Thermal Behavior of Benzoic Acid/Isonicotinamide Binary Cocrystals

Article abstract of DOI:10.1021/acs.cgd.5b00351

A comprehensive study of the thermal behavior of the 1:1 and 2:1 benzoic acid/isonicotinamide cocrystals is reported. The 1:1 material shows a simple unit cell expansion followed by melting upon heating. The 2:1 crystal exhibits more complex behavior. Its unit cell first expands upon heating, as a result of C-H...π interactions being lengthened. It then is converted into the 1:1 crystal, as demonstrated by significant changes in its X-ray diffraction pattern. The loss of 1 equiv of benzoic acid is confirmed by thermogravimetric analysis-mass spectrometry. Hot stage microscopy confirms that, as intuitively expected, the transformation begins at the crystal surface. The temperature at which conversion occurs is highly dependent on the sample mass and geometry, being reduced when the sample is under a gas flow or has a greater exposed surface area but increased when the heating rate is elevated. (Figure Presented).

Full text of DOI:10.1021/acs.cgd.5b00351

Article
pubs.acs.org/crystal
Thermal Behavior of Benzoic Acid/Isonicotinamide Binary Cocrystals
Asma Buanz,† Timothy J. Prior,‡ Jonathan C. Burley,§ Bahijja Tolulope Raimi-Abraham,†
Richard Telford,∥ Michael Hart,⊥ Colin C. Seaton,∥ Philip J. Davies,# Ian J. Scowen,¶ Simon Gaisford,*,†
and Gareth R. Williams*,†
†UCL School of Pharmacy, University College London, 29−39 Brunswick Square, London WC1N 1AX, U.K.
‡Department of Chemistry, University of Hull, Cottingham Road, Hull HU6 7RX, U.K.
§Laboratory of Biophysics and Surface Analysis, Boots Science Building, School of Pharmacy, University of Nottingham, University
Park, Nottingham NG7 2RD, U.K.
∥School of Chemistry and Forensic Sciences, Faculty of Life Sciences, University of Bradford, Bradford BD7 1DP, U.K.
⊥Diamond Light Source, Harwell Science & Innovation Campus, Didcot OX11 0DE, U.K.
#TA Instruments, 730−740 Centennial Court, Centennial Park, Elstree WD6 3SZ, U.K.
¶School of Chemistry, University of Lincoln, Brayford Pool, Lincoln LN6 7TS, U.K.
S
* Supporting Information
ABSTRACT: A comprehensive study of the thermal behavior of the 1:1 and
2:1 benzoic acid/isonicotinamide cocrystals is reported. The 1:1 material
shows a simple unit cell expansion followed by melting upon heating. The 2:1
crystal exhibits more complex behavior. Its unit cell first expands upon heating,
as a result of C−H···π interactions being lengthened. It then is converted into
the 1:1 crystal, as demonstrated by significant changes in its X-ray diffraction
pattern. The loss of 1 equiv of benzoic acid is confirmed by thermogravimetric analysis−mass spectrometry. Hot stage
microscopy confirms that, as intuitively expected, the transformation begins at the crystal surface. The temperature at which
conversion occurs is highly dependent on the sample mass and geometry, being reduced when the sample is under a gas flow or
has a greater exposed surface area but increased when the heating rate is elevated.
combining them with saccharin,7 and can result in improve-
INTRODUCTION
■
ments in processability.8,9 In favorable instances, two APIs may
Pharmaceutical cocrystals have attracted increasing research
be combined in a cocrystal.10 There are myriad examples of
attention in recent years to ameliorate problems with solubility
binary cocrystals in the literature,4 and increasingly ternary
and physical and chemical stability.1 The physical form of an
systems are also being studied.4,11 In order to drive effective
active pharmaceutical ingredient (API) is integral in the
determination of these properties, and control and under-
cocrystal formation, it is important that there are strong
intermolecular interactions between the crystal components.
standing of this is of vital importance in the medicine
These may take the form of H-bonds, π−π stacking
development process. Metastable and amorphous physical
forms offer rapid dissolution but suffer from the concomitant
interactions, or van der Waals forces. Increasingly, researchers
are developing sophisticated ways of assessing the compatibility
problem of potential conversion back to the most stable
of coformers and systematically predicting which combinations
crystalline form, jeopardizing the therapeutic efficacy of the
formulation. The preparation of salts (comprising the ionized
API and a counterion) can yield accelerated dissolution, and
will and will not form stable composite systems.12−16
Benzoic acid (BA) and isonicotinamide (isoNCT) have both
this route has been widely explored.2 Hydrates and solvates
been widely explored as coformers for pharmaceutical cocrystal
synthesis.17−21 They are known to form both 1:1 (BA/
have additionally received much attention.3
isoNCT)22 and 2:1 (BA2/isoNCT)18 cocrystals. Possible
Cocrystals, in which the neutral API is combined with a
neutral coformer in a single crystalline phase, have been shown
synthetic routes to these include solution methods,18,22 melt
crystallization,21,23 template synthesis using BA crystals as a
to be capable of improving physical, mechanical, dissolution,
substrate,21 and thermal inkjet printing.17 The structures of
both BA/isoNCT and BA2/isoNCT are given in Figure 1. In
each case, there are hydrogen bonds between the benzoic acid
and the pyridine function of the isonicotinamide. The 2:1
and bioavailability properties for a range of APIs.4 For instance,
the hygroscopicity of theophylline can be reduced through
forming cocrystals with oxalic acid.5 The bioavailability in dogs
of 2-[4-(4-chloro-2-fluorophenoxy)phenyl]pyrimidine-4-car-
boxamide, a developmental sodium channel blocker, has been
found to increase upon formation of a glutaric acid cocrystal.6
In addition, cocrystal formation can solve challenges associated
with bitter or unpleasant tasting APIs, for instance, by
Received: March 13, 2015
Revised: May 18, 2015
© XXXX American Chemical Society
A
DOI: 10.1021/acs.cgd.5b00351
Cryst. Growth Des. XXXX, XXX, XXX−XXX

Crystal Growth & Design
Article
Figure 1. Structures of (a) BA/isoNCT and (b) BA2/isoNCT.
material also contains hydrogen bonds between the BA COOH
group and the isoNCT amide group, plus C−H···π contacts,
while the 1:1 crystal features hydrogen bonds between the
amide groups of adjacent isoNCT units.
alterations in the ordering of the molecular components, and
thus how the intermolecular interactions influence and are
influenced by this transformation is of great interest. In
addition, understanding such interactions and transformations
will yield useful insight into the stability and design of
pharmaceutical cocrystals. In this work, we endeavored to
understand more about exactly how the BA2/isoNCT to BA/
isoNCT transformation occurs. In order to do so, a range of
analytical techniques were applied to study the thermal
behavior of both the 2:1 and 1:1 cocrystals and the thermally
induced transformation between them.
It has been shown that both the stoichiometry and
polymorphic form of binary pharmaceutical cocrystals can be
varied through the synthetic conditions. For instance, polymers
have been used to drive the formation of different polymorphs
of the 2:1 adefovir/succinic acid cocrystal.24 Synthesis at
different temperatures results in different polymorphic 1:1
cocrystals of benzoic acid and pentafluorobenzoic acid.25 The
1:1 form of a coumaric acid/nicotinamide cocrystal has been
reported to be converted into a 2:1 material upon making a
slurry in water;26 the 1:1 urea−succinic acid system can
similarly be converted to a 2:1 form when it is stirred in certain
solvents.27 Similar results been observed for the BA/isoNCT
case: both the 2:1 and 1:1 cocrystals are stable in water, but in
ethanol the 2:1 system is converted into a 1:1 material.20
Both Seaton and Buanz have separately reported that upon
heating to 142−143 °C, the BA2/isoNCT material is converted
to BA/isoNCT.17,21 This transformation involves the loss of
one-third of the unit cell contents and a very significant
reorganization of the coformer arrangement. It is hence
remarkable that this conversion can take place. A few thermally
induced cocrystal transitions are known beyond the BA/
isoNCT system: for instance, the caffeine/theophylline
cocrystal dissociates upon heating to yield a binary mixture of
pure caffeine and theophylline.28 Similarly, two 1:1 nicotina-
mide/pimelic acid cocrystal phases are known and can be
interconverted upon heating.29
MATERIALS AND METHODS
■
Synthesis. Benzoic acid (ACS reagent, ≥99.5%) and isonicotina-
mide (99%) were purchased from Sigma-Aldrich, U.K. BA and
isoNCT at 1:2 and 2:1 molar ratios were used to prepare BA/isoNCT
and BA2/isoNCT, respectively. BA (0.2322 g) and isoNCT (0.5409 g)
(for BA/isoNCT) or BA (0.2433 g) and isoNCT (0.1587 g) (for BA2/
isoNCT) were added to 50 mL of distilled water.18 Solutions were
prepared by heating to ca. 70 °C until complete dissolution had
occurred, then allowing them to cool to room temperature and storing
for at least 1 month. Crystals were collected by vacuum filtration and
stored over silica gel in a desiccator for at least 24 h before
characterization.
Characterization. Hot Stage Microscopy. Microscopy was
initially carried out using a Zeiss Axioplan-2 microscope with a
Linkham THMS600 hot stage controlled through the Axiovision
software, with a linksys 32 patch for hot stage control. All microscopy
was carried out under cross-polarization. Samples of BA2/isoNCT
were heated from 40 to 120 °C at 5 °C min−1 in an open crucible.
Further microscopy studies were undertaken using a LeicaADM 2700
M microscope connected to a FP82HT Mettler Toledo Instruments
heating stage unit and a FP90 Mettler Toledo Instruments central
processor unit. Samples were heated from 40 to 150 °C at 2 or 10 °C
min−1. The Studio86 Design capture software (version 4.0.1) was used
to record and capture thermal events in real time.
Time-Resolved in Situ X-ray Diffraction. Experiments were
performed on Beamline I12 of the Diamond Light Source, U.K.30
This beamline generates a continuous spectrum of X-rays over the
energy range from 50 to 150 keV. All experiments reported here were
conducted with the beam monochromated to an energy of ca. 53 keV
(λ = 0.2337 Å). The Oxford-Diamond In Situ Cell (ODISC)31 was
employed to heat the samples. ODISC comprises an IR-heated furnace
Liquid-mediated transformations between cocrystal forms
can be fairly easily understood in that dissolution/reprecipita-
tion mechanisms can facilitate significant changes in molecular
arrangements. Small changes in intermolecular interactions to
yield slightly different polymorphs upon heating are also
unsurprising, most likely being facilitated by thermal motions.
The thermal transformation of BA2/isoNCT to BA/isoNCT is,
insofar as the authors are aware, the only thermally mediated
cocrystal transition in which the stoichiometry of the material
very profoundly changes. The transformation necessitates huge
B
DOI: 10.1021/acs.cgd.5b00351
Cryst. Growth Des. XXXX, XXX, XXX−XXX

Crystal Growth & Design
Article
system, and reactions were performed using ca. 0.5 g of material placed
in a glass culture tube. The latter was mounted in a glassy carbon tube
to ensure effective heat transfer. A Thales Pixium RF4343 detector was
sited 2.5 m away from the reaction vessel. Diffraction patterns (4 s)
were collected every 5 s.
Data were analyzed using the Fit2D program32 to convert the 2D
data collected on the Pixium to 1D patterns.33 These patterns were
background-subtracted using in-house routines, visualized in 3D using
the Origin software (v9.1), and integrated using a second set of in-
house routines (these fit a background to the data and then use
Gaussian functions to model reflections and determine their areas). It
was also possible to carry out LeBail fits and partial Rietveld fits to the
observed data to extract structural information. The GSAS suite of
programs was used for this.34
Variable Temperature Raman Spectroscopy. A LabRam HR
Raman microscope/spectrometer was employed. A 50× lens was used
to focus a 785 nm wavelength laser onto the sample. A grating of 600
lines mm−1 was employed to disperse the scattered light onto a
Synapse 1024-pixel CCD thermo-electrically cooled to −70 °C.
Spectra were collected for 2 s every 2 °C between 25 and 199 °C. The
sample was maintained at a constant nominal temperature during data
acquisition, and ramping between temperatures was set to be 10 °C
min−1.
Differential Scanning Calorimetry. Differential scanning calorim-
etry (DSC) measurements were performed on Q2000 and rapid
heating (RHC) instruments (both TA Instruments). Calibrations for
cell constant and enthalpy were performed with indium (Tm = 156.6
°C, ΔHf = 28.71 J g−1) according to the manufacturer’s instructions.
Nitrogen was used as the purge gas at a flow rate of 50 mL min−1. TA
aluminum pans and lids (Tzero) and heating rates of 10 °C min−1
were used for all experiments unless stated otherwise. Data were
analyzed using the TA Instruments Universal Analysis 2000 software,
and all transformation temperatures are reported as extrapolated
onsets. The sample mass employed was ca. 5 mg for Q2000 DSC and
0.1 mg for RHC-DSC.
Thermogravimetric Analysis. Measurements were performed on a
PerkinElmer Pyris-6 analyzer. Samples were heated at various rates in
open ceramic pans using nitrogen as purge gas (flow rate = 20 mL
min−1). Data collection and analysis were performed using the Pyris
software (version 3.81).
Figure 2. HSM data collected under crossed-polar conditions during
the heating of BA2/isoNCT at 5 °C min−1. The red bar in each image
represents 1 mm. The 120 °C image has been presented before in a
previous study.21 The rings in the 100 °C image highlight the needle-
shaped crystals, which are present on the surface of the large columnar
crystals.
Supporting Information). If the HSM experiment is repeated
with the crystal surrounded by oil, small bubbles of what appear
to be a gas can be seen exiting the crystal during heating
(Figure S3, Supporting Information). The needles are believed
to comprise the 1:1 BA/isoNCT cocrystal, as they are seen to
persist above the melting point of BA (122 °C), as depicted in
Figure S4, Supporting Information. If heating is continued
above 125 °C, then the remaining material is observed to melt
gradually over the approximate temperature range 130−160 °C,
depending on the heating rate.
In Situ X-ray Diffraction. The 1:1 Cocrystal. The changes
observed in the diffraction pattern of the 1:1 BA/isoNCT
cocrystal upon heating from 30 to 179 °C at ca. 9.5 °C min−1
are depicted in Figure 3. As expected, the unit cell expands and
Thermogravimetric Analysis−Mass Spectrometry. Samples of the
2:1 BA2/isoNCT cocrystal (ca. 5 mg) were loaded onto platinum pans
and heated from room temperature to 180 °C at a constant heating
rate of 10 °C min−1 under a helium purge gas (50 mL min−1) using a
TA Instruments Discovery TGA. Evolved gas analysis was performed
throughout the experiment using mass spectrometry, with data
acquired in scan mode (m/e 5−155) using a Pfeiffer Vacuum
ThermoStar TM GSD 301 T mass spectrometer. Data were analyzed
using the TA Instruments Universal Analysis Software
RESULTS
■
Hot Stage Microscopy. The transformation of the 2:1
BA2/isoNCT material upon heating has been reported
previously.17,18,21 To obtain insight on how this process occurs
on the macroscopic scale, hot-stage microscopy (HSM) was
employed. HSM data collected at different temperatures during
heating at 5 °C min−1 are presented in Figure 2.
In the HSM data, large columnar crystals of BA2/isoNCT
can be seen at 40 °C. They begin to transform at around 80−85
°C, when small needle-shaped crystals can be seen to appear on
the surface of the brightly colored columns (an enlargement of
Figure 2 is provided in the Supporting Information (Figure S1)
to aid visualization). These needles increase in number as the
sample is heated further, until the transformation appears
complete at around 120 °C. Undertaking the HSM experiment
at different heating rates clearly shows that the onset of the
transformation both begins and appears to be complete at
higher temperatures with more rapid heating (see Figure S2,
Figure 3. Evolution of the diffraction pattern for the BA/isoNCT 1:1
cocrystal with temperature.
reflections shift to lower angle upon heating. It should be noted
that the intensity of the Bragg reflections fluctuates somewhat
between temperatures: this appears to be a random effect that is
tentatively ascribed to preferred orientation effects and the
individual crystals comprising the sample repacking during the
heating process.
The patterns may be satisfactorily indexed on expanded
versions of the known cell for BA/isoNCT.22 There is a steady
increase in the a parameter with rising temperature (2.3%
between 30 and 166 °C), a small increase (0.45%) in the b
parameter, and decreases in c and β (0.43% and 0.35%,
respectively). The cell volume increases in an approximately
linear fashion with temperature, as does the a parameter
C
DOI: 10.1021/acs.cgd.5b00351
Cryst. Growth Des. XXXX, XXX, XXX−XXX

Crystal Growth & Design
Article
(Figure 4). The unit cells calculated at selected temperatures
are detailed in the Supporting Information, Table S1. Partial
expands much more profoundly (by 4.9% between 32 and 150
°C, cf. the a and b parameters, which change by 0.3% and
−0.08%, respectively, over the same temperature range), as
depicted in Figure 6. This can easily be rationalized in light of
Figure 4. Variation in a parameter and unit cell volume observed on
heating BA/isoNCT.
Figure 6. Variation in c parameter and unit cell volume observed on
heating BA2/isoNCT.
Rietveld fits are given in Figure S5, Supporting Information.
Fitting of the observed data demonstrates that the structure of
the 1:1 cocrystal does not undergo substantial molecular
reorganization up to 161 °C. There is some evidence for a small
amount of the 2:1 material being present as a trace impurity in
the unheated BA/isoNCT.
At ca. 160 °C, the reflections begin to decrease in intensity as
the material melts. This process is complete by 174 °C, after
which there are no diffraction features present in the pattern.
The 2:1 Cocrystal. The evolution of the diffraction pattern of
the 2:1 BA/isoNCT cocrystal was also recorded upon heating
from room temperature to 179 °C at approximately 9 °C
min−1. The raw data collected are given in Figure 5. As the
sample is heated, the characteristic reflections of the 2:1 crystal
move to lower angle as a result of thermal expansion of the unit
cell (see Figure 5b).
Below 150 °C, the diffraction patterns can be indexed on
expanded versions of the previously reported cell for BA2/
isoNCT.18 LeBail fitting was performed at selected temper-
atures, and the unit cells obtained are given in Table S2,
Supporting Information. Partial Rietveld fits to the data may be
found in Figure S6, Supporting Information. There is a gradual
increase in the a parameter with rising temperature, while the b
parameter remains essentially constant, and the c parameter
the structure of BA2/isoNCT, which contains thick sections
extending in the ab plane. These are held together in the c
direction by C−H···π interactions. This weak interaction
lengthens considerably upon heating, which is manifested
principally in an expansion along the c-axis. The unit cell
volume and c parameter both increase with temperature in an
almost linear fashion (see Figure 6).
Above ca. 140 °C, some changes are apparent in the powder
diffraction patterns of the 2:1 cocrystal. Most notably, the 001
reflection broadens, and by 150 °C, this is clearly composed of
two peaks of roughly equal intensity that are very close
together. It is possible to fit this with a single peak, but the
pattern can better be fitted with a two-phase mixture (see Table
S2 and Figure S6f, Supporting Information). A LeBail fit with
two phases suggests that a smaller cell is evolving. Above 160
°C, the pattern is completely changed: at 161 °C, no trace of
the 2:1 material can be seen. Upon continued heating to 179
°C, the system melts, and no diffraction peaks are visible.
The diffraction patterns of both the 2:1 and 1:1 crystals at
161 and 166 °C are shown in Figure S7, Supporting
Information. The patterns obtained from both starting
materials are virtually identical, demonstrating the similarity
Figure 5. In situ XRD data recorded during heating of the 2:1 BA2/isoNCT cocrystal: (a) 3D plot, and (b) selected diffraction patterns recorded at
different temperatures.
D
DOI: 10.1021/acs.cgd.5b00351
Cryst. Growth Des. XXXX, XXX, XXX−XXX

Crystal Growth & Design
Article
Table 1. Unit Cell Parameters Calculated at 161 °C
starting material
a, Å
b, Å
c, Å
β, deg
V, Å3
BA/isoNCT
BA2/isoNCT
23.373(5)
23.343(7)
5.226(1)
5.225(2)
20.431(6)
20.497(9)
96.276(18)
96.370(30)
2480.5(4)
2484.6(6)
between crystalline components present in the two samples at
these temperatures. Both the BA/isoNCT and BA2/isoNCT
systems are found to have virtually identical C-centered cells at
161 °C (see Table 1). The pattern from the 2:1 material is
rather low intensity at 161 °C and above, but it is possible to
perform a partial Rietveld fit to the observed data using the
published crystal structure for the 1:1 cocrystal; this is shown in
the Supporting Information, Figure S8. In terms of normal
statistics, the fit is relatively poor, because of the narrow angular
range and earlier background subtraction. However, visual
inspection of the fit to the data shows that there is a single
phase present: the 1:1 cocrystal.
This transformation is consistent with the emergence of a
smaller unit cell when BA2/isoNCT is heated above 150 °C. A
loss of BA is needed for the interconversion to the 1:1
cocrystal. Structurally, both the a and c parameters change
significantly. Within the 2:1 cocrystal, hydrogen bonded layers
are stacked along [001]. BA projects above and below each
layer. Loss of some BA upon heating would be expected to
reduce the c parameter. Similarly there are C−H···π contacts
between the layers that extend in the xz plane; directionally,
these have no component along b. Thus, a and c change
substantially as BA is lost but the b parameter does not vary
much.
Raman Spectroscopy. Both BA/isoNCT and BA2/
isoNCT were investigated using variable-temperature (VT)
Raman spectroscopy, focusing on the lattice vibrations below
200 cm−1 since these will give most information on the
molecular arrangements in the solid state.35,36 The results of
this analysis are presented in Figure 8. It is very clear that at ca.
150 °C there is an abrupt change in the features of the
spectrum in the case of BA2/isoNCT (red). Comparison of
these data with those obtained on the 1:1 BA/isoNCT material
(black) confirms that above 150 °C the spectrum observed for
the 2:1 system is the same as that for the 1:1. Melting is
subsequently observed at around 165 °C. These data are in
good agreement with those obtained from XRD and HSM.
Thermogravimetric Analysis−Mass Spectrometry. To
confirm the conversion of the 2:1 cocrystal to the 1:1 form,
thermogravimetric analysis−mass spectrometry (TGA-MS) was
performed. The results are shown in Figure 9. Mass loss is
observed to commence at 80 °C. The mass spectrum shows
clear signals at m/e = 51, 77, 105, and 122. These are typical of
benzoic acid, thus confirming that BA is lost from the material.
The material appears to decompose completely under these
conditions, and it is not possible to ascertain with certainty
where the conversion from 2:1 to 1:1 ends and when
degradation begins. There does however appear to be a point
of inflection at around 131−132 °C, where there is mass loss of
32.6%. The calculated loss for the conversion of BA2/isoNCT
to BA/isoNCT is 31%, in good agreement with this observed
value.
Selected reflections from each material were integrated, and
the normalized intensity was plotted against temperature; the
resultant data are presented in Figure 7.
Exploring the Transition Temperature. The data
presented above clearly show the transformation of the 2:1
cocrystal into the 1:1 species upon heating, in accordance with
the literature.17,21 The transformation happens at different
temperatures depending on the method of analysis used, which
may be a result of sample size and shape variation. To
investigate this further, a series of DSC and TGA experiments
were performed. DSC data for the raw materials and both
cocrystals are presented in Figure S9, Supporting Information.
The DSC thermograms of the 1:1 cocrystal show a single
endotherm corresponding to melting at 162.6 0.17 °C. The
2:1 material exhibits a more complex thermogram, with an
endotherm at 143.11 0.14 °C followed by an exotherm and
then a second endotherm. These data are in full agreement with
the literature.17,18,21 In a previous report,17 it was suggested that
the first endotherm corresponds to melting of the BA2/isoNCT
material followed by recrystallization of BA/isoNCT and finally
the melting of the latter material. However, the HSM and XRD
experiments presented above demonstrate that complete
melting of the 2:1 cocrystal does not occur prior to the 1:1
system forming. We thus propose that the DSC events are
more subtle than a simple melt/recrystallization and that the
initial endotherm in the 2:1 DSC trace corresponds to the
removal of BA from the structure. This will leave defects in the
remaining structure, which then rearranges to give the 1:1
material in an exothermic process. The second endotherm is
attributed to the 1:1 melt.
Figure 7. Changes in intensity observed with heating for selected
reflections of the 2:1 and 1:1 BA/isoNCT cocrystals.
The area of the 2:1 001 reflection remains essentially
constant until 150 °C, when it declines rapidly to zero.
Concomitantly, the 1:1 200 reflection grows rapidly in
intensity, reaching a maximum at 166 °C before then declining
to zero again as the material melts. The normalized area vs
temperature curves for the two species cross at approximately
0.5. This could be regarded as indicative of a direct solid−solid
transition: as diffracted intensity is lost from the starting
material, it is gained by the product. That said, it was clear from
the HSM data that BA gas is evolved from the 2:1 cocrystal
during transformation, and this gas could be involved in
mediating the conversion.
The effect of varying the heating rate on the BA2/isoNCT
transition temperature was next explored (see Figure 10).
E
DOI: 10.1021/acs.cgd.5b00351
Cryst. Growth Des. XXXX, XXX, XXX−XXX

Crystal Growth & Design
Article
Figure 8. VT Raman data obtained on the BA/isoNCT and BA2/isoNCT cocrystals. Data for the 2:1 material are shown in red and those for the 1:1
in black.
Increasing the heating rate has the effect of raising the
extrapolated onset temperature at which the 2:1 to 1:1
transformation occurs in conventional DSC, from 140 °C at
2 °C min−1 to 141.6 0.5 °C at 10 °C min−1 and 143 0.5 °C
at 100 °C min−1. In rapid-heating DSC, two endotherms are
visible at lower heating rates (presumably the first corresponds
to the 2:1 → 1:1 transformation, and the second to the 1:1
melt). The extrapolated onset of the transformation endotherm
moves from 142.7 0.2 °C at 100 °C min−1 to 142.9 0.2 °C
(300 °C min−1), 143.1 0.6 °C (500 °C min−1) and 140.7
0.9 °C (1000 °C min−1). At 1000 °C min−1, the transformation
and melt endotherms have merged. It is therefore clear that the
Figure 9. TGA-MS data recorded on the 2:1 BA2/isoNCT cocrystal.
heating rate of the sample influences the temperature at which
Figure 10. DSC data recorded on BA2/isoNCT at different heating rates. Thermograms were obtained using (a) conventional DSC instrument and
(b) rapid-heating DSC instrument.
F
DOI: 10.1021/acs.cgd.5b00351
Cryst. Growth Des. XXXX, XXX, XXX−XXX

Crystal Growth & Design
Article
the 2:1 to 1:1 transition occurs. This was verified by TGA
experiments (see Figure S10, Supporting Information): the loss
of BA from BA2/isoNCT occurs at progressively higher
temperatures as the heating rate is increased.
isoNCT) and also by TGA-MS, in which BA is clearly present
in the purge gas. It is thus conclusively demonstrated that BA2/
isoNCT loses BA to form BA/isoNCT when it is subjected to
heating.
In addition, there are clear differences between the
conventional DSC and RHC-DSC data during heating at 100
°C. The peak of the transformation endotherm is at very
different temperatures in the two (146.3 °C in DSC and 150.1
°C in RHC-DSC), which is attributed to the variations in the
size of the pan used between the two (the RHC-DSC uses pans
of 1.6 mm in diameter and a sample size of just under 0.1 mg;37
in contrast the Tzero pans used for DSC have an inner
diameter of 5 mm and use a sample mass of ca. 5 mg). The
effect of the pan was further explored by heating BA2/isoNCT
at 10 °C min−1 in conventional DSC using different pans; the
results are presented in Figure S11, Supporting Information.
The temperature at which the 2:1 to 1:1 transition occurs
(Ttrans) is clearly dependent on the sample geometry. Ttrans is
lower when an open Tzero pan is used than with a closed pan
and is also lowered when using a hermetically sealed pan if a
pinhole is made. This suggests that the removal of evolved gas
from the crystal during transformation is very important: in
open or pinholed pans, where the sublimed BA can more easily
be carried away by the purge gas, the transformation happens at
a lower temperature. The fact that an increased heating rate
shifts Ttrans to elevated temperatures is consistent with this: the
removal of sublimed BA will be a kinetic process, and time is
required to clear it after it has been evolved.
The reason this transformation occurs can be understood by
a simple consideration of the thermodynamics in the two
materials. There will be a significant increase in entropy when
the conversion occurs, as a result of the freeing of an equivalent
of gaseous BA. There is also likely to be an enthalpic
contribution to the Gibbs free energy of the transformation.
BA/isoNCT contains classical H-bonds, which dominate the
intermolecular interactions. BA2/isoNCT also possesses some
H-bonds, but there are in addition many much weaker C−H···π
interactions. More H-bonds are possible in the BA/isoNCT
material. The existence of a 2:1 to 1:1 transformation agrees
with the equations calculated for the free energy of formation of
BA/isoNCT cocrystals by Sangster; these show the 1:1
cocrystal to be more thermodynamically stable than the 2:1
under all the conditions used in this work.23
This therefore permits explanation of why the 2:1 → 1:1
transition arises at different temperatures in different experi-
ments. In situ diffraction experiments use a relatively large
volume of sample, but the culture tube used for reaction is only
9 mm in internal diameter. Most of the sample surface is
confined by glass; it is thus difficult for the BA to escape from
the material, and a high transformation temperature (Ttrans) is
observed. In contrast, in HSM, the crystals are very loosely
packed with BA gas able to escape easily from any part of any
crystal, leading to a lower Ttrans. A similarly low Ttrans is notable
in the TGA-MS data, conducted in an open pan with a flow of
gas. In DSC and TGA experiments, Ttrans moves to higher
temperatures with more rapid heating rates. The use of open
DSC pans (or those with pin-holes) results in lower Ttrans than
when the pans are sealed. These effects can be explained
because there is likely to be an equilibrium between gaseous BA
and BA in the crystal structure. A gas flow aids the movement
of gaseous BA away from the crystals, shifting this equilibrium
in favor of BA/isoNCT.
DISCUSSION
■
In this paper, a comprehensive collection of data obtained upon
heating the 1:1 and 2:1 benzoic acid/isonicotinamide cocrystals
is reported. Previous reports17,18,21 have identified the existence
a thermally mediated 2:1 to 1:1 transition using DSC and
microscopy data. Here a range of in situ techniques have been
utilized to explore the transition in more detail. Hot stage
microscopy clearly shows that as the BA2/isoNCT material is
heated new crystals of BA/isoNCT begin to appear on its
surface at around 85 °C, and by 120 °C the entire crystal has
been completely transformed. In situ synchrotron diffraction
reveals that, while BA/isoNCT shows only an expansion in its a
parameter when heated, BA2/isoNCT exhibits much more
significant changes. Both its a and c parameter expand rapidly
as the temperature rises, and above 140 °C, it appears that a
two-phase mixture of materials is present, with a smaller unit
cell evolving. Above 160 °C, the pattern has completely
changed, and a partial Rietveld fit using the BA/isoNCT
structure confirms that the 2:1 material is completely
transformed to the 1:1. The unit cells calculated at 161 °C
are essentially identical, regardless of which cocrystal was
present before heating commenced. The changes observed in
the 2:1 diffraction pattern before the transformation occurs can
be rationalized by a consideration of the intermolecular
bonding in the two structures: the expansion in the c direction
arises from a weakening of C−H···π interactions in the 2:1
system. This causes the BA−isoNCT−BA slabs (see Figure 1)
to move apart, creating space and facilitating the departure of
BA molecules to yield the 1:1 material. The HSM data show
that gas bubbles are evolved from the structure, consistent with
this hypothesis.
CONCLUSIONS
■
A range of in situ techniques have been implemented to study
the thermal behavior of the 2:1 and 1:1 benzoic acid/
isonicotinamide cocrystals. Building on earlier studies, we
have performed detailed hot-stage microscopy, time-resolved X-
ray diffraction, differential scanning calorimetry, and thermog-
ravimentric analysis−mass spectrometry experiments to thor-
oughly characterize these systems, the thermal behavior of
which has previously not been studied in detail. The 1:1 system
shows a simple thermal expansion of the unit cell, followed by
melting. The 2:1 crystal shows more interesting behavior and is
converted to the 1:1 system upon heating. This transition
begins at the surface of the crystals. It is preceded by significant
changes in the unit cell, which can be rationalized in terms of
the breaking of C−H···π interactions required to convert the
2:1 to the 1:1 stoichiometry. The evolution of benzoic acid is
clearly evidenced by combined thermogravimetric analysis−
mass spectrometry. The temperature at which the phase
transition occurs is highly dependent on the sample environ-
ment and is reduced when the uncovered surface area is
increased or with a flow of gas. This temperature increases with
faster heating rates. There is believed to exist an equilibrium
between benzoic acid (BA) in the 2:1 crystal and evolved BA
gas, and hence the more rapidly the sublimed BA is cleared
The transition from 2:1 to 1:1 stoichiometry is verified by in
situ Raman spectroscopy (where very clear changes in the
spectrum occur upon heating, consistent with formation of BA/
G
DOI: 10.1021/acs.cgd.5b00351
Cryst. Growth Des. XXXX, XXX, XXX−XXX

Crystal Growth & Design
Article
(19) Boyd, S.; Back, K.; Chadwick, K.; Davey, R. J.; Seaton, C. C. J.
Pharm. Sci. 2010, 99, 3779−3786.
from the cocrystal surface the lower the temperature at which
the transformation occurs.
(20) Croker, D. M.; Davey, R. J.; Rasmuson, Å. C.; Seaton, C. C.
CrystEngComm 2013, 15, 2044−2047.
ASSOCIATED CONTENT
■
(21) Seaton, C. C.; Parkin, A.; Wilson, C. C.; Blagden, N. Cryst.
Growth Des. 2008, 8, 363−368.
S
* Supporting Information
(22) Aakeroy, C. B.; Beatty, A. M.; Helfrich, B. A. Angew. Chem., Int.
̈
An enlargement of Figure 2, additional HSM images, Rietveld
fits to selected time-resolved XRD datasets, comparisons of
XRD patterns collected from the 1:1 and 2:1 cocrystals at
different temperatures, and additional DSC and TGA data. The
Supporting Information is available free of charge on the ACS
Publications website at DOI: 10.1021/acs.cgd.5b00351.
Ed. 2001, 40, 3240−3242.
(23) Sangster, J. J. Phys. Chem. Ref. Data 1999, 28, 889−930.
(24) Jung, S.; Ha, J.-M.; Kim, I. Polymers 2013, 6, 1−11.
(25) Reddy, L. S.; Bhatt, P. M.; Banerjee, R.; Nangia, A.; Kruger, G. J.
Chem.Asian. J. 2007, 2, 505−513.
(26) Bevill, M. J.; Vlahova, P. I.; Smit, J. P. Cryst. Growth Des. 2014,
14, 1438−1448.
̈
AUTHOR INFORMATION
(27) Alhalaweh, A.; George, S.; BostrOm, D.; Velaga, S. P. Cryst.
■
Growth Des. 2010, 10, 4847−4855.
Corresponding Authors
(28) Eddleston, M. D.; Lloyd, G. O.; Jones, W. Chem. Commun.
2012, 48, 8075−8077.
*S.G. E-mail: s.gaisford@ucl.ac.uk. Tel: +44 (0) 20 7753 5853.
*G.R.W. E-mail: g.williams@ucl.ac.uk. Tel: +44 (0) 207 753
5868.
(29) Aitipamula, S.; Wong, A. B. H.; Chow, P. S.; Tan, R. B. H.
CrystEngComm 2012, 14, 8193−8198.
(30) http://www.diamond.ac.uk/Home/Beamlines/I12.html.
(31) Moorhouse, S. J.; Vranjes, N.; Jupe, A.; Drakopoulos, M.; D, O.
H. Rev. Sci. Instrum. 2012, 83, No. 084101.
Notes
The authors declare no competing financial interest.
(32) Hammersley, A. P. ESRF Internal Report 1998, ESRF98HA01T,
FIT2D V9.129 Reference Manual V3.1.
ACKNOWLEDGMENTS
■
(33) Hammersley, A. P.; Svensson, S. O.; Hanfland, M.; Fitch, A. N.;
Haeusermann, D. High Pressure Res. 1996, 14, 235−248.
(34) Larson, A. C.; Von Dreele, R. B. Los Alamos National
Laboratory Report 1994, General Structure Analysis System (GSAS).
(35) Al-Dulaimi, S.; Aina, A.; Burley, J. CrystEngComm 2010, 12,
1038−1040.
The authors gratefully thank the Diamond Light Source for
access to Beamline I12 (EE7782-1), Professor Dermot O’Hare
and Dr. Saul Moorhouse (University of Oxford) for access to
ODISC, Dr. Christina Reinhard (Diamond) for help and
support with experiments on I12, Dr. David Berry (Durham
University) and Prof. Nicolas Blagden (Lincoln University) for
support in collecting the hot-stage microscopic data, and Mr.
Alexander Prior for helpful discussions.
(36) Brillante, A.; Bilotti, I.; Della Valle, R. G.; Venuti, E.; Girlando,
A. CrystEngComm 2008, 10, 937−946.
(37) Danley, R.; Caulfield, P.; Aubuchon, S. Am. Lab. 2008, 40, 9−
11.
REFERENCES
■
(1) Blagden, N.; Coles, S. J.; Berry, D. J. CrystEngComm 2014, 16,
5753−5761.
(2) Stahl, P. H.; Wermuth, C. G. Pharmaceutical Salts: Properties,
Selection, and Use, 2nd ed.; Wiley VCH: Weinheim, Germany, 2011.
(3) Brittain, H. G. J. Pharm. Sci. 2011, 100, 1260−1279.
(4) Sun, C. C. Expert Opin. Drug Delivery 2013, 10, 201−213.
(5) Trask, A. V.; Motherwell, W. D. S.; Jones, W. Int. J. Pharm. 2006,
320, 114−123.
(6) McNamara, D. P.; Childs, S. L.; Giordano, J.; Iarriccio, A.;
Cassidy, J.; Shet, M. S.; Mannion, R.; O’Donnell, E.; Park, A. Pharm.
Res. 2006, 23, 1888−1897.
(7) Banerjee, R.; Bhatt, P. M.; Ravindra, N. V.; Desiraju, G. R. Cryst.
Growth Des. 2005, 5, 2299−2309.
(8) Sun, C. C.; Hou, H. Cryst. Growth Des. 2008, 8, 1575−1579.
(9) Karki, S.; Friscic, T.; Fabian, L.; Laity, P. R.; Day, G. M.; Jones,
W. Adv. Mater. 2009, 21, 3905−3909.
(10) Cheyney, M. L.; Weyna, D. R.; Shan, N.; Hanna, M.; Wojitas,
L.; Zaworotko, M. J. J. Pharm. Sci. 2011, 100, 2172−2181.
(11) Tothadi, S.; Desiraju, G. R. Chem. Commun. 2013, 49, 7791−
7793.
(12) Habgood, M. Cryst. Growth Des. 2013, 13, 4549−4558.
(13) Habgood, M.; Deij, M. A.; Mazurek, J.; Price, S. L.; ter Horst, J.
H. Cryst. Growth Des. 2010, 10, 903−912.
(14) Habgood, M.; Price, S. L. Cryst. Growth Des. 2010, 10, 3263−
3272.
(15) Mohamed, S.; Tocher, D. A.; Price, S. L. Int. J. Pharm. 2011,
418, 187−198.
(16) Issa, N.; Barnett, S. A.; Mohamed, S.; Braun, D. E.; Copley, R.
C. B.; Tocher, D. A.; Price, S. L. CrystEngComm 2012, 14, 2454−2464.
(17) Buanz, A. B. M.; Telford, R.; Scowen, I. J.; Gaisford, S.
CrystEngComm 2013, 15, 1031−1035.
(18) Seaton, C. C.; Parkin, A.; Wilson, C. C.; Blagden, N. Cryst.
Growth Des. 2009, 9, 47−56.
H
DOI: 10.1021/acs.cgd.5b00351
Cryst. Growth Des. XXXX, XXX, XXX−XXX