Quantifying homo- and heteromolecular hydrogen bonds as a guide for adduct formation

Article abstract of DOI:10.1002/chem.201103129

An investigation into the predictability of molecular adduct formation is presented by using the approach of hydrogen bond propensity. Along with the predictions, crystallisation reactions (1a-1j) were carried out between the anti-malarial drug pyrimethamine (1) and the acids oxalic (a), malonic (b), acetylenedicarboxylic (c), adipic (d), pimelic (e), suberic (f), azelaic acids (g), as well as hexachlorobenzene (h), 1,4-diiodobenzene (i), and 1,4-diiodotetrafluorobenzene (j); seven (1a to 1g) of these successfully formed salts. Five of these seven salts were found to be either hydrated or solvated. Hydrogen bond propensity calculations predict that hydrogen bonds between 1 and acids a-g are more likely to form rather than the H bonds involved in self-association, providing a rationale for the observation of the seven new salts. In contrast, propensity of hydrogen bonds between 1 and h-j is much smaller as compared to other bonds predicted for self-association/solvate formation, in agreement with the observed unsuccessful reactions. Copyright

Full text of DOI:10.1002/chem.201103129

FULL PAPER
DOI: 10.1002/chem.201103129
Quantifying Homo- and Heteromolecular Hydrogen Bonds as a Guide for
Adduct Formation
Amit Delori,[a] Peter T. A. Galek,[b] Elna Pidcock,[b] and William Jones*[a]
Abstract: An investigation into the
predictability of molecular adduct for-
mation is presented by using the ap-
proach of hydrogen bond propensity.
Along with the predictions, crystallisa-
tion reactions (1a–1j) were carried out
between the anti-malarial drug pyri-
methamine (1) and the acids oxalic (a),
malonic (b), acetylenedicarboxylic (c),
adipic (d), pimelic (e), suberic (f), aze-
laic acids (g), as well as hexachloroben-
zene (h), 1,4-diiodobenzene (i), and
1,4-diiodotetrafluorobenzene (j); seven
(1a to 1g) of these successfully formed
salts. Five of these seven salts were
found to be either hydrated or solvat-
ed. Hydrogen bond propensity calcula-
tions predict that hydrogen bonds be-
tween 1 and acids a–g are more likely
to form rather than the H bonds in-
volved in self-association, providing
a rationale for the observation of the
seven new salts. In contrast, propensity
of hydrogen bonds between 1 and h–j
is much smaller as compared to other
bonds predicted for self-association/sol-
vate formation, in agreement with the
observed unsuccessful reactions.
Keywords: computer chemistry
·
cocrystallization · hydrogen bonds ·
knowledge-based predictions · phar-
maceutical materials
Introduction
successes or failures in salt or co-crystal formation in order
to suggest or prioritise experimental work more effectively.
Identifying when solvate formation would be the likely out-
come would also be beneficial.
Molecular adducts (e.g., salts and co-crystals) are considered
as alternatives when the physiochemical properties of
a parent drug molecule are unsuitable or inadequate for sat-
isfactory formulation.[1] Considerable variation in such phar-
maceutically important physical properties as solubility,[2]
dissolution rate,[2i] bioavailability,[3] melting point,[2h,4] stabili-
ty[5] and tableting properties[6] can be achieved by such salt
or co-crystal formation. The synthesis of pharmaceutical
salts and co-crystals can, however, be difficult, especially if
the drug molecule lacks strong hydrogen bond donor and/or
acceptor functionality.[7] Competition with solvate formation
is also a factor that can influence the outcome. Frequently,
supramolecular chemists rely on their experience and empir-
ical knowledge for the judicial choice of an appropriate
counter molecule or ion, and the Cambridge Structural Da-
tabase (CSD) can be queried for specific, informative, exam-
ples.
Molecular adducts involving multiple neutral species, so-
called co-crystals, are of particular current interest.[1a] A dif-
ficulty, however, is a generally low success rate of co-crystal-
lisation reactions making co-crystal screening a time-con-
suming process. Several strategies have been developed for
choosing appropriate counter molecules (coformers).
Hunter and co-workers have proposed a computational
method for co-crystal screening,[8] and Price and colleagues
proposed difference in lattice energy of adduct and reactants
as a criteria for predicting adduct formation.[9] Additionally,
Fꢀbiꢀn and co-workers proposed that coformers chosen
based upon similarity in shape descriptors can increase the
success of a co-crystallisation reaction.[10]
To date the supramolecular synthon approach is the most
widely used guide for the formation of salts or co-crystals.[11]
This approach suggests that molecules with complementary
functional groups can interact to form an adduct. This syn-
thon approach can, however, be less applicable in cases of
multi-functional molecules,[12] which are frequently of inter-
est to the pharmaceutical industry. The synthon approach is
qualitative in nature, takes into consideration only the com-
plementarity of the functional groups and neglects other im-
portant factors, such as steric effects,[13] which can prove
vital for the effectiveness of a given interaction. The Cam-
bridge Crystallographic Data Centre has developed a hydro-
gen bond propensity tool[13b,14] that can be applied to predict
the possibility of new polymorphs or adducts based on the
potential H bonds that might form. On providing informa-
tion of molecular structure, the software searches the CSD
Supramolecular chemists, especially those working with
complex drug molecules in the pharmaceutical industry,
would benefit from being able to predict the likelihood of
[a] Dr. A. Delori, Prof. W. Jones
Department of Chemistry, University of Cambridge
Cambridge, CB2 1EW (UK)
Fax : (+44)1223-336-017
E-mail: wj10@cam.ac.uk
[b] Dr. P. T. A. Galek, Dr. E. Pidcock
Cambridge Crystallographic Data Centre
12 Union Road, Cambridge, CB2 1EZ (UK)
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/chem.201103129.
Chem. Eur. J. 2012, 00, 0 – 0
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
1
&
ÞÞ
These are not the final page numbers!

for those molecules/salts/co-crystals that have similar func-
tional groups and quantifies the propensity of all the possi-
ble hydrogen bonds. It takes into consideration the chemical
environment of the donor/acceptor groups as well as other
important factors, such as competition[15] (as a function of
total counts of donor and acceptor atoms), aromaticity and
steric crowding.[13b]
At the outset of any attempted “co-crystallisation reac-
tion” it is not clear whether there will be any new phase. If
there is a new phase it could be: a co-crystal; a salt; a salt or
co-crystal solvate; a salt or co-crystal hydrate; or a hydrate
or solvate of the reactants. In the work reported here the
hydrogen bond propensity tool is applied to assess the likely
outcome of attempted co-crystallisations.[16] Using data in
the CSD, the hydrogen bond propensity method can be
equally applied to ionised or neutral species; all that is re-
quired is an appropriate description of the chemical groups
as topological objects (see the Experimental Section for ex-
amples). Its application here can be interpreted as making
an intelligent estimate of the success of supramolecular syn-
thon formation by extracting crystallographic knowledge
stored in the CSD. We explore the predictability of salt, co-
crystal or solvate formation using the hydrogen bond pro-
pensity tool. For this purpose we have chosen the anti-ma-
larial drug pyrimethamine (1) as our model compound
(Scheme 1).
Pyrimethamine is a challenging multifunctional molecule
with numerous hydrogen bond forming groups. In addition,
its low solubility in water makes the formation of molecular
adducts an attractive proposition. In pyrimethamine the
most basic site is N1 with a calculated pKa value of 6.94. It
is established that the pKa of the reactants plays a very im-
portant role in the formation of a salt or co-crystal. Depend-
ing on the DpKa (i.e., difference in pKa of the most basic
site on the base and most acidic site on the acid) either
a salt or a co-crystal can form.[2g,3a,18] Co-crystal formation is
expected for DpKa <0 and salt for DpKa >3. A DpKa value
in the region 0–3 is less informative as to whether salt or co-
crystal will form.[3a,19]
*
Scheme 1. DpKa is the difference in pKa of most basic atom on 1 and
Results and Discussion
most acidic atom on the complementary acid.[17] Throughout the study
the same labels are used for: 1) neutral or ionic state of a molecule, and
2) reaction and reaction products. If a reaction product is solvated, then
the presence of solvent is explicitly mentioned.
Our first hydrogen bond propensity study was performed as-
suming neutral moieties and therefore the likelihood of co-
crystal formation. The species (a–j) studied alongside 1 are
listed in Scheme 1. We found that 7 out of 10 experiments
resulted in salt formation. The DpKa was calculated for the
reactions 1a to 1g, which produced adducts. For 1a to 1c,
DpKa was found to be greater than 3 (Scheme 1), suggesting
the likelihood of formation of salts, which was confirmed by
determining their crystal structures from X-ray data (see
later). But for adducts 1d to 1g, the DpKa is between 0–3 in
which case either a salt or a co-crystal might be expected.
As we will show, however, experimentally 1d to 1g were
also found to be salts. Jones and co-workers have previously
reported the formation of various salts of pyrimethamine by
mechanochemistry.[20] The difference between a salt or co-
crystal product from these reactions is simply the migration
À
of a proton from the acidic COOH donor to a basic ac-
ceptor. Nevertheless, we subsequently carried out a second
hydrogen bond propensity study with charged species. The
results from both studies will be compared with the ob-
served hydrogen bonding patterns found in the synthesised
adducts.
To put the resulting reaction products into context it is
useful to study the crystal structure of pure pyrimethamine
&
2
&
www.chemeurj.org
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 0000, 00, 0 – 0
ÝÝ
These are not the final page numbers!

Homo- and Heteromolecular Hydrogen Bonds
FULL PAPER
(CSD reference code: MUFMAB) reported by Muthiah and
co-workers.[21] Pyrimethamine molecules recognise each
À
other by N H···N dimeric hydrogen bonds utilising the most
À
basic nitrogen at position 1 and the NH2 group attached to
the carbon at 2 position from one side (motif 1: M1) and the
À
less basic nitrogen at 3 position and NH2 group attached to
Scheme 2. a), b) Heterosynthon between aminopyridine and carboxylic
acid with and without proton transfer, respectively; c) aminopyridine-hal-
ogen bond.
carbon at position 4 on the other (motif 2: M2), forming
tapes. The adjacent tapes further interact with each other
utilising motif 3 (M3) forming sheets. Interaction between
the tapes leads to the formation of a hydrogen-bonded
system of three fused cyclic hydrogen-bonded ring motifs
(M3M2M3) and is labelled as motif 4 (M4), as shown in
Figure 1.
with solvent, determines whether the desired adduct would
be realised. This might be described as a competition be-
tween the hydrogen bond networks of pure reactants versus
potential alternatives enabled through an additional counter
molecule/ion or solvent. This first study was intended to pre-
dict the possible formation of co-crystals, and presents
ranked propensity predictions for all unique hydrogen
bonds that could form between neutral species with an indi-
cation of whether the interaction is between moieties of
1 only, or involves 1 and a–j. It is then possible to quickly
visualise and compare the possible H bond outcomes. In our
calculations we also considered the molecules of solvent
(methanol) and water.
Predicted propensities for the successful reactions, 1a to 1g:
Table 1 reveals that in all adducts 1a to 1g, the four top-
ranked predictions were between 1 and the acids suggesting
the high likelihood of co-crystal formation. Moreover, it is
interesting to note that since the functional groups involved
in 1a to 1g are the same, in all the cases the first two predic-
tions were between atom N7 as the donor and O9, O11 as
the acceptors. The next two were between N8 and O9/O11.
Since the propensity predictions between N7 and O9/O11
(propensity range of 0.90–0.92) are noticeably higher than
between N8 and O9/O11 (propensity 0.83–0.86), the model
suggests a higher probability of formation of a hydrogen
bond between N7 and O9/O11. The next four bonds of high-
est propensity are predicted between solvent and acid or 1,
suggesting high likelihood of formation of solvates. Indeed 5
of these 7 reactions have yielded solvates (1a, 1c, 1f, 1g crys-
tallised as methanol solvates and 1–b as hydrate). Since we
have used methanol as the solvent in all co-crystallisations
there is a greater chance of formation of a methanol solvate
than a hydrate. The source of the water (in hydrated adduct,
such as 1b) might be atmospheric or small amounts present
in the methanol.
Figure 1. a) The molecules of 1 interact with each other by motifs 1 and 2
forming tapes. b) Adjacent tapes interact with each other by motif 3
forming sheets. c) Motif 4 (M4) is formed by the interaction between the
adjacent tapes. Throughout this study the term M4 is used for all quadru-
ple hydrogen-bonded motifs of three fused hydrogen-bonded motifs
(M3M2M3), irrespective of the nature of acceptor A.
On crystallising 1 with a–g, the aminopyridine group of
1 can interact with acids either by aminopyridinium–carbox-
ylate or aminopyridine-carboxylic acid heterosynthon, yield-
ing salts and co-crystals, respectively. Conformers h–j, can
À
interact with 1 by N H···X hydrogen bonds (Scheme 2).
We attempted the crystallisation reaction of 1 with a–j.
Adduct formation was successful between 1 and a–g, and
unsuccessful for 1 and h–j. Hydrogen bond propensity calcu-
lations for co-crystallisation reactions were then compared
with the outcome of our experiments and the results are
summarised in Table 1. Our hypothesis was that the relative
likelihood of individual hydrogen bonds to form between
molecules of 1 or a–j only, that is, self-association, or involv-
ing both 1 and a–j, giving a molecular adduct, or indeed
Predicted propensities for the unsuccessful reactions, 1h to
1j: Encouraged by the success of the propensity model in
predicting co-crystal formation with 1, we aimed to check
the ability of the propensity model to discriminate against
less suitable conformers for 1. We planned crystallisation of
1 with h–j. Since in 1h to 1j, only single hydrogen bonds
À
(N H···X) are available as alternatives to the aminopyridine
and carboxylic acid homodimers of 1 and acids, respectively,
difficulties in synthesising possible adducts were to be ex-
Chem. Eur. J. 2012, 00, 0 – 0
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
&
3
&
ÞÞ
These are not the final page numbers!

W. Jones et al.
Table 1. Predicted and experimental results of crystallisation of 1 and a–j as neutral species in the presence of methanol (solvent used for co-crystallisa-
tions) and water.
Reaction
Numbering scheme
of model taken
Bonds of highest propensity[a]
Type of
Predicted
result
Experimental
result
donor
acceptor
propensity
interaction[b]
N7
N7
N8
N8
O14
O14
N7
O9
0.92
0.92
0.86
0.86
0.81
0.81
0.77
A
A
A
A
B
B
B
O11
O11
O9
O11
O9
co-crystal
formation
salt formed
(methanol solvate)
1a
O14
N7
N7
N8
N8
O14
O14
N7
O9
0.91
0.91
0.85
0.85
0.79
0.79
0.77
A
A
A
A
B
B
B
O11
O11
O9
O9
O11
O14
co-crystal
formation
salt formed
AHCTUNGTRENN(GUN hydrate)
1b
N7
N7
N8
N8
O14
O14
N7
O9
O11
O9
O11
O9
O11
O14
0.91
0.91
0.86
0.86
0.80
0.80
0.77
A
A
A
A
B
B
B
co-crystal
formation
salt formed
(methanol solvate)
1c
N7
N7
N8
N8
N7
O14
O14
O9
O11
O9
O11
O14
O11
O9
0.90
0.90
0.83
0.83
0.77
0.77
0.77
A
A
A
A
B
B
B
co-crystal
formation
salt formed
(non-solvated)
1d
N7
N7
N8
N8
N7
O14
O14
O9
O11
O9
O11
O14
O9
0.90
0.90
0.83
0.83
0.77
0.77
0.77
A
A
A
A
B
B
B
co-crystal
formation
salt formed
(non-solvated)
1e
O11
N7
N7
N8
N8
O14
O14
N7
O9
O11
O9
O11
O11
O9
0.90
0.90
0.83
0.83
0.77
0.77
0.77
A
A
A
A
B
B
B
co-crystal
formation
salt formed
(methanol solvate)
1f
O14
pected. In 1h to 1j, the bonds of highest propensity (first
two for 1h, 1j and first three for 1i) are between 1 and meth-
anol or water. The next bond (3rd for 1h, 1j and 4th for 1i)
is for the self-association of molecules of 1. Since the pro-
pensity of bonds for co-crystal formation (14th bond for 1h,
1i; propensity 0.41, 0.33, respectively) and 11th for 1j (pro-
pensity 0.41; see Tables S8–S10 in the Supporting Informa-
tion for full propensity calculations for 1h to 1j) is signifi-
cantly lower compared to those of highest propensity (pro-
pensity 0.79–0.87), we can conclude that the propensity
&
4
&
www.chemeurj.org
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 0000, 00, 0 – 0
ÝÝ
These are not the final page numbers!

Homo- and Heteromolecular Hydrogen Bonds
Table 1. (Continued)
FULL PAPER
Reaction
Numbering scheme
of model taken
Bonds of highest propensity[a]
Type of
Predicted
result
Experimental
result
donor
acceptor
propensity
interaction[b]
N7
N7
N8
N8
O14
O14
N7
O9
0.90
0.90
0.83
0.83
0.77
0.77
0.77
A
A
A
A
B
B
B
O11
O11
O9
O11
O9
co-crystal
formation
salt formed
(methanol solvate)
1g
O14
N7
N7
N7
N8
N8
N7
O14
O13
O14
N1
O13
O14
N3
0.87
0.87
0.76
0.76
0.76
0.65
0.63
B
B
C
B
B
C
B
co-crystal
did not form
1h
self-association[c]
self-association[c]
self-association[c]
O13
N7
N7
N8
N7
N8
O14
O14
O13
O14
O13
N1
O14
O13
O14
0.79
0.79
0.67
0.67
0.67
0.56
0.55
B
B
B
C
B
B
B
co-crystal
did not form
1i
N7
N7
N7
N8
N8
O14
O14
O14
O13
N1
O14
O13
O14
O13
0.84
0.83
0.73
0.72
0.71
0.63
0.62
B
B
C
B
B
B
B
co-crystal
did not form
1j
[a] A full list of propensity predictions is given in Tables S1–S10 in the Supporting Information. [b] A represents interaction between 1 and a–j, B repre-
sents interaction of 1, a–j or solvent with solvent, and C represents self-assembly of molecules of 1 or a–j. [c] Self-association of molecules predicted: cal-
culations also indicate the possibility for 1 crystallizing as solvate.
model suggests little chance of 1 interacting with h–j to
form a co-crystal (Table 1). It is also possible that a co-crys-
tal might form even without interaction between the two
components; the propensity calculations are, of course, lim-
ited in their use for predicting adducts formed in this way.
The bonds of the highest propensity calculated between
1 and solvent, strongly suggests the possible formation of
solvate of 1. Reactants of 1h to 1j did not interact with each
other to form co-crystals, as shown by the comparison of
PXRD plots (Table S18 in the Supporting Information). The
PXRD suggested that the reactants simply crystallised out
separately without the formation of solvate. An attempt to
obtain a methanol solvate of pyrimethamine by crystallising
it from methanol failed as indicated by the comparison of
the PXRD plot of the material so obtained with that of pure
pyrimethamine. However, a new material was obtained by
grinding pyrimethamine in the presence of a small amount
of methanol. The material was subsequently characterised as
a methanol solvate of 1 by single-crystal XRD (Table S20 in
the Supporting Information).
Hydrogen-bond propensity calculations for adducts 1a to 1g
based on charged species: It was encouraging to note that
the propensity calculations for co-crystal formation between
the neutral molecules of 1 and acids (a–g) predicted the for-
mation of co-crystals. Subsequent experiments, however, re-
vealed the products 1a to 1g to be salts as a result of the
transfer of a proton from the acid to the most basic nitrogen
(N1) of 1, creating a short, strong hydrogen bond of length
in the range 1.78–1.90 ꢂ in 1a to 1g. We, therefore, repeated
the propensity calculations for 1a to 1g, representing the
molecules using charged functional groups, as appropriate,
consistent with the observed crystal structures. On the basis
of this evidence, for routine employment of the method as
Chem. Eur. J. 2012, 00, 0 – 0
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
&
5
&
ÞÞ
These are not the final page numbers!

W. Jones et al.
a screening tool, it would be advisable to investigate models
for both neutral and charged adducts. As mentioned previ-
ously, a priori anticipation of product as a salt or co-crystal
is difficult, especially for co-crystallisations with DpKa in the
range 0–3.
Seven new propensity models were trained by using modi-
fied functional groups consistent with the ions of the ob-
served crystal structures (Scheme 2, a). We again considered
the presence of methanol and water. Our view was that our
initial expectations would be confirmed, because by adding
opposite charges on 1 and species a–g the chances for inter-
action should increase. The calculations supported our as-
sumption. On comparing the values of the bonds of highest
propensity in Table 2 with Table 1, it is clear that the com-
puted propensity values involving charged species increased.
In Table 2, we have given the first seven predictions (a full
list is given in Tables S11–S17 in the Supporting Informa-
tion). Analysis of these first seven predictions revealed that
in 1a to 1g the bonds of highest propensity are between cat-
ionic 1 and acid anions, and hence our model successfully
predicts the formation of salts between 1 and a–g.
Crystal structures of the observed reaction products: Forma-
tion of adduct 1a to 1g was confirmed by single-crystal
XRD. The materials were further characterised by using
thermal techniques, such as thermogravimetric analysis
(TGA) and differential scanning calorimetry (DSC;
Table S19 in the Supporting Information)
Salt between pyrimethamine and oxalic acid, 1a: Salt 1a was
obtained as a methanol solvate on crystallising 1 and a from
a 1:1 mixture in methanol. In the salt, the ions of 1 and a rec-
ognise each other and form molecular tapes (Figure 2).
There are two types of ions of pyrimethaminium (labelled A
and B) and of deprotonated oxalic acid (labelled C (mono-
deprotonated) and D (doubly deprotonated), respectively).
In the tape network of 1a, the most basic protonated nitro-
Figure 2. Tape network observed in 1a. M5 and M6 are the motifs 5 and
6, respectively.
À
gen (N1) and the NH2 group at adjacent carbon (C2) of
both A and B type of cations of pyrimethaminium recognise
the complementary carboxylate group on anions C and D,
respectively, by utilising hydrogen-bonded motif 5 (M5). A
À
and B recognise each other by N H···N dimeric hydrogen-
bonded motif 6 (M6) with H···N distance of 2.12 and 2.51 ꢂ
À
by using nitrogen at the 3 position and NH2 group attached
to carbon at 4 and 2 position of A and B, respectively. It is
noteworthy that motif 6 was not observed in the structure of
À
À
1. In addition in the tape network the O H···O interactions
(H···OÀ distance 1.73) were also observed between the C
and D. The oxygen atom of methanol interacts with pyrime-
À
thaminium in the same layer by N H···O hydrogen bonds
(H···O distances 2.09, 2.11, 2.12 ꢂ).
Figure 3. Recognition pattern as observed in 1b.
Salt between pyrimethamine and malonic acid, 1b: In 1b,
the ions of 1 and b recognise each other to form molecular
tapes (Figure 3). In the salt there are two types of molecular
ions of pyrimethaminium (labelled A and B) and deproton-
ated malonic acid (labelled C and D; both monodeprotonat-
ed). As observed in 1a, A and B recognise the complemen-
tary carboxylate group of C and D, respectively, by motif 5
(M5) with H···OÀ distances 1.81, 1.82 and 1.95, 1.97 ꢂ. A
and B recognise each other by motif 2 (M2) with H···N dis-
À
tance of 2.21 and 2.33 ꢂ. Whereas the OH of the carboxyl-
&
6
&
www.chemeurj.org
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 0000, 00, 0 – 0
ÝÝ
These are not the final page numbers!

Homo- and Heteromolecular Hydrogen Bonds
FULL PAPER
Table 2. Predicted and experimental results of crystallisation of 1 and a–g as charged species.
Adduct
Numbering scheme
of model taken
Bonds of highest propensity[a]
Type of
Predicted
result
Experimental
result
donor
acceptor
propensity
interaction[b]
N7
N7
N8
N7
N1
O14
O13
O10
O9
1.00
0.99
0.99
0.98
0.98
0.98
0.97
A
A
A
A
A
B
O10
O11
O10
O10
O10
salt formed
(methanol solvate)
1a
salt formation
salt formation
salt formation
B
N7
N7
N8
N7
N1
O14
O13
O10
O9
1.00
0.99
0.99
0.98
0.98
0.98
0.97
A
A
A
A
A
B
O10
O11
O10
O10
O10
salt formed
AHCTUNGTRENN(GUN hydrate)
1b
B
N7
N7
N7
N7
N8
N8
N1
O10
O12
O9
O11
O10
O12
O10
1.00
1.00
0.99
0.99
0.99
0.99
0.98
A
A
A
A
A
A
A
salt formed
(methanol solvate)
1c
N7
N7
N8
N1
N7
O14
O13
O10
O9
1.00
0.99
0.99
0.98
0.98
0.98
0.96
A
A
A
A
A
B
O10
O10
O11
O10
O10
salt formed
(non-solvated)
1d
salt formation
B
N7
N7
N8
N1
N7
O14
O13
O10
O9
1.00
0.99
0.99
0.98
0.98
0.98
0.96
A
A
A
A
A
B
O10
O10
O11
O10
O10
salt formed
(non-solvated)
1e
salt formation
B
N7
N7
N7
N7
N8
N8
N1
O10
O12
O9
O11
O10
O12
O10
1.00
1.00
0.99
0.99
0.99
0.99
0.98
A
A
A
A
A
A
A
salt formed
(methanol solvate)
1f
salt formation
N7
N7
N8
N1
N7
O14
O13
O10
O9
1.00
0.99
0.99
0.98
0.98
0.98
0.96
A
A
A
A
A
B
O10
O10
O11
O10
O10
salt formed
(methanol solvate)
1g
salt formation
B
[a] A full list of propensity predictions is given in Tables S11–S17 in the Supporting Information. [b] A represents interaction between 1 and a–g, and B
represents interaction of 1, a–g or solvent with solvent.
Chem. Eur. J. 2012, 00, 0 – 0
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
&
7
&
ÞÞ
These are not the final page numbers!

W. Jones et al.
ic acid group of monodeprotonated malonic acid was found
À
À
to be involved in an intramolecular O H···O hydrogen
bond (H···OÀ distance 1.52, 1.56 ꢂ) the oxygen atoms of the
À
carboxylic acid formed N H···O hydrogen bonds (H···O dis-
tances 1.98, 2.17, 2.23, 2.38 ꢂ) with ions of 1.
Salt between pyrimethamine and acetylenedicarboxylic acid,
1c: Salt 1c was obtained as a methanol solvate on crystallis-
ing 1 and c from a 1:1 mixture in methanol. As observed in
1a and 1b, in 1–c also, we note that ions of 1 recognise the
carboxylate group of the deprotonated acid by motif 5 (Fig-
ure 4a). The pyrimethaminium cations recognise each other
Figure 5. a) Recognition pattern observed in 1d. b) Interactions observed
between the adjacent two layers.
À
group at C4 of 1 by N H···O hydrogen bonds with H···O
À
distance 2.04, 2.22 ꢂ, the OH group of carboxylic acid
functional group of d of one layer recognises the adjacent
layer by O H···O interactions (H···OÀ distance 1.68 ꢂ), as
À
À
shown in Figure 5b.
Salt between pyrimethamine and pimelic acid, 1e: In the 1:1
salt obtained between 1 and e, oppositely charged ions rec-
ognise each other by motif 5 (Figure 6a). The pyrimethami-
Figure 4. a) Tape network observed in 1c. b) Interactions observed be-
tween adjacent tapes.
by motif 2, as also observed in the crystal structure of 1 and
1b. The adjacent tapes further interact with each other by
À
N H···O hydrogen bonds with H···OÀ distance of 2.09 ꢂ
À
(Figure 4b). Whereas the oxygen atom of C and D mole-
À
cules of methanol interact with 1 by N H···O hydrogen
bonds (H···O distances 2.06, 2.20 ꢂ), the hydroxyl group of
À
C and D forms O H···O (H···OÀ distances 1.87 ꢂ) and O
H···O (H···O distances 1.83 ꢂ) hydrogen bonds with depro-
tonated acid and C respectively (Figure 4a).
À
À
Salt between pyrimethamine and adipic acid,[22] 1d: The
tape network observed in 1–d is remarkably different from
1a to 1c in the absence of interactions between the pyrime-
thaminium cations (Figure 5a). But as observed in 1a to 1c,
in 1d also the ions of 1 recognised carboxylate group of
À
anions of acid by motif 5. The NH2 group at C2 of 1 is also
À
À
involved in N H···O hydrogen bonds with carboxylate
group of another molecule of deprotonated adipic acid with
H···OÀ distance 2.21 ꢂ. Whereas the carbonyl group of car-
Figure 6. a) Tape network observed in 1–e. b) Interaction between the ad-
jacent tapes.
À
boxylic acid functional group of d interacts with NH2
&
8
&
www.chemeurj.org
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 0000, 00, 0 – 0
ÝÝ
These are not the final page numbers!

Homo- and Heteromolecular Hydrogen Bonds
FULL PAPER
nium cations recognise each other by motif 2 as also ob-
served in the crystal structure of 1, 1b and 1c. Whereas the
À
OH group of carboxylic acid functional group of e is found
À
to be involved in a O H···O hydrogen bond (H···OÀ dis-
À
tance 1.61 ꢂ) with another molecule of e in the same layer,
À
the carbonyl of carboxylic acid functional group forms N
H···O hydrogen bonds (H···O distance of 2.16, 2.22 ꢂ) with
À
the NH2 groups of molecules of 1 in adjacent layers, form-
ing motif 3. Interaction of a tape (shown in red in the Fig-
ure 6b) with the acid molecules in the layer above and
below (shown in blue and green in the Figure 6b) leads to
the formation of motif 4 (M4).
Salt between pyrimethamine and suberic acid, 1f: In the 2:1
adduct observed between 1 and f, the molecules of suberic
acid are doubly deprotonated. The pyrimethaminium cations
interact with each other by motif 2 forming dimers
À
(Figure 7). Further the NH2 group at C2 and C4 of 1, inter-
Figure 8. a) Tape network observed in 1g. b) interactions observed be-
tween the adjacent tapes.
Conclusions
Crystallisation reactions 1a to 1j were carried out between
pyrimethamine and various counter molecules (a–j), of
which 1a to 1g were successful in forming adducts. Hydro-
gen bond propensity calculations were performed along with
the experimental work to validate the use of this tool in pre-
dicting the formation of salts/co-crystals. Calculations cor-
rectly predicted the formation and non-formation of adducts
in 1a to 1g and 1h to 1j, respectively. All seven successful
adducts formed were found to be salts. Predictions were
made both assuming neutral species forming co-crystals and
charged species forming salts, with comparable results.
Though propensity calculations correctly predicted the
outcome of crystallisation reactions 1a to 1j, to verify the
general applicability of these calculations in predicting the
formation of adducts, comparison of predicted and experi-
mental results with molecules containing a wider range of
functional groups is required. This work is currently under-
way, with encouraging preliminary results. It is worth noting
that as the propensity calculations make use of the CSD, the
predictions will continue to grow in accuracy with the addi-
tion of more relevant crystallographic data.
Figure 7. Recognition pattern observed in 1f.
acted with an oxygen atom of the methanol molecules (la-
belled A) by motif 3 forming tetramer. It also leads to the
formation of motif 4 (M4), as also observed in the crystal
structure of 1 and 1e. The deprotonated acid molecules con-
nect the adjacent tetramers by motif 5. In addition, the
methanol molecules (labelled B and C) interact with the car-
À
À
boxylate oxygen of the acid by O H···O hydrogen bonds
(H···OÀ distances 1.81 and 1.86 ꢂ).
Salt between pyrimethamine and azelaic acid, 1g: In 1g, ions
of 1 recognise the carboxylate group of the deprotonated
aliphatic acid by motif 5 (Figure 8a). The pyrimethaminium
cations recognised each other by motif 2, as also observed in
the crystal structure of 1 and 1b, 1c, 1e and 1f. While the
carbonyl group of carboxylic acid functional group of azelaic
À
acid is involved in N H···O hydrogen bonds, forming
motif 3 (H···O distances 2.21, 2.31 ꢂ), the hydroxyl group of
Experimental Section
the carboxylic acid functional group recognises adjacent
layers by O H···O hydrogen bonds (H···OÀ distance
À
À
General: All the chemicals used in this study were obtained from Aldrich
and were used without further purification. The solvents employed for
the crystallisations were of spectroscopy grade of highest available purity.
All the crystallisation reactions were completed by dissolving 1 and a–j
in 1:1 ratio in a CH3OH solvent and slow evaporation of the obtained so-
lution. Single crystals of the salts 1a to 1g were obtained over a period of
48 h. In a typical preparation, 1 (0.1244 g; 0.5 mmol) and adipic acid
1.74 ꢂ), as shown in Figure 8b. In 1g motif 4 (M4) is also
observed, as in case of 1, 1e and 1f.
Chem. Eur. J. 2012, 00, 0 – 0
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
&
9
&
ÞÞ
These are not the final page numbers!

W. Jones et al.
(0.0731 g; 0.5 mmol) were dissolved in CH3OH (15 mL) by being gently
warmed on a hot plate. The resultant solution was kept for evaporation
at ambient conditions by protecting the conical flask from external me-
chanical disturbances. Within 48 h, colourless and good-quality crystals of
1d, were obtained that were suitable for studies by single crystal X-ray
diffraction methods. As reactions 1h to 1j were unable to give adducts
from solution, we also tried to form adducts by liquid-assisted-grinding,[23]
but LAG also failed to yield adducts.
CCDC 846669–846675 and 859637 contain the supplementary crystallo-
graphic data for this paper. These data can be obtained free of charge
from The Cambridge Crystallographic Data Centre via www.ccdc.cam.a-
c.uk/data_request/cif.
Hydrogen bond propensity prediction: Predictive models were prepared
by using a development version of Mercury 3.0 (pre-release)[16] and the
CSD v.5.32 (November, 2010). The neutral models used functional
groups as displayed in Scheme 3a–h and the charged models used groups
as shown in Scheme 3i–o. In total ten neutral models and seven charged
models were prepared for molecule 1 and co-formers a–j and a–g, respec-
tively. For each model roughly 1000 CSD structures were used as the
data source to fit the model (the least being 955 for model 1h and the
most being 1407 for model 1b; see Table S21 in the Supporting Informa-
tion for details). In these calculations functional groups of water and
methanol were also considered. Acceptance criteria for the propensity
models were that each functional group is represented in >250 struc-
tures. The minimum predictivity observed was >78%.
A methanol solvate of pyrimethamine was obtained by LAG and was
structurally characterised by single crystal XRD. The crystals of methanol
solvate of 1 were obtained by seeding methanol solution of 1 with seeds
obtained by LAG. All the LAG experiments were carried by grinding the
reactants (0.5 mmol) at 30 Hz for 30 min in the presence of methanol
(30 mL).
Crystal structure determination of 1a to 1g: Good quality single crystals
of 1a to 1g were chosen by viewing under microscope and glued to
a glass fibre and mounted on a goniometer of a Bruker single-crystal X-
ray diffractometer equipped with APEX CCD detector. The data collec-
tions were smooth in all the cases without any complications and all the
crystals were found to be stable throughout data collection period. The
intensity data were processed by using Bruker suite of programs,[24]
SAINT, followed by absorption correction by SADABS.[24] The structures
were solved using SHELXS and refined by least-square methods with
SHELXL.[24] All the non-hydrogen atoms were refined by anisotropic
methods and the hydrogen atoms were either refined or placed in the cal-
culated positions. All the structural refinements converged to good R fac-
tors (Table 3) and the intermolecular interactions were computed by
using PLATON software,[25] and are given in Table 4. The packing dia-
grams were generated by using Diamond, version 3.1e.[26]
Acknowledgements
We thank the Pfizer Institute for Pharmaceutical Materials Science for
funding (Fellowship for A.D.). We acknowledge Dr. Neil Feeder (Pfizer
Global R&D) and Dr. Colin Groom (CCDC) for helpful discussions. The
CCDC is thanked for providing a new version of the Mercury software
for the propensity calculations; Dr. James Chisholm for discussions re-
garding the use of these propensity calculations and Dr. John E. Davies
for collecting single-crystal X-ray data for the reported adducts.
Table 3. Crystallographic data for the adducts 1a to 1g.
1a
1b
1c
1d
1e
1f
1g
formula
2(C12H14N4Cl1): 2(C12H14N4Cl1): 2(C12H14N4Cl1): C12H14N4Cl1: C12H14N4Cl1:
C12H14N4Cl1:
0.5(C8H12O4):
3(CH4O)
C12H14N4Cl1:
C9H15O4:
CH4O
C2H1O4:
0.5(C2O4):
2(CH4O)
696.57
2(C3H3O4):
H2O
C4O4:
2(CH4O)
C6H9O4
C7H11O4
Mr
723.57
675.57
394.85
408.88
431.94
468.97
crystal habit
crystal colour
crystal system
space group
a [ꢂ]
b [ꢂ]
c [ꢂ]
a [8]
b [8]
blocks
blocks
needles
colourless
monoclinic
P21/c
13.866 (2)
7.0299 (1)
36.562(5)
90.00
109.03 (1)
90.00
3369.22(8)
4
1.332
180(2)
0.71073
0.246
50.02
rod-shaped
colourless
triclinic
rectangular blocks rectangular blocks rods
colourless
triclinic
colourless
triclinic
colourless
triclinic
P1
colourless
monoclinic
P21/c
colourless
triclinic
P1
¯
¯
¯
¯
¯
P1
P1
P1
10.545(2)
12.183(2)
14.248(3)
74.55(1)
80.24(1)
69.57(1)
1647.23(5)
2
1.404
180(2)
0.71073
0.259
64.12
9.377(2)
11.746(2)
16.110(3)
72.40(1)
85.34(1)
78.11(1)
1654.77(5)
2
1.452
180(2)
0.71073
0.263
55.78
8.080(1)
11.174(2)
12.100(2)
88.65(1)
70.51(1)
72.03(1)
975.79(3)
2
6.925(2)
11.188(3)
13.429(3)
92.83 (1)
92.21(1)
97.63 (1)
1028.77(5)
2
9.346(1)
15.398(2)
18.060(2)
90.00
113.70(1)
90.00
6.604 (2)
13.278(4)
15.205(5)
114.56 (2)
92.63 (2)
95.44 (2)
1201.76(6)
2
G [8]
V [ꢂ3]
Z
2379.65(5)
4
1calcd [gcmÀ3
T [K]
]
1.344
1.320
1.206
1.296
180(2)
0.71073
0.227
180(2)
180(2)
180(2)
l (MoKa
m [mmÀ1
2q range [8]
limiting indices
)
0.71073
0.218
0.71073
0.195
0.71073
0.198
]
58.26
55.74
54.96
55.66
À15ꢀhꢀ15
À18ꢀkꢀ18
À17ꢀlꢀ21
730
À12ꢀhꢀ12
À15ꢀkꢀ15
À21ꢀlꢀ21
756
À16ꢀhꢀ16
À8ꢀkꢀ8
À43ꢀlꢀ43
1416
À10ꢀhꢀ11 À9ꢀhꢀ9
À15ꢀkꢀ15 À14ꢀkꢀ14
À16ꢀlꢀ16 À17ꢀlꢀ17
À12ꢀhꢀ12
À19ꢀkꢀ19
À23ꢀlꢀ23
924
À8ꢀhꢀ8
À17ꢀkꢀ17
À18ꢀlꢀ19
500
F
N
416
432
reflns measured
unique reflns
reflns used
31494
11422
6180
442
0.955
0.0599
0.1191
0.065/À0.598
27346
7868
5577
444
1.023
0.0538
0.1300
1.063/À0.505
30778
5920
4750
419
1.024
0.0443
0.1086
0.337/À0.475
19164
5238
3879
248
1.066
0.0494
0.1221
15131
4878
3741
257
1.034
0.0569
0.1504
26096
5437
4582
266
1.064
0.0480
0.1325
0.425/À0.416
11961
5629
2843
296
0.912
0.0499
0.1012
0.193/À0.259
parameters
GOF on F2
R1 [I>2s(I)]
wR2
largest diff. peak/hole [eÀ ꢂÀ3
]
0.310/À0.328 0.498/À0.523
&
10
&
www.chemeurj.org
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 0000, 00, 0 – 0
ÝÝ
These are not the final page numbers!

Homo- and Heteromolecular Hydrogen Bonds
FULL PAPER
Table 4. Bond length and angles of the hydrogen bonds observed in adducts 1a to 1g.
Hydrogen bond
1a
1b
1c
1d
1e
1f
1g
À
D H···A
2.12 2.98 165 2.21 3.08 169 2.10 2.97 171
2.51 3.33 155 2.33 3.18 162 2.22 3.10 175
2.11 2.97 166 2.18 3.06 172 2.14 3.02 172
À
N H···N
2.04 2.91 169 1.98 2.83 160 2.06 2.89 156 2.04 2.91 168 2.16 3.02 166 2.05 2.91 168 2.21 2.94 140
2.09 2.93 160 2.17 2.83 132 2.20 2.91 137 2.22 2.93 137 2.22 2.91 135 2.17 2.84 133 2.31 3.17 166
2.11 2.93 155 2.23 3.04 152
À
N H···O
2.12 2.88 145 2.38 2.99 127
2.14 2.94 152 1.95 2.82 170 1.92 2.80 176 2.04 2.91 167 1.95 2.83 175 1.97 2.80 157 2.06 2.93 177
2.22 3.06 160 1.97 2.85 175 1.93 2.80 172 2.21 2.88 133
À
À
N H···O
2.09 2.85 145
1.83 2.71 173 1.81 2.71 171 1.81 2.69 178 1.80 2.68 177 1.78 2.65 169 1.86 2.72 174 1.81 2.68 170
1.85 2.73 178 1.82 2.73 172 1.90 2.78 176
À
N+ H···O
À
À
O H···O
1.83 2.73 169
1.91 2.70 168
1.74 2.64 170 1.52 2.47 162 1.87 2.71 169 1.68 2.55 167 1.61 2.49 172 1.81 2.66 176 1.74 2.59 174
À
À
O H···O
1.92 2.75 168 1.56 2.48 158
1.93 2.77 169
1.86 2.72 174 1.88 2.77 176
À
C H···O
2.57 3.34 139 2.28 3.20 160
2.53 3.49 161
À
À
C H···O
2.56 3.43 152 2.52 3.39 152 2.52 3.39 151
2.79 3.48 128
À
C H···Cl
ju, Chem. Commun. 2005, 1073; c) J. Galcera, E. Molins, Cryst.
Growth Des. 2009, 9, 327; d) R. Banerjee, P. M. Bhatt, N. V. Ravin-
dra, G. R. Desiraju, Cryst. Growth Des. 2005, 5, 2299; e) D. J. Good,
N. Rodrꢃguez-Hornedo, Cryst. Growth Des. 2010, 10, 1028; f) S. J.
Bethune, N. Huang, A. Jayasankar, N. Rodrꢃguez-Hornedo, Cryst.
Growth Des. 2009, 9, 3976; g) C. L. Cooke, R. J. Davey, Cryst.
Growth Des. 2008, 8, 3483; h) C. B. Aakerçy, S. Forbes, J. Desper, J.
Am. Chem. Soc. 2009, 131, 17048; i) R. Thakuria, A. Nangia,
CrystEngComm 2011, 13, 1759; j) S. Ghosh, P. P. Bag, C. M. Reddy,
Cryst. Growth Des. 2011, 11, 3489.
[3] a) M. L. Cheney, N. Shan, E. R. Healey, M. Hanna, L. Wojtas, M. J.
Zaworotko, V. Sava, S. J. Song, J. R. Sanchez-Ramos, Cryst. Growth
Des. 2010, 10, 394; b) M. B. Hickey, M. L. Peterson, L. A. Scoppet-
tuolo, S. L. Morrisette, A. Vetter, H. Guzmꢀn, J. F. Remenar, Z.
Zhang, M. D. Tawa, S. Haley, M. J. Zaworotko, O. Almarsson, Eur.
J. Pharm. Biopharm. 2007, 67, 112; c) D. P. McNamara, S. L. Childs,
J. Giordano, A. Iarriccio, J. Cassidy, M. S. Shet, R. Mannion, E.
OꢅDonnell, A. Park, Pharm. Res. 2006, 23, 1888.
[4] a) R. D. B. Walsh, M. W. Bradner, S. Fleischman, L. A. Morales, B.
Moulton, N. Rodrꢃguez-Hornedo, M. J. Zaworotko, Chem. Commun.
2003, 186; b) D. Braga, E. Dichiarante, G. Palladino, F. Grepioni,
M. R. Chierotti, R. Gobetto, L. Pellegrino, CrystEngComm 2010, 12,
3534.
[5] a) A. V. Trask, W. D. S. Motherwell, W. Jones, Cryst. Growth Des.
2005, 5, 1013; b) A. V. Trask, W. D. S. Motherwell, W. Jones, Int. J.
Pharm. 2006, 320, 114.
Scheme 3. Hydrogen bond definitions used for predicting: a)–h) various
co-crystals, and i)–o) salts of pyrimethamine. A=cyclic bonds with varia-
ble bond type (single, double, aromatic, delocalised); B=aromatic; C=
number of bonded atoms is 3.
ˇˇ ´
[6] a) S. Karki, T. Friscic, L. Fabian, P. R. Laity, G. M. Day, W. Jones,
Adv. Mater. 2009, 21, 3905; b) C. M. Reddy, K. A. Padmanabhan,
G. R. Desiraju, Cryst. Growth Des. 2006, 6, 2720.
ˇˇ ´
[7] a) S. Karki, T. Friscic, L. Fꢀbiꢀn, W. Jones, CrystEngComm 2010, 12,
[1] a) S. G. Fleischman, S. S. Kuduva, J. A. McMahon, B. Moulton,
R. D. B. Walsh, N. Rodrꢃguez-Hornedo, M. J. Zaworotko, Cryst.
Growth Des. 2003, 3, 909; b) P. Vishweshwar, J. A. McMahon, M. L.
Peterson, M. B. Hickey, T. R. Shattock, M. J. Zaworotko, Chem.
Commun. 2005, 4601; c) S. N. Black, E. A. Collier, R. J. Davey, R. J.
Roberts, J. Pharm. Sci. 2007, 96, 1053; d) K. K. Arora, N. G. Tayade,
4038; b) S. R. Perumalla, E. Suresh, V. R. Pedireddi, Angew. Chem.
2005, 117, 7930; Angew. Chem. Int. Ed. 2005, 44, 7752.
[8] a) D. Musumeci, C. A. Hunter, R. Prohens, S. Scuderi, J. F. McCabe,
Chem. Sci. 2011, 2; 883; b) C. A. Hunter, Angew. Chem. 2004, 116,
5424; Angew. Chem. Int. Ed. 2004, 43, 5310; c) C. B. Aakerçy, J.
Desper, M. M. Smith, Chem. Commun. 2007, 3936; d) T. S. Thakur,
G. R. Desiraju, Cryst. Growth Des. 2008, 8, 4031.
[9] N. Issa, P. G. Karamertzanis, G. W. A. Welch, S. L. Price, Cryst.
Growth Des. 2009, 9, 442.
[10] a) L. Fꢀbiꢀn, Cryst. Growth Des. 2009, 9, 1436; b) J. Fayos, Cryst.
Growth Des. 2009, 9, 3142.
[11] a) G. R. Desiraju, Angew. Chem. 1995, 107, 2541; Angew. Chem. Int.
Ed. Engl. 1995, 34, 2311; b) S. Varughese, Y. Azim, G. R. Desiraju,
J. Pharm. Sci. 2010, 99, 3743; c) R. Banerjee, B. K. Saha, G. R. De-
siraju, CrystEngComm 2006, 8, 680; d) B. Sarma, N. K. Nath, B. R.
ˇˇ ´
R. Suryanarayanan, Mol. Pharm. 2011, 8, 982; e) T. Friscic, W.
Jones, J. Pharm. Pharmacol. 2010, 62, 1547; f) J. R. G. Sander, D.-K.
ˇ
Bucar, R. F. Henry, G. G. Z. Zhang, L. R. MacGillivray, Angew.
Chem. 2010, 122, 7442; Angew. Chem. Int. Ed. 2010, 49, 7284;
g) J. A. Foster, M.-O. M. Piepenbrock, G. O. Lloyd, N. Clarke,
J. A. K. Howard, J. W. Steed, Nat. Chem. 2010, 2, 1037.
[2] a) J. F. Remenar, S. L. Morissette, M. L. Peterson, B. Moulton, J. M.
MacPhee, H. R. Guzmꢀn, ꢄ. Almarsson, J. Am. Chem. Soc. 2003,
125, 8456; b) P. M. Bhatt, N. V. Ravindra, R. Banerjee, G. R. Desira-
Chem. Eur. J. 2012, 00, 0 – 0
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
&
11
&
ÞÞ
These are not the final page numbers!

W. Jones et al.
Bhogala, A. Nangia, Cryst. Growth Des. 2009, 9, 1546; e) T. R. Shat-
tock, K. K. Arora, P. Vishweshwar, M. J. Zaworotko, Cryst. Growth
Des. 2008, 8, 4533.
Rajbanshi, Z. J. Li, J. Desper, CrystEngComm 2010, 12, 4231;
c) C. B. Aakerçy, M. E. Fasulo, J. Desper, Mol. Pharm. 2007, 4, 317.
[20] a) A. V. Trask, D. A. Haynes, W. D. S. Motherwell, W. Jones, Chem.
Commun. 2006, 51; b) V. Sethuraman, N. Stanley, P. T. Muthiah,
W. S. Sheldrick, M. Winter, P. Luger, M. Weber, Cryst. Growth Des.
2003, 3, 823; c) N. Stanley, V. Sethuraman, P. T. Muthiah, P. Luger,
M. Weber, Cryst. Growth Des. 2002, 2, 631.
[21] V. Sethuraman, P. T. Muthiah, Acta Crystallogr. Sect. E 2002, 58,
o817.
[22] P. Devi, P. T. Muthiah, T. N. Guru Row, V. Thiruvenkatam, Acta
Crystallogr. Sect. E 2007, 63, o4065.
[23] N. Shan, F. Toda, W. Jones, Chem. Commun. 2002, 2372.
[24] a) Siemens, SMART System, Siemens Analytical X-ray Instruments
Inc., Madison, WI (USA), 1995; b) G. M. Sheldrick, Siemens Area
Detector Absorption Correction Program (SADABS), University of
Gçttingen, Gçttingen, Germany, 1994; c) G. M. Sheldrick,
SHELXTL-PLUS program for crystal structure solution and refine-
ment, University of Gçttingen, Gçttingen, Germany.
[12] P. M. Bhatt, Y. Azim, T. S. Thakur, G. R. Desiraju, Cryst. Growth
Des. 2009, 9, 951.
[13] a) J. Chrisholm, E. Pidcock, J. van de Streek, L. Infantes, S. Mother-
well, F. H. Allen, CrystEngComm 2006, 8, 11; b) P. A. Wood, P. T. A.
Galek, CrystEngComm 2010, 12, 2485.
[14] a) P. T. A. Galek, L. Fabian, W. D. S. Motherwell, F. H. Allen, N.
Feeder, Acta Crystallogr. Sect. B 2007, 63, 768; b) P. T. A. Galek,
F. H. Allen, L. Fabian, N. Feeder, CrystEngComm 2009, 11, 2634;
c) P. T. A. Galek, L. Fabian, F. H. Allen, CrystEngComm 2010, 12,
2091; d) P. T. A. Galek, L. Fabian, F. H. Allen, Acta Crystallogr. Sect.
B 2010, 66, 237; e) P. T. A. Galek, E. Pidcock, P. A. Wood, I. J.
Bruno, C. R. Groom, CrystEngComm 2012, 14, 2391.
[15] L. Infantes, W. D. S. Motherwell, Z. Kristallogr. 2005, 220, 333.
[16] Propensity calculations were carried out with a newer version of
mercury software (mercury 2.4.1cfc), which is under development
by CCDC. For these calculations conquest version 1.13 was used as
database.
[25] A. L. Spek, PLATON, molecular geometry program, University of
Utrecht, The Netherlands, 1995.
[17] The pKa value of pyrimethamine was calculated by using sparc
online calculator http://sparc.chem.uga.edu/sparc/. The pKa values of
acids were obtained from: pKa Data Compiled by R. Williams.
[18] a) B. R. Bhogala, S. Basavoju, A. Nangia, CrystEngComm 2005, 7,
551; b) A. Delori, E. Suresh, V. R. Pedireddi, Chem. Eur. J. 2008, 14,
6967.
[26] Diamond–Crystal and Molecular Structure Visualization, Ver-
sion 3.1f, Crystal Impact, Brandenburg & Putz, Bonn, 2008.
Received: October 4, 2011
Revised: February 1, 2012
Published online: && &&, 2012
[19] a) M. L. Cheney, D. R. Weyna, N. Shan, M. Hanna, L. Wojtas, M. J.
Zaworotko, Cryst. Growth Des. 2010, 10, 4401; b) C. B. Aakerçy, A.
&
12
&
www.chemeurj.org
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 0000, 00, 0 – 0
ÝÝ
These are not the final page numbers!

Homo- and Heteromolecular Hydrogen Bonds
FULL PAPER
Ready, steady, go! Chemists working
on drug molecules would benefit from
being able to predict the likelihood of
successes or failures in salt or co-crys-
tal formation to prioritise experimental
work. We have used the hydrogen
bond propensity method (see figure)
to predict the success of ten molecular
adduct crystallisation reactions. The
results show good agreement with the
experimental observations.
Hydrogen Bonds
A. Delori, P. T. A. Galek, E. Pidcock,
W. Jones* ...................... &&&& — &&&&
Quantifying Homo- and Heteromolec-
ular Hydrogen Bonds as a Guide for
Adduct Formation
Chem. Eur. J. 2012, 00, 0 – 0
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
&
13
&
ÞÞ
These are not the final page numbers!