Chiral discrimination in diastereomeric salts of chlorine-substituted mandelic acid and phenylethylamine

Article abstract of DOI:10.1002/chir.20822

The crystal structures of diastereomeric salts of chloromandelic acid and phenylethylamine were determined and are presented herein. The structure comparison between less soluble salts and more soluble salts shows that weak interactions such as CH/π inter

Full text of DOI:10.1002/chir.20822

CHIRALITY 22:707–716 (2010)
Chiral Discrimination in Diastereomeric Salts of Chlorine-
Substituted Mandelic Acid and Phenylethylamine
1
QUAN HE,1 HASSAN GOMAA,1 SOHRAB ROHANI,1 JESSE ZHU, AND MICHAEL JENNINGS2
*
1Department of Chemical and Biochemical Engineering, the University of Western Ontario London, Ontario, Canada
2Department of Chemistry, the University of Western Ontario, London, Ontario, Canada
ABSTRACT
The crystal structures of diastereomeric salts of chloromandelic acid
and phenylethylamine were determined and are presented herein. The structure com-
parison between less soluble salts and more soluble salts shows that weak interactions
such as CH/p interactions and van der Waals gain importance and contribute to chiral
recognition when the hydrogen bonding patterns are very similar. Chirality 22:707–716,
C
VV
2010.
2010 Wiley-Liss, Inc.
KEY WORDS: crystal structure; chiral recognition; diastereomeric salt; X-ray
crystallographic analysis; hydrogen bonding
INTRODUCTION
between less and more soluble diastereomeric salts is
largely dependent on the existence and/or magnitude of
these three interactions. Among these interactions, the
hydrogen-bonding interaction is believed to account for
much of the chiral discrimination.5,9,13 The hydrogen-
bonding patterns in structures of two distereomeric salt
crystals are usually far from identical, whereas the hydro-
gen bonding patterns in less soluble salts with high resolu-
tion efficiency have similar characteristics.5,8
There is a growing interest in the pharmaceutical indus-
try for the production of pure enantiomers of drugs, since
a pair of enantiomers can have significant differences in
pharmacological activity. The resolution of a racemate via
diastereomeric salt formation first developed by Pasteur1
is one of the most attractive methods for obtaining pure
enantiomers of a given racemic acid or base on an indus-
trial scale. The advantages of such a method are the sim-
plicity of crystallization, the low cost of operation, the recy-
clability of resolving agents, and the availability of both
enantiomers. However, although much experience has
been garnered with more than 10,000 successful resolu-
tions having been reported, the details of diastereomeric
salt formation and the underlying chiral recognition mech-
anism in the resolution are not well understood. There is
not a universal criterion for the selection and design of
resolving agents. For a given racemate, the screening of
an efficient resolving agent still relies on trial and error,
which is very time-consuming and laboratory-intensive.2,3
Much research has been conducted to investigate the
stability difference between a pair of diastereomeric salts
on the basis of crystal structures.4–14 The rationale has
hinged on an expectation of getting valuable insight into
chiral discrimination as well as for predicting and design-
ing a suitable resolving agent for a given racemate. Cer-
tainly the success of a resolution process is due to the
large solubility difference between the resulting pair of dia-
stereomeric salts. Furthermore, the solubility difference is
generally related to the stability of the salts, and the stabil-
ity greatly depends on their crystal structures. Therefore,
chiral discrimination mechanism may be expounded by X-
Ray crystallographic analysis of the structures of the dia-
stereomeric salts.
Kinbara et al.4,5 systematically investigated the diaster-
eomeric salts of chiral primary amines with mandelic
acid derivatives and the diastereomeric salts of 2-arylalka-
noic acids with 1-aminoindan-2-ol as well as 2-amino-1, 2-
diphenylethanol. Their on-going efforts revealed that
there was a correlation between the efficiencies of the
optical resolutions and the crystal structures of the salts.
For successful resolutions, a characteristic hydrogen-
bonding network, consisting of a stable 21 column and
having planar boundary surfaces, was frequently found in
less soluble salt crystals. These crystals were considered
to be stabilized from the viewpoint of both their hydro-
gen-bonding and van der Waals interactions. Thus,
resolving agent can be designed to achieve stable hydro-
gen-bonding networks with target racemates using the
concept of crystal engineering.
Kobayashi et al.6,7 further focused on the role of CH/p
interaction in the less soluble salts and designed new
resolving agent, which potentially enhanced the CH/p
interactions. Investigation revealed that the unique proper-
ties of CH/p interaction largely contributed to the stabili-
Contract grant sponsor: Ningbo Nature Science Fund; Contract grant
numbers: No.2005B100085
*Correspondence to: Sohrab Rohani, Department of Chemical and Bio-
chemical Engineering, the University of Western Ontario London, On-
tario, N6G 4R3, Canada. E-mail: srohani@uwo.ca
Received for publication 30 July 2009; Accepted 23 October 2009
DOI: 10.1002/chir.20822
In general, there are three significant interactions in dia-
stereomeric salt crystals; these are hydrogen-bonding,
CH/p type interactions between aryl groups and finally
van der Waals interactions. The difference in stability
C
Published online 8 February 2010 in Wiley InterScience
(www.interscience.wiley.com).
VV
2010 Wiley-Liss, Inc.

708
HE ET AL.
Fig. 1. Structures of chloromandelic acids and (R)-1-phenylethylamine
zation of the crystals. They also found that except for the
Bialonska and Ciunik13 worked on diastereomeric salts
well-recognized 21 column hydrogen-bonding network, a in N-4-nitrobenzoylamino acid and strychnine system from
new closed globular cluster hydrogen-bonding network the standpoint of steric information in crystal structures.
existed in less soluble salts of arylethylamine with o-alkyl They proposed that the hydrophobic ‘‘lock and key’’ recog-
phenylphosphinothioic acids as well as with P-chiral alkyl- nition mode caused the surface to have more hydrophilic
phenylphosphinothioic acids. This kind of cluster hydro- properties, and the nature of hydrogen-bonding deter-
gen bonding network was composed of four phosphono- mined the crystallization sequence of diastereomeric salts.
thioate anion molecules and four ammonium cation
Langkilde et al.14 determined the crystal structures of
molecules to give a globular molecular cluster. The factors diastereoemric salts formed between (S)-latic acid and 1-
determining the pattern of hydrogen bonding network for phenylethylamine. The relative energies of salts were
the given system were also investigated.8–10
investigated by Hartree-Fock calculation leading to conclu-
Nishio’s and his coworkers11 systematically investigated sion that relative stabilities were attributed to interplay of
aromatic CH/p distance and access angel parameters in enthalpy and entropy effects.
diastereomeric salts on the basis of crystallographic data-
All the above progress implies that crystal structure
base study and proposed that the aromatic CH/p may play analysis of diastereomeric salts is one of the most promis-
an important role in determining the relative stability of ing methods for investigating and comparing the diaster-
crystals.
eomeric salts’ stabilities and is crucial for chiral discrimi-
Yoshioka et al.12 reported that excellent resolution of nation studies. The crystal structure investigation can
DL-phenylglycine with (1S)-(1)-camphor-10-sulfonic acid extract the common and valuable characteristic in less
is due to the free solubility of more soluble salt. Crystal soluble salts to interpret the successful resolutions.
structure examination of diastereomeric salts indicated Although, it is premature to give general rules for chiral
that the less soluble salt had a dense and stable molecular discrimination, the accumulation of kind studies hopefully
conformation consisting of strongly hydrophobic layers, move a step closer to allow systematic screening and
which resulted in low solubility in water. In contrast, the design of suitable resolving agents.
more soluble salt showed a coarser molecular conforma-
Mandelic acid derivatives are useful chiral intermediates
tion consisting of characteristic ‘‘holey’’ vacancy layers. and building blocks for the synthesis of biologically active
The honeycomb-like holes were important factor in the compounds, and they are also used as the resolving agents
high solubility of the more soluble salt.
in chiral resolution processes.15,16 Optically active (R)-1-
TABLE 1. Thermodynamic properties of diastereomeric saltsa
Salt
Melting point/8C
Heat of fusion/JK/mol
Solubilityb/g
Specific rotation [a]2D2/8
(R)-p-ClMAn˜(R)-PEA(less)
(S)-p-ClMAn˜(R)-PEA(more)
(R)-m-ClMAn˜(R)-PEA(less)
(S)-m-ClMAn˜(R)-PEA(more)
194.60
141.76
151.51
109.52
58.41
32.30
39.76
34.58
5.97
62.3
47. 7
248.0
50.5
245.5
51.8
142.15
aThe data in the table is the average value of three times measurements.
bWeight (g) of the solute dissolved in 100 g of methanol at 228C.
Chirality DOI 10.1002/chir

CHIRAL DISCRIMINATION IN DIASTEREOMERIC SALTS
709
TABLE 2. Crystal structure data of diastereomeric salts
(R)-p-ClMAn˜(R)-PEA
(S)-p-ClMAn˜(R)-PEA
(R)-m-ClMAn˜(R)-PEA
(S)-m-ClMAn˜(R)-PEA
Salts
(less soluble )
(more soluble)
(less soluble)
(more soluble)
Formula
C8H12N1-C8H6ClO32
C8H12N1-C8H6ClO32
C8H12N1-C8H6ClO32
C8H12N1-C8H6ClO32
Formula weight
Tempaerature (K)
Crystal system
Space group
307.76
150
307.76
150
Monoclinic
P21
10.4091
5.7635
13.2544
90
96.831
90
307.76
150
307.76
150
Monoclinic
P21
9.4779
5.9701
13.5755
90
91.166
90
Orthorhombic
P212121
6.8848
8.3979
26.9433
90
Orthorhombic
P212121
6.9214
8.4761
25.8423
90
˚
a(A)
˚
b(A)
˚
c(A)
a(8)
b(8)
g(8)
90
90
1557.8
1.312
4
90
90
1516.08
1.348
3
˚
V(A )
789.52
1.295
2
768.00
1.331
2
Dcalc (g/cm3)
Z
4
Rw
R
0.095
0.038
0.115
0.044
0.034
0.085
0.144
0.054
phenylethylamine (hereafter PEA) is well known to be a Chiral HPLC using (R, R) Whelk-O2 column (250 mm 3
good resolving agent for mandelic acid derivatives. It can 4.6 mm) from Regis Tech. Inc. (Morton Grove, IL).
effectively resolve mandelic acid, p-methoxymandelic acid,
Preparation of Less Soluble Salts
p-methylmandelic acid, and 2-naphthylglycolic acid by dia-
stereomeric salt formation.4–6 It is therefore straightfor-
ward to use PEA as a resolving agent during the course of
our study concerning the optical resolution of o-chloro-
mandelic acid (hereafter o-ClMA). However, contrary to
our expectation, we failed to resolve the o-ClMA by PEA
under our experimental conditions in spite of many efforts
to change the solvent and the crystallization conditions.
To elucidate the experimental result, resolutions of anal-
ogous racemates, m-chloromandelic acid (hereafter m-
ClMA), and p-chloromandelic acid (hereafter p-ClMA)
with PEA were performed. Crystal structures of the result-
ing diastereomeric salts of m-ClMA and p-ClMA with PEA
were obtained by single crystal X-ray diffraction. The com-
parison of crystal structures revealed the chiral recogni-
tion in concerning systems. The chemical structures of
concerned racemates and resolving agent are depicted in
Figure 1.
Less soluble salts were obtained from a general op-
tical resolution process. Add gradually (5.7 ml, 0.045
mol) (R)-PEA to a stirred solution of racemic m-ClMA or
p-ClMA (8.4 g, 0.045 mol) in 72 ml methanol form white
crystals. The slurry was heated to 333 K using a water
bath to dissolve the solid crystal. Maintain the solution at
333 K for 30 min and then cool the solution to 295 K
slowly. Keep the solution at 295 K for 60 min, collect the
precipitate and wash it twice with methanol. The filtered
precipitate is recrystallized in methanol to give the opti-
cally pure salt (R)-m-ClMAÁ(R)-PEA (1.45 g, 21% yield) or
(R)-p-ClMAÁ(R)-PEA (3.1 g, 45% yield).
Preparation of More Soluble Salts
More soluble salts were synthesized by enantio-
meric pure (S)-m-ClMA or (S)-p-ClMA with PEA. To
a solution (S)-m-ClMA or (S)-p-ClMA (2.0 g, 0.01 mol) in
10 ml 2-propanolto, (R)-PEA (1.3 ml, 0.01 mol) was added
gradually to afford white crystals. The crystals were col-
lected by filtration and washed with 2-propanol twice to
give (S)-m-ClMAÁ(R)-PEA (1.74 g, yield 53%) or (S)-p-
ClMAÁ(R)-PEA (1.85 g, yield 56%).
EXPERIMENTAL
General
(RS)-p-ClMA with 99% purity and (RS)-m-ClMA with 97%
purity was purchased from Alfa Aesar, A Johnson Matthey
Company (Ward Hill, MA). PEA with optical purity of 99%
ee was purchased from Sigma-Aldrich Canada (Oakville,
ON) and used without further purification.
The melting point and heat of fusion of diastereomeric
salts were determined by Mettler Toledo DSC 822e differ-
ential scanning calorimeter (Greifensee, Switzerland) with
heating rate 58C/min.
TABLE 3. Hydrogen-bonding geometry of
˚
(R)-p-ClMAÁ(R)-PEA [A and 8]
DÀÀH. . .A
d(DÀÀH) d(H. . .A) d(D. . .A) <(DHA)
O9-H9A. . .O11a
N13-H13A. . .O12
N13-H13B. . .O11b
N13-H13C. . .O12c
0.82
0.89
0.89
0.89
2.09
2.00
1.99
2.04
2.848 (2)
2.863 (2)
2.869 (3)
2.878 (3)
154
157
167
156
Specific rotations of salts were measured by Autopol IV
Digital Polarimeter Rudolph America (Hackettstown, NJ)
at 589 nm, equipped with a quartz cell of 100 mm path
Symmetry transformations used to generate equivalent atoms:
ax 2 1/2, 2y 1 5/2, 2z
bx 2 1/2, 2y 1 3/2, 2z
length. Optical purity of (R)-p-ClMA was determined by cx 1 1/2, 2y 1 3/2, 2z
Chirality DOI 10.1002/chir

710
HE ET AL.
TABLE 4. Hydrogen-bonding geometry of
TABLE 5. Hydrogen-bonding geometry of
˚
˚
(S)-p-ClMAÁ(R)-PEA [A and 8]
(R)-m-ClMAÁ(R)-PEA [A and 8]
DÀÀH. . .A
d(DÀÀH) d(H. . .A) d(D. . .A) <(DHA)
DÀÀH. . .A
d(DÀÀH) d(H. . .A) d(D. . .A) <(DHA)
O9-H9A. . .O9a
0.84
0.84
0.91
0.91
0.91
2.26
2.09
1.89
1.83
1.88
2.939 (2)
2.826 (2)
2.798 (2)
2.731 (2)
2.779 (2)
139
146
172
169
171
O9ÀÀH9A. . .O11a
N13ÀÀH13A. . .O12
N13ÀÀH13B. . .O11b
N13ÀÀH13CÀÀO9c
N13ÀÀH13C. . .O12c
0.84
0.91
0.91
0.91
0.91
2.13
1.94
1.99
2.41
2.01
2.890 (2)
2.812 (2)
2.873 (2)
3.045 (2)
2.882 (2)
150.5
160.3
164.1
127.3
159.0
O9-H9A. . .O11a
N13-H13A. . .O12b
N13-H13B. . .O11
N13-H13C. . .O12c
Symmetry transformations used to generate equivalent atoms:
a–x 1 1, y 1 1/2, 2z 1 1
Symmetry transformations used to generate equivalent atoms:
ax 2 1/2, 2y 2 1/2, 2z
bx, y 2 1, z
bx 2 1/2, 2y 1 1, 2z
c2x 1 2, 2y 2 1/2, 2z 1 1
cx 1 1/2, 2y 1 1, 2z
Solubility Determination
Comparison of Diastereomeric Salt Crystal Structures
The solubilities of more and less soluble salts were
Since the stability of a salt highly depends on its crystal
measured by gravimetric method. The diastereomeric structure, a crystallographic study of the pairs of the dia-
salts in the excessive amount were dissolved in methanol stereomeric salts was performed to clarify a chiral discrim-
in 10 ml vials, which were immersed in a constant temper- ination mechanism. Single crystals of both less and more
ature bath and were shaken gently for a few hours to soluble salts were obtained for X-ray crystallographic anal-
reach the equilibrium. The saturated solution was care- ysis. Crystals of less soluble salts of (R)-p-ClMAÁ(R)-PEA
fully taken out using a syringe equipped with a filter, and and (R)-m-ClMAÁ(R)-PEA are colorless rods crystallizing
the solution was weighed accurately. Then the solution in orthorhombic space group P212121, and the unit cell
was put in vacuum oven to evaporate the solvent thor- contains four (R)-p-ClMA or (R)-m-ClMA anion and four
oughly, and the weight of residue solute was obtained. (R)-PEA cations. Crystals of more soluble salts (S)-p-
Thus, the solubility of salt in methanol was determined. ClMA,Á(R)-PEA, and (S)-m-ClMAÁ(R)-PEA are colorless
Three parallel measurements were performed for each plates in monoclinic space group P21, and the unit cell con-
sample, and the average value was preported in this arti- tains two (S)-p-ClMA or (S)-m-ClMA anions and two (R)-
cle. The accuracy of balance was 60.01 mg.
PEA cations. Crystallographic lattice parameters are sum-
marized in Table 2, and all hydrogen bonds are listed in
Table 3–6. Figure 2 graphically illustrates the atomic-num-
bering schemes. Detailed crystallographic analysis of both
salts of [p-ClMA][PEA] have been presented in our previ-
ous work20,21 and the crystallographic data of salts of
[m-ClMA][PEA] are presented herein.
The crystal structure of the less soluble salt (R)-p-
ClMAÁ(R)-PEA, revealed a typical packing structure with
an acid-base salt structure containing (R)-p-ClMA anions
and the (R)-PEA cations, supported by electrostatic inter-
action, hydrogen-bonding, and van der Waals interactions.
The ammonium hydrogens formed three hydrogen bonds
with carboxylate oxygens of adjacent (R)-p-ClMA, namely
N13-H13A-O12, N13-H13B-O11(ii), and N13-H13C-O12(iii).
Thus, a columnar hydrogen-bonding network was formed
around a two-fold screw axis parallel to the a-axis formed
Crystal Structure Determination and Refinement
Single crystals were grown from a concentrated metha-
nol or 2-propanol solution of diastereomeric salts by slow
evaporation at room temperature. The single crystal was
mounted on a glass fiber, and data were collected at low
temperature 21238C on a Nonius Kappa-CCD area detec-
tor diffractometer with COLLECT17 using monochromatic
Mo-Ka radiation (k 5 0.71073 E). Crystal cell refinement
and data reduction were carried out using HKL2000
DENZO-SMN.18 The absorption correction was applied
using HKL2000 DENZO-SMN. The SHELXTL/PC V6.1419
for Window NT suite of programs was used to solve and
refine the structure.
TABLE 6. Hydrogen-bonding geometry of
RESULTS AND DISCUSSION
Thermodynamic Properties of Diastereomeric Salts
˚
(S)-m-ClMAÁ(R)-PEA [A and 8]
DÀÀH. . .A
d(DÀÀH) d(H. . .A) d(D. . .A) <(DHA)
The thermodynamic data of diastereomeric salts of both
p-ClMA and m-ClMA with PEA are summarized in Table 1.
The thermodynamic properties are related to the relative
solubilities of the salts leading to efficient optical separa-
tions by crystallization. The melting point and heat of
fusion of less soluble salts are significantly higher than
those of more soluble salts, which imply that less soluble
salts are thermodynamically more stable than the corre-
sponding more soluble salts. It is this stability difference,
which results in the success of chiral resolutions.
O9-H9A. . .O9a
0.84
0.84
0.91
0.91
0.91
2.4
2.06
1.92
1.90
1.92
3.014 (6)
2.841 (3)
2.818 (3)
2.771 (3)
2.801 (3)
130.1
155.1
171.4
160.6
162.1
O9-H9A. . .O12a
N13-H13A. . .O11
N13-H13B. . .O11b
N13-H13C. . .O12c
Symmetry transformations used to generate equivalent atoms:
a–x, y 2 1/2, 2z 1 1
b–x 1 1, y 1 1/2, z
c–x 1 1, y 2 1/2, 2z 1 1
Chirality DOI 10.1002/chir

CHIRAL DISCRIMINATION IN DIASTEREOMERIC SALTS
711
Fig. 2. Atomic-numbering molecule schemes of diastereomeric salts. [Color figure can be viewed in the online issue, which is available at www.inter
science.wiley.com.].
by ammonium hydrogens and carboxylate oxygens. A
schematic illustration of the 21 column structure is shown
in Figure 3. Additionally, there was another hydrogen
bond O9-H9A-O11 (i) between the hydroxy hydrogen of
(R)-p-ClMA and an adjacent carboxylate group, which
interlinked the two neighboring 21 columns. These hydro-
gen bonds formed a 2-dimensional network perpendicular
to the c-axis as shown in Figure 4a, extending throughout
the crystal. Viewing down either the a-axis or b-axis
revealed planar boundary surfaces among the hydrogen-
bonding layers as shown in Figures 4b and 4c, respec-
tively. This characteristic hydrogen-bonding network was
commonly observed in other salts of primary amine and
mandelic acid derivatives investigated by Kinbara et al.
and Kobayashi and Saigo.4–6 They proposed that for a suc-
cessful resolution it was necessary for the less soluble
salts to form this kind of hydrogen-bonding layer, consist-
ing of stable 21 columns and having planar boundary surfa-
ces. These planar boundary surfaces interact through van
der Waals forces between layers and thus close packing
can be realized in this crystal. Therefore, the increased
stability of these salts (less soluble) is evident from the
crystal structure coupling the hydrogen-bonding with the
van der Waals interactions.
In the crystal structure of the more soluble salt (S)-p-
ClMAÁ(R)-PEA, except the typical three hydrogen bonds
between ammonium hydrogens and carboxylate oxygens
as well as O9-H9A-O11(i) between the hydroxyl hydrogen
of (S)-p-ClMA and the carboxylate oxygen atom of another
(S)-p-ClMA, an additional hydrogen bond, namely O9-
H9A-O9(i) between the hydroxy hydrogen of (S)-p-ClMA
Fig. 3. Schematic illustration of 21 column hydrogen-bonding pattern
[Color figure can be viewed in the online issue, which is available at
www.interscience.wiley.com.].
and the hydroxy oxygen of another (S)-p-ClMA was
Chirality DOI 10.1002/chir

712
HE ET AL.
Fig. 5. Crystal structure of more soluble salt (S)-p-ClMAÁ(R)-PEA
Fig. 4. Crystal structure of less soluble salt (R)-p-ClMAÁ(R)-PEA
[Color figure can be viewed in the online issue, which is available at
www.interscience.wiley.com.].
[Color figure can be viewed in the online issue, which is available at www.
interscience.wiley.com.].
Chirality DOI 10.1002/chir

CHIRAL DISCRIMINATION IN DIASTEREOMERIC SALTS
713
observed as shown in Figure 5a, but interlink between ad-
jacent 21 columns was different. Ignoring the hydrogen
bonds between hydroxy hydrogen and carboxylate oxy-
Fig. 6. Crystal structure of less soluble salt (R)-m-ClMAÁ(R)-PEA
[Color figure can be viewed in the online issue, which is available at
www.interscience.wiley.com.].
observed. The pattern of this hydrogen-bonding network
was slightly different from the one in the corresponding
less soluble salt. The similar 21 column structure was
Fig. 7. Crystal structure of more soluble salt (S)-m-ClMAÁ(R)-PEA
[Color figure can be viewed in the online issue, which is available at
www.interscience.wiley.com.].
Chirality DOI 10.1002/chir

714
HE ET AL.
In the crystal structure of the more soluble salt (S)-m-
ClMAÁ(R)-PEA, as can be seen in Figure 7, the pattern of
this hydrogen-bonding network was exactly the same as
that observed in the more soluble salt of (S)-p-ClMAÁ(R)-
PEA. The hydrophobic boundary surface is not as planar
as that in less soluble salt but it is more even than that in
more soluble salt of (S)-p-ClMAÁ(R)-PEA.
Comparing the above structures, we found similar 21
column hydrogen-bonding networks in both less soluble
salts and the more soluble salt for the meta-ClMA and the
para-ClMA. This kind hydrogen-bonding pattern was con-
sidered to exist more often in less soluble salts and be im-
portant interactions responsible for stabilizing the salts of
arylethylamine with mandelic acid. But in our case, basi-
cally the hydrogen-bonding network does not showed sig-
nificant difference and only hydrogen bond linking the 21
column was slightly different between less soluble salts
and more soluble salts. As a consequence, hydrogen bond
interactions seemly did not play the most important role in
crystal stabilization. On the other hand, the difference in
planar boundary surfaces was dramatic in p-ClMA species
while it was not significant in m-ClMA species. Only crys-
Fig. 8. Aromatic group packing in hydrophobic layer of less soluble
salts [Color figure can be viewed in the online issue, which is available at
www.interscience.wiley.com.].
gen, there were a zig-zag hydrogen-bonding line between
hydroxy hydrogens and hydroxy oxygens. On the other
hand, as shown in Figures 5b and 5c, the boundary surfa-
ces in the more soluble salt was not planar as opposed to
what was observed in the less soluble salt. The substituent
on the phenyl ring projected out of the layer and made the
boundary surface uneven. So, close packing of the layers
is not as efficient as that in the less soluble salt, and, con-
sequently, the van der Waals interactions between the
layers are less strong.
In the crystal of less soluble salt (R)-m-ClMAÁ(R)-PEA,
the 21 columnar hydrogen-bonding pattern was found,
which was similar to that of less soluble salt (R)-p-
ClMAÁ(R)-PEA. However, the interlinks between 21 col-
umns were different, they consisted of the hydrogen bond
O9-H9A-O11(i) between the hydroxy hydrogen of (R)-m-
ClMA and the carboxylate oxygen of another (R)-m-ClMA
as well as N13-H13C-O9(iii) between the hydroxy oxygen
and the ammonium hydrogen. In the same manner, the
less soluble salt also had the same planar boundary surfa-
ces as that in less soluble salt of (R)-p-ClMAÁ(R)-PEA.
Thus, the hydrophobic layers stacked effectively by van
der Waals interaction. Hydrogen pattern and crystal pack-
ing are displayed in Figure 6.
Fig. 9. Aromatic group packing in hydrophobic layer of more soluble
salts [Color figure can be viewed in the online issue, which is available at
www.interscience.wiley.com.].
Chirality DOI 10.1002/chir

CHIRAL DISCRIMINATION IN DIASTEREOMERIC SALTS
715
tal packing by van der Waals was seemly not enough to chlorine does not project outward and makes this (R)-o-
lead to large stability difference in diastereomeric salts ClMA molecule shorter than that of PEA, which may
under investigation. Therefore, we believed that other fac- result in an uneven boundary surface. This remains a con-
tors may play a role in stabilization of the structure except jecture as the crystal structure of diastereomeric salts of o-
the van der Waals effect and hydrogen-bonding interac- ClMA with PEA could not be obtained as it consistently
tion.
Examining the modes of aromatic ring packing showed
gave an oily solution.
the distinctive differences between the less and more solu-
ble salts as depicted in Figures 8 and 9. In the crystal
structure of the less soluble salt (R)-p-ClMAÁR-PEA, the ar-
omatic rings of (R)-p-ClMA were parallel to one another,
so are the aromatic rings of R-PEA. The aromatic ring of
(R)-p-ClMA was nearly at right angles to the R-PEA aro-
matic ring, and the interplanar angle between aromatic
planes was 89.18. An interplanar angle of 89.18 was also
found in the less soluble salt (R)-m-ClMAÁ(R)-PEA. These
aromatic rings formed a T-shape packing network in
hydrophobic regions, which is generally believed to be en-
ergetically favorable to crystal stability.22–24 By compari-
son, the interplanar angle of the more soluble para-species
was 59.88 and that of the meta-species was 53.78. The aro-
matic ring of PEA was inclined relative to the aromatic
ring of the chloromandelic acid, and, thus, no T-shaped
packing was observed. These off-angle packing arrange-
ments are less favorable to the stability of crystal than that
with interplanar angle of 908. In addition, the aromatic
CH/p distance of less soluble salts were slightly shorter
than those of more soluble salts in both cases. These
observations are in agreement with the conclusion pro-
posed in literature.17 Interestingly, in the concerning
cases, it was seemly that CH/p interactions contributed
most to chiral recognition even though we did not intro-
duce extended aromatic rings favorable to CH/p interac-
tions to the resolving agents.
Summarizing the above crystal structure studies, the
primary interaction, hydrogen-bonding did not show sig-
nificant differences between less soluble salts and more
soluble salts, which should contribute to the stabilities of
salts in the similar manner. The T-shaped aromatic rings
packing observed in both less soluble salts is consistent
with improving the stability of the crystal structure. The
planar boundary surfaces in less soluble salts were also
favorable to the close packing and thus beneficial to the
stability of salts. As a result, we believe that chiral recogni-
tion was realized by the combination of two secondary
interactions and allowed efficient resolutions to be
achieved in our concerning cases.
CONCLUSIONS
The investigation on diastereomeric salts crystal struc-
tures of chlorine-substituted mandelic acid and phenyle-
thylamine demonstrates that hydrogen-bonding pattern,
planarity of boundary surface between hydrogen-bonding
layers and aromatic groups packing mode are three crucial
factors to the stabilities of salts. The observed planar
boundary surface and T-shape aromatic rings packing in
crystals can efficiently stabilize less soluble salts, which
lead to significant solubility difference between less and
more soluble salts and thus results in successful resolu-
tions. The chiral recognition in diastereomeric resolution
is ascribed to the CH/p interactions and van der Waals dif-
ference between less soluble salts and more soluble salts
when hydrogen-bonding exhibit identical features in sys-
tems under investigation.
REFERENCES
1. Jacques J, Collet A, Wilen, SH. Enantiomers, racemates and resolu-
tions. New York: Wiley; 1981.
2. Kozma D. CRC Handbook of Optical resolutions via diastereomeric
salt formation. Boca Raton: CRC Press; 2001.
3. Collins AN, Sheldrake GN, Crosby J. Chirality in industry. Chichester,
England: John Wiley & Sons; 1992.
4. Kinbara K, Sakai K, Hashimoto Y, Nohira H, Saigo K. Chiral discrimi-
nation upon crystallization of the diastereomeric salts of 1-arylethyl-
amines with mandelic or p-methoxymandelic acid: interpretation of
the resolution efficiencies on the basis of the crystal structures.
J Chem Soc Perkin Trans 2 1996;12:2615–2622.
5. Kinbara K. Design of resoving agents based on crystal engineering.
Synlett 2005;5:732–743.
6. Kobayashi Y, Saigo K. The Role of CH/p Interaction in the stabiliza-
tion of less-soluble diastereomeric salt crystals. Chem Rec 2007;7:47–
56.
7. Kobayashi Y, Saigo K. Periodic ab initio approach for the cooperative
effect of CH/pInteraction in crystals: relative energy of CH/p and
hydrogen-bonding interactions.
15060.
J Am Chem Soc 2005;127:15054–
8. Kobayashi Y, Hiroaki H, Maeda J, Saigo K. Factors determining the
pattern of a hydrogen-bonding network in the diastereomeric salts of
1-arylethylamines with enantiopure P-chiral acids. Chirality 2008;
20:577–584.
As for the failure of resolution of o-ClMA with PEA, the
underlying reason escapes us due to no the X-ray quality
crystals of the diastereomeric salts of o-ClMA with PEA. In
the case of the ortho-ClMA, the chloride atom is too close
to the carboxylate group to allow deprotonation or the
intra-molecular interactions become more important than
the inter-molecular interactions may be the possible rea-
sons. Having observed the planar boundary surface in
both less soluble salts (R)-p-ClMAÁ(R)-PEA and (R)-m-
ClMAÁ(R)-PEA, with the chlorine atom pointing toward the
boundary surface, it would make its molecular length
roughly the same as that of the neighboring PEA mole-
cules and could lead to an even surface. The o-position
9. Kobayashi Y, Morisawa F, Saigo K. Enantiopure O-substituted phenyl-
phosphonothioic acids: chiral recognition ability during salt crystalliza-
tion and chiral recognition mechanism. J Org Chem 2006;71:606–615.
10. Kobayashi Y, Maeda J, Saigo K. Synthesis and chiral recognition abil-
ity of o-ethyl (2-naphthyl) phosphonothioic acid. Tetrahedron Asym-
metry 2006;17:1617–1621.
11. Suezawa H, Ishihara S, Umezawa Y, Tsuboyama S, Nishio M. The aro-
matic CH/p hydrogen bond as an important factor in determining the
relative stability of diastereomeric salts relevant to enantiomeric reso-
lution-a crystallographic database study. Eur
23:4816–4822.
J Org Chem 2004;
12. Yoshioka R, Hiramatsu H, Okamura K, Tsujioka I, Yamada S. Crystal
structure-solubility relationships in optical resolution by diastereo-
meric salt formation of DL-phenylglycine with (1S)-(1)-camphor-10-
sulfonic acid. J Chem Soc Perkin Trans 2 2000;10:2121–2128.
Chirality DOI 10.1002/chir

716
HE ET AL.
13. Bialonska A, Ciunik Z. Hydrophobic ‘lock and key’ recognition of N-4-
nitrobenzoylamino acid by strychnine. Acta Cryst 2006;B62:1061–
1070.
20. He Q, Jennings M, Rohani S, Zhu J, Gomaa H. (R)-Phenylethylammo-
nium (S)-4-chloromandelate. Acta Cryst 2008;E64(3):o559.
21. He Q, Jennings M, Rohani S, Zhu J, Gomaa H. (R)-Phenylethylammo-
nium (R)-4-chloromandelate. Acta Cryst 2007;E63(11):o4199
14. Langkilde A, Oddershede J, larsen S. Salts of lactic acid and 1-phenyl-
ethylamine, their structures and relative stabilities. Acta Cryst
2002;B58:1044–1050.
22. Hunter CA, Sanders JKM. The nature of p–p interactions. J Am Chem
Soc 1990;112:5525–5534.
15. Adams AD, Jones AB, Berger JP. Preparation of 2-aryloxy-2-arylalka-
noic acids for diabetes and lipid disorders. WO2002064094, 2002.
23. Nishio M, Hirota M, Umezawa Y. CH p interaction: evidence, nature,
and consequences. Weinheim, Berlin: Wiley-VCH; 1998.
16. Deng JG, Wu TF, Cui X, Chen YC, Jiang YZ. Process for preparing op-
tical-purity Fluoxetine.CN1294120A, 2001.
24. Nishio M. CH/p hydrogen bonds in crystals. Cryst Eng Comm
2004;6:130–158.
17. ‘‘Collect’’ data collection software, Nonius BV, 1999.
APPENDIX
18. Otwinowski Z, Minor W. Processing of X-ray diffraction data collected
in oscillation mode, methods in enzymology.Macromolecular Crystal-
lography Part A276:307–326. Carter CW Jr, Sweet RM, editors, New
York: Academic Press; 1997.
The crystal data of meta-CIMA and PEA is available for
reference at CCDC 716272 and 716273. The data of para-
CIMA and PEA is available for reference at CCDC 667265
and 680635.
19. Sheldrick GM. A short history of SHELX. Acta Cryst 2008;A64:112–
122.
Chirality DOI 10.1002/chir