An investigation of the hydrogen-bond preferences and co-crystallization behavior of three didonor compounds

Article abstract of DOI:10.1021/cg2002145

We assess the suitability of the three didonor compounds as building blocks for ternary co-crystals of the type (didonor)(monoacceptor)₂. A Cambridge Structural Database (CSD) survey was carried out to analyze the hydrogen-bond connectivity and develop a strategy for the preparation of the desired co-crystal. Six specific compounds were selected and crystals were grown from 1:1 and 1:2 solutions of didonor compounds (m-hydroxybenzoic acid, p-hydroxybenzoic acid, and racemic mandelic acid) and acceptor compounds (acridine, triphenylphosphine oxide, and nicotinamide) leading to three co-crystals (m-hydroxybenzoic acid) · (triphenylphosphine oxide) ₂ (1), ((RS)-mandelic acid) · (acridine) (2) and (p-hydroxybenzoic acid) · (nicotinamide) (3). Characterization by single-crystal structure determination confirms the success of this design strategy.

Full text of DOI:10.1021/cg2002145

ARTICLE
pubs.acs.org/crystal
An Investigation of the Hydrogen-Bond Preferences and
Co-crystallization Behavior of Three Didonor Compounds.
Andreas Lemmerer,*,† Daniel A. Adsmond,*,‡ and Joel Bernstein*,§,^
†Molecular Sciences Institute, School of Chemistry, University of the Witwatersrand, Johannesburg 2050, South Africa.
‡Ferris State University, Department of Physical Sciences, Big Rapids, Michigan, 49307, USA.
§Department of Chemistry, Ben-Gurion University of the Negev, Beer Sheva, Israel, 84105.
S Supporting Information
b
ABSTRACT: We assess the suitability of the three didonor
compounds as building blocks for ternary co-crystals of the type
(didonor)(monoacceptor)2. A Cambridge Structural Database
(CSD) survey was carried out to analyze the hydrogen-bond
connectivity and develop a strategy for the preparation of the
desired co-crystal. Six specific compounds were selected and
crystals were grown from 1:1 and 1:2 solutions of didonor
compounds (m-hydroxybenzoic acid, p-hydroxybenzoic acid,
and racemic mandelic acid) and acceptor compounds (acridine,
triphenylphosphine oxide, and nicotinamide) leading to three
co-crystals (m-hydroxybenzoic acid) (triphenylphosphine oxi-
3
de)2 (1), ((RS)-mandelic acid) (acridine) (2) and (p-hydro-
3
xybenzoic acid) (nicotinamide) (3). Characterization by
single-crystal structure determination confirms the success of this design strategy.
3
’ INTRODUCTION
Scheme 1. Prototypical Schematic Ternary Co-crystal,
Where D Represents a Hydrogen-Bond Donating Group and
A Represents an Accepting Group
When it comes to planning the supramolecular synthesis of
neutral multicomponent molecular complexes, the supramole-
cular chemist must consider the entire hydrogen-bonding func-
tionality of the molecules under investigation.1 Classical
hydrogen bonds which include OꢀH O, OꢀH N, NꢀH
3 3 3
3 3 3
3
O, and NꢀH N,2 are the likely candidates to be used in
3 3
3 3 3
carboxylic acids. Our strategy here is to crystallize a didonor
compound having two chemically different hydroxyl groups with
two chemically different acceptor compounds, each appended
with one of the strong acceptors listed above. If each of the two
different donors preferentially hydrogen bonds to one of the two
different acceptors, a ternary complex would form. Under favor-
able solubility conditions, this complex would precipitate as a
ternary co-crystal (Scheme 1). Ideally, we would be able to select
a set of two donors and two acceptors that would pair up in a
predictable manner when appended to molecules as shown in
Scheme 1.
Here we describe our investigation of the hydrogen-bond
behavior of a set of three functional groups known for their
donating behavior and three functional groups known for their
accepting behavior in the context of a limited set of molecules.
We investigate the crystallization behavior of binary solutions
such a synthesis and most likely to succeed.3 For example, a
recent analysis of the CSD4 showed that hydroxyl groups (found
in alcohols, phenols, and carboxylic acids) hydrogen bond to
aromatic nitrogen and primary amide acceptors with extremely
high frequency. These hydrogen bonds have been recognized as
robust supramolecular heterosynthons suitable for co-crystal
synthesis.4 In addition, it has been found that heteromeric
interactions, those between different functional groups, are more
likely to occur than between the same functional group
(homomeric interactions).5
A large number of co-crystals take advantage of the strong
hydrogen-bond attraction between a hydroxyl group donor
(alcohols, phenols, and carboxylic acids) and a strong hydro-
gen-bond acceptor such as a phosphine oxide, pyridine, or
amide.6 Our ultimate goal is to take advantage of these reliable
functional group attractions in designing ternary co-crystals
(containing 3 different compounds in the same crystal). Aaker€oy
and co-workers7a have published several examples of ternary co-
crystals in recent years where a diacceptor compound having two
chemically different acceptors is crystallized with two different
Received: February 15, 2011
Revised:
March 17, 2011
Published: March 17, 2011
r
2011 American Chemical Society
2011
dx.doi.org/10.1021/cg2002145 Cryst. Growth Des. 2011, 11, 2011–2019
|

Crystal Growth & Design
ARTICLE
Scheme 2. Possible Stoichiometric Variants of Binary
Didonor/Monoacceptor Co-crystals
Chart 1. Six Neutral Molecules Investigated in This Report
and Their Three Co-crystals (1ꢀ3)
Scheme 3. Competition between Two Donor and Two
Acceptor Groups in Binary Didonor/Diacceptor Co-crystals
compared to those observed in other co-crystals involving the same
compounds deposited in the Cambridge Structural Database (CSD).
Trends in hydrogen-bond behavior of these six compounds are
sought.
containing a didonor compound and a strong acceptor com-
pound. For this work we selected three didonor compounds,
each containing a carboxylic acid as the first donor. Two of these
compounds (m-hydroxybenzoic acid and p-hydroxybenzoic
acid) contain a phenol as the second donor, while the third
compound (RS)-mandelic acid contains an alcohol as the second
donor group. These didonor compounds were co-crystallized
with compounds selected for their known accepting ability:
triphenylphosphine oxide (with a single phosphine oxide ac-
ceptor), acridine (with a single-pyridine acceptor), and nicoti-
namide (with both a pyridine and an amide carbonyl acceptor).
Although the primary amide also acts as a donor, its principal role
in these structures is that of an acceptor and we will refer to it here
as a hydrogen-bond acceptor.
A binary co-crystal containing a didonor molecule and a
monoacceptor molecule (such as triphenylphosphine oxide or
acridine) would be expected to show one of the three connectiv-
ity patterns illustrated in Scheme 2. Preferential hydrogen-
bonding of both donors to the monoacceptor molecule would
result in a 1:2 co-crystal with the connectivity shown. Preferential
hydrogen-bonding of only one of the two donors to the mono-
acceptor molecule would result in a 1:1 co-crystal displaying one
of the two connectivity patterns shown in Scheme 2, where the
remaining donor hydrogen bonds to a neighboring didonor
molecule. In addition, one might be able to exert a level of
control on the final co-crystal stoichiometry simply by adjusting
the mole ratio of the components in the crystallizing solution.
Alternatively, one can study the hydrogen-bond preferences of
a didonor compound by co-crystallizing it with a diacceptor
compound where A1 and A2 are chemically different acceptor
groups (such as nicotinamide) setting up a competition between
two donors and two acceptors. The preference of one of the two
1:1 co-crystal patterns shown in Scheme 3 would provide insight
into the hydrogen-bond preferences of the groups involved.
In this study, we co-crystallized didonors both with monoacceptor
and diacceptor compounds and report three new co-crystals resulting
’ EXPERIMENTAL SECTION
Materials. m-hydroxybenzoic acid, p-hydroxybenzoic acid, (RS)-
mandelic acid, triphenylphosphine oxide, acridine and nicotinamide
were purchased from commercial sources (Aldrich) and used without
further purification. All solvents used in the study were of AR (99.9%)
quality.
Co-crystal 1 (1:2 m-Hydroxybenzoic acid/Triphenylpho-
sphine Oxide). A 1:2 stoichiometric amount of m-hydroxybenzoic
acid (55 mg, 0.40 mmol) and triphenyphosphine oxide (224 mg, 0.80
mmol) was dissolved in 13 mL of toluene. Colorless prisms were
obtained upon slow evaporation of the solution at room temperature.
Identical 1:2 co-crystals (by FTIR, KBr pellet) were obtained from 1:1
and from 1:2 solutions of 20% acetone/80% toluene.
Co-crystal 2 (1:1 (RS)-Mandelic acid/Acridine). A 1:1 stoi-
chiometric amount of (RS)-mandelic acid (61 mg, 0.40 mmol) and
acridine (72 mg, 0.40 mmol) was dissolved in 9 mL of 33% acetone/67%
cyclohexane. Greenish yellow prisms were obtained upon slow evapora-
tion of the solution at room temperature. Identical 1:1 co-crystals (by
FTIR, KBr pellet) were obtained from a 1:2 solution of 33% acetone/
67% cyclohexane.
Co-crystal 3 (1:1 p-Hydroxybenzoic Acid/Nicotinamide). A
1:1 stoichiometric amount of p-hydroxybenzoic acid (100 mg, 0.819
mmol) and nicotinamide (113 mg, 0.819 mmol) was dissolved in 5 mL
of methanol and colorless crystals grown by slow evaporation over a few
days at room temperature.
Table 1 summarizes the results obtained from the full set of co-
crystallization experiments (co-crystal stoichiometries reported with
didonor compound listed first.) Asterisks denote that the single-crystal
structure has been determined and is being submitted for publication
elsewhere. Stoichiometries of co-crystals without X-ray structures were
determined by proton NMR.
Single-Crystal X-ray Diffraction (SCXRD). Intensity data were
collected on a Bruker SMART 1K CCD area detector diffractometer
with graphite monochromated Mo KR radiation (50 kV, 30 mA) and
performed at T = 293 K. The collection method involved ω-scans of
width 0.3ꢀ. Data reduction was carried out using the program SAINTþ,8
and empirical absorption corrections were made using the program
from the six compounds listed above: (m-hydroxybenzoic acid)
3
(triphenylphosphine oxide)2 (1), ((RS)-mandelic acid) (acridine)
3
(2) and (p-hydroxybenzoic acid) (nicotinamide) (3). The hydro-
3
gen-bonding interactions observed in these three co-crystals are
2012
dx.doi.org/10.1021/cg2002145 |Cryst. Growth Des. 2011, 11, 2011–2019

Crystal Growth & Design
ARTICLE
Table 1. Summary of All Didonor/Acceptor Co-crystallization Results
m-hydroxybenzoic acid
(RS)-mandelic acid
p-hydroxybenzoic acid
TPPO
co-crystal 1 (1:2)
oil (no crystals)
crystals from oil (identity uncertain)
acridine
1:1(2 forms)*, 1:2*, 3:2, 1:4.5 hemihydrate
XAQQIQ5f
co-crystal 2 (1:1)
JILZOU3b with (R)-mandelic acid
1:1
nicotinamide
co-crystal 3 (1:1)
Table 2. Crystallographic Data for 1ꢀ3
1
2
3
formula
mw
C
43H36O5P2
C21H17NO3
331.36
C13H12N2O4
260.25
694.66
T (K)
293(2)
293(2)
293(2)
cryst size (mm3)
cryst syst
space group (No.)
a (Å)
0.20 ꢁ 0.20 ꢁ 0.20
monoclinic
P21/n (14)
11.600(1)
13.569(1)
23.303(2)
90.567(2)
3667.7(6)
4
0.30 ꢁ 0.20 ꢁ 0.20
0.40 ꢁ 0.10 ꢁ 0.06
monoclinic
C2/c (15)
30.962(4)
7.371(1)
11.223(2)
107.777(6)
2438.8(5)
8
monoclinic
P21/c (14)
10.320(3)
21.133(7)
7.825(3)
102.105(6)
1668.5(9)
4
b (Å)
c (Å)
β (deg)
V (Å3)
Z
F(calcd) (Mg mꢀ3
)
1.258
1.319
1.418
μ(MoꢀKR) (mmꢀ1
)
0.164
0.089
0.107
theta range for data collection (deg)
no. of reflns collected
no. of unique data [R(int)]
data with I > 2σ(I)
1.74 to 25.50
17873
1.93 to 25.49
9017
2.76 to 25.50
6314
6823 [0.0784]
2938
3096 [0.0391]
1837
2263 [0.0271]
1614
final R (I > 2σ(I))
0.0483
0.0488
0.0412
final wR2 (all data)
0.1579
0.1554
0.1307
CCDC No.
812396
812397
812398
SADABS.8 The crystal structures were solved in the WinGX9 suite of
programs by direct methods using by direct methods using SHELXS-
97.10 Non-hydrogen atoms were first refined isotropically followed by
anisotropic refinement by full matrix least-squares calculations based on
F2 using SHELXL-97.10 Thereafter, all hydrogen atoms attached to N
and O atoms were located in the difference fourier map and their
coordinates refined freely with isotropic parameters 1.5 times (O) or 1.2
times (N) those of the heavy atoms to which they are attached. All C-H
hydrogen atoms were placed at idealized positions and refined as riding
atoms with isotropic parameters 1.2 times those of the heavy atoms to
which they are attached. Diagrams and publication material were
generated using ORTEP-3,11 PLATON,12 and DIAMOND.13 Further
crystallographic data are summarized in Table 2.
Cambridge Structural Database. All searches were done on the
version 5.31 Database with the November 2009 and February 2010
updates.14 The filters applied to all searches are: 3D coordinates
determined, R-factor e0.075, not disordered, no errors, no ions, no
powder structures, only organics. The searches were performed with the
six molecules as the target molecule as a query. The search for m-
hydroxybenzoic co-crystals gave 13 hits, which was reduced to 11 after
removing the crystal structures of the search compound. The search for
p-hydroxybenzoic co-crystals gave 27 hits, which were reduced to 16
after removing duplicate entries, hydrates and any molecular salt
structures. The search for nicotinamide co-crystals gave 22 hits, which
was reduced to 11 after removing crystal structures of the target
molecule, hostꢀguest complexes and compounds with covalent bonds
to the pyridine. The search for mandelic acid gave 21 hits, which were
reduced to 5 hits by removing molecular salts, structures of the target
molecule, clathrates and hydrates. The search for triphenylphosphine
oxide co-crystals gave 88 hits, which were reduced to 45 by removing
crystal structures of the target compound, hydrates, structures with
bondstometals, structureswithhalogenbonding orCꢀH X hydrogen
3 3 3
bonding only, duplicates, and complexes with inorganic acids. The search
for acridine co-crystalsgave26hits, whichwere reduced to16byremoving
the structures of the parent compound, and those that do not form any
strong hydrogen bonds (halogen bonding, CꢀH X are excluded).
3 3 3
’ RESULTS
The crystal structure of 1 has one molecule of m-hydroxy-
benzoic acid and two molecules of triphenylphosphine oxide
(TPPO) in the asymmetric unit (Figure 1). The carboxylic acid
and phenol protons both hydrogen bond to the phosphine oxide
of the TPPO molecules to form a trimeric supermolecule
through OꢀH O hydrogen bonds (Figure 2a). The trimeric
3 3 3
supermolecules pack in ribbons along [101] and are connected
by four CꢀH O interactions between the aromatic rings of
3 3 3
the TPPO and the O atoms of m-hydroxybenzoic acid and TPPO
(Figure 2b). Table 3 summarizes the hydrogen-bond geometries.
The crystal structure of 2 has one molecule each of mandelic
acid and acridine in the asymmetric unit (Figure 1). Two
mandelic acid molecules are connected into a dimer by a
R22(10) ring formed by a OꢀH O hydrogen bond from the
3 3 3
alcohol O3 to the carbonyl O2 (Figure 3a). The dimer hydrogen
2013
dx.doi.org/10.1021/cg2002145 |Cryst. Growth Des. 2011, 11, 2011–2019

Crystal Growth & Design
ARTICLE
Figure 2. (a) Two hydrogen-bonding interactions in co-crystal 1 form a
three-membered supermolecule. (b) Packing of the supermolecules,
showing the CꢀH O interactions.
3 3 3
R22(8) are connected through NꢀH O hydrogen bonding
3 3 3
forming a R24(8)ring. The phenol proton hydrogen bonds to the
pyridine N atom to form infinite ribbons along [101] (Figure 4a).
The ribbons layer above each other along the b-axis (Figure 4b).
Table 3 summarizes the individual hydrogen-bond geometries.
Figure 1. Asymmetric unit and atomic numbering scheme of co-crystals
1ꢀ3, and 50% displacement ellipsoids. Only the symmetry independent
hydrogen bonds are shown. For clarity, the H atoms not involved in
hydrogen bonding are omitted in 1.
’ DISCUSSION
Here we present three examples of didonor co-crystals, each
having a preferred stoichiometry that is independent of the
solution stoichiometry. Two of the co-crystal pairs we have
chosen show a preference for 1:1 co-crystal formation, while
one shows a preference for 1:2 co-crystal formation. Individual
CSD searches for co-crystals involving each of the six molecules
show how often a particular type of heteromeric or homomeric
interaction is observed in the total number of co-crystals
(Table 4). The hydrogen-bonding contact criteria were the
default values used in MERCURY15 and short contacts were
also included by inspection if hydrogen-bond donors appeared
not to be used. All interactions were catalogued.
1:2 m-Hydroxybenzoic Acid/Triphenylphosphine Oxide
Co-crystal 1. Co-crystal 1, in which both the carboxylic acid
and phenol groups of m-hydroxybenzoic acid form hydrogen
bonds with two independent TPPO molecules, provides an
example of the 1:2 co-crystal represented in Scheme 2. This
bonds to two acridine molecules through the carboxylic acid
proton to the pyridine N. This forms a four-membered super-
molecule. The crystal packing has alternating layers of mandelic
acid and acridine along the b-axis such that the acridine molecules
of adjacent interdigitated interdigitated and feature π
π
3 3 3
interactions between them (Cg Cg distance: 3.897(8) Å)
3 3 3
(Figure 3b). Table 3 summarizes the hydrogen-bond geometries.
The crystal structure of co-crystal 3 has one molecule each of
p-hydroxybenzoic acid and nicotinamide in the asymmetric unit.
The p-hydroxybenzoic acid and nicotinamide molecules hydro-
gen bond using four NꢀH O and two OꢀH O hydrogen
3 3 3
3 3 3
bonds. The carboxylic acid and the amide groups hydrogen bond
to each other forming R22(8) rings. Pairs of neighboring rings
2014
dx.doi.org/10.1021/cg2002145 |Cryst. Growth Des. 2011, 11, 2011–2019

Crystal Growth & Design
ARTICLE
Table 3. Geometrical Parameters for Hydrogen Bonds in 1ꢀ3
compda
d(DꢀH) (Å)
d(H A) (Å)
d(D A) (Å)
—(DꢀH A) (deg)
3 3 3
3 3 3
3 3 3
1
O1ꢀH1 O4
0.87(4)
0.84(4)
0.93
1.73
2.571(3)
2.679(3)
3.616(5)
3.400(5)
3.378(4)
3.350(5)
162(4)
167(5)
177
3 3 3
O3ꢀH3 O5
1.85(4)
2.69
3 3 3
C24ꢀH24 O1i
3 3 3
C36ꢀH36 O2ii
0.93
2.51
160
3 3 3
C37ꢀH37 O4ii
0.93
2.52
153
3 3 3
C41ꢀH41 O2iii
0.93
2.50
153
3 3 3
2
O1ꢀH1 N1
0.82
1.79
2.573(2)
2.839(3)
3.372(4)
3.355(3)
160
162(3)
138
3 3 3
O3ꢀH3 O2iii
0.93(3)
0.93
1.94(3)
2.63
3 3 3
C4ꢀH4 O3iv
3 3 3
C6ꢀH6 O1v
0.93
2.61
138
3 3 3
3
O1ꢀH1 O4
0.95(3)
0.89(3)
0.85(3)
0.86(3)
1.67(30
1.87(3)
2.08(4)
2.16(4)
2.617(2)
2.743(2)
2.924(2)
2.965(2)
169(3)
168(2)
172(3)
155(3)
3 3 3
O3ꢀH3 N2vi
3 3 3
N1ꢀH1S O2
3 3 3
N1ꢀH1A O2vii
3 3 3
a Symmetry transformation codes: (i) ꢀx þ 3/2, y þ 1/2, ꢀz þ 3/2; (ii) ꢀx þ 3/2, y ꢀ 1/2, ꢀz þ 3/2; (iii) ꢀx þ 1, ꢀy þ 1, ꢀz þ 1; (iv) x, y, z þ 1;
(v) ꢀx, ꢀy þ 1, ꢀz þ 1; (vi) x þ 1/2, ꢀy þ 1/2, z þ 1/2; (vii) ꢀx þ 1, y, ꢀz þ 3/2.
1:2 co-crystal forms by slow evaporation of 1:1 as well as 1:2
solutions of the reactants. We searched the CSD beforehand to
gain insight into the hydrogen-bonding behavior of the two
starting materials. We found 11 m-hydroxybenzoic acid co-crystal
structures in the CSD (reduced set) yielding 12 symmetry
independent m-hydroxybenzoic acid molecules. In 9 of the 12,
both the carboxyl proton and the phenol proton of m-hydro-
xybenzoic acid hydrogen bonds to crystallographically indepen-
dent acceptor molecules. In one structure, only the carboxyl
proton hydrogen bonds to the acceptor molecule and in another,
only the phenol donates to the acceptor molecule. In short, there
is no clear dominance of the carboxyl proton or the phenolic
proton of m-hydroxybenzoic acid in binding to the acceptor
compound. The most prevalent hydrogen bond is a carboxyl to
aromatic nitrogen appearing 9 times in the 11 structures. It is
closely followed by the phenol to aromatic nitrogen hydrogen
bond which appears seven times. The same set of structures
includes 14 symmetry independent acceptor molecules, eleven
which act as diacceptors (accepting twice from a donor molecule)
and only three which act as monoacceptors. One of those three,
HONTOU, (which the acceptor compound is 4-phenylpyridine)
provides the only example in the database of m-hydroxybenzoic
acid displaying the 1:2 pattern shown in Scheme 2. Our own 1:2
m-hydroxybenzoic acid/acridine co-crystal also displays this
pattern. The second monoacceptor structure, SUVZEO, forms
a 1:1 co-crystal of the type displayed in Scheme 2. The phenol
hydrogen bonds to the N-oxide acceptor while the carboxylic
acids bond to each other in an R22(8) dimer. The third structure
containing monoaccepting molecules, HONVIQ, is a 3:2 co-
crystal with an ADADA pattern in which the terminal molecules
accept only once (even though they have two terminal nitrogens),
whereas the central molecule accepts twice. Eight of the 11
structures that have diacceptor compounds display 1:1 hydro-
gen-bond patterns with alternating didonor and diacceptor
molecules of the type represented in Scheme 3. Only three of
the diacceptor compounds in these structures were asymmetrical
(nicotinamide, isonicotinamide, and caffeine), differentiating
between the two acceptors. Co-crystals with such compounds
could potentially provide valuable insight into the hydrogen-bond
preferences of the donor/acceptor set. All three of these examples
have an amide carbonyl and an aromatic nitrogen. In both the
isonicotinamide and caffeine structures, the phenol hydrogen
bonds to the amide while the carboxylic acid hydrogen bonds to
the aromatic nitrogen. In nicotinamide, the opposite happens: the
carboxylic acid forms a R22(8) ring with the amide while the phenol
hydrogen bonds to the aromatic nitrogen. While this set of asym-
metric m-hydroxybenzoic acid/diacceptor structures is much too
small to draw any conclusions about the hydrogen-bond preferences
of the carboxyl and phenolic protons of 3HBA, it clearly shows that
on this particular didonor compound, the pairing between the two
donors and two acceptors can go either way. Interestingly, in the
m-hydroxybenzoic acid/pyrazine co-crystal, both of the equivalent
aromatic nitrogens accept from carboxylic acids while the phenols are
left to hydrogen bond to the acid carbonyl. This results in an
unpredicted 2:1 pattern where two didonor molecules alternate with
one diacceptor molecule in an infinite (DDA)n chain.
As shown by our CSD search, TPPO has a rich history of
co-crystal formation (45), especially with carboxylic acid (17) or
phenol/alcohol donors (13/4). This could be due to the weak
hydrogen bonding interactions that TPPO can undergo with
itself, and hence can be used to obtain co-crystals with a
compound that has been difficult to crystallize.16 The four
polymorphs of TPPO17 all have weak CꢀH OdP hydrogen
3 3 3
bonding, and hence molecules with strong donor functionalities
will hydrogen bond to the OdP group preferentially and will not
have to compete much with a CꢀH donor.16
1:1 Mandelic Acid/Acridine Co-crystal 2. Mandelic acid,
which has both acid and alcohol donor groups, uses the COOH
3
N heteromeric interaction from the acid to the pyridine N
3 3
atom of acridine to form co-crystal 2 with acridine; whereas the
alcohol hydrogen bonds to the carbonyl of the acid functional
group forming an R22(10) hydrogen-bonded ring. This preference
is somewhat surprising especially when one considers that this
same 1:1 co-crystal forms in a 1:2 solution having a 2-fold excess
2015
dx.doi.org/10.1021/cg2002145 |Cryst. Growth Des. 2011, 11, 2011–2019

Crystal Growth & Design
ARTICLE
Figure 4. (a) 1D ribbon formed by the OꢀH O, NꢀH O, and
3 3 3
3 3 3
OꢀH N hydrogen bonds in co-crystal 3. (b) Ribbons have a wavelike
3 3 3
shape and pack in a parallel arrangement.
hydrogen bonds to an alcohol oxygen in a C(5) chain. Neither
the homomeric acid R22(8) dimer nor the homomeric C(2)
alcohol chain is observed in any of the three polymorphs. The
structure of (S)-mandelic acid has two molecules in the asym-
metric unit. In both molecules, the alcohol hydrogen bonds to
the acid carbonyl and the carboxylic acid proton hydrogen bonds
to the alcohol oxygen as seen in the polymorphs of the racemic
compound. In the structure of (S)-mandelic acid, however, the
R22(10) ring is replaced by a C(5) chain. There are only five
mandelic acid co-crystals in the CSD reduced set. In two of these,
the mandelic acid is racemic and in three of them the mandelic
acid is present as a single enantiomer. In all five co-crystal
structures the guest compound has at least two good hydrogen
acceptors, providing the minimum requirements for one of the
Scheme 3 hydrogen-bond patterns. In all five structures, the
carboxylic acid proton hydrogen bonds to the guest molecule (in
four of the structures the acceptor is an aromatic nitrogen and in
one it is a phosphine oxide). No trend, however, is observed in
the hydrogen bonding of the alcohol proton. In two structures
the alcohol proton also hydrogen bonds to the guest molecule, in
two others it hydrogen bonds to the acid carbonyl in a C(5)
chain, and in one structure the alcohol hydrogen bonds to an acid
OH. In AVIPEA, the alcohol of (R)-mandelic acid hydrogen
bonds to a phosphine oxide as does the carboxyl proton. In
LUNPAL, the alcohol of racemic mandelic acid hydrogen bonds
to an anti lone pair of an amide carbonyl which is also involved in
a centrosymmetric R22(10) amide dimer. Effectively in this
structure the amide dimer has inserted itself into the R22(10)
seen previously, forming a centrosymmetric R44(18) ring. The
carboxyl proton, in this structure, hydrogen bonds to an aromatic
nitrogen. In JILZOU the alcohol proton of (R)-mandelic acid
hydrogen bonds to the acid carbonyl rather than to the carbonyl
of a primary amide in the host. In PIKLEA, the alcohol of (S)-
mandelic acid also hydrogen bonds to the acid carbonyl in
preference to a secondary amide carbonyl. In OFOKEA the
alcohol proton of the racemic mandelic acid hydrogen bonds to
the acid OH in a 2:1 co-crystal with bipyridine where both of the
aromatic nitrogens are hydrogen bonded to carboxylic acid protons.
Figure 3. (a) four-membered supermolecule created using OꢀH
O
3 3 3
and OꢀH N hydrogen bonds of 2. (b) Crystal packing of the
3 3 3
supermolecules of co-crystal 2 showing the π π interactions between
3 3 3
the acridine molecules.
of the best available acceptor, the aromatic nitrogen. There are
three polymorphs of (RS)-mandelic acid19 reported in the CSD.
In all three polymorphs the alcohol proton hydrogen bonds to
the acid carbonyl in an R22(10) ring and the carboxylic acid proton
2016
dx.doi.org/10.1021/cg2002145 |Cryst. Growth Des. 2011, 11, 2011–2019

Crystal Growth & Design
ARTICLE
Table 4. Heteromeric Interactions between Different Functional Groups on Different Molecules in Co-crystals Involving the Six
Molecules
no. of structures no. and type of homomeric interactions in co-crystals total no. of heteromeric interactions to various acceptor atoms
m-hydroxybenzoic acid
11
acid dimer: 1
COOH N: 9
3 3 3
amide dimer: 1
COOH OdCꢀNH2:1
3 3 3
OdCꢀNꢀH OdCꢀOH:2
3 3 3
PhꢀOH N: 7
3 3 3
PhꢀOH O: 4
3 3 3
TPPO
45
OH OH: 3
Ph-OH OdPPh3: 13
3 3 3
3 3 3
Ph3PdO HO-R: 4
3 3 3
Ph3PdO HN-CdO: 6
3 3 3
Ph3PdO HN-R: 5
3 3 3
Ph3PdO HOOC: 17
3 3 3
PhꢀOH OdC: 1
3 3 3
mandelic acid
5
amide dimer: 1
acid dimer: 1
COOH N: 4
3 3 3
COOH OdP: 1
3 3 3
RꢀOH OdCꢀOH: 2
3 3 3
RꢀOH O(H)ꢀCdO: 1
3 3 3
RꢀOH OdCꢀNH2: 1
3 3 3
RꢀOH OdP: 1
3 3 3
R-(H)O HN-CdO: 1
3 3 3
acridine
16
17
OꢀH OꢀH: 1
COOH N: 11
3 3 3
3 3 3
PhꢀOH N: 2
3 3 3
RꢀOH N: 1
3 3 3
NꢀH N: 2
3 3 3
R-NH OdCꢀOH: 1
3 3 3
p-hydroxybenzoic acid
acid dimer: 6
COOH N: 6
3 3 3
OꢀH OꢀH: 1
COOH OdCꢀNH2: 3
3 3 3
3 3 3
COOH OdCꢀR: 1
3 3 3
HOꢀCdO HNꢀCdO: 4
3 3 3
HO-CdO HNꢀR: 2
3 3 3
PhꢀOH N: 10
3 3 3
PhꢀOH OdCꢀOH: 2
3 3 3
PhꢀOH OdCꢀR: 3
3 3 3
PhꢀOH OdCꢀNH2: 1
3 3 3
Ph-(H)O HꢀNꢀCdO: 1
3 3 3
nicotinamide
11
9 amide dimers
5 amide tetramers
1 amide chains
COOH N: 8
3 3 3
PhꢀOH N: 2
3 3 3
OdCꢀNꢀH OdCꢀOH: 3
3 3 3
OdCꢀNꢀH O(H)ꢀPh: 2
3 3 3
COOH OdCꢀNH2: 1
3 3 3
RꢀN-H OdCꢀOH: 1
3 3 3
RꢀOH OdCꢀOH: 1
3 3 3
a The number of interactions can exceed the sample size of co-crystal structures used, as most co-crystals have more than one heteromeric interaction.
To date, mandelic acid has produced no 1:2 co-crystals displaying the
general pattern shown in Scheme 2. It is interesting to note that even
1:2 solutions of racemic mandelic and acridine result in 1:1 co-crystals.
Acridine, of which there are 16 co-crystals in the reduced set,
has predominantly been co-crystallized with acid containing
itself. Acridine itself can also act as a good π-acceptor in the
formation of πꢀπ charge-transfer complexes.18
1:1 p-Hydroxybenzoic Acid/Nicotinamide Co-crystal 3.
The co-crystal between p-hydroxybenzoic acid and nicotinamide
is a competition experiment between the two donor functional-
ities on the p-hydroxybenzoic acid molecule (phenol and car-
boxylic acid) and the two acceptor functionalities of the
nicotinamide molecule (pyridine and amide carbonyl). An essen-
tial component of such a competition experiment is that the same-
molecule functional groups be isolated from each other to prevent
intramolecular hydrogen bonding from interfering in the compe-
tition. This isolation can be easily accomplished by placing groups
molecules (11*COOH N) and only with two phenols and
3 3 3
one alcohol. Nineteen of the 20 unique acridine molecules in
these structures act as hydrogen-bond acceptors. Similarly, as was
indicated above for TTPO, the known polymorphs of acridine18
feature CꢀH N or π-stacking interactions, and hence acri-
3 3 3
dine is eminently suitable suitable for co-crystallization experi-
ments with good hydrogen bond donors, as it lacks good donors
2017
dx.doi.org/10.1021/cg2002145 |Cryst. Growth Des. 2011, 11, 2011–2019

Crystal Growth & Design
ARTICLE
Table 5. Hydrogen-Bonding between Carboxylic Acid and
Phenol Donors and Aromatic N and Primary Amide
Acceptors in Four Crystal Structures
hydrogen-bonding functional groups each on both molecules,
prefers 1:1 co-crystal formation.
When the set of CSD co-crystals involving m-hydroxybenzoic
acid is combined with the set of CSD co-crystals involving p-
hydroxybenzoic acid, there are a total of 28 co-crystals. Out of
these 28 structures, there are 19 symmetry-independent hydro-
xybenzoic acid molecules where both the carboxyl and phenolic
protons hydrogen bond to the acceptor molecule. This provides
strong support for compounds containing both a carboxylic acid
and a phenol being good candidates for didonor compounds in
ternary co-crystals of the type shown in Scheme 1. There are
three examples of hydroxybenzoic acid co-crystals where only the
acid donates to an acceptor molecule and seven examples, where
only the phenol donates to an acceptor molecule. When only the
acid donates to the acceptor molecule, the phenol donates to an
acid carbonyl and when only the phenol donates, the acid forms
an R22(8) acid dimer. In all three instances where only the acid
bonds to the acceptor molecule, it is to an aromatic nitrogen.
When only the phenol bonds to the acceptor molecule, it bonds
to an aromatic nitrogen (four times), an amide (once), a ketone
(once), and an N-oxide (once).
Of the 19 hydroxybenzoic acids where both acid and phenol
groups hydrogen bond to the acceptor molecule, nine are
bonded to a symmetrical diacceptor compound and seven are
bonded to an asymmetrical diacceptor compound. Again, it is the
diacceptors with two different acceptor groups that aid in sorting
out hydrogen-bond preferences. Although five of the asymme-
trical diacceptor molecules have both an aromatic nitrogen and
an amide, there was no hint of a selectivity difference between the
phenol and the carboxylic acid. In two structures, the acid bonds
to the N, whereas the phenol bonded to the amide and in three
structures the exact opposite is seen. It is interesting to note that
p-hydroxybenzoic acid bonds to tetramethylpyrazine only
through the carboxyl proton in one structure and only through
the phenolic proton in another. In short, on the basis of the
available evidence, it appears that given a carboxylic acid, phenol,
amide, and aromatic nitrogen there is no preferred pairing of the
donors and acceptors. In 2/3 of the examples of the hydro-
xybenzoic acid co-crystals, both the carboxylic acid and the
phenol groups bond to the acceptor molecule and in 1/3 only
one of the two donors bonds to the acceptor molecule. The only
apparent difference in hydrogen-bonding behavior between m-
hydroxybenzoic acid and p-hydroxybenzoic acid is in the fre-
quency of structures in which the phenol group exclusively
hydrogen bonds to the acceptor molecule leaving the carboxylic
acid to form an R22(8) acid dimer. Although this occurs in only 1
out of 11 (9%) m-hydroxybenzoic acid co-crystals, the phenol
dominance is seen in 7 out of 17 (41%) p-hydroxybenzoic acid
co-crystals. It is useful for comparison purposes that three of the
acceptor compounds forming phenol exclusive co-crystals with
p-hydroxybenzoic acid also form analogous co-crystals with m-
hydroxybenzoic acid. In one of these, only the carboxylic acid
bonds to the acceptor molecule and in the other two, both the
acid and the phenol bond to the acceptor molecule.
in a meta or para relationship to each other on an arene backbone.
The carboxyl group and the phenol may be combined in m-
hydroxybenzoic acid or p-hydroxybenzoic acid, whereas the
pyridine and primary amide groups may be combined in nicoti-
namide or isonicotinamide. The pair of compounds selected for 3
completes the set of 4 possible pairings of these compounds, the
other three previously appearing in the CSD (Table 5).
In the literature, nicotinamide has been predominantly co-
crystallized with carboxylic acid containing molecules (9 out of
11 nicotinamide co-crystals found in the CSD)20 and is a good co-
crystal former. This same set of nicotinamide co-crystals features
COOH N eight times, PhꢀOH N twice and the homo-
3 3 3
3 3 3
meric amide dimer nine times. Hence, both functional groups on
p-hydroxybenzoic acid can be expected to partner (compete
between each other) with the pyridine N of the nicotinamide. In
addition, six COOH N andten PhꢀOH N hydrogen bonds
3 3 3
3 3 3
are observed in the set of 17 p-hydroxybenzoic acid co-crystals
structures found in the CSD. This further illustrates the ability of
both donor types to hydrogen bond to aromatic nitrogens. In co-
crystal 3 it is the phenol proton of p-hydroxybenzoic acid that
hydrogen bonds to the pyridine N atom, whereas the carboxylic
acid forms a heterodimer with the amide functional group of
nicotinamide. The latter interaction is observed once in the set of
nicotinamide:carboxylic acid co-crystals found in the CSD, and is
encountered more frequently in co-crystals with isonicotinamide
(an isomer of nicotinamide) and carboxylic acids.16 The fact that
the phenol competes with the acid and that the acid competes with
homomeric amide dimer formation is a seldom seen case; the first
reported one is the isomerically related co-crystal (p-hydroxyben-
zoic acid):(isonicotinamide).17 This co-crystal features identical
heteromeric interactions and a ribbon structure as in 3, however
the ribbon is corrugated and not flat as in 3 due to the different
position of the N atom in the pyridine ring. The authors called such
a product a minor adduct of a supramolecular reaction (the major
’ CONCLUSION
adduct would have been one that featured the COOH N and
In the end, with m-hydroxybenzoic acid and p-hydroxybenzoic
acid, we have two donors on the same molecule that exhibit a
number of examples of hydrogen bonding to each of two
acceptors, the amide carbonyl and the aromatic nitrogen, but
there appears to be little selectivity within this set of four
3 3 3
amide dimer). Co-crystals of m-hydroxybenzoic acid with nicoti-
namide exhibit the same interactions,5f whereas isonicotinamide
and m-hydroxybenzoic acid has the (expected) amide dimer and
COOH N hydrogen bond.18 Co-crystal 3, which has two good
3 3 3
2018
dx.doi.org/10.1021/cg2002145 |Cryst. Growth Des. 2011, 11, 2011–2019

Crystal Growth & Design
ARTICLE
functional groups even when both donors are appended to the
same aromatic ring. Either of these two didonor compounds
could be used in designing a ternary co-crystal of the type shown
in Scheme 1, but if the two monoacceptor molecules contained
an amide and an aromatic nitrogen respectively, it would be
nearly impossible to predict which donor would bond to which
acceptor. In our third didonor compound, racemic mandelic acid,
we have a compound with a carboxylic acid which has a strong
affinity for aromatic nitrogens and amide carbonyls and an
alcohol, which has unclear loyalties. In the mandelic acid
polymorphs and in co-crystal 2, the alcohol shows a preference
for bonding to the mandelic acid carbonyl in a centrosymmetric
R22(10) ring, even in crystals grown from a solution with a 2-fold
excess of acridine. This preference would have to be broken
before this compound would be a viable candidate for ternary co-
crystal formation. Using one of the enantiomers of mandelic acid
instead of the racemate would break the need for forming the
centrosymmetric dimer and possibly increase the affinity of the
alcohol for a competing acceptor. The chances of ternary co-
crystal formation might also be improved by increasing the
distance between the acid and alcohol on the didonor molecule.
In conclusion, we have demonstrated that through (a)
synthesizing didonor/monoacceptor co-crystals and didonor/
diacceptor co-crystals of the types shown in Schemes 1 and 2
from solutions of varying mole ratios and (b) using the Cam-
bridge Structural Database for investigating the hydrogen-bond
histories of the compounds of interest, we can gain valuable
insight into the viability of specific didonor compounds for
potential ternary co-crystal synthesis.
(2) (a) Perrin, C. L.; Nielson, J. B. Annu. Rev. Phys. Chem. 1997,
48, 511–544. (b) Desiraju, G. R. Acc. Chem. Res. 2002, 35, 565–573. (c)
Aaker€oy, C. B.; Seddon, K. R. Chem. Soc. Rev. 1993, 397–407.
(3) (a) Aaker€oy, C. B.; Beatty, A. M. Aust. J. Chem. 2001,
54, 409–421. (b) Friꢀsꢀciꢁs, T.; Jones, W. Faraday Discuss 2007,
136, 167–178.
(4) Shattock, T. R.; Arora, K. K.; Vishweshwar, P.; Zaworotko, M. J.
Cryst. Growth Des. 2008, 8, 4533–4545.
(5) (a) Lemmerer, A.; Bernstein, J. CrystEngComm 2010,
12, 2029–2033. (b) He, Guangwe.; Chow, P. S.; Tan, R. B. H. Cryst.
Growth Des. 2009, 9, 4529–4532. (c) Stahly, G. P. Cryst. Growth Des.
2009, 9, 4212–4229. (d) Childs, S. L.; Zaworotko, M. J. Cryst. Growth
Des. 2009, 9, 4208–4211. (e) He, G.; Chow, P. S.; Tan, R. B. H. Cryst.
Growth Des. 2009, 9, 4529–4532. (f) McMahon, J. A.; Bis, J. A.;
Vishweshar, P.; Shattock, T. R.; McLaughlin, O. L.; Zaworotko, M. J.
Z. Kristallogr. 2005, 220, 340–350.
(6) Aaker€oy, C. B.; Schultheiss, N. In Making Crystals by Design;
Braga, D., Grepioni, F., Eds.; Wiley-VCH: Weinheim, Germany, 2007,
pp 209ꢀ240.
(7) (a) Aakero€y, C. B.; Desper, J.; Urbian, J. F. Chem. Commun.
2005, 2820–2822. (b) Aaker€oy, C. B.; Beatty, A. M.; Helfrich, B. A.
Angew. Chem., Int. Ed. 2001, 40, 3240–3242. (c) Aakero€y, C. B.; Salmon,
D. J. CrystEngComm 2005, 7, 439–448. (c) Aaker€oy, C. B.; Desper, J.;
Smith, M. M. Chem. Commun. 2007, 3936–3938. (d) Bhogala, B. R.;
Nangia, A. New. J. Chem. 2008, 32, 800–807.
(8) SAINTþ, version 6.02 (includes XPREP and SADABS);Bruker
AXS Inc.: Madison, WI.
(9) Farrugia, L. J. J. Appl. Crystallogr. 1999, 32, 837–838.
(10) Sheldrick, G. M. Acta Crystallogr., Sect. A 2008, 64, 112–122.
(11) Farrugia, L. J. J. Appl. Crystallogr. 1997, 30, 565.
(12) Spek, A. L. J. Appl. Crystallogr. 2003, 36, 7–13.
(13) Brandenburg, K. Diamond, version 2.1e; Crystal Impact GbR:
Bonn, Germany.
(14) Allen, F. H. Acta Crystallogr. Sect. B 2002, 58, 380–386.
(15) C. F. 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, 466–470.
(16) Etter, M. C.; Baures, P. W. J. Am. Chem. Soc. 1988,
110, 639–640.
(17) (a) Brock, C. P.; Schweizer, W. B.; Dunitz, J. D. J. Am. Chem.
Soc. 1985, 107, 6964–6970. (b) Spek, A. L. Acta Crystallogr., Sect. C:
Struct. Commun. 1987, 43, 1233–1235.(c) Lenstra, A. T. Private com-
munication refcode (TPEPHO12); Cambridge Crystallographic Data
Centre: Cambridge, England.
’ ASSOCIATED CONTENT
S
Supporting Information. X-ray crystallographic informa-
b
tion files CIF (1ꢀ3); H NMR and IR data for 1and 2(PDF);andcif,
pdf files, and co-crystal breakdown of the CSD searches. This material
is available free of charge via the Internet at http://pubs.acs.org.
’ AUTHOR INFORMATION
Corresponding Author
*Address: Molecular Sciences Institute, School of Chemistry, Uni-
versity of the Witwatersrand, Private Bag 3, 2050, Johannesburg,
South Africa. E-mail: Andreas.Lemmerer@wits.ac.za. Fax: þ27-11-
717-6749. Tel: þ27-11-717-6711.
Present Addresses
^Faculty of Natural Sciences, New York University Abu Dhabi, P.
O. Box 129188, Abu Dhabi, United Arab Emirates.
(18) Mei, X.; Wolf, C. Cryst. Growth Des. 2004, 4, 1099–1103.
(19) (a) Wei, K.-T.; Ward, D. L. Acta Crystallogr., Sect B 1977,
33, 797–800. (b) Fischer, A.; Profir, V. M. Acta Crystallogr., Sect. E 2003,
59, o1113–o1116.(c) Mughai, R. K.; Gillon, A. L.; Davey, R. J. Private
Communication, 2006, DLMAND03; Cambridge Crystallographic
Data Centre: Cambridge, England.
(20) Bꢁathori, N. B.; Lemmerer, A.; Venter, G. A.; Bourne, S. A.;
Caira, M. R. Cryst. Growth Des. 2011, 11, 75–87.
(21) Vishweshwar, P.; Nangia, A.; Lynch, V. M. CrystEngComm
2003, 5, 164–168.
(22) Aaker€oy, C. B.; Beatty, A. M.; Helfrich, B. A. J. Am. Chem. Soc.
2002, 124, 14425–14432.
’ ACKNOWLEDGMENT
This work was supported in part by Grant 2004118 from the
United StatesꢀIsrael Binational Science Foundation (Jerusalem).
A.L. thanks the South African National Research Foundation for a
postdoctoral scholarship (SFP2007070400002), Oppenheimer
Memorial Trust, and the Molecular Sciences Institute for funding.
’ REFERENCES
(1) (a) Burrows, A. D. Struct. Bonding (Berlin) 2004, 108, 55–96. (b)
Fan, E.; Vicent, C.; Geib, S. J.; Hamilton, A. D. Chem. Mater. 1994,
6, 1113–1117. (c) Desiraju, G. R. Chem. Commun. 1997, 1475–1482. (d)
Infantes, L.; Motherwell, W. D. S. Chem. Commun. 2004, 1166–1167.
2019
dx.doi.org/10.1021/cg2002145 |Cryst. Growth Des. 2011, 11, 2011–2019