Density and stability differences between enantiopure and racemic salts: Construction and structural analysis of a systematic series of crystalline salt forms of methylephedrine

Article abstract of DOI:10.1021/cg200030s

A data set of systematically related solid-state structures of pharmaceutical relevance has been created and used to investigate structural impact on physical properties in 20 pairs of enantiopure and racemic methylephedrinium salts. The structures are described and compared through graph-set analysis and the crystal packing similarity features of Mercury CSD 2.3. The commonest graph-set motif, C²₂ (9), was found to be present in 22 of the 37 independent structures and was flexible enough to include both carboxylate and sulfonate functionalities. An equivalent C ¹₂ (7) motif was present in all six halide structures investigated. Analysis of molecular structure found three common methylephedrinium cation conformations, while analysis of cation packing found six isostructural groups, each containing at least two salt structures and based on one of three common cation packing motifs. Melting points and crystallographically obtained densities were examined in detail for the 13 enantiopure and racemic structural pairs found to be chemically identical to each other. While average densities conform to Wallach's rule, 6 of the 13 individual pairings do not. This does not support the structural justification normally given for Wallach's rule. One of the three observed common cation packing motifs is highly associated with failure of Wallach's rule, as are significant differences in hydrogen bonding between the enantiopure and racemic structures.

Full text of DOI:10.1021/cg200030s

ARTICLE
pubs.acs.org/crystal
Density and Stability Differences Between Enantiopure and Racemic
Salts: Construction and Structural Analysis of a Systematic Series of
Crystalline Salt Forms of Methylephedrine
Alan R. Kennedy,*,† Catriona A. Morrison,*,† Naomi E. B. Briggs,† and William Arbuckle‡
†WestCHEM, Department of Pure and Applied Chemistry, University of Strathclyde, Glasgow G1 1XL, Scotland
‡MSD, Newhouse, Motherwell, Lanarkshire ML1 5SH, Scotland
S Supporting Information
b
ABSTRACT: A data set of systematically related solid-state structures of pharmaceu-
tical relevance has been created and used to investigate structural impact on physical
properties in 20 pairs of enantiopure and racemic methylephedrinium salts. The
structures are described and compared through graph-set analysis and the crystal
packing similarity features of Mercury CSD 2.3. The commonest graph-set motif, C22
(9), was found to be present in 22 of the 37 independent structures and was flexible
enough to include both carboxylate and sulfonate functionalities. An equivalent C12 (7)
motif was present in all six halide structures investigated. Analysis of molecular
structure found three common methylephedrinium cation conformations, while analysis of cation packing found six isostructural
groups, each containing at least two salt structures and based on one of three common cation packing motifs. Melting points and
crystallographically obtained densities were examined in detail for the 13 enantiopure and racemic structural pairs found to be
chemically identical to each other. While average densities conform to Wallach’s rule, 6 of the 13 individual pairings do not. This
does not support the structural justification normally given for Wallach’s rule. One of the three observed common cation packing
motifs is highly associated with failure of Wallach’s rule, as are significant differences in hydrogen bonding between the enantiopure
and racemic structures.
’ INTRODUCTION
structural rationalization behind Wallach’s rule is that array
structure is different for enantiopure and racemic versions of
compounds due to variation in their packing ability. Racemic
compounds can crystallize in any of the 230 space groups,
whereas enantiopure compounds are limited to crystallizing in
only 65 chiral space groups.9 In theory, as enantiopure crystals
are only allowed to use proper symmetry operators (i.e., rotations
and translations), they are unable to be as tightly packed as their
equivalent racemic crystals (that can also use improper symmetry
operators) and this results in a less dense compound. A higher
level discussion of the structural consequences of symmetry type
is given by Brock and Dunitz.10
Experimental evidence for Wallach’s rule has always been
somewhat patchy. Jacques et al.4 compiled early work, including
that of Wallach himself, and showed that the racemic form was
more dense than the enantiopure in only half of the cases cited. A
more complete (129 pairs of structures) and modern compila-
tion of literature data by Brock et al.9 did find evidence that
racemic crystals are systematically slightly more dense (and have
greater thermodynamic stability) than enantiopure forms. How-
ever, they also point out that their data set is inherently biased
due to the differing chiral purities of the compounds used to
Pharmaceutical salt formation is one of the easiest and most
cost-effective ways to optimize and manipulate the material
properties of a new drug1 without altering the desired active
pharmaceutical ingredients (API).2 Salts of molecules with chiral
centers are highly relevant as more than half of all patented drugs
are chiral;3 however, these are often marketed as racemates as this
is the form initially produced during preparation in achiral
environments.4 Much attention has of course been paid to the
characterization of enantiopure drugs in comparison to the
racemic drug, due to the well-known different pharmacological
effects of the two drug forms. Perhaps less well-known to the
general chemistry community is that chirality affects the bulk
physical properties (for instance, solubility, dissolution rate, and
melting point) as well as the biological mechanisms of drug
molecules.5
Kitaigorodskii’s “principle of close packing” states that close
packing ultimately equates to thermodynamic stability and that
therefore the densest material will be the most thermodynami-
cally stable.6 Together with optimizing hydrogen-bonding and
other directional intermolecular interactions, the need for pack-
ing efficiency is often taken as the driving force behind crystal
formation. Wallach’s rule states that racemic crystals are denser
than their chiral counterparts.7 Taking Kitaigoroskii and Wallach
together gives the notion that “racemic compounds have lower
enthalpies than (their equivalent) pure enantiomers”.8 The
Received: January 10, 2011
Revised:
February 8, 2011
Published: March 09, 2011
r
2011 American Chemical Society
1821
dx.doi.org/10.1021/cg200030s Cryst. Growth Des. 2011, 11, 1821–1834
|

Crystal Growth & Design
ARTICLE
Figure 1. Enantiopure and racemic methylephedrine.
obtain the original samples. In all cases, close examination of the
individual racemic and enantiopure pairs shows Wallach’s rule to
be false on many occasions.
With respect to pharmaceutically relevant salts, some notable
experimental and theoretical studies have been carried out on
enantiopure and racemic salt pairs, looking at differences in
packing arrangements and lattice energies.11 Hampering further
understanding is the lack of relatively large data sets of system-
atically related salt structures. This is true of general salt forms of
pharmaceutical or pharmaceutical-like molecules (a notable, and
here highly relevant, exception to this is the work by Davy
et al.)12 and is even more apparent where pairs of enantiopure
and racemic structures are required. This paper contributes data
to the debate by reporting structures for 20 enantiopure and
racemic salt pairs of methylephedrine, Figure 1, and by analyzing
the observed differences in their crystal structures and packing.
Methylephedrine is a member of the phenylethylamine struc-
tural family, of which there are many members that have
pharmaceutical significance, for example, the anti-asthma drug
salbutamol, the decongestant ephedrine, and the stimulant
methamphetamine. There are few known literature structural
studies of methylephedrine. The structure of the enantiopure
free base is deposited in the Cambridge Crystallographic Data
Centre (CCDC), reference UCAWOL.13 A structure for racemic
methylephedrine has not been achieved, and indeed in our hands
recrystallization of the racemic compound gave a conglomerate.
The only other related work present in the CCDC is the 1934
study by Gossner, which investigated the halide salts.14 Because
of the early year of these data collections, the work does not
report any atomic coordinates and therefore the structures are
redetermined herein.
Twenty pairs of salt structures were obtained from enantio-
pure, (1R,2S)(-), and racemic, ((), methylephedrine with 20
pharmaceutically relevant counterions. The molecular structures
of the 20 counterions are shown in Figure 2. This choice allows
for direct comparison between the enantiopure and racemic pairs
as all salts were synthesized by the same technique and crystals
were grown under broadly similar conditions. Analysis of the
hydrogen bonding and the crystal packing will be presented to
determine similarities between structures. Their crystallographi-
cally measured densities will be examined to allow Wallach’s rule
to be tested and finally the experimental melting points will be
inspected.
Figure 2. Molecular structures of the 20 acids used in salt formation.
solution was filtered and left to produce crystals by slow evaporation and
cooling. Where unsatisfactory single crystals were recovered, these were
redissolved and the resultant solution was filtered into a tube of
approximate diameter 5 mm, to allow for slower evaporation and crystal
growth.
Crystallographic Data Collection and Processing. Samples
for single crystal diffraction studies were obtained as above. Measure-
ments were recorded at low temperature by Bruker-Nonius CCD and
Oxford Diffraction diffractometers with Mo KR radiation (λ = 0.7107 Å)
or Cu KR radiation (λ = 1.5418 Å). Refinement of atomic coordinates
and thermal parameters was to convergence and by full-matrix least-
squares methods on F2 within SHELX-97.15 Where heavy atoms were
present (S, Cl, Br, I) refinement of Flack parameters determined
absolute structure. Selected crystallographic data and refinement para-
meters are given in Tables 1, 2 and 3 and full details are deposited as cif
files. These can be obtained free of charge from The Cambridge
Crystallographic Data Centre via www.ccdc.cam.ac.uk/data_request/
cif by quoting deposition numbers CCDC 806488-806524.
Melting Point Determination. Melting points were collected in
triplicate using a Buchi B-545 automatic melting point apparatus.
’ EXPERIMENTAL SECTION
’ RESULTS AND DISCUSSION
For this work, salt synthesis was attempted with both the
enantiopure and racemic forms of the base methylephedrine and
38 acids. In 20 cases crystals of both forms were achieved,
resulting in the possibility of directly comparing the structural
effect of using enantiopure versus racemic bases. Of the 20 pairs
of structures, 13 are enantiopure/racemic pairs with chemically
identical formula. The remaining seven pairs consist of four
which are chemically different (mostly due to differing hydration
states, but for salts P1 and P2 through formation of tri-iodide
Materials. All experiments were carried out using commercially
available materials, apart from (()-methylephedrine, which was formed
from a 50:50 mixture of the commercially available (1R,2S)(-) and
(1S,2R)(þ) methylephedrine bases.
Salt Synthesis and Crystallization. All salts were prepared by
the same method. This involved the reaction of a partially dissolved
aqueous solution of the free base with a 10% excess equivalent of the
appropriate acid in aqueous solution to form a solution of the salt. The
solution was stirred for 30 min with gentle heating to 50 °C. Finally, the
1822
dx.doi.org/10.1021/cg200030s |Cryst. Growth Des. 2011, 11, 1821–1834

Crystal Growth & Design
ARTICLE
1823
dx.doi.org/10.1021/cg200030s |Cryst. Growth Des. 2011, 11, 1821–1834

Crystal Growth & Design
ARTICLE
1824
dx.doi.org/10.1021/cg200030s |Cryst. Growth Des. 2011, 11, 1821–1834

Crystal Growth & Design
ARTICLE
1825
dx.doi.org/10.1021/cg200030s |Cryst. Growth Des. 2011, 11, 1821–1834

Crystal Growth & Design
ARTICLE
1826
dx.doi.org/10.1021/cg200030s |Cryst. Growth Des. 2011, 11, 1821–1834

Crystal Growth & Design
ARTICLE
Table 4. Graph-Set Analysis for Chemically Identical Salt Pairs of Benzoate Derivatives (a-anion, c-cation, w-water)a
compound
hydrogen bond
graph-set
network growth
compound
hydrogen bond
graph-set
network growth
A1
a
OHc COO-
>a < b
C22(9)
a
A2
a
OHc COO-
>a < b
C22(9)
a
3 3 3
3 3 3
b
a
NH COO-
b
a
NH COO-
3 3 3
3 3 3
B1
OHc COO-
>b < c
>a < c
>a < d
>b < d
>a < b
>c < d
>a < b
D21(2)
D22(5)
D22(8)
D22(8)
C22(9)
C22(9)
C22(9)
a
B2
OHc COO-
>b < c
>a < c
>a < d
>b < d
>a < b
>c < d
>a < b
D22(2)
D22(5)
D22(8)
D22(8)
C22(9)
C22(9)
C22(9)
b
3 3 3
3 3 3
b
c
NH COO-
b
c
NH COO-
3 3 3
3 3 3
OHw COO-
OHw COO-
3 3 3
3 3 3
d
OHw NO2
d
OHw NO2
3 3 3
3 3 3
C1
D1
a
OHc COO-
a
a
C2
D2
a
OHc COO-
b
a
3 3 3
3 3 3
b
a
NH COO-
b
a
NH COO-
3 3 3
3 3 3
OHc COO-
>c < d
>a < b
C22(9)
R22(9)
OHc COO-
>a < b
C22(9)
3 3 3
3 3 3
b
c
NH COO-
b
NH COO-
3 3 3
3 3 3
OHc COO-
3 3 3
d
a
NH COO-
3 3 3
E1
OHc COO-
>a < b
C22(9)
b
E2
a
OHc COO-
>a < b
C22(9)
b
3 3 3
3 3 3
b
NH COO-
b
NH COO-
3 3 3
3 3 3
a The entry under “network growth” gives the crystallographic direction in which the hydrogen-bonding network propagates. Graph-set analysis for the
other salts is presented in the Supplementary Information.
To aid further discussion, the structures will be separated into
groups depending on their counterion type and initially dis-
cussed in terms of structure similarities through graph set analysis
and network growth. The structures will also be analyzed to
determine different cation conformations and crystal packing
similarities throughout the data set. Finally, the melting points
and densities will be examined to see if the pairs of chemically
identical salts conform to Wallach’s rules.
Substituted Benzoate Salts. There are five salt pairs with
benzoic acid derived counterions, which crystallize to produce
chemically identical species for the (1R,2S)(-)-methylephe-
drinium and (()-methylephedrinium salts. The crystallo-
Figure 3. Schematic of the common C22(9) graph set motif. Y = C or SO.
graphic details are shown in Table 1, compounds A1 to E2.
The graph set analysis for these five pairs is shown in Table 4. A
common feature is that all contain the C22(9) motif, see Figure 3,
involving the cation’s OH and NH groups as donors and both
O-atoms of the anion COO- group as the acceptors. All 10
structures grow using the C22(9) chain to give one-dimensional
(1D) hydrogen-bonded networks, either along the crystallo-
graphic a or crystallographic b directions; see Figure 4. Com-
parison of the hydrogen-bonding details shows that the
enantiopure-racemic pairs all have identical graph-set analyses
for each salt pair, except that of 4-nitrobenzoate (D1 and D2).
This latter difference is related to the fact that there are two
anions and two cations present per asymmetric unit of D1. The
carboxylate group of one anion forms hydrogen bonds with the
NH and OH of two different cations, while the crystallogra-
phically independent anion interacts with NH and OH func-
tionalities of a single methylephedrinium. Thus, the structure
contains both ring, R22(9), and chain, C22(9), motifs, see Figure 5,
and the C22(9) chain is present in all 10 of these benzoate salt
structures. Kinbara et al. reported that they found significant
differences in the hydrogen-bonding motifs adopted between
enantiopure and racemic primary amines and carboxylate
anions.11b We do not observe the same specific hydrogen-bonding
motif, nor significantly do we see any systematic difference
between the enantiopure and chiral salts with respect to hydro-
gen-bonding motifs.
rather than iodide counterions) and three cases where the
racemic salt spontaneously resolved to form a conglomerate.
Three spontaneous resolutions from 20 are slightly higher than
the often quoted 5-10% occurrence of this phenomenon but fits
with the suggestion that salt forms are more prone to sponta-
neously resolve than neutral molecules.4,18 Also of note here is
that in our hands the free base itself spontaneously resolves to
form a conglomerate.
There are several features general to all 37 independent
crystal structures. All form hydrogen-bonded cation-anion
pairs, but there are no direct cation-cation hydrogen-bonded
pairs. Anion-anion hydrogen-bonding interactions are pre-
sent in all structures with an organic anion that possesses a
classic hydrogen-bond donor. The salt structures containing
hydrogen-maleate and hydrogen-malonate anions conform to
Etter’s rules,16 which predict that intramolecular bonds will
form in preference to intermolecular bonds, as all use their
COOH groups as internal hydrogen-bond donors. Of the 37
structures, six were isolated as monohydrates and one as a
hemihydrate. In all structures except the two ethane-1,2-
disulfonate salts G1 and G2, the acid and base react to give
a one to one salt. In the ethane-1,2-disulfonate structures, the
acid is dideprotonated and thus two to one cation to anion
salts are produced. None of the reported products have
neutral free acid molecules present, and so all are true salts
and not cocrystals.
1827
dx.doi.org/10.1021/cg200030s |Cryst. Growth Des. 2011, 11, 1821–1834

Crystal Growth & Design
ARTICLE
Figure 4. Growth of a 1D hydrogen-bonded C22(9) chain in the a direction for (1R,2S)(-)-methylephedrinium 2-chlorobenzoate.
Figure 5. Illustration of the C22(9) and R22(9) graph-sets found for
(1R,2S)(-)-methylephedrinium 4-nitrobenzoate.
Seven further benzoate-based salt structures are available here
for comparison. In the three conglomerate forming cases R, S,
and T where only the enantiopure structure was accessible, the
common C22(9) chain was observed as above. Note that only four
para substituted benzoates were used here and that it was three of
these four that formed conglomerates. Only the nitro derivative
formed a racemic phase, while the chloro, hydroxyl, and methyl
derivatives showed spontaneous resolution. The implication is
that similarly shaped counterions all behaved in like manner with
respect to spontaneous resolution.
In a further four cases, enantiopure and racemic benzoate
structures are available, but these salts did not form chemically
identical pairs. The enantiopure species are monohydrates, while
the racemic compounds are anhydrous (see Table 2 compounds
N1 to O2) and as such identical hydrogen-bonding motifs are not
seen for the enantiopure-racemic pairs. However, there are
similarities to the previously discussed structures. The structures
of (()-methylephedrinium 3-chlorobenzoate, (1R,2S)(-)-
methylephedrinium benzoate monohydrate and (()-methyle-
phedrinium benzoate all grow to form 1D ribbons via the C22(9)
chain as seen before. However, (1R,2S)(-)-methylephedrinium
3-chlorobenzoate monohydrate (N1) forms a two-dimensional
(2D) sheet with network growth in a second direction through
solvent separated anions along the crystallographic a direction
forming a C22(6) chain; see Figure 6. This structure is thus
different from all other benzoate salts studied herein, in that it
Figure 6. Hydrogen-bonds propagate in the a and b directions, forming
a 2D sheet in (1R,2S)(-)-methylephedrinium 3-chlorobenzoate.
(F1 and F2) have the same hydrogen-bonding present for both
structures of the pair. This is in direct contrast to what was found
for the benzoate salts. One similarity throughout 9 of the 10
structures (the four chemically identical paired structures and
(1R,2S)(-)-methylephedrinium 4-hydroxybenzenesulfonate,
Q1) is that only two of the three sulfonate oxygen atoms (or
three of the four sulfate oxygen atoms) are utilized as hydrogen-
bond acceptors. This allows for a C22(9) graph-set to be present as
the means of network propagation for the salts of benzenesulfo-
nate, 1,2-ethanedisulfonate, and hydrogen sulfate monohydrate.
Despite the change from COO to RSO3 functionality, this motif
is analogous to the C22(9) chain seen with the benzoate salts.
However, the methanesulfonate salts do not have the C22(9)
chain present and instead display discrete D22(8) motifs. Only for
the structure of (()-methylephedrinium 4-hydroxybenzenesul-
fonate hemihydrate, Q2, are all crystallographically independent
sulfonate oxygen atoms involved in hydrogen bonding.
The crystal structure for compound (1R,2S)(-)-methylephe-
drinium 1,2-ethanedisulfonate, G1, has two base molecules per
asymmetric unit compared to one per asymmetric unit for the
equivalent (()-methylephedrinium salt, G2. The structure of G2
forms hydrogen bonds with two cations and two anions in a ring
formation, R44(24); see Figure 7. In the G1 structure, instead of
the hydrogen-bonding forming a ring, the ethanedisulfonate ion
lies along the crystallographic a direction and forms a second
C22(9) chain making the overall network a two-dimensional (2D)
sheet. This formation of a ring is also seen in the structure of (()-
methylephedrinium hydrogen-sulfate monohydrate where the
does not form direct N-H OOC hydrogen-bonds. Indeed,
3 3 3
N1 is even more of an outlier, in that all of the other salts in this
paper make direct hydrogen bonds between NH and the formally
charged group of the anion.
Sulfonate and Hydrogen-Sulfate Salts. Of the five enantio-
-
pure-racemic pairs of YSO3 (Y = R or OH) salt structures
obtained, four are of pairs of salts that are chemically identical.
However, the graph-set analysis detailed in the Supporting
Information shows that only the salts of benzenesulfonate
1828
dx.doi.org/10.1021/cg200030s |Cryst. Growth Des. 2011, 11, 1821–1834

Crystal Growth & Design
ARTICLE
Figure 7. Ring formation in (()-methylephedrinium 1,2-ethanedisulfoante, R44(24), and (()-methylephedrinium hydrogen-sulfate monohydrate,
R44(12).
and racemic salt pairs are chemically identical. The “iodide” salts
differ as the (1R,2S)(-)-methylephedrinium salt is a mixed
iodide, tri-iodide structure, while the racemic salt was isolated
as the simple iodide. The bromide and chloride pairs form
identical hydrogen-bonding patterns and all six propagate via a
C12(7) chain. All, excluding the iodide-tri-iodide structure, form
1D ribbons. In the (1R,2S)(-)-methylephedrinium iodide tri-
iodide structure, P1, network propagation is in the a and c
directions via C12(7) chains, giving the same hydrogen-bond
motif as present in the other halides. The tri-iodide anions form
channels but make no hydrogen-bonding interactions, as illu-
strated in Figure 9.
Different Conformations of the Cation. Within the salt
structures the cation adopts one of three different conformations.
These three different conformations are illustrated in Figure 10.
Four torsion angles can be used to describe these conformations,
and these encompass the two stereocenters at carbon atoms C7
and C8. Between each of the three classes, significant variation is
based largely on torsion angle 4. Overlay of the cations that
represent the three different conformations was performed using
Figure 8. Propagation of of hydrogen-bonding in (1R,2S)(-)-methy-
lephedrinium malonate along the crystallographic a direction.
anions and water molecules hydrogen bond to form two R44(12)
rings; see Figure 7.
the structural overlay feature in Mercury CSD 2.3.17 This is
illustrated in Figure 11 and further details are tabulated in the
Supporting Information. Of those structures where more than one
crystallographically independent cation exists in the asymmetric
unit, only G1 contains two different conformers. Conformation
(a) isthe most common, beingpresent in 23 cations, with a torsion
angle 4 range of -26.27 (in D1) to -63.75° (in L2). Conforma-
tion (b) occurs in 16 of the cations and has a range of 67.19 (in F2)
to 91.29° (in J2). Conformation (c) is the least reoccurring, being
present in only seven cations and has a torsion angle 4 range of
175.65 (in T1) to -170.62° (in G1). Interestingly, only arrays of
enantiopure cations exist in this third conformation, that with the
amine proton anti to the phenylethylene chain. Related work on
ephedrine salts found two cation conformations.11d,12a,18
Dicarboxylic Acid Derivatives. We describe two enantio-
pure-racemic pairs of dicarboxylic acid salt structures and both
are chemically identical, namely, the hydrogen-maleate and
hydrogen-malonate salts. The two hydrogen-maleate salts (J1
and J2) have identical hydrogen-bonding present, with the
hydrogen-malonates’ hydrogen-bonding differing due to the
(1R,2S)(-)-methylephedrinium malonate salt having two ion
pairs per asymmetric unit, which results in the growth of two
parallel units along the crystallographic a direction; see Figure 8.
All four salts conform to Etter’s rules,16 with the presence of
internal hydrogen bonds giving the graph-sets S11 (7) and S11(6).
The common C22(9) chain seen for both the benzoate derivatives
and the sulfonate salts is absent here. This is because one of the
oxygen atoms of the carboxylate is involved in intramolecular
hydrogen bonding. In the case of the hydrogen-maleate salts,
C22(12) chains propagate through using the oxygen atoms at
either end of the maleate anion. The hydrogen-malonate pairs
have different hydrogen bonding from each other, with the
enantiopure salt showing a discrete graph-set and the racemic
salt propagating through C22(7) chains with both the NH and OH
of the cation hydrogen bonding to the same oxygen of the anion.
Halide Salts. Three pairs of halide salt structures were
obtained. For the chloride and bromide salts, the enantiopure
Crystal Packing Similarities. Investigation into similarities in
the crystal packing of the 37 structures was performed by using
the “crystal packing similarity” module in Mercury CSD 2.3.17
This tool takes a reference molecule and then examines the three-
dimensional (3D) geometry of the cluster of surrounding
molecules; this then allows the program to determine the
number of similar molecules within clusters from several struc-
tures, and their geometric similarity. An excellent description of
this technique is given in ref 19. Analysis was performed for the
group of 37 structures and the results are summarized in the tree
diagram, Figure 12. This was constructed by looking for similar
1829
dx.doi.org/10.1021/cg200030s |Cryst. Growth Des. 2011, 11, 1821–1834

Crystal Growth & Design
ARTICLE
Figure 9. Network growth along the c direction for (1R,2S)(-)-methylephedrinium iodide tri-iodide. View is down the a axis. The tri-iodide anions are
shown as purple bars.
Figure 10. Three conformations were found for the methylephedrinium cation. Conformation (a) is illustrated by compound A1, (b) by H1, and (c) by
N1.
In the tree diagram, less similar structures are named at the top of
the diagram with the similarity increasing on moving downward.
The tree branches convey similar substructures. The colors of the
large circles highlight different cation molecular conformations, as
discussed above, to show reoccurrences throughout the tree. The
small colored dots indicate distinct cation to cation pair geometries,
or packing motifs. As can be seen from the tree diagram all but four
structures (highlighted with small green dots) have a common
packing motif for two cations. The three different packing motifs
are illustrated in Figure 13, as motif X (21 conformer, yellow dots),
motif Y (translation conformer, blue dots), and motif Z (racemic
conformer, red dots). Motif Y, the “translation conformer” is by far
the most common, occurring in 28 of the structures, with the
“racemic conformer” Z occurring in 3 structures, and the “21
conformer” X occurring in only 2 isostructural structures. The
remaining four structures, K1, M2, O1 and O2, are outliers in the
Figure 11. (a-c) Overlays of the three different cation conformations.
data set anddo not fit into any of the three different groups and also
packing in various cluster sizes ranging from 2 to 15 cations. Note
that the analysis looked only at the largest molecular component
of the salts, that is, the cation, in order to exclude interference
from the various anions and water molecules, with a distance and
torsion angle tolerance of 20%.19 Structures that have 15 out of
15 similar molecules within a structure are considered to be
isostructural, at least with respect to cation packing.
do not relate to each other. All of the three different cation
molecular conformations observed are able to adopt the “transla-
tion conformer,” with the other two packing motifs being adopted
only by cations with molecular conformation (a).
The tree diagram shows that there are several structures that
are isostructural with respect to cation packing. These structures
are located at the bottom of the tree with 15 matching neighbors
1830
dx.doi.org/10.1021/cg200030s |Cryst. Growth Des. 2011, 11, 1821–1834

Crystal Growth & Design
ARTICLE
Figure 12. Tree diagram generated from crystal packing similarity search on the cation fragment.
Figure 13. Pair-packing conformations observed for cations.
in the similarity packing search. Four of the six isostructural
groups (those of compounds L1, M1, C2 N2, E1 S1 T1, and A2
E2) have both identical orientations of the cations and identical
hydrogen-bonding present within the structures. However, with-
in the isostructural pair B1 and C1 the anion orientation is
different; see Figure 14. This allows the 2-nitrobenzoate struc-
ture B1 to also accommodate a water molecule. The additional
water molecule results in additional hydrogen bonds being
present, as well as the C22(9) chain. These additional interactions
with the water molecule do not result in an additional dimension
of network propagation; instead, an anion-water-anion motif
gives an additional C22(9) chain parallel to the first. Finally, for the
isostructural pair of H2 and I2 (MeSO3 and SO4H anions) the
addition of a water molecule and extra hydrogen-bonding cap-
ability in I2 gives a significant change in hydrogen bonding. This
results in a 3D network for the hydrogen-sulfate monohydrate
structure, compared to the more common 1D motif of the
methanesulfonate structure. The analysis of structures B1, C1,
H2, and I2 suggests that here hydrogen bonding is a consequence
of the cation crystal packing array and not the other way around.
This conclusion was reached as the cation packing is isostructural
and does not change despite the addition of water molecules and
major alterations to the hydrogen-bonding motifs.
Density Comparisons of Identical Salt Pairs. Comparison of
the densities and melting points of the enantiopure and racemic salt
pairs has been attempted only for the 13 pairs that form chemically
identical salts, Figure 15. The average density of the enantiopure salts
is 1.300 g/cm3, whereas the average density for the racemic salts is
approximately 1.5% greater at 1.319 g/cm3. This is in line with
previous studies.9 However, looking individually at the 13 salt pairs,
only 7 conform to Wallach’s rules,7 where the racemic compound is
more dense than the enantiopure compound. Of the remainder, there
are four pairs where the density is effectively identical (differences
of less than 0.005 g/cm3) and two salt pairs (2-nitrobenzoate and
1831
dx.doi.org/10.1021/cg200030s |Cryst. Growth Des. 2011, 11, 1821–1834

Crystal Growth & Design
ARTICLE
Figure 14. Comparison of packing for isostructural pairs, shown both with (right-hand side) and without (left-hand side) the anion present. Green and
red indicate cations from different structures.
1,2-ethanedisulfate, samples B and G) where the enantiopure
compound is considerably denser than the racemic compound.
Interestingly, H2, here the only racemic salt to crystallize in a chiral
space group, does obey Wallach’s rule. Note that in all cases here,
slow crystallization of the racemic compound gave only racemic
crystalline phases — implying that the racemic phase is thermo-
dynamically more stable than the enantiopure phase. To investi-
gate potential structural reasons for the exceptions to Wallach’s
rule, compounds B and G were examined for differences between
their structures and those of the other salts. Of the 13 chemically
identical pairs, G is unique in being the only one with a significant
difference in hydrogen bonding between its enantiopure and
racemic forms; see graph-set analysis in the Supporting Informa-
tion. Note that the hydrogen bonding in enantiopure G1 propa-
gates in two dimensions rather than the 1D chain seen for G2. We
suggest that this difference in hydrogen bonding lies at the root of
the observed difference in packing efficiency. This ties in with
previous suggestions that significant differences in hydrogen
bonding may lie at the root of observed differences in behavior
of racemic and enantiopure salts.11b However, B1 and B2 have
identical hydrogen-bonding motifs (the common C22(9) graph
set) so differences in hydrogen bonding is not the explanation
here. B is somewhat unusual in being one of only two hydrates in
the group of 13 pairs, but potentially of more interest is that B2 is
one of the three structures that adopt the “Z” racemic cation pair
packing motif. The other two “Z” motifs are for A2 and E2.
Interestingly, both of these structures are in the group with
essentially identical densities for the enantiopure and racemic
structures. This suggests that the racemic packing motif Z is less
packing efficient than the homochiral motifs X and Y. Packing
efficiencies in both compounds B and G are illustrated by the void
calculations shown in Supporting Information. A final point is that
the three conglomerate forming compounds, R1, S1, and T1, do
not appear to have enantiopure phases with significantly higher
densities than those of comparable species. Indeed, the density of
R1 (1.217 g/cm3) is the lowest recorded here. Thus, the “extra”
stability of these salts is not reflected in their densities.
Melting Point Comparisons of Identical Salt Pairs. Melting
point is a simple parameter of interest both as a crucial material
characteristic for working and tableting pharmaceutical
compounds1 and as a feature correlated to other vital parameters
such as lattice energy and solubility.11d,12,20 Experimental melt-
ing points could only be achieved for 12 of the 13 identical salt
pairs, as both the salts of methanesulfonate were hygroscopic and
therefore a reliable value could not be measured. The average
melting point for the 12 enantiopure salts is 127.0 °C, with the
average racemic melting point being greater with a value of
133.0 °C. This is consistent with the racemic phases being
thermodynamically favored over the enantiopure phases. Indivi-
dual pair-evaluation indicates that the melting point is higher for
the racemic compound in 8 of 12 cases. However, the exceptional
samples are not all the same as those identified by the density
measurement comparison, as illustrated in Figure 15. The
compounds B and G were highlighted above as having signifi-
cantly higher densities for the enantiopure compound. Of these,
only the 1,2-ethanedisulfonate salts also have a higher melting
point for the enantiopure salt, G1, over its racemic equivalent,
G2. G1 and G2 were the compounds where a change in
hydrogen-bonding motif (and network dimensionality) was
noted. However, for B the racemic compound has a much higher
melting point than the enantiopure form. Care needs to be taken
here as the B salts are hydrates, and the melting points recorded
will presumably be those of anhydrous forms of unknown
structure. The 3-fluorobenzoate salts C1 and C2 are the other
1832
dx.doi.org/10.1021/cg200030s |Cryst. Growth Des. 2011, 11, 1821–1834

Crystal Growth & Design
ARTICLE
Figure 15. Density and melting points of 13 identical salt pairs. Top diagram shows absolute values; bottom diagram shows differences.
pair to have a much higher melting point for the enantiopure
form than the chiral form. The structural analysis sheds little light
on this, but it should be noted that C1 and C2 have essentially
identical densities and thus do not strictly obey Wallach’s rule.
Thus, it can be said that both pairs that have high melting point
enantiopure forms are also observed to disobey Wallach’s rule.
give conglomerates, and four pairs produced salts that crystal-
lized as different chemical entities. All three of the conglomerates
had anions derived from para-substituted benzoic acids but did
not appear to be structurally different from the other salts or to
have exceptionally high densities.
All the structures analyzed formed hydrogen-bonded cation-
anion pairs, with no cation-cation pairs present. In all cases
except N1, the cation-anion contact involved the cation’s NH
groups as a hydrogen-bond donor to the anion. The commonest
graph-set motif of C22(9) was found to be present in 22 of the 37
independent structures and was flexible enough to include both
OCO and OSO functionalities. The equivalent C12 (7) motif was
present in all six halide structures. Salts of dicarboxylic acids did
’ CONCLUSION
This paper investigates 20 pairs of enantiopure and racemic
methylephedrinium salts. Of the 20 pairs, 13 produced enantio-
pure and racemic crystal phases of identical chemical makeup,
three pairs had racemic species that spontaneously resolved to
1833
dx.doi.org/10.1021/cg200030s |Cryst. Growth Des. 2011, 11, 1821–1834

Crystal Growth & Design
ARTICLE
not form the C22(9) motif. The network growth varied with 29
salts producing 1D hydrogen-bonded chains, five salts producing
2D hydrogen-bonded sheets, and three salts producing 3D
hydrogen-bonding networks. The methylephedrinium cation is
seen to adopt one of three different conformations and there are
also three common cation pair packing motifs, with 13 salt
structures adopting 6 different 3D isostructural groupings with
respect to cation packing. Hydrogen-bonding differences (and
indeed differing hydration states) are seen within some of the
otherwise isostructural cation arrays.
(2) Paulekuhn, G. S.; Dressman, J. B.; Saal, C. J. Med. Chem. 2007,
50, 6665.
(3) (a) Borman, S. Chem. Eng. News 1990, 68, 9. (b) Millership, J. S.;
Fitzpatrick, A. Chirality 1993, 5, 573.
(4) Jacques, J.; Collet, A.; Wilen, S. H. Enantiomers, Racemates, and
Resolutions; Wiley: New York, 1981.
(5) Li, Z. J.; Grant, D. J. W. J. Pharm. Sci. 1997, 86, 1073.
(6) Kitaigorodskii, A. I. In Organic Chemical Crystallography; Con-
sultands Bureau: New York, 1961.
(7) Wallach, O. Liebigs Ann. Chem. 1895, 286, 90.
(8) Eliel, E. L. Stereochemistry of Carbon Compounds; McGraw-Hill:
New York, 1962.
For the chemically identical enantiopure-racemic pairs, melt-
ing points and densities were collated, principally as a test of the
validity of Wallach’s rules.7 In general, the overall trend in
densities agrees with Wallach’s rules with the average racemic
density being higher than the average enantiopure density. This
though may be a somewhat counterfeit result; we can access
crystals in cases where the stability of the enantiopure form is
significantly less than that of the racemic form (as enantiopure
starting material is available), but we cannot access structures of
the racemic form if the enantiopure form is significantly more
stable. As our comparison can be made only when both forms are
available, there will always be a bias toward more stable (and
hence presumably more dense) average values for the set of
racemic compounds. Close examination of the individual pairs
finds that exceptions to Wallach’s rule are common, with 6 of 13
pairs failing to have a more dense racemic structure. Overall, we
find little evidence to support the notion that packing racemic
ions with achiral ones must give a denser structure than packing
enantiopure and achiral ions. One of the three observed cation
packing motifs (the racemic motif Z) is closely associated with
failure of Wallach’s rule but does not explain all the observed
instances, for instance, that of compound G, where there is a
significant difference between the hydrogen-bonding motifs of
G1 and G2. Thus, we have highlighted two different details of
array structure that may explain inefficient packing.
(9) Brock, C. P.; Schweizer, W. B.; Dunitz, J. D. J. Am. Chem. Soc.
1991, 113, 9811.
(10) Brock, C. P.; Dunitz, J. D. Chem. Mater. 1994, 6, 1118.
(11) (a) Zingg, S. P.; Arnett, E. M.; McPhail, A. T.; Bothnerby, A. A.;
Gilkerson, W. R. J. Am. Chem. Soc. 1988, 110, 1565. (b) Kinbara, K.;
Hashimoto, Y.; Sukegawa, M.; Nohira, H.; Saigo, K. J. Am. Chem. Soc.
1996, 118, 3441. (c) D’Oria, E.; Karamertzanis, P. G.; Price, S. L. Cryst.
Growth Des. 2010, 10, 1749. (d) Li, Z. J.; Ojala, W. H.; Grant, D. J. W.
J. Pharm. Sci. 2001, 90, 1523. (e) Li, Z. J.; Zell, M. T.; Munson, E. J.;
Grant, D. J. W. J. Pharm. Sci. 1999, 88, 337.
(12) (a) Collier, E. A.; Davey, R. J.; Black, S. N.; Roberts, R. J. Acta
Crystallogr. 2006, A62, 498. (b) Black, S. N.; Collier, E. A.; Davey, R. J.;
Roberts, R. J. J. Pharm. Sci. 2007, 96, 1053.
(13) Staples, R. J.; Cho, E. Z. Kristallogr.-New Cryst. Struct. 2005,
220, 543.
(14) Gossner, N. Z. Kristallogr., Kristallgeom., Kristallphys., Kristall-
chem. 1934, 89, 417.
(15) Sheldrick, G. M. Acta Crystallogr. 2008, A64, 112.
(16) Etter, M. C. Acc. Chem. Res. 1990, 23, 120.
(17) Macrae, C. F.; Bruno, I. J.; Chisholm, J. A.; Edgington, P. R.;
McCabe, P.; Pidcock, E.; Rodriguez Monge, L.; Taylor, R.; van de Streek,
J.; Wood, P. A. J. Appl. Crystallogr. 2008, 41, 466.
(18) Duddu, S. P.; Grant, D. J. W. Pharm. Res. 1994, 11, 1549.
(19) Childs, S. L.; Wood, P. A.; Rodriguez-Hornedo, N.; Sreenivas
Reddy, L.; Hardcastle, K. I. Cryst. Growth Des. 2009, 9, 1869.
(20) (a) Guerrieri, P.; Rumondor, A. C. F.; Li, T.; Taylor, L. S. AAPS
PharmSciTech 2010, 11, 1212. (b) Anderson, B. D.; Conradi, R. A.
J. Pharm. Sci. 1985, 74, 815.
’ ASSOCIATED CONTENT
S
Supporting Information. Details of single crystal char-
b
acterizations as cif files and details of graph-set analyses, geome-
tries of the three molecular conformations found, tabulated
density and melting point data and a graphical comparison of
packing efficiency in compounds B and G. This information is
available free of charge via the Internet at http://pubs.acs.org.
’ AUTHOR INFORMATION
Corresponding Author
*(A.R.K.) Fax: (þ44) 141-548-4822. E-mail: a.r.kennedy@
strath.ac.uk (C.A.M.) Fax: (+44)141-548-4822. E-mail: catriona.
morrison@strath.ac.uk..
’ ACKNOWLEDGMENT
The authors thank MSD and the EPSRC (CTA fund) for
funding towards a studentship for CAM. Thanks are also due to
the National Crystallography Service at the University of South-
ampton for data collection on six of the reported structures.
’ REFERENCES
(1) (a) Stahl, P. H.; Wermuth, C. G., Eds. Handbook of Pharmaceu-
tical Salts: Properties, Selection and Use; VHCA: Zurich, 2008. (b) Gould,
P. L. Int. J. Pharm. 1986, 33, 201.
1834
dx.doi.org/10.1021/cg200030s |Cryst. Growth Des. 2011, 11, 1821–1834