Influence of Bio-Isosteric Replacement on the Formation of Templating Methanol and Acetonitrile Solvates in Lophines

Article abstract of DOI:10.1021/acs.cgd.6b00655

Bio-isosteric replacement is a frequently used tool in medicinal chemistry. While the pharmacological activity is not influenced by the exchange of substituents, the solid-state characteristics and formation of different crystal forms may well be altered dramatically, jeopardizing the processability and safety of the drug compound. In this study we investigate a series of triphenylimidazole (TPI) derivatives as model compounds with the bio-isosteric exchange of only one halogen position (F, Cl, Br, I). Crystallization from two industrially used solvents (methanol and acetonitrile) reveals solvate formation of all TPIs, for which the basic hydrogen bonded motif does not change. The three-dimensional packing depends on the size of the substituent and changes from fluoro- to chloro- and bromo-substitution but remains the same for the larger iodo-substituent. From acetonitrile, only F-TPI and Cl-TPI form an isomorphic channel solvate, which in both cases desolvates reversibly to an isomorphic crystal form. Due to the halogen atom lining of the channels, bromine and iodine are too large to generate a stable packing. This study illustrates the importance of understanding the influence of bio-isosteric substitution on the solid state, in order to best utilize this common tool.

Full text of DOI:10.1021/acs.cgd.6b00655

Article
pubs.acs.org/crystal
Influence of Bio-Isosteric Replacement on the Formation of
Templating Methanol and Acetonitrile Solvates in Lophines
Thomas Kitchen,† Connor Melvin,† Mohd Nadzri Mohd Najib,† Andrei S. Batsanov,‡
and Katharina Edkins*,†
†School of Medicine, Pharmacy and Health, Durham University Queen’s Campus, University Boulevard, Stockton-on-Tees, TS17
6BH, United Kingdom
‡Department of Chemistry, Durham University, South Road, Durham, DH1 3LE, United Kingdom
S
* Supporting Information
ABSTRACT: Bio-isosteric replacement is a frequently used tool in medicinal chemistry.
While the pharmacological activity is not influenced by the exchange of substituents, the
solid-state characteristics and formation of different crystal forms may well be altered
dramatically, jeopardizing the processability and safety of the drug compound. In this study
we investigate a series of triphenylimidazole (TPI) derivatives as model compounds with
the bio-isosteric exchange of only one halogen position (F, Cl, Br, I). Crystallization from
two industrially used solvents (methanol and acetonitrile) reveals solvate formation of all
TPIs, for which the basic hydrogen bonded motif does not change. The three-dimensional
packing depends on the size of the substituent and changes from fluoro- to chloro- and
bromo-substitution but remains the same for the larger iodo-substituent. From acetonitrile,
only F-TPI and Cl-TPI form an isomorphic channel solvate, which in both cases desolvates
reversibly to an isomorphic crystal form. Due to the halogen atom lining of the channels,
bromine and iodine are too large to generate a stable packing. This study illustrates the
importance of understanding the influence of bio-isosteric substitution on the solid state,
in order to best utilize this common tool.
thus its pharmacological activity in patients. In addition to
crystal forms that contain the desired molecule, solvates
incorporate solvent molecules from the crystallization proc-
ess.13 Solvates can be interpreted as intermediates formed
during the crystallization process,14,15 and can introduce traces
of unwanted solvent into the pharmaceutical product, which
can prove harmful or toxic to the patient. Thus, the levels of
residual solvent in these applications are regulated.16 On the
other hand, solvates represent a route to obtain novel
unsolvated crystal forms, which are not accessible by direct
crystallization.17−22
The inclusion of solvent molecules into crystal forms has
been widely studied, not only in the pharmaceutical field. The
influence of solvate formation has been studied in a co-crystal
system that has been crystal-engineered to undergo photo-
dimerization. It was found that the inclusion of acetonitrile
leads to a molecular arrangement allowing photodimerization,
while the methanol solvate hinders this.23 In other studies,
inclusion has been utilized to separate xylenes24 showing a clear
preference of one guest over another,25 or to store volatile
guests.26,27 Inclusion complexes have recently become the focus
of attention to study the incorporated molecules, either in an
INTRODUCTION
■
The development of new pharmaceutical compounds to treat
emerging and pandemic diseases such as ebola or long-standing
problems such as bacterial infections is among the key topics of
pharmaceutical research. It demands a steady stream of novel
pharmaceutical compounds to tackle the disease in question.
One of the tools used in the area of medicinal chemistry is bio-
isosteric replacement, which is the replacement of substituents
of a lead molecule with other chemical groups without changing
the overall biological activity.1,2 This enables the medicinal
chemist to improve specific properties, such as solubility,
crystallinity, and metabolic stability, which subsequently will
improve the overall processability of the new drug compound.
This concept has been applied to such diverse drug classes as
cannabinoids3 and HIV antivirals.4 The selection of replace-
ment groups is nowadays heavily helped by in silico studies and
property predictions.5−7
Even though this approach has been used for over 60 years,8
the implications of bio-isosteric replacements of substituents,
which sometimes differ vastly in size and/or electronics, on the
crystallization properties of the new molecule have been widely
ignored. Small organic molecules show a high probability of
crystallizing in multiple crystal forms, which may differ in
physicochemical characteristics such as melting point or
solubility.9−12 The latter characteristic is especially important
for pharmaceuticals, as solubility and especially water solubility
is directly linked to bioavailability of the drug compound and
Received: April 30, 2016
Revised: June 12, 2016
Published: June 20, 2016
© 2016 American Chemical Society
4531
DOI: 10.1021/acs.cgd.6b00655
Cryst. Growth Des. 2016, 16, 4531−4538

Crystal Growth & Design
Article
unusual conformation,28 or to obtain structural information on
EXPERIMENTAL SECTION
■
a hard to crystallize compound.29,30
Materials. Benzil, p-halogenbenzaldehyde (F, Cl, Br, I),
ammonium acetate, potassium carbonate, and all organic solvents
were purchased from Sigma-Aldrich and used without further
purification.
Methods. Synthesis. All TPIs were synthesized using the literature
method.39 Equimolar amounts of benzil and substituted benzaldehyde
were suspended in acetic acid with a 10-fold amount of ammonium
acetate. The suspension was then heated in a microwave to 180 °C for
5 min and cooled to room temperature. The resulting solution was
then poured into ice water and neutralized with K2CO3 forming a
beige precipitate. This was collected by filtration and washed with H2O
before recrystallization.
Crystallization Methods. Approximately 10−20 mg compound was
used in each crystallization experiment. For slow cooling and fast
cooling experiments, sufficient solvent was added to produce a hot
saturated solution. These were then left to cool to room temperature
in an insulating wooden block (slow cooling) or quenched in an ice
and water bath (fast cooling). For evaporation, the material was
dissolved in sufficient solvent to produce an undersaturated solution at
room temperature and allowed to stand to evaporate. Precipitation was
conducted by producing a hot saturated solution and adding at least
double the volume of cold (room temperature) diethyl ether as
antisolvent. The crystals forming were immediately removed from
solution in the case of the fast cooling and precipitation experiments,
while evaporation and slow cooling experiments were left for 1 day
before analysis.
Considering the variation in size and interaction potential in
bio-isosteric groups, the choice of replacement group will
influence the crystal structure of the substances, as has recently
been shown for a series of trityl compounds.31 In addition,
change of halogen substituents from chloro via bromo to iodo
on phloroglucinol has been shown to influence the
incorporation mode of water into the crystal structure.32 We
are thus systematically investigating the influence that the
change of only one substituent has on the crystallization
behavior and polymorphism of model compounds.33 In order
to reduce the possibility of strong interactions between the
molecules, we use 2,4,5-triphenylimidazoles (lophine, TPI,
Figure 1) as our model, which can only donate and accept one
Differential Scanning Calorimetry. Differential scanning calorim-
etry (DSC) was performed with a Q2000 (TA Instruments, New
Castle, DE, USA). Approximately 1 to 3 mg of sample was accurately
Figure 1. Molecular structure of 2,4,5-triphenylimidazole (TPI)
derivatives (lophines) used in this study.
weighed (microbalance, Sartorius, Gottingen, Germany) into sealed Al
̈
pans and submitted to a ramp scan in the range from 25 to 370 °C
with a heating rate of 10 °C min−1. Dry nitrogen was used as the purge
gas (purge: 50 mL min−1). The instrument was calibrated for
temperature and energy in the respective temperature range with pure
indium (purity 99.999%, mp 156.6 °C, heat of fusion 28.45 J g−1).
Thermogravimetric Analysis. Measurements were performed with
a Q500 thermogravimetric system (TA Instruments, New Castle, DE,
USA). Samples of approximately 2 to 6 mg were accurately weighed
into 50 μL platinum pans and subjected to non-isothermal TGA at a
heating rate of 10 °C min−1. Nitrogen was used as purge gas at a
purging rate of 60 mL min−1.
X-ray Diffractometry. Powder X-ray diffraction patterns were
recorded using an Empyrean diffractometer (Panalytical, Almelo,
Netherlands) in Bragg−Brentano geometry with Cu Kα radiation,
graphite monochromator, 0.2 mm fixed Soller slits, and PIXcel
detector. The X-ray tube was operated at 40 kV and 40 mA, while the
samples were prepared on a low-background silicone slide sample
holder as dry powder.
hydrogen bond and show thus a reduced complexity compared
to pharmaceutically used compounds. Lophine derivatives are
used as core structures in medicinal chemistry,34 as well as for
photophysical applications.35−37
During our work on the polymorphism of halogenated TPIs,
we investigated a series of 2-(4′-p-X-phenyl)-4,5-diphenylimi-
dazole (X-TPI; X = F, Cl, Br, and I). These substituents are
frequently exchanged for bio-isosteric replacement,38 and do
not significantly change the overall electronics of the resulting
compounds. However, the substituents vary considerably in
size, with F having the smallest van der Waals radius with 1.47
Å, Cl with 1.75 Å, Br with 1.85 Å to I with 1.98 Å. In the
present study we report our results of crystallization experi-
ments with two standard solvents, methanol and acetonitrile,
which are frequently used in pharmaceutical processes.
Figure 2. Thermoanalytical data of the four TPI methanol solvates: (a) TG thermograms and (b) DSC thermograms.
4532
DOI: 10.1021/acs.cgd.6b00655
Cryst. Growth Des. 2016, 16, 4531−4538

Crystal Growth & Design
Article
Table 1. Weight Loss of the Methanol Solvates as Determined by TG Analysis
weight loss range
weight loss
[%]
molar ratio
calc. weight loss monomolar
[%]
melting unsolvated form (onset)
[°C]
[°C]
F-TPI
Cl-TPI
Br-TPI
I-TPI
52−92
9.31
8.42
8.32
6.80
1.007
0.949
1.063
0.962
9.250
8.831
7.867
7.053
252.9
262.7
261.2
268.0
70−102
70−118
73−100
Single crystals of F-TPI and Br-TPI methanol solvates were picked,
coated in perfluoro polyether oil, and mounted on MiTeGen sample
holders. These were placed directly into a precooled cryostream on a
Bruker (Karlsruhe, Germany) D8 Quest diffractometer using Mo Kα
radiation (0.71073 Å). Single crystals of Cl-TPI acetonitrile solvate
were picked, coated in perfluoro polyether oil, and mounted on
MiTeGen sample holders. The diffraction experiment was carried out
on a Rigaku (Sevenoaks, UK) Saturn 724+ diffractometer at Station
I19 of the Diamond Light Source synchrotron (undulator, λ = 0.6889
Å, ω-scan, 1.0° per frame). The data from a four-component
nonmerohedral twin were processed using Bruker Apex-II software,
deconvoluted using CELL_NOW,40 and corrected for absorption
using TWINABS 2012/1.41 The structure was solved using
Olex.solve42 and refined using ShelXL.43 All non-hydrogen atoms
could be located from Fourier difference maps and were refined
anisotropically. Hydrogen atoms were calculated as geometrical riding
models on the heavy atom and refined isotropically.
Table 2. Crystallographic Data of the Methanol Solvates of
F-TPI and Br-TPI as well as the Acetonitrile Solvate of Cl-
TPI as Determined by X-ray Single Crystal Diffraction
crystal form
formula
F-TPI ×
Br-TPI ×
Cl-TPI × AcN
MeOH
MeOH
C21H15N2F ×
CH4O
C21H15N2Br ×
CH4O
C21H15N2Cl ×
0.25C2H3N
Mr (g mol−1
)
346.39
monoclinic
P2
407.31
monoclinic
P21/c
341.06
monoclinic
Pn
crystal system
space group
T (K)
120
120
100
a (Å)
13.164(1)
5.9105(6)
23.800(2)
90
5.9477(3)
11.6799(7)
27.003(2)
90
8.854(1)
19.875(3)
19.924(3)
90
b (Å)
c (Å)
α (°)
β (°)
104.421(3)
90
91.541(2)
90
93.712(2)
90
γ (°)
V (Å3)
Z
RESULTS AND DISCUSSION
1793.4(3)
4
1875.2(2)
4
3498.6(9)
8
■
Methanol Solvates. Crystallization of all TPI derivatives
from methanol yielded a crystal form containing solvent of
crystallization. The analysis of the respective TG thermograms
(Figure 2a) reveals a weight loss corresponding to 1 mol
methanol per mol TPI (Table 1) for all crystal forms. The
weight loss of F-TPI starts almost immediately with a slow
zero-order loss of solvent, which is most likely the evaporation
of trapped and surface-bound solvent. The weight loss speeds
up at temperatures above 60 °C resulting in a one-step solvent
release. The solvates of Cl-, Br-, and I-TPI show slightly higher
stability with the start of the weight loss not below 70 °C. Cl-
TPI shows a single mass loss step comparable to F-TPI, while
Br-TPI and I-TPI show multiple steps. All desolvation events
are in the same temperature range as the boiling point of
methanol, which indicates the high stability of these crystal
forms.
density (calc.,
1.283
1.443
1.295
g cm−3
)
abs coeff., (mm−1
)
0.087
728
2.205
831
0.208
1420
F(000)
crystal size (mm3) 0.237 × 0.231
× 0.12
0.619 × 0.188 × 0.5 × 0.02 × 0.02
0.093
θ range for data
collection
2.066−25.999
1.90−26.00
0.993−24.415
index ranges
−15 ≤ h ≤ 15 −7 ≤ h ≤ 7
−10 ≤ h ≤ 10
−23 ≤ k ≤ 23
−23 ≤ l ≤ 23
34297
−7 ≤ k ≤ 7
−29 ≤ l ≤ 29
20125
−14 ≤ k ≤ 14
−34 ≤ l ≤ 34
51111
reflections col.
independent refl.
7045
3697
5902
refinement
method
full-matrix least-squares on F2
Data/restraints/
parameter
7045/1/477
3697/0/236
1.0312
34287/2/896
The corresponding DSC thermograms (Figure 2b) show the
desolvation as endothermic events in temperature regions
comparable to those detected by TG. Interestingly, Br-TPI and
I-TPI show double peaks, which might be due to partial
desolvation followed by melting of the remaining intact
solvated crystal form, i.e., passing of the peritectic. This
observation fits well with the two-step desolvation detected by
TG. All TPI solvates desolvate into a stable crystalline form,
which melts at temperatures above 250 °C.
The methanol solvates of F-TPI and Br-TPI could be
crystallized in sufficient quality to obtain single crystal
structures. The crystallographic data of these two crystal
forms are listed in Table 2. While the most common packing
motif of imidazole derivatives is the hydrogen bonded
chain,35,44 this motif is broken in the case of the methanol
solvates. In both crystal forms, the methanol molecule
intercalates into the chains accepting one hydrogen bond
from one imidazole and donates a second to the next imidazole
(Figure 3a). The TPI molecules are oriented counter-parallel
while the chains show a zigzag ribbon like conformation with
goodness of fit on 1.021
1.012
F2
R indices [I > 2σI] R1 = 0.0583
wR2 = 0.1091
R1 = 0.0330
wR2 = 0.1127
R1 = 0.0427
wR2 = 0.1255
R1 = 0.0444
wR2 = 0.1043
R1 = 0.0509
wR2 = 0.1068
R indices (all data) R1 = 0.1137
wR2 = 0.1260
the edge being formed by the methanol molecules (Figure 3b).
These then stack on top of each other. The next stacks are
connected through weaker C−H···π interactions of adjacent
nonsubstituted phenyl rings. Even though the bromine
substituents are closer than twice their van der Waals radius
(distance: 3.5472(5) Å with 2 × vdW radius: 3.7 Å), the
orientation of the TPI molecules, and therefore the geometry of
the Br−Br interaction, does not directly point toward halogen
bonding, and thus this interaction is less likely to be a driving
force of this packing motif.
Interestingly, the packing of the stacks is different for F-TPI
compared to Br-TPI. While the fluoro-substituent is small
enough to allow the phenyl rings to properly interlock giving a
4533
DOI: 10.1021/acs.cgd.6b00655
Cryst. Growth Des. 2016, 16, 4531−4538

Crystal Growth & Design
Article
It can be assumed, however, that the hydrogen-bonded chains
persist in all four crystal forms. This is corroborated by the very
similar desolvation temperatures of all four crystal forms
indicating that the desolvation process has a comparable
activation energy, i.e., the energy necessary to break the
interactions between the solvent and the host molecules. If the
basic packing motif would change drastically between the four
solvates, a wider variation of desolvation temperatures would be
anticipated alongside different desolvation kinetics.
The powder X-ray diffractograms of all four TPI derivatives
show certain similarities (Figure 5). While Br-TPI and I-TPI
Figure 3. Molecular structure of Br-TPI methanol monosolvate: (a)
hydrogen bonded chains with intercalated methanol and (b) stacking
of the chains in the crystal structure, hydrogen atoms are omitted for
clarity. Color code: carbon (gray), hydrogen (light gray), nitrogen
(blue), oxygen (red), bromine (brown).
dense packing, the bromine atom is too large for this motif.
This results in two distinct changes of the packing: first,
adjacent layers in the crystallographic a-direction of Br-TPI
move farther apart, and second, the layers shift in the
crystallographic b-direction of Br-TPI so that the bromine
atoms orient along channels. This results in a higher symmetry
of the crystal structure as well as a slightly different long-range
packing. While the F-TPI methanol solvate crystallizes in
identical layers, i.e., AAAA, Br-TPI methanol solvate shows
alternating layers, i.e., ABAB, as can be seen by the shift in
opposite directions of the Br−Br pairs (Figure 4). This
Figure 5. Powder X-ray diffractograms of the methanol solvates of the
four TPI derivatives. Calculated patterns of the single crystal structures
in light gray.
can be regarded as isostructural, which can be detected as a shift
of the peak positions without changing the overall pattern,
there is no similarity between these two, F-TPI and Cl-TPI. It
can thus be assumed that the packing of Br-TPI is the most
efficient for the compounds with large substituents in this
series, while the smaller ones can pack in different arrange-
ments of the basic hydrogen bonded planes.
Upon desolvation, all methanol solvates transform into an
unsolvated crystal phase (Figure 6). All of these show low
crystallinity and a comparably high amorphous background, but
Figure 4. Molecular packing in the methanol solvates of F-TPI vs Br-
TPI.
correlation is reminiscent of polytypism, even though these two
crystal forms change in their chemical entities (F vs Br
substitution) and thus the term can, strictly speaking, not be
used.45
The crystallographic symmetry of the F-TPI methanol
solvate generates disorder in the hydrogen bonded chains.
The asymmetric unit contains three methanol molecules, of
which two lie on special positions on the twofold axis. Thus,
their OH-groups show disorder over two positions generating
disorder in the position of the proton of the imidazole ring.
Even though the third methanol molecule can be refined
without disorder due to its general position, the hydrogen
bonded chain forces its OH-group to be disordered as well.
Since refinement of the structure in P1 resolves the disorder, it
can be assumed that the crystallographic symmetry would be
broken if hydrogen atom positions could be located accurately,
e.g., by using neutron diffraction.
Since only two of the four solvates could be crystallized for
single crystal diffraction, it is difficult to make definite
statements about the packing of the other two crystal forms.
Figure 6. Powder X-ray diffractograms of the crystal forms after
desolvation of the methanol monosolvates.
4534
DOI: 10.1021/acs.cgd.6b00655
Cryst. Growth Des. 2016, 16, 4531−4538

Crystal Growth & Design
Article
Figure 7. TG and DSC thermograms of acetonitrile solvates of F-TPI and Cl-TPI.
reveal distinctly different peak patterns and intensities proving
them to be different crystal forms to the solvates. However, it
can be observed that the desolvated crystal forms of F-TPI and
Cl-TPI are isomorphous. This is surprising, as the methanol
solvates were not structurally related. The Br-TPI and I-TPI
crystal forms, on the other hand, reveal no similarity in their
desolvated modification unlike their methanol solvates. It can
thus be assumed that the desolvation and concomitant
reorganization process is significantly influenced by the
different sizes of the substituent and similar packing arrange-
ments are not energetically preferable.
Acetonitrile Solvates. Crystallization of F-TPI from
acetonitrile yielded a crystal form, which TG indicated to be
a solvate (Figure 7a). The weight loss in the range from 60 to
150 °C of 3.194% corresponds well to a stoichiometry of a
tetarto-solvate (0.25 mol acetonitrile to 1 mol F-TPI, calculated
mass loss: 3.162%), while the broad desolvation temperature
range suggests slow desolvation kinetics. The corresponding
differential scanning calorimetry trace shows a well-matched
desolvation endotherm in the same temperature range followed
by a melting endotherm at 252.3 °C (onset) of the
corresponding unsolvated crystal form.
The thermal data of the Cl-TPI solvate (Figure 7b) shows a
lower mass loss of 2.942%, as detected by TGA, which again
corresponds to a tetarto solvate (calculated weight loss
3.009%). The desolvation is almost undetectable by DSC,
which is probably due to the broad range of this event from
50 °C to almost 200 °C. The respective heat flow will be so low
that it is almost indistinguishable from the baseline. However, a
small but reproducible endothermic peak can be detected from
150 to 167 °C, which is most likely connected to the
desolvation event. Melting subsequently occurs at 262.5 °C.
The change from chlorine to bromine (1.85 Å van der Waals
radius) as the substituent proves too big a change for the
acetonitrile solvate formation, and the resulting crystal form is
unsolvated (Figure S1). The X-ray powder pattern confirms the
presence of a crystal form that is not structurally related to the
acetonitrile solvates of F-TPI or Cl-TPI as observable by the
distinctly different peak positions. The same is true for iodo-
substituted TPI, which shows no detectable weight loss in the
TGA measurements (Figure S2) and has distinctly different
peak positions in the powder pattern. This I-TPI unsolvated
crystal form shows no similarity to any of the other crystal
forms from acetonitrile described above for the other halogens.
Although the investigated solvates of F-TPI and Cl-TPI show
a strongly anisotropic crystal growth with a needle-shaped
habit, we were able to solve the crystal structure of the Cl-TPI
solvate on instrument I19 at the Diamond Light Source,
Didcot, UK. The crystal form crystallizes in the monoclinic
spacegroup Pn with four molecules of Cl-TPI and one molecule
of acetonitrile in the asymmetric unit. The Cl-TPI molecules
arrange in the well-reported infinite chains through hydrogen-
bonding along the imidazole rings, which are oriented along the
crystallographic a-axis and also along the preferred axis of
growth (Figure S3). In between four chains of host molecules
are channels filled with the solvent, which again is oriented
along the crystallographic a-axis (Figure 9). These channels are
lined alternatingly with a ring of chlorine atoms and a layer of
phenyl rings interacting with the nitrogen atom of the solvent
through C−H···N hydrogen bonds, which are generally
regarded to be weak. The channels, however, are stabilized by
Cl···Cl halogen bonds (Figure S4), 3.38−3.63 Å in length, and
at a bond angle of 155−162°.
Changing the TPI substituent from fluorine to chlorine
(vdW radius 1.47 Å to 1.77 Å) seems not to prevent the
formation of this solvate. Powder X-ray diffraction proved that
the Cl-TPI solvate is isostructural to that of F-TPI (Figure 8).
The diameter of these pores is approximately 4.93 Å, which is
only 1.39 Å more than double the van der Waals radius of
chlorine. This might explain the desolvation behavior observed
by TGA over a wide temperature range. The channels
containing the acetonitrile have a comparable structure to a
string of pearls, in which the pearls represent sufficient void
Figure 8. Powder X-ray diffractograms of the crystal forms crystallizing
from acetonitrile.
4535
DOI: 10.1021/acs.cgd.6b00655
Cryst. Growth Des. 2016, 16, 4531−4538

Crystal Growth & Design
Article
smaller obstacle to the diffusion of acetonitrile through the
pores. This explains the narrower temperature range of the
desolvation observed by TGA, while the initial stability and
higher onset temperature of the weight loss can be explained by
the energy necessary to overcome the intermolecular
interactions between the solvent and the host molecules in
both cases.
In order to show the close relationship of the solvate to the
respective desolvated crystal form of F-TPI and Cl-TPI, both
solvates were desolvated by heating up to 190 °C using the
TGA and following the weight change to ensure complete
desolvation. PXRD analysis confirmed the presence of a new
crystal form (Figures 10), which shows peak shift to higher
angles and thus smaller d-spacings when compared to the
original solvated crystal structure. Even though these shifts are
comparably large, the overall powder patterns change little. As
the solvated crystal forms, the desolvated forms of F-TPI and
Cl-TPI are isostructural. Another indication of the isostructur-
ality of both solvates with their respective desolvates is the
absence of pseudomorphosis or a moving front indicating
single-crystal to single-crystal transition during desolvation, as
has been proven by thermomicroscopy. Final proof, however,
would only be possible by indexing the powder pattern and
solving the crystal structure, which is the scope of ongoing
work. If the two solvate/desolvate pairs are indeed isostructural,
this would corroborate our hypothesis of the solvent leaving the
crystal structure by diffusion along the channels. Neither of the
new crystal forms is related to those observed for Br-TPI or I-
TPI.
Figure 9. Crystal structure of Cl-TPI acetonitrile solvate along the
crystallographic a-axis. Hydrogen atoms omitted for clarity; atomic
displacement ellipsoids drawn at 50%, color code: carbon (gray),
nitrogen (blue), chlorine (green).
space for acetonitrile to diffuse through (i.e., the spaces lined
with phenyl rings), while the string spaces represent the
constriction by the chlorine atoms. When looked at statically,
the solvent molecule would not be able to diffuse through these
channels and desolvation would be accompanied by destruction
of the crystal structure and correlated pseudomorphosis (as
seen for the methanol solvates) or at least cracking of the
crystal in order to release the solvent. However, as has recently
been shown for a porous organic cage structure, crystalline
materials are not static and molecules within the structure can
move around their position.46 The authors of this study show
that certain gases can induce a higher nominal porosity of the
material by accessing void space within the structure that is
only accessible for very short periods of time.
The unsolvated crystal forms were then stored in a saturated
atmosphere of acetonitrile to probe the reversibility of the
desolvation process (Figures 11 and S5). After 7 days of
Since the F-TPI acetonitrile solvate shows an isostructural
powder pattern to Cl-TPI (Figure 10), the crystal structure has
the same string-of-pearls channels with constriction through the
fluorine atoms. If we assume that the overall crystal structure
does not change and the pore size stays approximately the
same, the pore diameter would be 2 Å more than the van der
Waals radius of the lining fluorine atoms, and thus present a
Figure 11. Powder X-ray patterns of solvated, desolvated, and
resolvated acetonitrile solvate of F-TPI.
storage, the PXRD pattern of the materials showed full
resolvation to the acetonitrile solvates, while storage for a
further 5 days only improves the crystallinity of the sample, as
observed by the increased sharpness of the peaks. The
resolvation upon exposure to saturated solvent atmosphere
could be explained by at least partial retention of the porous
structure in the desolvated crystal structure, which contracts in
order to fill the voids and thus minimizes the crystal lattice
energy. This would facilitate the reuptake of solvent without the
necessity of a phase change. Such behavior has been shown for
Figure 10. Powder X-ray patterns of acetonitrile solvates and
respective desolvates of F-TPI and Cl-TPI.
4536
DOI: 10.1021/acs.cgd.6b00655
Cryst. Growth Des. 2016, 16, 4531−4538

Crystal Growth & Design
Article
pharmaceutical hydrates,47,48 and can pose problems during
manufacture. In this case, the formation of an isostructural
desolvate poses a way to generate a new anhydrous crystal
form.17,49
Desolvation did not occur when the solvates were stored for
7 days under dry conditions (phosphorus pentoxide) at
ambient temperature, which shows that a certain amount of
activation energy is necessary to start the desolvation. However,
storing the solvate at ambient conditions, i.e., at approximately
50% relative humidity at 23 °C, led to a transition to the
desolvate over a period of 2 weeks. It is thus likely that water
plays an important role in the desolvation of these solvates, and
this is the focus of ongoing studies.
Unsolvated Crystal Forms. Each of the TPI derivatives
shows two unsolvated crystal forms, either from desolvation of
the solvates or, in the case of Br-TPI and I-TPI, by direct
crystallization from acetonitrile (Figure S6). The F-TPI and Cl-
TPI desolvates from both solvates are the same modification.
This is surprising, as it has been shown above that the
acetonitrile solvate and its desolvate are related or even
isostructural. It can be assumed that, in this case, the desolvate
is so stable it forms during the pseudomorphosis and connected
reorganization of the crystal lattice upon release of methanol
from the methanol solvate. In the case of Br-TPI, the desolvate
of the methanol solvate results in a different crystal form than
direct crystallization from acetonitrile. This shows the
significance of using solvate formation during crystal form
discovery, as has been frequently shown for organic
compounds.17,20,47,49 I-TPI results in the same crystal form
independent from the solvent/solvate, which again points
toward one energetically favored modification.
and iodine substitution proves to be too much of a steric
hindrance for the packing to be stable, and thus Br-TPI and I-
TPI crystallize in nonsolvated crystal forms. Two different
unsolvated crystal forms can be obtained for Br-TPI by
desolvation of the solvates or direct crystallization from
acetonitrile, while F-TPI, Cl-TPI, and I-TPI crystallize in the
same unsolvated crystal form. Overall, there seems to be related
behavior between fluoro- and chloro-substitution as well as
bromo-and iodo-substitution with little correlation between
chloro- and bromo-substitution. This is an interesting finding
and should be verified with additional data.
Investigation of this series of model compounds for bio-
isosteric replacement with reduced complexity shows the
significant influence of the size of the substituent on the
crystallization of different modifications. While this can be a big
advantage to engineer the appropriate crystal form or particle
size, e.g., through desolvation and correlated pseudomorphosis,
it can also pose the problem of residual solvent or unwanted
solvate formation and thus increase the number of processing
steps needed. It is clear from this study that the simple
exchange of one substituent for its bio-isostere can have
massive influence on the solid-state properties of a compound
and that more research into this field is needed.
ASSOCIATED CONTENT
* Supporting Information
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/acs.cgd.6b00655.
■
S
Additional thermal and crystallographic data (PDF)
Accession Codes
Even though this study shows clearly the influence the simple
exchange of one substituent has on the crystallization
characteristics of the resulting compounds, the data presented
here are only the early stage of the investigation, and more
information is needed to draw definite conclusions of the
substituent influence on the solid-state behavior. What is
blatantly clear is that the simple exchange of a halogen
substituent on a simple model compound with reduced
complexity can significantly influence the solid-state landscape
and the crystal forms available. While the change from fluorine
to chlorine seems to have less of an effect, exchanging chlorine
with bromine effects a more radical change in the solid-state
behavior. The iodo-substituent then follows the behavior of
bromine.
CCDC 1449859 and 1477419−1477420 contain the supple-
mentary crystallographic data for this paper. These data can be
obtained free of charge via www.ccdc.cam.ac.uk/data_request/
cif, or by emailing data_request@ccdc.cam.ac.uk, or by
contacting The Cambridge Crystallographic Data Centre, 12,
Union Road, Cambridge CB2 1EZ, UK; fax: +44 1223 336033.
AUTHOR INFORMATION
Corresponding Author
*E-mail: katharina.fucke@durham.ac.uk.
■
Author Contributions
The manuscript was written through contributions of all
authors. All authors have given approval to the final version of
the manuscript.
Notes
CONCLUSION
■
The authors declare no competing financial interest.
In conclusion, we have reported the solvate formation of a
series of bio-isosterically substituted triphenylimidazoles (2-(4′-
p-X-phenyl)-4,5-diphenylimidazole, X-TPI; X = F, Cl, Br, and
I) in two industrially used solvents, methanol and acetonitrile.
There is a clear difference of solvate formation depending on
the size of the substituent as well as the incorporation motif of
the solvent. Methanol intercalates into the characteristic
hydrogen-bonded imidazole chains, a motif that most likely
persists independently from the substituent. However, while
the packing in two dimensions stays the same, the substituent’s
size has an influence on the packing in the third dimension.
Solvate formation from acetonitrile can only be observed for
the smaller halogen substituents (F-TPI and Cl-TPI), which
form a channel solvate that can be desolvated and resolvated,
apparently without changing the overall crystal structure. Since
the channels are lined with the halogen substituent, bromine
ACKNOWLEDGMENTS
■
The authors would like to thank Diamond Light Source,
Didcot, UK for the award of instrument time on Station I19
(MT 8682) and the instrument scientist for their kind support.
REFERENCES
■
556.
(1) Showell, G. A.; Mills, J. S. Drug Discovery Today 2003, 8, 551−
(2) Olesen, P. H. Curr. Opin. Drug Discovery Dev. 2001, 4, 471−8.
(3) Mugnaini, C.; Pasquini, S.; Corelli, F. Curr. Med. Chem. 2012, 19,
4794−4815.
(4) Tuyishime, M.; Danish, M.; Princiotto, A.; Mankowski, M. K.;
Lawrence, R.; Lombart, H. G.; Esikov, K.; Berniac, J.; Liang, K.; Ji, J. J.;
Ptak, R. G.; Madani, N.; Cocklin, S. Bioorg. Med. Chem. Lett. 2014, 24,
5439−5445.
4537
DOI: 10.1021/acs.cgd.6b00655
Cryst. Growth Des. 2016, 16, 4531−4538

Crystal Growth & Design
Article
(5) Lusher, S. J.; McGuire, R.; van Schaik, R. C.; Nicholson, C. D.; de
Vlieg, J. Drug Discovery Today 2014, 19, 859−868.
(6) Ertl, P. Curr. Opin. Drug Discovery Dev. 2007, 10, 281−288.
(7) Devereux, M.; Popelier, P. L. A. Curr. Top. Med. Chem. 2010, 10,
657−668.
(8) Friedman, H. L. NAS-NRS Publication 1951, 295−358.
(9) Haleblian, J.; McCrone, W. J. Pharm. Sci. 1969, 58, 911−929.
(10) Grant, D. J. W. Theory and origin of polymorphism. In
Polymorphism in Pharmaceutical Solids, Brittain, H. G., Ed.; Marcel
Dekker Inc.: New York, 1999; Vol. 95, pp 1−33.
(11) Bernstein, J. Polymorphism in Molecular Crystals; Clarendon
Press: Oxford, 2002.
(12) Brittain, H. G. Polymorphism in Pharmaceutical Solids; Marcel
Dekker Inc.: New York, 1999.
(13) Griesser, U. J. The importance of solvates. In Polymorphism,
Hilfiker, R., Ed.; Wiley-VCH: Weinheim, Germany, 2006; pp 211−
233.
(38) Thornber, C. W. Chem. Soc. Rev. 1979, 8, 563−580.
(39) Wolkenberg, S. E.; Wisnoski, D. D.; Leister, W. H.; Wang, Y.;
Zhao, Z. J.; Lindsley, C. W. Org. Lett. 2004, 6, 1453−1456.
(40) Sheldrick, G. M., CELL_NOW; University of Gottingen, 2003.
̈
(41) Blessing, R. Acta Crystallogr., Sect. A: Found. Crystallogr. 1995,
51, 33−38.
(42) Dolomanov, O. V.; Bourhis, L. J.; Gildea, R. J.; Howard, J. A. K.;
Puschmann, H. J. Appl. Crystallogr. 2009, 42, 339−341.
(43) Sheldrick, G. M. Acta Crystallogr., Sect. A: Found. Crystallogr.
2008, A64, 112−122.
(44) Martinez-Carrera, S. Acta Crystallogr. 1966, 20, 783−789.
(45) Guinier, A.; Bokij, G. B.; Boll-Dornberger, K.; Cowley, J. M.;
Durovic, S.; Jagodzinski, H.; Krishna, P.; de Wolff, P. M.; Zvyagin, B.
B.; Cox, D. E.; Goodman, P.; Hahn, T.; Kuchitsu, K.; Abrahams, S. C.
Acta Crystallogr., Sect. A: Found. Crystallogr. 1984, 40, 399−404.
(46) Manurung, R.; Holden, D.; Miklitz, M.; Chen, L.; Hasell, T.;
Chong, S. Y.; Haranczyk, M.; Cooper, A. I.; Jelfs, K. E. J. Phys. Chem. C
2015, 119, 22577−22586.
(14) Fucke, K.; Howard, J. A. K.; Steed, J. W. Chem. Commun. 2012,
(47) Stephenson, G. A.; Groleau, E. G.; Kleemann, R. L.; Xu, W.;
Rigsbee, D. R. J. Pharm. Sci. 1998, 87, 536−542.
(48) Sundaramurthi, P.; Suryanarayanan, R. J. Pharm. Sci. 2014, 103,
3095−3106.
(49) Iwata, K.; Kojima, T.; Ikeda, Y. Cryst. Growth Des. 2014, 14,
3335−3342.
48, 12065−12067.
(15) Davey, R. J.; Brychczynska, M.; Sadiq, G.; Dent, G.; Pritchard,
R. G. CrystEngComm 2013, 15, 856−859.
(16) Food and Drug Administration: Guideline Q3C Impurities:
Residual Solvent, 1997.
(17) Minkov, V. S.; Beloborodova, A. A.; Drebushchak, V. A.;
Boldyreva, E. V. Cryst. Growth Des. 2014, 14, 513−522.
(18) Braun, D. E.; Gelbrich, T.; Kahlenberg, V.; Griesser, U. J.
CrystEngComm 2015, 17, 2504−2516.
(19) Zencirci, N.; Griesser, U. J.; Gelbrich, T.; Kahlenberg, V.; Jetti,
R. K. R.; Apperley, D. C.; Harris, R. K. J. Phys. Chem. B 2014, 118,
3267−3280.
(20) Bhattacharya, S.; Saha, B. K. Cryst. Growth Des. 2013, 13, 606−
613.
(21) Williams, P. A.; Hughes, C. E.; Buanz, A. B. M.; Gaisford, S.;
Harris, K. D. M. J. Phys. Chem. C 2013, 117, 12136−12145.
(22) Williams, P. A.; Hughes, C. E.; Martin, J.; Courvoisier, E.;
Buanz, A. B. M.; Gaisford, S.; Harris, K. D. M. J. Phys. Chem. C 2016,
120, 9385−9392.
(23) Hirano, S.; Toyota, S.; Toda, F.; Fujii, K.; Uekuasa, H. Angew.
Chem., Int. Ed. 2006, 45, 6013−6016.
(24) Nassimbeni, L. R.; Bathori, N. B.; Patel, L. D.; Su, H.; Weber, E.
Chem. Commun. 2015, 51, 3627−3629.
(25) Samipillai, M.; Nassimbeni, L. R.; Weber, E. J. Chem. Crystallogr.
2014, 44, 293−300.
(26) Jacobs, A.; le Roex, T.; Nassimbeni, L. R.; Toda, F. Org. Biomol.
Chem. 2006, 4, 2452−2457.
(27) Amombo Noa, F. M.; Bourne, S. A.; Su, H.; Nassimbeni, L. R.
Cryst. Growth Des. 2016, 16, 1636−1642.
(28) Downing, G. A.; MacNicol, D. D.; Frampton, C. S. Chem.
Commun. 1998, 2085−2086.
(29) Inokuma, Y.; Arai, T.; Fujita, M. Nat. Chem. 2010, 2, 780−783.
(30) Yoshioka, S.; Inokuma, Y.; Hoshino, M.; Sato, T.; Fujita, M.
Chem. Sci. 2015, 6, 3765−3768.
(31) Knepper, I.; Seichter, W.; Skobridis, K.; Theodorou, V.; Weber,
E. CrystEngComm 2015, 17, 6355−6369.
(32) Saha, B. K.; Nangia, A. Chem. Commun. 2005, 3024−3026.
(33) Edkins, R. M.; Hayden, E.; Steed, J. W.; Fucke, K. Chem.
Commun. 2015, 51, 5314−5317.
(34) Cote, B.; Boulet, L.; Brideau, C.; Claveau, D.; Ethier, D.;
Frenette, R.; Gagnon, M.; Giroux, A.; Guay, J.; Guiral, S.; Mancini, J.;
Martins, E.; Masse, F.; Methot, N.; Riendeau, D.; Rubin, J.; Xu, D.; Yu,
H.; Ducharme, Y.; Friesen, R. W. Bioorg. Med. Chem. Lett. 2007, 17,
6816−6820.
(35) Edkins, R. M.; Probert, M. R.; Fucke, K.; Robertson, C. M.;
Howard, J. A.; Beeby, A. Phys. Chem. Chem. Phys. 2013, 15, 7848−
7853.
(36) Edkins, R. M.; Probert, M. R.; Robertson, C. M.; Howard, J. A.
K.; Beeby, A. RSC Adv. 2014, 4, 5351−5356.
(37) Nakashima, K.; Taguchi, Y.; Kuroda, N.; Akiyama, S.; Duan, G.
L. J. Chromatogr., Biomed. Appl. 1993, 619, 1−8.
4538
DOI: 10.1021/acs.cgd.6b00655
Cryst. Growth Des. 2016, 16, 4531−4538