Insights into the crystallisation process from anhydrous, hydrated and solvated crystal forms of diatrizoic acid

Article abstract of DOI:10.1002/chem.201404693

Diatrizoic acid (DTA), a clinically used X-ray contrast agent, crystallises in two hydrated, three anhydrous and nine solvated solid forms, all of which have been characterised by X-ray crystallography. Single-crystal neutron structures of DTA dihydrate and monosodium DTA tetrahydrate have been determined. All of the solid-state structures have been analysed using partial atomic charges and hardness algorithm (PACHA) calculations. Even though in general all DTA crystal forms reveal similar intermolecular interactions, the overall crystal packing differs considerably from form to form. The water of the dihydrate is encapsulated between a pair of host molecules, which calculations reveal to be an extraordinarily stable motif. DTA presents functionalities that enable hydrogen and halogen bonding, and whilst an extended hydrogen-bonding network is realised in all crystal forms, halogen bonding is not present in the hydrated crystal forms. This is due to the formation of a hydrogen-bonding network based on individual enclosed water squares, which is not amenable to the concomitant formation of halogen bonds. The main interaction in the solvates involves the carboxylic acid, which corroborates the hypothesis that this strong interaction is the last one to be broken during the crystal desolvation and nucleation process.

Full text of DOI:10.1002/chem.201404693

DOI: 10.1002/chem.201404693
Full Paper
&
Pharmaceuticals |Hot Paper|
Insights into the Crystallisation Process from Anhydrous,
Hydrated and Solvated Crystal Forms of Diatrizoic Acid
[a]
[b, c]
[d]
[b, d]
Katharina Fucke,* Garry J. McIntyre,
Marie-Hꢀlꢁne Lemꢀe-Cailleau, Clive Wilkinson,
[c]
[b]
[b]
Alison J. Edwards, Judith A. K. Howard, and Jonathan W. Steed*
Abstract: Diatrizoic acid (DTA), a clinically used X-ray con-
trast agent, crystallises in two hydrated, three anhydrous
and nine solvated solid forms, all of which have been
characterised by X-ray crystallography. Single-crystal
neutron structures of DTA dihydrate and monosodium DTA
tetrahydrate have been determined. All of the solid-state
structures have been analysed using partial atomic charges
and hardness algorithm (PACHA) calculations. Even though
in general all DTA crystal forms reveal similar intermolecular
interactions, the overall crystal packing differs considerably
from form to form. The water of the dihydrate is encapsul-
ated between a pair of host molecules, which calculations
reveal to be an extraordinarily stable motif. DTA presents
functionalities that enable hydrogen and halogen bonding,
and whilst an extended hydrogen-bonding network is real-
ised in all crystal forms, halogen bonding is not present in
the hydrated crystal forms. This is due to the formation of
a hydrogen-bonding network based on individual enclosed
water squares, which is not amenable to the concomitant
formation of halogen bonds. The main interaction in the
solvates involves the carboxylic acid, which corroborates the
hypothesis that this strong interaction is the last one to be
broken during the crystal desolvation and nucleation
process.
Introduction
from physical and chemical stability to solubility and rate of
dissolution. The latter two are important especially for pharma-
ceutical compounds, as the drug has to dissolve in aqueous
body fluids in order to be taken up into cells and tissues to
reach the target biological receptors. Thus, low solubility di-
rectly affects the bio-availability and must be addressed during
drug formulation.
Pharmaceutically relevant small organic molecules tend to
[
1]
exist in more than one crystal form. These materials can
represent different crystalline arrangements of the compound
itself leading to polymorphs, which differ by the conformation
of the molecule, by the overall packing, or by a combination
[
2]
of both. In addition, solvent molecules from the crystallis-
ation process can be included into the crystal lattice, leading
to hydrates in the case of incorporated water and solvates for
Polymorphism in pharmaceutical materials is widely
[
5]
studied,
and normally every newly developed drug
[5c]
compound is subjected to extensive crystal form screening
[
3]
any other solvent. Both hydrates and solvates are a special
in order to discover as many of the experimentally
accessible polymorphs and solvates, particularly hydrates, as
[
4]
case of co-crystal formation. Different crystal forms can
exhibit quite varied physico-chemical characteristics, ranging
[6]
possible.
However, this approach is money- and time-consuming and
it does not guarantee the isolation of all possible stable crystal
[
a] Dr. K. Fucke
School of Medicine, Pharmacy and Health
Durham University
University Boulevard, Stockton-on-Tees, TS17 6BH (United Kingdom)
E-mail: Katharina.Fucke@durham.ac.uk
[7]
forms, as cases such as that of Ritonavir show. Thus, a more
fundamental understanding of the factors leading to different
polymorphic and solvated crystal forms is an ongoing key
[5a,8]
objective.
[
b] Dr. G. J. McIntyre, Dr. C. Wilkinson, Prof. J. A. K. Howard, Prof. J. W. Steed
On the molecular level, drug compounds generally contain
a range of functional groups, which allow them to interact
with the target protein, receptor or enzyme, to show a pharma-
cological effect. As a result, pharmaceuticals are relatively com-
plicated small molecules, and to date little is known about the
factors influencing their polymorphism and hydrate/solvate
formation. In particular the interplay of the mutual interactions
of the functional groups on the drug molecules themselves
and their interactions with solvent molecules during crystallis-
Department of Chemistry, Durham University
South Road, Durham, DH1 3LE (United Kingdom)
E-mail: Jon.Steed@durham.ac.uk
[
c] Dr. G. J. McIntyre, Dr. A. J. Edwards
The Bragg Institute
Australian Nuclear Science and Technology Organisation
Locked Bag 2001, Kirrawee DC NSW 2234 (Australia)
[
d] Dr. M.-H. Lemꢀe-Cailleau, Dr. C. Wilkinson
Institut Laue-Langevin
6
Rue Horowitz, 38042 Grenoble Cedex (France)
[9]
ation from solution or the melt is rarely investigated.
Extensive information of this type, however, as has been
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/chem.201404693.
Chem. Eur. J. 2015, 21, 1036 – 1047
1036
ꢂ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
[
10]
obtained for the widely studied carbamazepine, will lead to
an understanding of why and how different solid forms can be
crystallised, and how certain forms can be avoided.
higher than that of the free acid. Another new application of
the contrast molecule is its immobilisation on the surface of
gold nanoparticles for IV administration as contrast for en-
[24]
Two directional intermolecular interactions play a key role
during crystallisation and in the crystalline state, namely hydro-
hanced computed tomography.
The DTA molecule has two amide NH groups and one
carboxylic acid OH functionality as hydrogen-bond donors, as
well as two amide and one carboxylic acid carbonyl groups as
hydrogen-bond acceptors. This balanced ratio of donors and
acceptors suggests that the substance might be expected to
crystallise in extensive hydrogen-bonded networks. The three
iodo substituents present the additional possibility of halogen-
bonding interactions, particularly with halogen-bond accepting
solvents. Folen et al. have reported the dihydrate form of DTA
as well as an additional anhydrous modification, which in this
[
11]
gen bonding and halogen bonding. Hydrogen bonding has
been found to stabilise crystal structures and to play a major
[
12]
role in how a molecule will arrange in the solid state. Due to
its accessibility and relative low cost, most studies into the
characteristics of hydrogen bonds are based on X-ray single-
crystal diffraction; however, accurate location of hydrogen
[13]
atoms requires the use of single-crystal neutron diffraction.
Accurate neutron structural data are thus of key importance in
[
14]
determining the experimental charge density. They are also
used to investigate the energetic contribution of the hydrogen
bond to the overall crystal lattice energy by using the Partial
[25]
study will be referred to as Form I. We have previously re-
ported three dimethylsulfoxide solvates of DTA giving insight
[
15]
[26]
Atomic Charges and Hardness Algorithm (PACHA).
into the desolvation of the compound during crystallisation,
[27]
In addition to hydrogen bonding, halogen bonds are
increasingly recognised as a key directional interaction in mo-
as well as the crystal structure of the disodium salt. We now
report a wide-ranging study on DTA polymorphism and solvate
formation in order to elucidate the factors governing the inter-
play of hydrogen bonding and halogen bonding in particular
and shed light onto the desolvation and crystallisation process
in this versatile system.
[
16]
lecular crystal structures. Even though these interactions are
generally considerably weaker than hydrogen bonds, they
have been shown in recent years to comprise a significant
[
17]
stabilising interaction, with interesting work reported on the
competition of hydrogen versus halogen bonding in molecular
[
18]
crystals.
Halogen bonding has been shown to persist in
[
19]
solution-state anion receptors
and can drive gelation Results
[
17a]
behaviour.
The crystal forms
In this study, we seek to understand the interplay of hydro-
gen and halogen bonding and their influence on the crystal
forms of an X-ray contrast molecule, diatrizoic acid (DTA,
Scheme 1), which is pharmaceutically relevant and listed in the
Diatrizoic acid was found to exist in two hydrated forms: the
[25]
known dihydrate and a new form that contains one water
molecule per four host molecules (a tetartohydrate). In addi-
tion, three anhydrous modifications, named forms I, II and III in
the order of their discovery, were obtained. Form I has been
[25]
previously reported by Folen and co-workers and crystallises
only from the melt, while Forms II and III can only be obtained
by thermal dehydration of the dihydrate. In addition, solvates
containing dimethyl formamide, methanol, tetrahydrofuran,
dioxane/water and dimethyl sulfoxide have been isolated.
DTA dihydrate
DTA is very sparingly soluble in water at room temperature,
while the solubility increases significantly upon heating to
boiling point. On cooling this solution, prismatic block-
shaped crystals of the dihydrate grow within a couple of
hours, and their crystal size depends on the cooling rate.
Large, high-quality crystals (Figure 1a) were obtained by very
Scheme 1. Molecular structure of diatrizoic acid.
[
20]
[21]
European and United States Pharmacopoeias as the dihy-
drated form. Specifically monographed crystal forms are gener-
ally prevalent and stable enough for manufacture, and since
DTA is poorly water soluble, the formation of a hydrate indi-
cates severe problems for pharmaceutical manufacturing, be-
cause hydrates show thermodynamically the lowest solubility
[22]
in water of all crystal forms of a specific compound. This
problem has been circumvented by the formulation of DTA as
a pharmaceutically acceptable salt form, as sodium (diatrizoate
sodium, DTS), meglumine or l-lysinium salt or a mixture there-
of, as a commercially available intravenous preparation. While
Figure 1. Photomicrographs of the diatrizoic acid hydrated crystal forms:
a) DTA dihydrate, b) DTS tetrahydrate initial blades and c) DTS tetrahydrate
final blocks.
[23]
the sodium salt also crystallises in a highly hydrated form,
the solubility of the charged species in water is considerably
Chem. Eur. J. 2015, 21, 1036 – 1047
www.chemeurj.org
1037
ꢂ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Table 1. X-ray crystallographic data of the crystal forms.
Parameter
formula
DTA
dihydrate
DTS
tetrahydrate
DTA Form I
DTA tetarto
hydrate
DTA DMF
disolvate
DTA THF
monosolvate
DTA dioxane/
water monosolvate
DTA methanol
monosolvate
C
2
11
H
9
O
4
N
2
I
3
·
C
11
H
8
O
4
N
O
2
I
3
Na·
C
11
H
9
O
4
N
2
I
3
C
11
H
9
O
4
N
2
I
3
·
C
11
H
9
I
3
N
2
O
4
·
C
C
11
H
9
I
O
3
N
2
O
4
·
C
11
H
9
I
3
N
2
O
4
·
11 9 3 2 4
C H I N O ·
H
2
O
4.25H
1421.88
triclinic
ꢀ
P1
120
2
2
0.25H O
622.91
2(C
3
H
7
NO)
4
H
8
H
2
O·C
720.04
monoclinic
P2 /n
4
H
8
O
2
4
CH O
645.96
triclinic
ꢀ
P1
120
M
r
649.93
triclinic
ꢀ
P1
613.90
monoclinic
C2/c
760.11
orthorhombic
Pbca
686.03
monoclinic
P2 /n
crystal system
space group
T [K]
monoclinic
C2/c
120
1
1
120
120
120
120
150
a [ꢁ]
b [ꢁ]
c [ꢁ]
a [8]
7.8264(6)
9.4397(8)
13.302(1)
89.673(2)
75.570(2)
71.091(2)
897.4(1)
2
11.754(2)
20.646(4)
9.353(2)
18.688(3)
90
117.020(5)
90
20.70(2)
9.407(8)
18.68(2)
90
116.440(8)
90
21.321(3)
9.277(1)
25.724(1)
90
90
90
8.893(1)
18.818(2)
12.767(1)
90
90.188
90
11.940(4)
11.878(4)
15.642(5)
90
100.925(3)
90
9.4497(1)
13.9552(2)
14.6126(2)
85.214(1)
87.695(1)
72.367(1)
1829.83(4)
4
13.091(3)
15.129(4)
73.62(2)
84.56(2)
63.50(1)
2024.1(8)
2
b [8]
g [8]
3
V [ꢃ ]
3215(1)
8
3257(5)
8
5088(1)
8
2136.5(4)
4
2178(1)
4
Z
R
1
[%]
2.16
2.12
4.59
7.40
2.51
4.40
4.42
2.46
slow cooling of a saturated aqueous solution containing minor
amounts of ethanol as solubiliser from boiling point to room
temperature.
gen bonding from N1 of an amido group to the non-proton-
ated oxygen atom O2 of the carboxylic acid group of the next
capsule (Figure 2b). The second stacking motif involves
hydrogen bonding from N2 of the second amido group to O3
of the amido group of an adjacent capsule resulting in stacks
along the crystallographic a-axis.
DTA dihydrate crystallises in the triclinic space group P 1¯ with
one formula unit in the asymmetric unit (Z’=1, Table 1;
Supporting Information, Table S1). The host molecule adopts
a conformation in which both amide substituents as well as
the carboxylic acid group are rotated out of the aromatic ring
plane, with the oxygen atoms of the amide groups and the
hydroxyl function of the acid group being on the same side of
the ring in a syn conformation. Four water molecules are sand-
wiched between pairs of DTA hosts in a hydrogen-bonded
square (Figure 2a). Two of these donate hydrogen bonds to
the carbonyl groups of DTA, while the other two accept one
hydrogen bond from the acid group of the host forming a
hydrogen-bonded capsule, resembling the water squares
encapsulated in tetrapodal urea complexes characterised by
This crystal packing results in the formation of layers which
are strongly hydrogen bonded. Interestingly, in the direction of
the crystallographic b-axis, the packing of the layers is only
stabilised through weaker CÀH···O interactions, and thus the
association between layers cannot be as strong as the inter-
actions within the layers.
A single-crystal neutron-diffraction structure determination
of DTA dihydrate was undertaken at 120 K using the Laue
diffractometer KOALA at the Bragg Institute (Lucas Heights,
Australia). Due to the low symmetry of this crystal structure
ꢀ
(triclinic, P1) the Laue diffraction patterns measured for
[
28]
neutron diffraction.
a single orientation of the crystal on the f axis gave only a re-
stricted amount of data. However, the structural model could
be refined to give a stable solution without restraints and all
hydrogen atoms could be identified reliably from the Fourier
difference maps (Figure 2a).
The stacking of this basic DTA-water-capsule into the 3D
crystal network is stabilised through two different hydrogen
bonds. Firstly, the capsules stack along (0 0 1) through hydro-
The hydrogen atom positions
in the neutron structure are sim-
ilar to the positions calculated
for the X-ray structure, but with
significant differences in the
bond lengths. As expected, the
terminal CH groups of the host
3
molecule show rather large
atomic anisotropic displacement
parameters (ADPs) for the hy-
drogen positions indicating ro-
tational mobility in these
groups, which could not be
modelled as static disorder and
Figure 2. a) Discrete water square found “encapsulated” between two DTA molecules in DTA dihydrate and
b) stacking motifs found in the structure of DTA dihydrate as derived from the single-crystal neutron structure.
could be of dynamic origin. The
(
Ellipsoids are drawn at 30% probability.
CÀH bonds vary between
Chem. Eur. J. 2015, 21, 1036 – 1047
www.chemeurj.org
1038
ꢂ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
0
.99(6) and 1.07(5) ꢃ and thus these values are close to the
are the strongest and hence are expected to be broken last
during desolvation during the crystal nucleation process. It is,
however, surprising how short this interaction is, with only
1.45 ꢃ distance between the hydrogen atom and the acceptor
oxygen atom. It is possible that this hydrogen atom can move
between the donor and the acceptor oxygen at elevated tem-
geometrically constrained calculated CÀH bonds of the X-ray
model (0.98 ꢃ).
As all of the remaining hydrogen atoms in the X-ray crystal
structure were located from the difference Fourier maps and
refined freely, the atomic positions are a compromise between
experimental electron density and electrostatic energy optimi-
[30]
peratures, as has been shown for a dimethylurea co-crystal,
[
29]
sation employed in the algorithm of olex.refine.
For this
but since the neutron structure has been determined at 120 K,
the hydrogen atom is frozen in a more localised position.
Because of the considerably stronger interactions between the
host and the water molecules compared to the inter-host
homomeric interactions, DTA dihydrate is in some sense the
reason, the directionality of the XÀH bond fits well with that
found in the neutron structure, whilst the bond distances vary
up to 0.3 ꢃ (Table S2, Supporting Information). This fact has
a major influence on the hydrogen-bonding energies and
strengths of these interactions in the subsequent calculations.
The atomic coordinates derived from the neutron structure
determination were used for semi-empirical lattice energy cal-
[15a]
opposite to theophylline monohydrate,
in which the
homomeric interactions are considerably stronger than those
between the host and the water molecules. As a result
theophylline monohydrate is relatively unstable and readily de-
[
15c]
culations using PACHA.
Even though the results of these
[15a]
calculations have to be interpreted with caution due to the al-
gorithm treating the atoms as point charges and calculating
only dipolar interactions excluding higher order effects, the ex-
tracted energies give a good approximation of the ranking of
interaction strength, and cope well with many-atom structures
at modest computational cost. The results of the PACHA
calculations are listed in Table 2. Starting from the host–host
hydrates leaving the overall structural network intact.
It is
also likely that since the water square embedded within the
DTA dimeric capsule represents such a robust supramolecular
motif, these capsular structures may exist in solution before
any stable crystal nuclei form.
Thermo-microscopic investigations showed that the dihy-
drate crystals remain unchanged between room temperature
and 1208C. Upon further heating, the crystals start to move
and jump whilst undergoing pseudomorphosis, indicating radi-
[31]
Table 2. Geometric and energetic data for DTA dihydrate.
cal changes to the crystal structure. A preparation of the di-
hydrate in paraffin oil shows bubble formation accompanying
the pseudomorphosis verifying the release of water, which is
complete at about 1358C. Dehydration at such high
temperatures is unusual for organic hydrates without cation
coordination of the incorporated water and demonstrates the
extraordinary stability of this crystal form. Upon further
heating sublimation into fine needles can be observed above
[
a]
Interaction
D···A
H···A
DÀH···A
Energy
À1 [b]
distance [ꢃ] distance [ꢃ] angle [8] [kJmol
]
[
a]
O2ÀH2···O1W
N1-H1···O2
2.49(2)
2.80(2)
2.82(1)
2.68(3)
1.45(3)
1.81(3)
1.82(2)
1.72(3)
1.77(3)
1.73(2)
177(2)
175(1)
172(1)
165(1)
174(2)
163(2)
À56.4
À16.6
À15.1
À28.5
À28.5
À34.0
[
[
[
[
a]
a]
a]
a]
N2ÀH2a···O3
O1WÀH1Wa···O2W
O1WÀH1Wb···O2W 2.72(2)
O2WÀH2Wa···O4
2.67(2)
3008C, while above 3208C the sample melts and decomposes
[a] Hydrogen–acceptor distance derived from the neutron single-crystal
concomitantly.
data. [b] Interaction energies as calculated by the PACHA algorithm.
The analysis of the thermogravimetry (TG) (Figure 3) con-
firms the high stability of DTA dihydrate. The thermogram
shows no weight loss of the sample up to 808C, while a follow-
ing single step of 5.68% weight loss represents the release of
the crystal water (calculated value for a dihydrate is 5.54%).
interactions of the DTA dihydrate, both homomeric
hydrogen bonds donated by amido NÀH moieties to either an
adjacent amido oxygen atom or the carboxylic acid group
show energies in the range of medium-to-strong hydrogen
[
12b]
bonds.
Interestingly, the hydrogen bond with the carboxylic
acid group as acceptor is only slightly stronger than the one
having an amide carbonyl group as acceptor. All hydrogen
bonds involving the water molecules are considerably stronger
than the homomeric host-to-host interactions. The hydrogen
bonds between O1W and O2W in the water square have a
stabilising energy of almost double the value compared to the
homomeric interactions.
The heteromeric hydrogen bonds between the incorporated
water molecules O2W and the host carbonyl oxygen atoms are
even stronger. The strongest interaction, however, is the hydro-
gen bond donated from the carboxylic acid OH group of the
host to the oxygen atom O1W of the water molecule, which is
Figure 3. Thermogravimetric (upper curve) and differential scanning
[26]
consistent with our previous work on DMSO solvates of DTA,
in which the hydrogen bonds donated from the acid of DTA
calorimetric (lower curve) thermograms of DTA dihydrate, both recorded at
À1
108C min
.
Chem. Eur. J. 2015, 21, 1036 – 1047
www.chemeurj.org
1039
ꢂ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
The dehydration temperature detected in the TG is lower than
that observed by thermo-microscopy which is due to the
dynamic dry helium flow around the sample. The dehydrated
sample is stable to 2908C, above which a second weight loss
indicates decomposition during which volatile decomposition
products are released.
possible to obtain single crystals suitable for structure deter-
mination of either of these two crystal forms, however, their
powder X-ray diffraction (PXRD) patterns and FTIR spectra are
presented in the Supporting Information (Figures S1 and S2).
Due to the high absorption of the DTA molecule caused by
the three iodine substituents, the PXRD patterns of Forms II
and III are very low in intensity. However, clear changes in
peak positions can be seen, proving that a phase transition oc-
curred. The same change can be detected by FTIR spectrosco-
py. The most striking changes are in the OH and NH vibrational
The observed thermal behaviour of DTA dihydrate is reflect-
ed in its differential scanning calorimetry (DSC) trace (Figure 3).
The thermogram shows a dehydration endotherm in the
temperature range from 80 to 1508C, which coincides with the
mass loss found by TG. The crystal form present after this de-
hydration event was verified by FTIR measurements to be
a new anhydrous form, Form II, as compared to the spectra of
the dihydrate and the previously published anhydrous Form I.
Upon further heating, Form II transforms in an exothermic
event into anhydrous Form III, a second new crystal form. Exo-
thermic transitions generally imply a monotropic relationship
À1
region around 3500 cm , which changes significantly from the
dihydrate to its dehydrated Form II. Surprisingly, Form III has
a sharp peak in this region, which points towards a very dis-
tinct hydrogen-bonding pattern involving the carboxylic acid
or the amido NH groups. This assumption is supported by the
changes detected in the carbonyl stretch vibrations at around
À1
1700 cm . Another region of changes is between 1700 and
[
32]
À1
between the two crystal forms.
With the thermal data listed in Table 3, a semi-schematic
1600 cm , in which the aromatic ring vibrations are located.
For phase identification, even though the exact nature of the
vibrations is hard to establish, the fingerprint region offers
unique patterns for each of the three crystal forms.
[32–33]
energy/temperature diagram was constructed (Figure 4).
It
is clearly visible that the G-isobars of the two polymorphs run
in parallel over the whole temperature range and do not inter-
sect. Thus, these two crystal forms are monotropically related
to each other, and Form III is the thermodynamically stable
polymorph at all temperatures. Unfortunately, it was not
DTA tetartohydrate
In addition to the DTA dihydrate, a second hydrated structure
of DTA was crystallised from 1-propanol (Table 1). The single-
crystal X-ray data can be successfully modelled as containing
0.25 molecules of water making this crystal form a tetarto-
hydrate. The crystal structure contains the host molecule in
a different conformation to the dihydrate, in which the two
amide carbonyl oxygen atoms lie on opposite sides of the
ring. This trans conformation causes the assembly of a different
hydrogen-bonded network, based on stacks along the
crystallographic b-axis, consisting of centrosymmetric dimers
connected through hydrogen bonds from the amide N2 to the
amide carbonyl O3 (Figure 5). The connection of these dimers
is also found in the structures of the dihydrate, and involves
two hydrogen bonds from the amide N1 to the acid group O2
connecting two of the host molecules in a centrosymmetric
fashion. The remaining hydrogen-bond donor of the tetarto-
hydrate, O1 of the carboxylic acid group, interacts with the
remaining hydrogen-bond acceptor O4. This last interaction
connects the stacks along the crystallographic c-axis. The
incorporated water molecule is relatively far away from any
potential hydrogen-bond donor or acceptor, and hydrogen
atom positions could not be located in the Fourier maps.
However, electrostatic optimisation of the hydrogen positions
suggests long-distance interactions between the water mole-
cule and the amido carbonyl oxygen atom O4, which would
give a bridging connection between two host stacks along the
crystallographic a-axis.
Table 3. Thermal data of the solid-state transitions of the anhydrous
crystal forms of DTA.
Transition
temperature
Heat of
Melting point
(onset) [8C]
Heat of
fusion
[kJmol ]
transition
À1
À1
(
onset) [8C]
[kJmol
]
Form II
Form III
210.8Æ0.4
À7.3Æ0.3
306.7Æ2.3
31.6Æ1.7
The X-ray crystal structure of the tetartohydrate was deter-
mined using synchrotron radiation. The structural model is not
of the highest precision due to paucity of observed data
caused by the small crystal size, thus the PACHA calculations
have to be interpreted cautiously. Again, only the energetic
trends will be considered. The DTA dimers are both connected
Figure 4. Semi-schematic energy/temperature diagram of the DTA
anhydrous Forms II and III.
Chem. Eur. J. 2015, 21, 1036 – 1047
www.chemeurj.org
1040
ꢂ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
that the water molecules act
mainly as ‘space fillers’ and that
the interactions involving them
are not essential for the stability
of the crystal structure.
Since the tetartohydrate and
Form I are structurally very close-
ly related, it is difficult to deter-
mine with certainty, which crys-
tal form was originally described
[25]
by Folen et al.
Even though
sufficiently large quantities of
the pure modification could not
be obtained in order to measure
experimental PXRD patterns, the
calculated patterns show such
similarity such that it would be
very difficult to differentiate be-
tween the two forms with exper-
imental data from laboratory-
based X-ray diffractometers (Fig-
ure S4, Supporting Information).
In addition, as the only method
to produce pure Form I is by
Figure 5. Molecular packing of the DTA tetartohydrate and the anhydrous Form I viewed a) along (1 0 0) and
b) along (0 1 0).
melting the DMF solvate in a paraffin oil suspension to reach
the peritectic melting point, samples of Form I for IR spectros-
copy have large quantities of paraffin oil as impurities, which
makes a comparison of the IR spectra impossible. Form I is
originally also characterised by TG, but the data are only pre-
sented from a starting temperature of 408C. Taking into ac-
count the possible non-stoichiometric behaviour of the tetarto-
hydrate and the calculated overall mass loss of only 0.7%, the
release of the water could be easily missed or misinterpreted
as loss of surface-attached solvent.
Table 4. Geometric and energetic data of DTA tetartohydrate and Form I.
[
a]
À1
Interaction
D···A distance [ꢃ]
DÀH···A angle [8]
Energy [kJmol ]
Tetartohydrate
O1···O4
N1···O2
N2···O3
O1W···O4
Form I
[
a]
2.51(2)
2.95(2)
2.82(2)
3.16(3)
178
À51.9
À28.9
À36.2
À18.5
175.7(9)
151.8(9)
[
a]
172
O1···O4
N1···O2
N2···O3
2.55(1)
2.817(9)
2.921(9)
160(6)
150.1(6)
174.3(8)
À51.1
À31.9
À29.5
[
a] Angle after electrostatic optimisation of the water hydrogen atom
DTS tetrahydrate
positions.
The monosodium salt of diatrizoic acid has been previously re-
[23]
ported by Tonnessen et al. and shows a higher solubility in
water than DTA and can be crystallised by precipitation from
aqueous solution with anti-solvent, for example, acetone. The
initial crystals formed are fine blades (Figure 1b), which over
time ripen to blocks (Figure 1c). This proved to be due to Ost-
wald ripening and not to the existence of different crystal
forms and a phase transition, as demonstrated by PXRD. Ther-
momicroscopy shows that these blocks undergo a very strong
pseudomorphosis accompanying dehydration between 75 and
798C to the extent that the DTS tetrahydrate crystals not only
transform into powder but the particles additionally lose all of
their initial macroscopic integrity during this event (Figure 6).
Having such a marked pseudomorphosis upon phase transition
indicates a radical change in the crystal structure during the
thermal event. Surprisingly, the dehydration does not take
place at very high temperatures, as would be expected for the
through medium-strength hydrogen bonds (Table 4), while the
interaction involving the acid group is considerably stronger.
This correlates well with the energies found for the dihydrate
and the DMSO solvates. The interactions involving the water
molecule are rather weak compared to those found in the di-
hydrate, which is attributed to the long distance to the inter-
acting atoms. However, this low energy can also explain that
a crystal form closely related to the tetartohydrate can be
grown from the melt completely water-free. This form
(Table S1, Supporting Information, and Figure 5) corresponds
to the anhydrous crystal form described by Folen and co-work-
[
25]
ers (Form I). In light of the close correlation of the crystal
packing as well as the lack of interaction between the host
and the hydrate molecules, it is possible that the tetarto-
hydrate shows non-stoichiometric behaviour.
[34]
The interaction energies of the tetartohydrate (Table 4) and
the considerable distance of the included water molecule to
any interacting functional group of the host molecule indicate
strongly coordinating hydrates of organic sodium salts.
Upon further heating, melting occurs at 2788C. This event re-
sembles a gradual amorphisation or glass-transition, in which
Chem. Eur. J. 2015, 21, 1036 – 1047
www.chemeurj.org
1041
ꢂ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
dehydration steps but also
during the decomposition step.
This water has to be crystal
water rather than decomposi-
tion product and would explain
Figure 6. Photomicrographs of the dehydration of DTS tetrahydrate upon heating. The first micrograph is taken at
the discrepancy between the
À1
748C with the following images taken every 5 s at a heating rate of 108Cmin
.
calculated and the experimental
water loss. However, it cannot
the crystals lose their birefringence without obvious formation
of a liquid phase and there is rapid decay of the morphology
of the particles. This behaviour is similar to that of polymers,
whilst small molecular crystals generally show a sharp melting
event, and it is presumably due to concomitant decomposi-
tion. Additionally, bubbles form in the resulting melt, which
indicates either decomposition or residual water entrapped in
the crystals during the dehydration process being released
upon melting.
be excluded that the sample loses crystal water during storage
or when the sample is initially put into the dry purge gas of
the TG, and it is possible that the lower-than-expected mass
loss is due to a combination of these processes.
The DSC trace of DTS tetrahydrate matches the unstable
behaviour observed by TG. Two endothermic events can be
detected, the first at approximately 908C, which correlates
with the first dehydration. The second event starts at 1908C,
matching the second mass loss observed by TG. Above this
temperature, a very pronounced exothermic event represents
the decomposition of the sample and is in good agreement
with the results of the TG experiment.
The DTS tetrahydrate is very unstable compared to DTA
dihydrate and the TG trace shows the release of the incorpo-
rated water starting even at room temperature (Figure 7). This
Block-shaped single crystals of the DTS tetrahydrate turned
out to be twinned and represent presumably an aggregation
of the initially crystallising blades. However, the single-crystal
X-ray structure was redetermined for further use as an initial
model for the neutron structure refinement. We could confirm
[23]
the model published earlier by Tonnessen et al. with two for-
mula units in the asymmetric unit (Z’=2). The authors report
a tetrahydrate with partially disordered water sites. In the
present determination the water positions are also disordered,
but the positions could be refined with a higher occupancy,
raising the water stoichiometry to 4.25 water molecules per
DTS unit. For the purposes of the present discussion, however,
the name tetrahydrate will be retained. Full occupancy of all
water positions would result in a stoichiometry of 4.5 water
molecules per DTS unit. Considering the relatively low stability
of the hydrate in dry atmosphere, as observed by TG, it is pos-
sible that the crystal water is partially released depending on
the handling of the sample prior to the diffraction experiment.
It additionally points towards at least partial non-stoichiometric
characteristics of this hydrate.
Figure 7. Thermogravimetric (upper curve) and differential scanning calori-
metric (lower curve) thermograms of DTS tetrahydrate, both recorded at
À1
108C min
.
indicates, that the crystal form can be dehydrated by lowering
the relative humidity (RH), while samples of the tetrahydrate
can be stored unchanged as dry powder under ambient condi-
tions (~40% RH). TGA analysis shows an initial mass loss step
of 6.38%, corresponding to 2.4 moles of water. At about 858C
this mass loss process is complete and the TG trace indicates
that the sample is in equilibrium until a second mass loss of
From the crystal structure, it is not obvious which water
molecules are lost in the first dehydration step observed by
TG. It is plausible that the non-coordinated water molecules
O1W and O2W could leave more readily, especially since they
are located in very small channels along (0 1 0) in the crystal
structure, which allow a more facile egress than from the coor-
dination sites of the sodium cations. This assumption is also
based on the energetically weaker interaction of hydrogen
bonds versus ion coordination. On the other hand, approxi-
mately 2.4 moles of water leave the crystal lattice on the first
dehydration step, thus the sodium coordination has to be
broken for at least some of the water molecules. These sites
are most likely the water molecules O4W and O5W, which
2.08% (0.75 moles of water) occurs in the range 170 to 2108C.
This second event was not observed by thermomicroscopy,
which could be due to the small particle size of the corre-
sponding dehydrated crystal form. A third mass loss is detect-
ed above 2408C representing decomposition of the sample.
Compared to the DTA dihydrate, this decomposition takes
place about 408C lower and indicates the lower thermal stabili-
ty of the sodium salt. The total mass loss before decomposi-
tion adds up to 3.15 moles, considerably lower than the 4
moles anticipated for a tetrahydrate. TG-MS measurements
revealed that water is released from the sample during both
[23]
bridge two sodium cations. Tonnessen et al. report in their
structure only half occupancy of one of these sites, while the
present determination exhibits full occupancy for both water
molecules, which is presumably due to rapid sample handling
Chem. Eur. J. 2015, 21, 1036 – 1047
www.chemeurj.org
1042
ꢂ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
and hence less dehydration before the single-crystal diffraction
experiment.
tion of the host molecule (protonated as well as deprotonated)
through hydrogen bonds from the amide NH (N1 and N101) to
the carboxylic acid group (O2 and O102) is realised in both
forms. But whilst this motif links the DTA-water capsules in the
dihydrate along the crystallographic c-axis (Figure 2), the
dimers in the DTS tetrahydrate are separated by the water-
coordinated sodium cations. The second motif is a water
molecule (O3W) bridging between the two amide carbonyl
groups of the same molecule (O103 and O104). This motif,
however, is only realised for one of the two crystallographically
independent host molecules in the DTS tetrahydrate. The
amido carbonyl oxygen (O4) of the other molecule is involved
in hydrogen bonding to three different water molecules:
a bridging water molecule (O1W) between the amido carbonyl
and sodium-coordinated oxygen atom of the carboxylate of
the same molecule (Figure 8), a bridging water molecule
Due to the complex nature of this crystal structure and the
presence of relatively heavy atoms, such as sodium and
particularly iodine, not all hydrogen atom positions could be
determined by X-ray crystallography. Thus, single-crystal neu-
tron diffraction data was collected on the Laue diffractometer
[35]
VIVALDI at the Institut Laue-Langevin (Grenoble, France).
These data enabled the accurate location of all hydrogen atom
positions and their anisotropic refinement. As observed for
DTA dihydrate, the bond lengths involving hydrogen atoms
are longer in the neutron structure compared to the X-ray
structure. In addition, all hydrogen atom positions could be
deduced from the difference Fourier maps. Again, the terminal
methyl groups show rather large ADPs for the hydrogen atoms
consistent with rotational disorder. In addition, the hydrogen
atoms on three of the water molecules (O1W, O2W and O4W)
show large ADPs, which in the case of the former two can be
explained by higher mobility of the water molecules in the
crystal lattice, as they are not coordinated to sodium cations.
O4W, however, is a bridging water molecule between two cat-
ions, thus its mobility should be reduced. The neutron beam
time available permitted data collection for rotation around
just one direction in the crystal. For a triclinic lattice in the
Laue technique this will lead to a paucity of data along and
near that axis which can result in exaggeration of the displace-
ment parameters in that direction. For information about the
hydrogen bond lengths, refer to the Supporting Information
(Table S3).
Due to the coordination of both the host and the water
molecules to the sodium cations, the DTS tetrahydrate exhibits
a more complex crystalline network than DTA dihydrate
Figure 8. Section of the neutron single-crystal structure of the DTS
tetrahydrate with disorder of the water molecule including O8W resolved.
Atomic displacement ellipsoids are drawn at the 30% probability level.
(
Table 5). However, similar motifs of hydrogen bonding can be
found in both crystal forms. The centrosymmetric dimer forma-
(
O6W) between the carbonyl oxygen and the carboxylate of an
Table 5. Geometric data of the DTS tetrahydrate.
adjacent anion and a water molecule (O8Wb) bridging to an-
other water molecule (O8Wa). This leaves one amido carbonyl
group free for linking to the next DTS dimer through hydrogen
bonding to the remaining free amido NH group (N2 and
N102).
Interaction D···A distance
ꢃ]
DÀH···A distance
D<C->H···A angle
[8]
[
[ꢃ]
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
a]
a]
a]
a]
a]
a]
a]
a]
a]
a]
a]
a]
a]
a]
a]
a]
a]
N1···O2
2.87(1)
1.85(2)
1.70(2)
1.93(1)
1.76(1)
2.52(5)
1.92(5)
1.85(2)
1.97(3)
2.30(2)
1.88(2)
1.88(2)
1.90(2)
2.01(2)
1.94(2)
1.89(2)
1.60(3)
2.08(4)
172(1)
174(1)
165(1)
174(1)
178(3)
164(3)
169(2)
174(2)
146(2)
169(3)
169(2)
167(2)
173(1)
156(1)
164(2)
167(3)
160(4)
N101···O102 2.76(1)
N2···O103
N102···O3
O1W···O1
O1W···O4
2.929(7)
1.794(7)
3.48(3)
2.86(3)
Even though the neutron single crystal experiment of DTS
resulted in the determination of an accurate structural model,
crystal lattice energy calculations using the PACHA algorithm
were not feasible due to the high complexity of the
coordination network.
O2W···O101 2.81(2)
O3W···O103 2.92(2)
O3W···O104 3.14(1)
O5W···O2
O5W···O7W 2.84(1)
O6W···O4 2.86(1)
O6W···O102 2.94(1)
O7W···O2 2.88(2)
2.82(2)
The solvates
In addition to hydrated crystal forms, a series of DTA solvates
were obtained by slow evaporation or slow cooling of
solutions of DTA in a range of hydrogen-bond acceptor and
donor solvents. The structures of three dimethylsulfoxide
(DMSO) solvates, namely tetra-, di- and monosolvate, have
been communicated previously and show an intriguing series
of snapshots of the desolvation process during the crystal
O7W···O104 2.83(1)
O8WA···O1W 2.75(2)
O8WB···O4
2.96(3)
[a] Hydrogen–acceptor distance derived from the neutron single-crystal
data. Single-atom descriptors (X) corresponding to one DTA molecule,
while the 10X atom descriptors correspond to the second molecule.
Chem. Eur. J. 2015, 21, 1036 – 1047
www.chemeurj.org
1043
ꢂ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
[
26]
nucleation process. This series of solvates demonstrates that
of the three hydrogen-bond donor sites (carboxylic acid and
two amide groups) the carboxylic acid is involved in the
strongest hydrogen bonds and remains solvated as the DMSO
bound to the other donors is removed.
Additional solvated crystal forms were obtained from DMF,
THF, dioxane and methanol. Crystallographic data are listed in
Table 1.
The DMF disolvate can be crystallised by evaporation of
DMF solutions under ambient conditions, upon which large
prisms form (Figure S5, Supporting Information) that are
relatively stable at room temperature. The single-crystal X-ray
structure was determined in the orthorhombic space group
Pbca with one formula unit in the asymmetric unit (Z’=1). The
DTA molecule adopts an antiparallel conformation of the
amide side chains, similar to the DMSO tetra- and disolvates.
Hydrogen bonding between the amide groups results in
a brickwork-like network in the crystallographic c-plane, similar
to that observed in the DMSO monosolvate. The solvent mole-
cules are located in between two layers of this network. These
show disorder over two positions for both molecules, with an
occupancy of approximately 50:50 for one molecule and 30:70
for the other. The overall orientation of the envelope of the
two solvent sites is, however, unchanged and the carbonyl
oxygen atom remains almost unchanged in position. Consis-
tent with the DMSO solvates, one DMF molecule interacts with
the carboxylic acid group of the host molecule through hydro-
gen bonding (Figure 9, Table 6). The second DMF molecule, on
the other hand, interacts with the host molecule with a near
linear O···I3 halogen bond and a somewhat more angular inter-
action to the iodine atom I2 (CÀI···O angles of 173 and 1678,
respectively, averaged over both solvent positions). These halo-
gen bonds hold the DMF molecule in place in between the
layers of DTA. Even though DMF and DMSO are related in size
and shape, and although the DMF disolvate and the DMSO dis-
olvate crystallise in the same space group, the overall packing
of the two crystal structures is different. The DMSO disolvate
shows only two interactions between the DTA molecules, one
being a NÀH···O hydrogen bond between two amide groups,
the other the pseudo-centrosymmetric dimer formation by an
I···O halogen bond involving the carboxylic acid group. All
other hydrogen-bonding groups interact with the incorporated
solvent molecules. The DMF disolvate, on the other hand, con-
sists of a brickwork-like layered structure of the DTA molecules
Figure 9. Brickwork-like crystal packing diagrams of the DMF disolvate:
a) individual hydrogen-bonded layer viewed along the crystallographic c
axis and b) location of the incorporated solvent (blue) in between the
layers viewed along the crystallographic b axis. Element colours: carbon
(
grey), nitrogen (blue), oxygen (red), iodine (purple), hydrogen (light grey).
ADPs are drawn at 30% probability.
disolvate and the DMSO monosolvate, even though the host
molecule exhibits a syn-conformation of the amido substit-
uents, unlike the anti-conformation found in the related
solvates. The overall packing is brickwork-like similar to the
DMF disolvate, but due to the syn-conformation of the host
molecule, the orientation of the brickwork follows the ABAB
motif found in the DMSO monosolvate (Figure S7, Supporting
Information). In fact, the THF monosolvate and the DMSO
monosolvate are isostructural, even though the former crystal-
lises in a lower-symmetry space group.
(
Figure 9a), in between which the solvent molecules are locat-
The methanol monosolvate (Figure 10) was obtained by
cooling a supersaturated methanol solution of DTA to 48C. The
ed (Figure 9b), which is comparable to that found in the
DMSO monosolvate. This network of the host molecules is sta-
bilised by NÀH···O hydrogen bonds between the amide groups
of adjacent molecules, and further stabilised by I···O halogen
bonds involving the carboxylic acid group. The network shows
the DTA molecules orienting in an AABB fashion, which is
different to the ABAB orientation of the DMSO monosolvate.
Cooling a THF solution from room temperature to 48C
results in the formation of a THF monosolvate. This molecular
complex was characterised by X-ray crystallography in the
ꢀ
crystal structure was determined in the triclinic space group P1
with two formula units in the asymmetric unit (Z’=2). The
overall packing resembles that of Form I of solvent-free DTA
(see above). The DTA molecules form infinite chains connected
through two different kinds of hydrogen bonds. The first mode
connects two DTA molecules by two hydrogen bonds between
an amido NÀH and the carboxylic acid non-protonated oxygen
atom into centrosymmetric dimers. These connect through
hydrogen bonds of the free amido functionalities to the next
monoclinic spacegroup P2 /n with one formula unit in the
dimer, forming
a second centrosymmetric dimer, which
1
asymmetric unit (Z’=1). It is structurally related to the DMF
connects the host molecules in an offset stack. The methanol
Chem. Eur. J. 2015, 21, 1036 – 1047
www.chemeurj.org
1044
ꢂ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Table 6. Bond distances, geometries and energies of the solvated crystal
structures of DTA.
À1
Interaction
D···A distance [ꢃ]
DÀH···A angle [8]
Energy [kJmol
]
DMF disolvate
O1···O1SA
N1···O3
N2···O4
I1···O2
I2···O1SC
I2···O1SD
I3···O1SC
I3···O1SD
[
a]
2.528(5)
2.796(5)
2.807(6)
3.044(3)
3.27(2)
2.97(1)
2.92(2)
3.07(1)
169(2)
159.9(2)
168.1(2)
À53.1
À34.5
À28.7
À2.6
[
[
a]
a]
À5.8
À7.7
THF monosolvate
O1···O1S
N1···O4
N2···O3
I1···O2
I2···O2
Methanol monosolvate
2.54(1)
167(6)
167(1)
173(1)
À36.6
À25.5
À27.8
À5.9
2.80(1)
2.81(1)
3.184(8)
3.063(6)
À2.7
O2···O2M
O102···O1M
O1M···O4
O2M···O103
N1···O2
N102···O102
N2···O3
N101···O104
I1···O4
I3···O103
I103···O103
Dioxane/water monosolvate
O1···O2S
N2···O2
2.509(4)
2.567(3)
2.678(4)
2.718(5)
2.889(5)
2.842(4)
2.866(5)
2.734(5)
3.111(3)
3.032(3)
3.038(3)
167(7)
176(6)
176(6)
175(8)
171.0(3)
164.7(3)
146.3(3)
165.1(3)
À44.3
À38.9
À32.8
À35.2
À32.6
À36.2
À27.7
À33.0
Figure 10. Representation of the hydrogen bonding observed in the DTA
methanol monosolvate. Element colours: carbon (grey), nitrogen (blue),
oxygen (red), iodine (purple), hydrogen (light grey). Atomic displacement
ellipsoids are drawn at 30% probability.
[
[
[
c]
c]
c]
[
d]
one formula unit in the asymmetric unit (Z’=1). The host mol-
ecules adopt a syn-conformation. They assemble in centrosym-
metric dimers, which are stabilised through two hydrogen
bonds formed between the NÀH of one amido group and the
non-protonated oxygen atom of the carboxylic acid. The
second amido group donates as well as accepts one hydrogen
bond involving the incorporated water molecule, while the
protonated oxygen atom of the acid group hydrogen bonds to
the dioxane. In addition to these strong interactions, two halo-
gen bonds are observed from I1 to an amido oxygen atom O4
and from I3 to the water molecule. This packing results in
large cavities, which contain the solvent, and thus resembles
the packing of the DTS tetrahydrate. Unlike the DTA tetarto-
hydrate, the dioxane/water monosolvate does not incorporate
water molecules purely as space fillers. The hydrogen atoms of
the water molecules interact extensively with the host as well
as the other solvent molecules, and are thus an inherent part
of the structure.
–
À5.9
[d]
–
2.559(5)
2.841(5)
2.744(5)
2.844(5)
2.841(5)
2.986(3)
3.034(4)
168.4(5)
160.9(4)
147(5)
À52.2
À26.7
À46.6
À12.1
À21.8
À5.3
O3···O1W
N1···O1W
O1W···O1S
I1···O4
161.8(4)
[
b]
165.6
[
c]
I3···O1W
À10.9
[a] Averaged for two positions of disorder. [b] Angle after electrostatic op-
timisation of the water hydrogen atom positions. [c] Interaction energy is
a sum of the CÀH···O hydrogen bond and halogen bond. [d] Interaction
energy not to be deconvoluted from concomitant hydrogen bonds.
guest is connected to these chains through hydrogen bonds
from the protonated carboxylic acid oxygen atom to its hy-
droxyl oxygen atom, which furthermore hydrogen bonds to
the oxygen atoms of the amido group of the next host. This
links the individual chains of host molecules. Halogen bonds
can be detected from I1, I3 and I103 to the carbonyl oxygen
atoms of the amido groups of neighbouring DTA molecules.
These interactions are an additional stabilising factor between
the stacks of host molecules. Interestingly, this crystal form is
the most unstable of all reported, as it readily transforms to
the dihydrate, even when stored at low temperatures in the
methanol mother liquor. When taken out of solution, the
crystals crack and show strong pseudomorphosis due to loss
of solvent at ambient conditions.
Table 6 lists the interaction geometries and energies as
calculated by PACHA. It is clear that the strongest interaction
in all solvates involves the carboxylic acid group, followed by
homomeric interactions of the DTA through the amide groups.
In the case of the dioxane/water solvate, the second
strongest interaction involves the incorporated water mole-
cule, supporting the statement that the water is not only
a space-filler in this solvate.
Structural relationships
Finally, a mixed water/dioxane monosolvate (ratio of 1:1:1
water/dioxane/DTA) could be crystallised from dioxane solution
at 48C. The X-ray crystal structure was determined by synchro-
tron radiation using beam line I19 at Diamond Lightsource,
Taking into account that DTA crystallises in three non-solvated
and nine solvate forms, it is apparent that even though there
are the same number of hydrogen-bond donors and acceptors
in the DTA molecule, the compound has a strong propensity
to include solvent. This tendency could be due to the awkward
[
36]
Oxfordshire, UK in the monoclinic space group P2 /n with
1
Chem. Eur. J. 2015, 21, 1036 – 1047
www.chemeurj.org
1045
ꢂ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
shape of the molecule with an asymmetric tripodal arrange-
Conclusions
[37]
ment of the substituents whilst having a generally flat core;
and even though three non-solvated crystal forms exist, it is
possible that the solvent molecules stabilise intermittent
crystal forms, that is, the solvates, during the nucleation
process of a crystallisation from solution.
In the present study we report and analyse nine crystal forms
of diatrizoic acid (DTA) and one crystal form of its monosodium
salt (DTS); eight of all crystal forms have been characterised by
X-ray single-crystal diffraction. The structures of the dihydrate
and tetrahydrated sodium salt have been additionally investi-
gated by single-crystal neutron diffraction. All of the non-ionic
structures have been examined by means of non-empirical
lattice energy calculations (PACHA) to probe the interaction
energies of the different intermolecular interactions.
Another interesting finding is that when DTA interacts
purely with water, that is, in the case of the dihydrate and the
tetartohydrate, no halogen bonds are formed, in stark contrast
to the non-aqueous solvates. Even though there are short con-
tacts between oxygen and iodine atoms in the dihydrate,
these are not shorter than the sum of the van der Waals radii,
and PACHA calculations do not suggest that this is a stabilising
The presence of nine solvated crystal forms (including three
DMSO solvates previously reported) in addition to the three
unsolvated modifications clearly shows that diatrizoic acid
interaction. This result indicates that the contact is
a
consequence of the final packing arrangement rather than
being a driving force towards it. For the DTS tetrahydrate, no
halogen bonds were observed, which is likely due to the much
stronger forces of the ionic interactions shaping the crystal
packing.
exhibits a strong tendency to include hydrogen-bond
acceptor as well as donor solvents. However, this propensity
cannot be attributed, as generally assumed, to an imbalance of
[9b]
hydrogen-bond donor and acceptor groups,
molecule contains three of each.
as the
As found for the DMSO solvate series, the solvent molecules
in all of the solvates reported in the present work interact with
the host molecules by the carboxylic acid group. The only
exception is the tetartohydrate, for which the water molecule
fulfils a purely space-filling role without any interaction with
the host. In addition, all solvents hydrogen bond to the host
through an oxygen atom. This finding consolidates the hy-
pothesis that the strongest hydrogen-bond donor will interact
It was found that the strongest interaction in all diatrizoic
acid crystal forms involves the carboxylic acid moiety, which in
most cases donates a hydrogen bond to the incorporated sol-
vent. This strong interaction is likely to be retained during the
gradual desolvation of pre-crystallisation clusters in solution, in
which the interactions with lower energy are broken first to
form new homomeric hydrogen bonds in the growing nucleus,
while the strongest interaction is more likely to be retained.
While halogen bonding provides additional stabilisation for
all of the solvates, the incorporation of water apparently
switches off this interaction. This is likely due to more stable
hydrogen bonding, such as the formation of a capsule in the
dihydrate structure, in which a square of four water molecules
are enclosed between two host molecules, which then pack
into the final crystal structure in a way that is incompatible
with halogen bonding interactions. Even though short contacts
between the iodine residues and amide carbonyl groups are
present, these do not contribute to the overall stabilising
crystal lattice energy. Halogen bonds are observed, however,
for the mixed dioxane/water ternary complex.
[
38]
with the strongest hydrogen-bond acceptor. It is very likely
that during the nucleation process, this interaction is the last
one to be broken in most hydrogen-bond acceptor solvents as
[
26]
observed in the case of the DMSO solvates.
The conformation of the molecule does not follow any clear
trend when compared across all of the crystal structures. In the
case of the dihydrate, the DMSO disolvate and the THF mono-
solvate, the host adopts a syn-conformation, in which both car-
bonyl groups of the amide chains lie on one side of the rigid
core. All other crystal structures show an anti-conformation.
Considering the crystal packing, the DTA DMSO monosolvate
and the DTA THF monosolvate are isostructural, whilst the DTA
brickwork-like network can also be found in the DTA DMF dis-
olvate. The DTS tetrahydrate and the DTA dioxane water mon-
osolvate show a close relationship in their packing, whilst the
DTA tetartohydrate and the anhydrous Form I are isostructural.
Even though these similarities do not follow a clear trend, for
example, the DTA DMF disolvate is closer related to the DMSO
monosolvate than to the DMSO disolvate, this suggests that
certain packing motifs, such as the brickwork-like hydrogen
bonded layers, are stable enough to be realised in several crys-
tal forms. Considering the importance of solvates in finding
Diatrizoic acid represents a valuable case study in under-
standing the factors leading to the formation of solvate and
non-solvated polymorphic crystal forms of pharmaceutical
compounds. This study shows the clear preference of interac-
tion between host and solute by strong hydrogen-bond donor
and acceptor groups, even when different hydrogen-bonding
sites are present. This is invaluable information to feed back
into polymorph prediction algorithms, in order to optimise
these and generate purely in silico polymorph and solvent
[40]
screening tools.
[
3a]
[39]
new polymorphs, as recently shown for furosemide, for
which the DMSO solvate desolvates to a different anhydrous
crystal form than the solvates from THF, dioxane and DMF, Acknowledgements
understanding the interplay between host molecules and
solvent guests, as well as the influence of the solvents on
conformation and packing can hardly be overestimated.
We would like to thank the Institut Laue-Langevin, Grenoble,
France, and ANSTO, Lucas Heights, Australia, for the allocation
of neutron beam time (proposal numbers 5-12-250 and P1289,
respectively). In addition, we are grateful for the opportunity
Chem. Eur. J. 2015, 21, 1036 – 1047
www.chemeurj.org
1046
ꢂ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
to measure the crystal structure of the DTA tetartohydrate and
the DTA dioxane/water monosolvate on the beamline I19 at
Diamond Light Source, Didcot, United Kingdom, with beam
time allocated to the regional team of Newcastle and Durham
Universities. We thank the Engineering and Physical Sciences
Research Council for funding grant number EP/F063229/1.
[17] a) L. Meazza, J. A. Foster, K. Fucke, P. Metrangolo, G. Resnati, J. W. Steed,
Nat. Chem. 2013, 5, 42–47; b) A. Priimagi, G. Cavallo, P. Metrangolo, G.
Resnati, Acc. Chem. Res. 2013, 46, 2686–2695; c) D. Yan, D.-K. Bu cˇ ar, A.
Delori, B. Patel, G. O. Lloyd, W. Jones, X. Duan, Chem. Eur. J. 2013, 19,
8213–8219.
[18] C. B. Aakerçy, S. Panikkattu, P. D. Chopade, J. Desper, CrystEngComm
2
013, 15, 3125–3136.
[
[
[
19] M. Cametti, K. Raatikainen, P. Metrangolo, T. Pilati, G. Terraneo, G. Res-
nati, Org. Biomol. Chem. 2012, 10, 1329–1333.
20] Pharmacopoeia Europaea 8.1, Deutscher Apotheker Verlag, Stuttgart&-
Govi-Verlag GmbH, Eschborn, 2014.
21] United States Pharmacopoeia Vol 37, The Stationary Office, Norwich,
England, 2014.
Keywords: crystallography · halogen bonds · lattice energy
calculation · neutron diffraction · polymorphism
[
[
[
1] J. Bernstein, Polymorphism in Molecular Crystals, Clarendon Press,
Oxford, 2002.
[
[
22] R. K. Khankari, D. J. W. Grant, Thermochim. Acta 1995, 248, 61–79.
23] L. E. Tønnesen, B. F. Pedersen, J. Klaveness, Acta Chem. Scand. 1996, 50,
2] D. J. W. Grant in Polymorphism in Pharmaceutical Solids, Vol. 95 (Ed.:
H. G. Brittain), Marcel Dekker Inc., New York, 1999, pp. 1–33.
3] a) U. J. Griesser in Polymorphism (Ed.: R. Hilfiker), Wiley-VCH, Weinheim,
Germany, 2006, pp. 211–233; b) K. R. Morris in Polymorphism in Phar-
maceutical Solids, Vol. 95 (Ed.: H. G. Brittain), Marcel Dekker Ltd, New
York, 1999, pp. 125–181.
603–608.
[
[
[
[
[
24] C. Peng, K. G. Li, X. Y. Cao, T. T. Xiao, W. X. Hon, L. F. Zheng, R. Guo, M. W.
Shen, G. X. Zhang, X. Y. Shi, Nanoscale 2012, 4, 6768–6778.
25] V. A. Folen, G. Schwartzman, M. Maienthal, W. L. Brannon, J. Assoc. Of.
Anal. Chem. 1978, 61, 72–75.
26] K. Fucke, J. A. K. Howard, J. W. Steed, Chem. Commun. 2012, 48, 12065–
[
[
4] J. W. Steed, Trends Pharmacol. Sci. 2013, 34, 185–193.
5] a) J. Bernstein, Cryst. Growth Des. 2011, 11, 632–650; b) C.-H. Gu, V.
Young, D. J. W. Grant, J. Pharm. Sci. 2001, 90, 1878–1890; c) S. Byrn, R.
Pfeiffer, M. Ganey, C. Hoiberg, G. Poochikian, Pharm. Res. 1995, 12, 945–
1
2067.
27] K. Fucke, M. J. G. Peach, J. A. K. Howard, J. W. Steed, Chem. Commun.
012, 48, 9822–9824.
2
28] a) D. R. Turner, M. Henry, C. Wilkinson, G. J. McIntyre, S. A. Mason, A. E.
Goeta, J. W. Steed, J. Am. Chem. Soc. 2005, 127, 11063–11074; b) D. R.
Turner, M. B. Hursthouse, M. E. Light, J. W. Steed, Chem. Commun. 2004,
954; d) L.-F. Huang, W.-Q. Tong, Adv. Drug Delivery Rev. 2004, 56, 321–
334.
[
[
6] See ICH guideline Q6 A Point 3.3.1.
1354–1355.
7] a) S. L. Morissette, S. Soukasene, D. Levinson, M. J. Cima, O. Almarsson,
Proc. Natl. Acad. Sci. USA 2003, 100, 2180–2184; b) J. Bauer, S. Spanton,
R. Henry, J. Quick, W. Dziki, W. Porter, J. Morris, Pharm. Res. 2001, 18,
[
29] O. V. Dolomanov, L. J. Bourhis, R. J. Gildea, J. A. K. Howard, H. Pusch-
mann, J. Appl. Crystallogr. 2009, 42, 339–341.
[
30] a) A. O. F. Jones, M. H. Lemꢀe-Cailleau, D. M. S. Martins, G. J. McIntyre,
I. D. H. Oswald, C. R. Pulham, C. K. Spanswick, L. H. Thomas, C. C. Wilson,
Phys. Chem. Chem. Phys. 2012, 14, 13273–13283; b) C. C. Wilson, N.
Shankland, A. J. Florence, J. Chem. Soc. Faraday Trans. 1996, 92, 5051–
859–866.
[
[
8] a) A. Caridi, S. A. Kulkarni, G. Di Profio, E. Curcio, J. H. ter Horst, Cryst.
Growth Des. 2014, 14, 1135–1141; b) R. J. Davey, S. L. M. Schroeder, J. H.
ter Horst, Angew. Chem. 2013, 125, 2220–2234; Angew. Chem. Int. Ed.
5
057.
31] Z. Skoko, S. Zamir, P. Naumov, J. Bernstein, J. Am. Chem. Soc. 2010, 132,
4191–14202.
2013, 52, 2166–2179.
[
9] a) L. Infantes, L. Fabian, W. D. S. Motherwell, CrystEngComm 2007, 9,
1
6
5
5–71; b) C. H. Gorbitz, H.-P. Hersleth, Acta Crystallogr. 2000, B56, 526–
34.
[
[
[
32] A. Burger, R. Ramberger, Mikrochim. Acta 1979, 72, 259–271.
33] A. Burger, R. Ramberger, Mikrochim. Acta 1979, 72, 273–316.
34] S. N. Ivashevskaya, J. van de Streek, J. E. Djanhan, J. Bruening, E. Alig, M.
Bolte, M. U. Schmidt, P. Blaschka, H. W. Hoeffken, P. Erk, Acta Crystallogr.
[
10] C. A. Hunter, J. F. McCabe, A. Spitaleri, CrystEngComm 2012, 14, 7115–
117.
7
[
11] a) G. R. Desiraju, J. Chem. Sci. 2010, 122, 667–675; b) A. Gavezzotti, Mo-
lecular Aggregation: Structure Analysis and Molecular Simulation of Crys-
tals and Liquids, Oxford University Press, New York, USA, 2007; c) E. Cor-
radi, S. V. Meille, M. T. Messina, P. Metrangolo, G. Resnati, Angew. Chem.
2
009, B65, 212–222.
[
35] a) C. Wilkinson, J. A. Cowan, D. A. A. Myles, F. Cipriani, G. J. McIntyre,
Neutron News 2002, 13, 37–41; b) G. J. McIntyre, M. H. Lemꢀe-Cailleau,
C. Wilkinson, Physica B 2006, 385–386, 1055–1058.
2
000, 112, 1852–1856.
12] a) J. Bernstein, R. E. Davis, L. Shimoni, N.-L. Chang, Angew. Chem. 1995,
07, 1689–1708; Angew. Chem. Int. Ed. Engl. 1995, 34, 1555–1573;
[
[
[
[
[
36] H. Nowell, S. A. Barnett, K. E. Christensen, S. J. Teat, D. R. Allan, J. Syn-
chrotron Radiat. 2012, 19, 435–441.
37] K. M. Anderson, M. R. Probert, A. E. Goeta, J. W. Steed, CrystEngComm
[
1
b) G. A. Jeffrey, An Introduction to Hydrogen Bonding, 1st ed., Oxford
Univeristy Press, Oxford, 1997; c) S. Schultes, C. de Graaf, H. Berger, M.
Mayer, A. Steffen, E. E. J. Haaksma, I. J. P. de Esch, R. Leurs, O. Kramer,
MedChemComm 2012, 3, 584–591.
2
011, 13, 83–87.
38] a) C. B. Aakerçy, N. Schultheiss, J. Desper, J. Mol. Struct. 2010, 972, 35–
0; b) M. C. Etter, Acc. Chem. Res. 1990, 23, 120–126.
4
39] V. S. Minkov, A. A. Beloborodova, V. A. Drebushchak, E. V. Boldyreva,
Cryst. Growth Des. 2014, 14, 513–522.
[
[
[
13] a) K. Fucke, J. W. Steed, Water 2010, 2, 333–350; b) C. C. Wilson, Z. Kris-
tallogr. 2005, 220, 385–398.
14] S. Grabowsky, D. Jayatilaka, R. F. Fink, T. Schirmeister, B. Engels, Z. Anorg.
Allg. Chem. 2013, 639, 1905–1921.
15] a) K. Fucke, G. J. McIntyre, C. Wilkinson, M. Henry, J. A. K. Howard, J. W.
Steed, Cryst. Growth Des. 2012, 12, 1395–1401; b) K. Fucke, K. M. Ander-
son, M. H. Filby, M. Henry, J. Wright, S. A. Mason, M. J. Gutmann, L. J.
Barbour, C. Oliver, A. W. Coleman, J. L. Atwood, J. A. K. Howard, J. W.
Steed, Chem. Eur. J. 2011, 17, 10259–10271; c) M. Henry, ChemPhys-
Chem 2002, 3, 561–569.
40] a) S. L. Price, Chem. Soc. Rev. 2014, 43, 2098–2111; b) S. Mohamed, D. A.
Tocher, S. L. Price, Int. J. Pharm. 2011, 418, 187–198; c) D. E. Braun, P. G.
Karamertzanis, S. L. Price, Chem. Commun. 2011, 47, 5443–5445; d) S. L.
Price, Acc. Chem. Res. 2009, 42, 117–126; e) J. Kendrick, G. A. Stephen-
son, M. A. Neumann, F. J. J. Leusen, Cryst. Growth Des. 2013, 13, 581–
589; f) M. A. Neumann, F. J. J. Leusen, J. Kendrick, Angew. Chem. 2008,
1
20, 2461–2464; Angew. Chem. Int. Ed. 2008, 47, 2427–2430; g) M. D.
Eddleston, K. E. Hejczyk, E. G. Bithell, G. M. Day, W. Jones, Chem. Eur. J.
013, 19, 7874–7882.
2
[
16] a) C. B. Aakerçy, M. Baldrighi, J. Desper, P. Metrangolo, G. Resnati, Chem.
Eur. J. 2013, 19, 16240–16247; b) M. Baldrighi, G. Cavallo, M. R. Chierot-
ti, R. Gobetto, P. Metrangolo, T. Pilati, G. Resnati, G. Terraneo, Mol.
Pharm. 2013, 10, 1760–1772.
Received: August 1, 2014
Published online on November 4, 2014
Chem. Eur. J. 2015, 21, 1036 – 1047
www.chemeurj.org
1047
ꢂ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim