esters, ureas, carbamates, ethers aziridines, phosphonates,
aldehydes, ketones, carboxylic acids, alcohols, and non-
steroidal anti-inflammatory drugs (NSAIDs).22–24 Separa-
tions on this CSP may be expected to utilize a combination
of p-acidity, p-basicity, hydrogen bond donor/acceptor
sites, and steric interactions for chiral recognition pur-
poses. This stationary phase presents two stereogenic cen-
ters and the accepted mechanistic significance of this CSP
suggests that the bulky substituents on vicinal stereogenic
centers confer considerable rigidity to apparently hold the
interaction sites in spatial positions that might enhance
the chiral recognition properties.
Enzymes, receptors, and other natural binding sites
within biological systems interact with different enantio-
mers in decisively different ways. As a result of such chiral
recognition, a drug enantiomer may differ in its pharmaco-
logical and/or toxicological profiles.1–3 In many cases,
only one isomer in a chiral compound is responsible for
the desired activity, while the other isomer may exhibit no
therapeutic value and may potentially cause unsuspected
adverse effects.4,5 Because of the different biological activ-
ities of enantiomers, the preparation of highly enantiopure
compounds is of vital importance.6,7 In recent years, vari-
ous chromatographic and electrophoretic methods have
been employed for the separation of enantiomers.8 Among
the many known methods of chiral separation, high-per-
formance liquid chromatography (HPLC) with chiral sta-
tionary phases (CSP) is a widely used approach for enan-
tiomeric separation at the analytical or preparative scale.9
Most chiral recognition models are based on the ‘‘three-
point interaction’’ theory frequently referred to as the
three-point rule.10 According to this rule, chiral recogni-
tion geometrically requires a minimum of three simultane-
ous interactions between the CSP and at least one solute
enantiomer, at least one of these interactions being stereo-
chemically dependent.11,12 This allows the enantiomers to
be resolved directly by a CSP, by forming short-lived dia-
stereoisomeric molecular complexes of nonidentical stabil-
ity.13–20 Interactions may be either bonding or repulsive.
It has been suggested that enantiomeric separation by
CSPs results from spatial factors and not due to differ-
ences in size or functionality of the analyte.21 However,
the precise details of chiral recognition mechanisms are
C
The model compound, mandelic acid (Fig. 1B) is gain-
ing increasing importance as a multifunctional chiral build-
ing block in preparative chemistry.25 For this reason, an
economical method for control of optical purity of this
class of compounds is highly desirable. A number of ana-
*Correspondence to: Ritu Aneja, Department of Biology, Georgia State
University, Atlanta, GA 30303. E-mail: raneja@emory.edu or Satinder
Ahuja, Ahuja Consulting, 1061 Rutledge Court, Calabash, NC 28467.
E-mail: sutahuja@atmc.net
Received for publication 10 December 2008; Accepted 17 June 2009
DOI: 10.1002/chir.20767
Published online 9 September 2009 in Wiley InterScience
(www.interscience.wiley.com).
VV
2009 Wiley-Liss, Inc.

480
ANEJA ET AL.
with automatic injector and column heater connected to a
Hitachi model L-3000 diode array detector was used for
the entire course of study. The mobile phase was hexane
with 10% 2-propanol/acetic acid mixture in the ratio 9:1.
The apparent pH of this mobile phase was 3.6. The study
was performed isocratically. The flow rate was 1 ml/min
and the column was run at ambient temperature (258C).
The compounds were used at a concentration of 1 mg/ml
unless appropriately diluted to improve peak quality.
RESULTS AND DISCUSSION
Mandelic acid and its analogs under study were
grouped in various categories depending upon the nature
of the substituents (R, A, B, and C). Group 1 (Table 1) lists
representative compounds with R being a phenyl group
with activating or deactivating substituents, B 5 H, A 5
OH/CH2OH/COOH/COOMe/COOEt/CH3/C2H5, and C
5 COOH/CH3/C2H5. Our data reveal that compounds 2–6
are retained onto the CSP. It is noticeable that all the
retained compounds possess electron-releasing substitu-
ents on the phenyl ring as a common feature. The reten-
tion can be explained by the increased strength of bonded
interactions of the analyte enantiomers with the CSP such
that they do not resolve in the chosen solvent system. The
following interactions seem evident. These are: (a) p do-
nor–p acceptor interactions between the electron-rich aryl
system of the analyte and the electron deficient dinitroben-
zamido (DNB) system of the CSP, (b) hydrogen bond
between the acidic dinitrobenzamido NꢀꢀH proton of the
CSP and the carbonyl O of the analyte, and (c) hydrogen
bond between the carbonyl O of the DNB group of the
CSP and hydroxyl H of the analyte, (Fig. 2). In compounds
2–6, another hydrogen-bonded interaction is thought to
occur between the hydroxyl H and the O of the NO2
system.
Fig. 1. Stationary phase 4-(3,5-dinitrobenzamido)tetrahydrophenan-
threne system covalently linked to 3-propyl silica surface (5 lm) in a Whelk-
O-1 column (Brush or Pirkle type CSP). Color represents substructure com-
ponents involved in noncovalent interactions. A: (i) Blue: pi-pi interaction.
(ii) Red, green, and pink: hydrogen bond donor and acceptor. B: (i) Molecu-
lar structure of mandelic acid. (ii) Mandelic acid analogs, where R, A, B, C
represents variable substituents. [Color figure can be viewed in the online
issue, which is available at www.interscience.wiley.com.]
This suggests that the p donor–p acceptor interaction
between the electron-rich aryl system of the analyte and
the electron deficient DNB system of the CSP is one of
the most significant interaction and plays an important
role in maintaining the steric rigidity of the molecule to
facilitate the rest of the interactions. Additionally, a substi-
tuted phenyl ring would favor a stronger p–p interaction
via the DNB system of the CSP. The enantiomers of com-
pounds 1 and 9 do not resolve (a 5 1) as they possess
nonsubstituted phenyl rings that do not offer sufficient
interactions for enantiodifferentiation to occur with the
present mobile phase. It may be noted that in compounds
10–14, the phenyl ring of the analyte is a p-acceptor sys-
tem that is expected to interact with the electron-rich tetra-
hydrophenanthrene system of the CSP. The other interac-
tions are similar and these compounds are also retained
by the CSP (Fig. 3).
logs of mandelic acid designed by preferentially varying
the substituents (R, A, B, and C) were chromatographed
on the CSP to evaluate the role of the CSP. They have
been discussed in separate groups (Tables 1–4) with
respect to their separation factor (a) and capacity ratio
(k0). From an analysis of the relationships between analyte
structure and chromatographic behavior, we have pro-
posed chiral recognition models to rationalize the experi-
mental observations to account for elution orders and the
magnitude of the observed separation factor, a. We
believe that the present approach would have far reaching
utility as a means of obtaining enantiomerically pure, con-
figurationally known mandelic acid analogs.
EXPERIMENTAL
Materials and Methods
Of particular interest is a better separation factor (a 5
1.4) when we have heterodisubstituted phenyl rings in
A (R,R)-Whelk-O 1 4.6 mm 3 250 mm column packed compounds 7 and 8. These analytes have OH and OMe
with 5 l particles was obtained from Regis Technologies. groups in ortho positions to each other which might facili-
All samples were obtained from Sigma-Aldrich Chemical tate intramolecular hydrogen bonding, thereby, disrupting
Co., all mobile phase solvents were obtained from Fisher- the interaction with the CSP at that face. This interaction
Scientific. A Varian model 5000 binary chromatograph may lead to a reduction in the number of interactions of
Chirality DOI 10.1002/chir

ENANTIOMERIC SEPARATION OF MANDELIC ACID AND ITS ANALOGS
481
k01
TABLE 1. Representative analytes with R 5 phenyl group with activating or deactivating substituents,
A 5 OH/CH2OH/COOH/COOMe/COOEt/CH3/C2H5, B 5 H, C 5 COOEt, CH3, C2H5
Analyte
R
A
B
C
a
Peak Width
1
2
3
4
5
6
7
8
9
Mandelic acid
4-Hydroxy mandelic acid
3-Hydroxy mandelic acid
3,4-Di-hydroxy mandelic acid
3-Methoxy mandelic acid
4-Methoxy mandelic acid
3-Hydroxy-4-methoxy mandelic acid
3-Methoxy-4-hydroxy mandelic acid
3-Phenyl lactic acid
Phenyl
OH
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
COOH
COOH
COOH
COOH
COOH
COOH
COOH
COOH
COOH
COOH
COOH
COOH
COOH
1
v broad
Retained
Retained
Retained
Retained
Retained
Narrow
Narrow
Narrow
Retained
Retained
Narrow
Retained
Narrow
Narrow
Wide
n/a
n/a
n/a
n/a
n/a
n/a
0.08
0.08
n/a
n/a
n/a
n/a
n/a
1.63
2.01
1.71
2.09
0.81
n/a
4-hydroxyphenyl
3-hydroxyphenyl
3,4-dihydroxyphenyl
3-methoxy phenyl
4-methoxy phenyl
3-hydroxy-4-methoxyphenyl OH
3-methoxy-4-hydroxyphenyl OH
Benzyl
OH
OH
OH
OH
OH
n/a
n/a
n/a
n/a
n/a
1.41
1.4
1
OH
OH
OH
OH
OH
OH
OH
CH3O
CH2OH
CH2OH
CH2OH
10 4(Trifluoromethyl)mandelic acid
11 (2,5-Difluorophenyl)hydroxy acetic acid 2,5-difluorophenyl
4-(trifluoromethyl)phenyl
n/a
n/a
1
12 4-Chloromandelic acid
13 4-Bromomandelic acid
14 Methyl mandelate
15 Ethyl mandelate
16 a-Methoxyphenyl acetic acid
17 Tropic acid
4-chloro phenyl
4-bromo phenyl
Phenyl
Phenyl
Phenyl
Phenyl
Phenyl
Phenyl
n/a
COOMe 1.06
COOEt
COOH
COOH
CH3
1.07
1.28
1.04
1.7
Narrow
Narrow
Narrow
18 2-Phenyl-1-propanol
19 b-Ethylphenethyl alcohol
C2H5
1
the analyte with the CSP and thus increase the magnitude maintain interactions with the CO of the CSP, leading to
of separation factor. In the case of compound 12, our data the onset of separations (a 5 1.04). This is further sup-
suggest that chloro-substituted phenyl ring system ported by an analysis of compounds 18 and 19. In com-
behaves similar to the parent model compound, mandelic pound 18 (A 5 CH2OH, C 5 CH3), a possibility of intramo-
acid; most likely this is due to the negative inductive and lecular hydrogen bonding exists between the OH of
positive mesomeric effects of the chlorine atom on the CH2OH and H of CH3 which places the H of CH2OH in
phenyl ring system. In the case of bromo-substituted phe- close proximity to potentiate hydrogen-bonded interac-
nyl ring in compound 13, the large size of Br atom and the tions with the carbonyl O of the CSP (Fig. 4). These inter-
availability of empty molecular orbitals on Br might favor actions dictate a strong chiral recognition process leading
conditions conducive to some additional interactions caus- to higher a value (a 5 1.7).
ing retention to occur with the present solvent system.
To rationalize the observed chromatographic data for
Our results suggest that compound 16 resolves on the Group 2 analytes, we suggest that the aryl portions of this
CSP with a 5 1.28. This suggests that that the methylation group of compounds would be utilized as p-donor sites
of OH on mandelic acid might distort the hydrogen- forming van der Waals (p–p) complexes with the 3,5-DNB
bonded interactions between the hydroxyl H of the analyte moiety of the CSP. The data depict that the magnitude of
and the carbonyl O of the DNB group on CSP, thereby a increases (as does the capacity ratio, k0) as the p-donor
causing chiral recognition to occur. It is noteable that nature of the aryl unit is increased (Compound
upon introduction of a methylene group (CH2) between 20<21<22). This may arise from a strengthening of the
the stereogenic center and hydroxyl group on mandelic vander waals p2p interaction involved in the chiral recog-
acid (compound 17), we begin to obtain separations. This nition process. It suggests that the OH of the analyte inter-
may be due to the exclusion of hydrogen-bonded interac- acts by hydrogen bonding with the 3,5-dinitrobenzamido
tions that were present between hydroxyl H of the analyte NH of the CSP. In addition, a tertiary interaction can occur
and the carbonyl O of the DNB group on the CSP. It sug- between the carbinyl hydrogen at the less basic 3,5-DNB
gests that the OH of CH2OH no longer remains in plane to carbonyl oxygen. In compound 20, it appears that due to
TABLE 2. Representative analytes with R 5 phenyl group with activating or deactivating substituents,
A 5 OH, B 5 H, C 5 COOH, CN, CH2OH
Analyte
R
A
B
C
a
Peak Width k01
1
Mandelic acid
20 Mandelonitrile
21 4-Methoxy mandelonitrile
Phenyl
Phenyl
4-methoxy phenyl
OH
OH
OH
H
H
H
H
H
H
COOH
CN
CN
1
1
1.2
v broad
Narrow
Narrow
Narrow
Narrow
Wide
n/a
n/a
2.72
6.79
1.77
3.4
22 Hydroxy(3-hydroxy-4-methoxyphenyl)acetonitrile 3-hydroxy-4-methoxy phenyl OH
23 2-Phenyl-1,2 ethanediol
24 1(2-Naphthyl)1,2-ethanediol
CN
1.35
Phenyl
Naphthyl
OH
OH
CH2OH 1.07
CH2OH 1.2
Chirality DOI 10.1002/chir

482
ANEJA ET AL.
TABLE 3. Representative analytes with R 5 phenyl group with activating or deactivating substituents,
A 5 alkyl chain, B 5 H, C 5 COOH
Analyte
2-Phenyl butyric acid
2(o-tolyl)-butyric acid
2(o-tolyl)-valeric acid
4-Methyl-2(o-tolyl)valeric acid
4-Isobutyl-a-methylphenyl acetic acid
2(4-Nitrophenyl)propionic acid
2-Fluoro-a-methyl-4-biphenyl acetic acid
6-Methoxy-a-methyl-2-naphthalene acetic acid
R
A
B
C
a
Peak Width
k01
25
26
27
28
29
30
31
32
Phenyl
C2H5
C2H5
C3H7
C4H9
CH3
CH3
CH3
CH3
H
H
H
H
H
H
H
H
COOH
COOH
COOH
COOH
COOH
COOH
COOH
COOH
1.14
1.07
1.09
1.2
1.17
1.04
1.19
1.97
Narrow
Narrow
Narrow
Narrow
Narrow
Narrow
Narrow
Narrow
0.52
0.43
0.39
0.3
0.43
1.89
0.82
3.29
2-methyl phenyl
2-methyl phenyl
2-methyl phenyl
4-isobutyl phenyl
4-nitro phenyl
2-fluoro biphenyl
6-methoxy naphthyl
the strong CN multiple bonding, the CN group does not This holds true for our data as we have observed in the
participate in interactions with the CSP. Therefore, when course of our study that conformational rigidity induced at
COOH is replaced by CN in compound 20, there are insuf- any of the carbons attached to the stereocenter of the ana-
ficient interactions required to cause chiral recognition, lyte results in better chiral recognition.
thus mandelonitrile fails to resolve. However, in com-
pound 21, when R 5 substituted phenyl that is, 4-methoxy of Group 3 (Table 3) present the possibility for three si-
phenyl, reasonable separations (a 5 1.25) are obtained. multaneous attractive interactions. First, p-donor–
Our data suggest that typical separation of enantiomers
a
This is possibly due to the strengthening of the p donor– acceptor interaction is expected to occur between the
acceptor interactions that present greater chiral recogni- highest occupied molecular orbital (HOMO) of the elec-
tion processes resulting in the separation of enantiomers tron-rich phenyl moiety and the lowest unoccupied molec-
on the CSP. This data can also be rationalized in the light ular orbital (LUMO) of the electron deficient dinitrobenza-
of steric rigidity due to bulky groups in compound 22 (R mido moiety of the CSP. Second, a hydrogen bond
5 3-OMe, 4-OH phenyl). We have suggested in the earlier between hydroxyl H of the carboxylate group to the amido
discussion that OH and OMe in ortho position to each nitrogen of 3,5-DNB of CSP. Third, we envision an interac-
other might result in intramolecular hydrogen bonding. In tion between the "A" (alkyl chain) group of the analyte
addition, the H of CH3 group on the phenyl ring can possi- and the electronegative oxygen of the NO2 group of the
bly serve as a potential H-bond receptor site for O of 3- DNB and/or the electron-rich tetrahydrophenanthrene
NO2 group of the DNB system on the CSP. In compound system of the CSP (Fig. 5). This interaction presumably
23, when CN is replaced by CH2OH, we begin to observe arises from a hyperconjugative effect, which is a field
separations. It may be inferred from the chromatographic effect and arises due to anomalous electron release pat-
data that the interactions via the alcoholic OH group of the terns of alkyl groups. Implicit in our present and prior chi-
analyte are weaker than the acidic OH function as in man- ral recognition models is the tenet that chiral recognition
delic acid (compound 1). In compound 24, when the R is most efficient when low-energy conformation of the
group is changed to naphthyl, we observe good separa- most strongly retained analyte enantiomer and CSP are
tions (a 5 1.25) plausibly, because of better p–p interac- involved during interaction.
tions between naphthyl group and the 3,5-DNB system of
From the data in Table 3 (compounds 26–28, A 5 C2H5,
the CSP. This suggests that the elution orders of the enan- C3H7, C4H9), it is seen that the magnitude of separation
tiomers is related to the absolute configuration by a chiral factors are essentially dependent on the length of the alkyl
recognition rationale that also provides insight into the ef- subsituent on the chiral center of the analyte. This sug-
ficiency of the chiral recognition process.
gests that increasing the chain length results in augmenta-
As in achiral molecular recognition, chiral recognition is tion of the interaction probably due to an optimization of
believed to be facilitated by conformational bias and/or the distance between O of 3-NO2 group of the DNB sys-
rigidity and by spatial localization of interaction sites.14 tem and the electropositive H of the alkyl group. Our data
TABLE 4. Representative analytes with R 5 phenyl group with activating or deactivating substituents,
A 5 OH, B 5 H, C 5 COOH, alkyl
Analyte
Mandelic acid
sec-phenethyl alcohol
1-Phenyl-1-propanol
4-Methoxy-a-methylbenzyl alcohol
1(2-Methoxyphenyl)-2-propanol
a-Ethylphenethyl alcohol
1-Phenyl-2-propanol
R
A
B
C
a
Peak Width
k01
1
Phenyl
Phenyl
Phenyl
OH
OH
OH
OH
OH
OH
OH
H
H
H
H
H
H
H
COOH
CH3
C2H5
CH3
CH3
C2H5
CH3
1
1
v broad
Narrow
Narrow
Narrow
Narrow
Narrow
Narrow
n/a
n/a
0.86
1.49
1.07
n/a
n/a
33
34
35
36
37
38
1.24
1.05
1.05
1
4-methoxy phenyl
2-methoxy benzyl
Benzyl
Benzyl
1
Chirality DOI 10.1002/chir

ENANTIOMERIC SEPARATION OF MANDELIC ACID AND ITS ANALOGS
483
TABLE 5. List of analytes that achieved separation using the Whelk-O1 column
Analyte
R
A
B
C
a
Peak Width k01
17 Tropic acid
Phenyl
CH2OH
CH3
OH
OH
OH
OH
C2H5
OH
C3H7
CH3
OH
C2H5
CH3
C4H9
OH
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
COOH
COOH
CH3
CH3
COOMe 1.06
COOEt 1.07
COOH
CH2OH 1.07
COOH
COOH
CH2OH 1.2
COOH
COOH
COOH
CN
C2H5
COOH
CN
COOH
COOH
CH3
1.04
1.04
1.05
1.05
Narrow
Narrow
Narrow
Narrow
Narrow
Narrow
Narrow
Narrow
Narrow
Narrow
Wide
Narrow
Narrow
Narrow
Narrow
Narrow
Wide
2.09
1.89
1.49
1.07
1.63
2.01
0.43
1.77
0.39
0.82
3.4
30 2(4-Nitrophenyl)propionic acid
35 4-Methoxy-a-methylbenzyl alcohol
36 1(2-Methoxyphenyl)-2-propanol
14 Methyl mandelate
15 Ethyl mandelate
26 2(o-tolyl)-butyric acid
4-nitro phenyl
4-methoxy phenyl
2-methoxy benzyl
Phenyl
Phenyl
2-methyl phenyl
Phenyl
2-methyl phenyl
2-fluoro biphenyl
Naphthyl
1.07
23 2-Phenyl-1,2 ethanediol
27 2(o-tolyl)-valeric acid
1.09
1.19
31 2-Fluoro-a-methyl-4-biphenyl acetic acid
24 1(2-Naphthyl)1,2-ethanediol
25 2-Phenyl butyric acid
29 4-Isobutyl-a-methylphenyl acetic acid
28 4-Methyl-2(o-tolyl)valeric acid
21 4-Methoxy mandelonitrile
34 1-Phenyl-1-propanol
Phenyl
1.14
1.17
1.2
0.52
0.43
0.3
4-isobutyl phenyl
2-methyl phenyl
4-methoxy phenyl
Phenyl
1.2
2.72
0.86
1.71
6.79
0.08
0.08
0.81
3.29
OH
CH3O
1.24
1.28
1.35
1.40
1.41
1.7
16 a-Methoxyphenyl acetic acid
22 Hydroxy(3-hydroxy-4-methoxyphenyl)acetonitrile 3-hydroxy-4-methoxy phenyl OH
Phenyl
Narrow
Narrow
Narrow
Narrow
Narrow
8
7
3-Methoxy-4-hydroxy mandelic acid
3-Hydroxy-4-methoxy mandelic acid
3-methoxy-4-hydroxyphenyl OH
3-hydroxy-4-methoxyphenyl OH
Phenyl
18 2-Phenyl-1-propanol
32 6-Methoxy-a-methyl-2-naphthalene acetic acid
CH2OH
CH3
6-methoxy naphthyl
COOH
1.97
suggest that introduction of an electron donating methyl pounds where group C 5 CN or CH2OH. On the basis of
group at position 2 of the phenyl (R) group reduces the observed chromatographic data, we suggest that interac-
stability of the p donor–acceptor complex by steric repul- tions via COOH are not significantly responsible for con-
sion at the site of interaction with O of 3-NO2 group of the ferring stereoselectivity in chiral recognition processes.
DNB system on CSP. This is evident from the reduction in Interestingly, in compound 34, when group C is changed
a values from 1.14 in compound 25 to 1.07 in compound
26. However, introduction of an electron withdrawing nitro
group on the phenyl ring in compound 30 also reduces the
stability of the diastereomer formed on the CSP. This may
be due to a repulsive interaction between the two electron
deficient phenyl rings causing a destabilization of the p do-
nor–acceptor interaction. In compound 29, we suggest that
the bulky isobutyl-substituted phenyl is forced out of plane
to facilitate p donor–acceptor interaction between the elec-
tron deficient DNB and the electron-rich phenyl ring. This
would allow the alkyl group (CH3) to come in close prox-
imity to the O of 3-NO2 group of the DNB system resulting
in better interactions at that face. The same explanation
can be extended to rationalize separations in compound 31
and 32. In addition to the above discussed interactions,
compound 32 possesses an electron-releasing naphthyl
group which offers much stronger p–p interaction with the
DNB system on the CSP resulting in increased a values
compared to the rest of the compounds in this series.
The data suggest that replacement of COOH group in
mandelic acid (compound 1) by an alkyl group (as in
Group 4 analytes, see Table 4) disrupts the very strong H-
bonding interaction and all analytes of this group elute
from the column. Furthermore, the data indicate that chro-
matographic behavior of compound 33 is similar to com-
pound 1 on the CSP. This is reflected in the elution of
both of these compounds from the CSP without separation
Fig. 2. Proposed chiral recognition model between the CSP and analy-
tes (2–6, 12–15). The colored pairs shown with dotted lines represent part-
ners for noncovalent interactions. [Color figure can be viewed in the
online issue, which is available at www.interscience.wiley.com.]
(a 5 1). This behavior has also been potentiated by com-
Chirality DOI 10.1002/chir

484
ANEJA ET AL.
Fig. 4. Proposed chiral recognition model between the CSP and analy-
tes (18, 19). The colored pairs shown with dotted lines represent partners
for noncovalent interactions. [Color figure can be viewed in the online
issue, which is available at www.interscience.wiley.com.]
in the analyte does not enhance the interactions on the
CSP. Contrary to expectations, compound 37 also behaves
similar to compound 38 and 33. This may be attributed to
the negligible repulsive interactions between the CSP and
the ethyl group attached to the chiral carbon bearing the
tropylium ring system. Additionally, the tropylium ring
system might lead to reduced p–p interactions. Therefore,
Fig. 3. Proposed chiral recognition model between the CSP and analy-
tes (10–14). The colored pairs shown with dotted lines represent partners
for noncovalent interactions. [Color figure can be viewed in the online
issue, which is available at www.interscience.wiley.com.]
from CH3 (as in compound 33) to a more bulky C2H5, we
observe that the enantiomers separate very nicely (a 5
1.24). It suggests that in compound 34, C2H5 group offers
a stereochemically dependent chiral recognition process
via a steric repulsion with the propyl chain on silica col-
umn due to the bulky ethyl group. However, this interac-
tion is absent when a methyl group is present (compound
33, C 5 CH3). Thus, a difference in steric bulk between
the two alkyl substituents, on the stereogenic center
seems to be important for enantioselectivity. As the phenyl
ring is changed to 4-methoxyphenyl in compound 35, we
observe a separation due to a differential stereochemically
dependent interaction being offered by the 4-OMe group
on the phenyl ring. The 4-OMe group most likely partici-
pates in hydrogen-bonded interactions with O of 4-NO2 of
the DNB system on CSP (similar to compound 18). How-
ever, replacement of methyl group with COOH (as in com-
pound 6) shows that the COOH hydrogen bonding interac-
tion plays a very significant role in retaining compound 6
onto the CSP. The behavior of compound 38 is similar to
compound 33, thus indicating that incorporation of a meth-
Fig. 5. Proposed chiral recognition model between the CSP and analy-
tes (25–32). The colored pairs shown with dotted lines represent partners
for noncovalent interactions. [Color figure can be viewed in the online
ylene group between the chiral center and the phenyl ring issue, which is available at www.interscience.wiley.com.]
Chirality DOI 10.1002/chir

ENANTIOMERIC SEPARATION OF MANDELIC ACID AND ITS ANALOGS
485
peptides antibiotic chiral stationary phases Il. Farmaco 2002;57:513–
529.
compound 36 is not retained for longer by the CSP (com-
pound 36, k0 5 1.07 compared with compound 35, k0 5
1.49). However, separation is observed in compound 36
because the 2-methoxybenzyl group has better p-donor
properties.
7. Ding GS, Liu Y, Cong RZ, Wang JD. Chiral separation of enantiomers
of amino acid derivatives by high-performance liquid chromatography
on a norvancomycin-bonded chiral stationary phase. Talanta 2004;62:
997–1003.
Table 5 provides ready reference to all compounds that
were resolved well by this column. It should be noted that
in this study, all the compounds for which separation was
obtained had either a hydroxyl group or a carboxylic acid
group attached to the chiral carbon. No separation was
obtained for any compound where both groups were
directly attached to the chiral carbon simultaneously
except when heterodisubstituted phenyl rings are present
as in compound 7 and 8. In summary, chiral recognition
models based upon three-point interaction has helped us
rationalize the separation of a large number of madelic
acid analogs.
8. Allenmark S, Schurig V. Chromatography on chiral carriers. J Mater
Chem 1997;7:1955–1963.
9. Cass QB, Batigalhia F. Enantiomeric resolution of a series of chiral
sulfoxides by high-performance liquid chromatography on polysaccha-
ride-based columns with multimodal elution. J Chromatogr A 2003;
987:445–452.
10. Dalgliesh CE. The optical resolution of aromatic amino-acids on paper
chromatograms. J Chem Soc 1952;3940–3942.
11. Pirkle WH. Chromatographic separation of enantiomers on rationally
designed chiral stationary phases. In: Ahuja S, editor. Chromatogra-
phy and separation chemistry. Washington, D.C.: ACS; 1986. p 101.
12. Ahuja S. Understanding chiral chromatography. In: Ahuja S, editor.
Chiral separation by chromatography. Washington, D.C.: ACS; 2000.
p 112.
In summary, there appears to be a relationship between
the structure and separation factor for mandelic acid ana- 13. Pirkle WH, House DW, Finn JM. Broad spectrum resolution of optical
isomers using chiral high-performance liquid chromatographic bonded
phases. J Chromatogr 1980;192:143–158.
logs, however, more than one chiral recognition process
seems to be important. This may be the consequence of
greater conformation flexibility. Although chiral recogni-
tion mechanisms are usually inferred from a body of chro-
matographic data and cannot be rigorously proven, we
consider that this approach is capable of producing a first
14. Pirkle WH, Finn JM. Chiral high-pressure liquid chromatographic sta-
tionary phases. 3. General resolution of arylalkylcarbinols. J Org
Chem 1981;46:2935–2938.
15. Pirkle WH, Schreiner JL. Chiral HPLC (high-performance liquid chro-
matographic) stationary phases. 4. Separation of the enantiomers of
approximation of the actual mechanism by which enantio-
selection is achieved. Furthermore, the ability to resolve a
wide assortment of mandelic acid analogs by the CSP
used in this study (or its improved successors) will facili-
tate the assignment of absolute configuration to these
compounds and provide means for preparation of small
samples of enantiomerically pure materials for chiroptic
studies.
bi-.beta. naphthols and analogs. J Org Chem 1981;46:4988–4991.
16. Pirkle WH, Finn JM, Hamper BC, Schreiner JL, Pribish JR. A useful
and conveniently accessible chiral stationary phase for the liquid chro-
matographic separation of enantiomers. In: Eliel E., Otsuka S, editors.
Asymmetric reactions and processes in chemistry. Washington, D.C.:
ACS Symposium Series, No. 185. 1982. p 245.
17. Pirkle WH, Welch CJ, Hyun MH. A chiral recognition model for the
chromatographic resolution of N-acylated 1-aryl-1-aminoalkanes. J Org
Chem 1983;48:5022–5026.
18. Pirkle WH, Welch CJ, Mahler GS, Meyers AL, Fuentes LM, Boes M.
Chromatographic separation of the enantiomers of N-acylated hetero-
cyclic amines. J Org Chem 1984;49:2504–2506.
ACKNOWLEDGMENTS
The authors thankfully acknowledge Drs. Anthony W.
Salotto and Ellen Weiser of Pace University for their con-
tributions.
19. Pirkle WH, Tsipouras A. Direct liquid chromatographic separation of
benzodiazepinone enantiomers. J Chromatogr 1984;29:291–298.
20. Pirkle WH, Pochapsky TC. Chiral molecular recognition in small
bimolecular systems: a spectroscopic investigation into the nature of
diastereomeric complexes. J Am Chem Soc 1987;109:5975–5982.
LITERATURE CITED
1. Hefnawy MM, Aboul-Enein HY. Enantioselective high-performance
liquid chromatographic method for the determination of baclofen in
human plasma. Talanta 2003;61:667–673.
21. Cram D. Preorganization-from solvents to spherands. J Angew Chem
Int Ed Engl 1986;25:1039–1057.
22. Welch CJ, Szczerba T, Perrin SR. Some recent HPLC separations of
the enantiomers of pharmaceuticals and other compounds using the
Whelk-O 1 CSP. J Chromatogr 1997;758:93–98.
2. Nation RL. Chirality in new drug development. Clinical pharmacoki-
netic considerations. Clin Pharmacokinet 1994;27:249–255.
3. Aboul-Enein HY, Abou-Basha LI, editors. Chirality and drug hazards.
In: The impact of stereochemistry on drug development and use,
Chapter 1. New York: Wiley; 1997. p 1.
23. Pirkle WH, Welch CJ, Lamm B. Design, synthesis, and evaluation of
an improved enantioselective naproxen selector. J Org Chem 1992;57:
3854–3860.
4. Ward TJ, Farris AB. Chiral separations using the macrocyclic antibiot-
ics: a review. J Chromatogr A 2001;906:73–89.
24. Pirkle WH, Welch CJ. An improved chiral stationary phase for the
chromatographic separation of underivatized naproxen enantiomers.
J Liq Chromatogr 1992;15:1947–1955.
5. Ward TJ. Chiral media for capillary electrophoresis. Anal Chem 1994;
66:632A–640A.
6. Aboul-Enein HY, Ali I. Optimization strategies for HPLC enantiosepa-
ration of racemic drugs using polysaccharides and macrocyclic glyco-
25. Gu¨nther K. Thin-layer chromatographic enantiomeric resolution via
ligand exchange. J Chromatogr 1988;448:11–30.
Chirality DOI 10.1002/chir