Two novel pharmaceutical cocrystals of a development compound - Screening, scale-up, and characterization

Article abstract of DOI:10.1021/cg201270s

A streamlined methodology for cocrystal screening and scale-up has been developed. There are two parts to this procedure: (1) a high throughput screening approach for the rapid identification of potential cocrystal leads and (2) a new procedure for solven

Full text of DOI:10.1021/cg201270s

Article
pubs.acs.org/crystal
Two Novel Pharmaceutical Cocrystals of a Development Compound −
Screening, Scale-up, and Characterization
Dennis H. Leung,*,† Sachin Lohani,*,† Richard G. Ball,‡ Nicole Canfield,§ Yaling Wang,#
Timothy Rhodes,§ and Annette Bak†
†Basic Pharmaceutical Sciences, ‡Analytical Chemistry, §Analytical Sciences, and #Center for Materials Characterization and Engineering,
Merck & Co., Inc., Rahway, New Jersey 07065, United States
S
* Supporting Information
ABSTRACT: A streamlined methodology for cocrystal
screening and scale-up has been developed. There are two
parts to this procedure: (1) a high throughput screening
approach for the rapid identification of potential cocrystal
leads and (2) a new procedure for solvent selection, rapidly
mapping the critical region in the ternary phase diagram, and
determining a cocrystallization procedure for scale-up. Using
this approach, we have discovered and scaled-up two new
cocrystal forms of the pharmaceutically active compound 1 for
the treatment of type II diabetes. This approach provided
more than 2 g of cocrystal material suitable for biopharmaceutical evaluation. The physicochemical characterization and single
crystal X-ray diffraction data of these cocrystals are also presented and discussed.
unwanted hydrates at high relative humidity such as for
anhydrous carbamazepine and caffeine.15,16
INTRODUCTION
■
More than 40% of drugs have poor aqueous solubility and that
fraction has been increasing over the past 20 years.1,2
Therefore, new methods of solubilization and improvement
of physicochemical properties are required to increase the
probability of success of new drug candidates. The selection of
a suitable form of an active pharmaceutical ingredient (API)
can be critical for the development and performance of a
dosage form. This is especially true for APIs with absorption-
limited exposure due to low aqueous solubility. Cocrystals are
emerging as an important alternative to polymorphs and salts
during the solid phase selection APIs. Several case studies have
been published that indicate that pharmaceutical cocrystals can
have better solubility and/or dissolution rate over the
corresponding free form.3−9 More specifically pharmaceutical
cocrystals may offer comparable dissolution and exposure
advantages to amorphous material, and this approach can also
be used for neutral molecules where salt formation is not
feasible.3,5,10
The availability of a wide variety of pharmaceutical
acceptable cocrystal formers (CCFs) provides an opportunity
to engineer the desired physicochemical properties into the
phase of an API,12,17 a key advantage over salt formation which
is limited to ionizable APIs and pharmaceutically acceptable
counterions.18 However, this diversity of CCFs and inter-
molecular interactions results in a large experimental space and
despite recent advances in cocrystal engineering, our ability to
predict cocrystal formation a priori from chemical structures of
CCFs and APIs remains limited; cocrystal screening today is at
best semiempirical.18 Accordingly, when incorporating cocrys-
tals into a solid phase selection strategy for APIs, two issues
become crucial to support ever shrinking drug-development
timelines: first, how to efficiently sample the large experimental
space to identify the combinations of API and CCFs that form
cocrystals (phase screening); second, how to efficiently scale-up
potential cocrystal hits for initial physicochemical and
pharmacokinetic (PK) evaluation (phase evaluation).
Other advantages of cocrystals have also been reported. For
example, cocrystals may be less prone to pseudopolymorphism
(e.g., solvates and hydrates) and polymorphism compared to
monocomponent crystals and salts.11−14 A possible explanation
may be that the best hydrogen bond donor/acceptor sites are
occupied by the coformer. In addition, a survey on cocrystal
chemical stability indicates either neutral or favorable impact of
cocrystal formation. The latter may be the case where instability
is dependent on orientation in the crystal lattice and hence
mainly applies to solid-state stability.6,7,15 Cocrystal formation
can also be beneficial in cases when the free form converts to
To increase the success rate of cocrystal screening,
researchers have recommended case by case selection of
CCFs for a given API driven by an understanding of the
hydrogen bond patterns (Etter’s general rules),19 a survey of
the Cambridge Structural Database (CSD),18,20 and a rational
screen design based on understanding of the ternary phase
diagram (TPD).21 In addition, several recent reports have
Received: September 26, 2011
Revised: December 23, 2011
Published: January 11, 2012
© 2012 American Chemical Society
1254
dx.doi.org/10.1021/cg201270s | Cryst. Growth Des. 2012, 12, 1254−1262

Crystal Growth & Design
Article
system (GADDS), a Bruker AXS HI-STAR area detector at a distance
of 15.05 cm as per system calibration, a copper source, an automated
x-y-z stage, and a 0.5 mm collimator. Data were collected over a 3−40°
2θ range at a step size of 0.05° 2θ. HPLC data were obtained using an
Agilent 1100 HPLC equipped with a fused-core C-18, 4.6 × 100 mm
column with a 2.7 μm particle size using a UV/vis detector at 210 nm
wavelength. Separations were conducted using a mobile phase of a
mixture of water with 0.1% phosphoric acid and acetonitrile in a
gradient elution at 1.8 mL/min at 40 °C. Single-crystal X-ray diffraction
data were collected on an Oxford Diffraction diffractometer. Structures
were solved by direct methods31 and refined using full-matrix least-
squares on F2.32 Procedures involving sonication utilized a Fisher
Scientific FS140 benchtop sonication bath.
Cocrystallization Screening Procedure. A 96-well plate of 1-mL
glass vials was charged with API 1 and CCF solids using an automated
Symyx Powdernium workstation. The API (25 mg) was dispensed into
each well followed by 1 or 2 mol equiv of a specific CCF. The last row
of the 96-well plate was reserved for reference samples of the
individual solid components. The appropriate solvents (30 μL) were
then added using a Thermo Scientific Matrix2 Electronic Multichannel
Pipette. The 96-well plate was then sealed using a clear silicone cap
mat and then subjected to sonication for 2 h. The 96-well plate was
unsealed and the solids were transferred to a 96-well UV plate for
automated XRPD analysis. Fumaric, L-tartaric, citric, maleic, succinic,
benzoic, L-malic, L-ascorbic, adipic, L-glutamic, D-glucuronic, and
hippuric acids were evaluated as potential CCFs with 1 using the
conditions described above. Methanol, ethanol, dichloromethane,
acetone, dimethyl sulfoxide, ethyl acetate, and water were evaluated as
potential cocrystallization solvents using the conditions described
above.
demonstrated that grinding or sonicating a mixture of CCF +
API in the presence of small quantities of solvent can be a
remarkably successful method for producing cocrystals.22−26
However, cocrystal screening methods such as slurry
conversion, evaporation, and liquid assisted grinding/sonication
may not always be applicable to cocrystal scale-up. In general,
relatively few studies have been reported on cocrystal scale-up,
and solution cocrystallization remains by far the most
commonly used scale-up technique.27,28 However, solution
cocrystallization can require significant development, which
may be prohibitive for use within the short timelines in
discovery and early development in the pharmaceutical
industry.
Herein we discuss a streamlined methodology for cocrystal
screening and scale-up. Our approach involves slurry/
sonication-based cocrystal screening and a methodology for
rapid identification of cocrystallization solvent and the region of
stability of cocrystal (henceforth referred to as the critical
region) on the TPDs of API + solvent + CCF system to enable
scale-up (ca. 1−2 g) of cocrystal hits for initial phase evaluation,
physicochemical characterization, and PK studies. We illustrate
the use of this approach through a case study that describes the
discovery and scale-up of two cocrystals of the pharmaceutically
active compound 1 (Figure 1) with benzoic acid (BA) and
Procedure for Identifying the Critical Region of the TPD.
Excess API 1 and cocrystal former were mixed in 0.5−3 mL of the
binary solvent mixture. The resulting suspensions were stirred for 2−5
days using a magnetic stir-bar at 25 °C, and solution phase
composition was analyzed by HPLC daily (after 2 days) until
equilibrium was reached as judged by <5% change in the composition
between two consecutive readings. The identity of the solid phase after
equilibration was determined using XRPD and/or Raman spectroscopy.
Unique vibration peaks in the Raman spectra at 639 cm−1, 195 cm−1,
and 491 cm−1, were used for in situ identification of 1, BA, and the 1-
BA cocrystal respectively. A suitable Raman peak could not be
identified for the 1-MA cocrystal. Therefore, X-ray powder diffraction
was used for phase identification for the 1-MA eutectic point
experiments. For cases where the solid phase did not conform to the
expectation, the system composition was adjusted such that at
equilibrium the solid composition conformed either to 1 + cocrystal
(referred to as EU1) or CCF + cocrystal (referred to as EU2). After
the desired solid phase was confirmed to exist at equilibrium, the
solution phase compositions of 1 and CCF were determined using
HPLC.
Procedure for Scaling-up 1-BA Cocrystal. Compound 1 (2.0 g) +
benzoic acid (BA) (6.6 g) + THF (50 mL) were combined and heated
to 35 °C to fully dissolve all solids. This was followed by the addition
of 33.5 mL heptane to obtain an initial volume fraction of THF in
heptane (vfTHF/Hep) = 0.67. The resulting solution was cooled to 25 °C
followed by addition of cocrystal seeds. This was followed by the
addition of 166.5 mL of heptane (antisolvent) at 28.3 mL/h to obtain
the final vfTHF/Hep = 0.25. The resulting solids were filtered and dried
under a vacuum at 40 °C for 2 h followed by overnight drying at 25 °C
under a vacuum. The isolated yield of the 1-BA cocrystal was 90.7% in
terms of 1. Chemical purity was determined by HPLC and NMR,
while phase purity was determined by XRPD and DSC, and was
determined to be acceptable (data not shown).
Figure 1. Chemical structure of compound 1.30
maleic acid (MA). Furthermore, the detailed physicochemical
characterization and the crystal structure of the two cocrystals
are also presented and discussed with an emphasis on the future
development potential of the two cocrystals (abbreviated and
preliminary characterization results have been discussed
elsewhere29).
EXPERIMENTAL SECTION
■
Materials. Reagents and solvents were obtained from commercial
sources and used as received unless otherwise noted. Compound 1 was
obtained from the Process Research Department (Merck Research
Laboratories, Rahway, NJ, USA) as a crystalline free base and used as
received.
Experimental Techniques. Cocrystals were characterized by
differential scanning calorimetry (DSC), thermogravimetric analysis
(TGA), Fourier transform infrared spectroscopy (FT-IR), X-ray
powder diffraction (XRPD), and single crystal X-ray diffraction
analysis. High performance liquid chromatography (HPLC) was used
to measure the concentration of various molecular species in solution,
and Raman spectroscopy was used to measure the amount of various
molecular species in the solid state.
DSC data were collected using a TA Instruments Q2000 DSC using
a 10 °C/min ramp rate from 25−300 °C. The melting points were
determined by identifying endothermic events in the DSC data. TGA
data were collected using a TA Instruments Q5000 TGA using a
10 °C/min ramp rate from 25−300 °C. Fourier transform infrared
spectroscopic (FT-IR) data were obtained using a Thermo-Nicolet
Nexus 670 FT-IR equipped with an attenuated total reflectance (ATR)
accessory. XRPD data were obtained using a Bruker AXS D8 Discover
X-ray diffractometer equipped with a general area diffraction detection
Procedure for Scaling-up 1-MA Cocrystal. Compound 1
(2.0 g) + maleic acid (MA) (2.0 g) + THF (37 mL) were combined
and heated to 35 °C to fully dissolve all of the solids. The solution was
cooled to 25 °C and cocrystal seeds were added. The antisolvent
heptane (55.5 mL) was then slowly added at 18.2 mL/h to obtain the
final vfTHF/Hep = 0.4. The resulting solids were filtered and dried under
a vacuum at 40 °C for 2 h followed by overnight drying at 25 °C under
1255
dx.doi.org/10.1021/cg201270s | Cryst. Growth Des. 2012, 12, 1254−1262

Crystal Growth & Design
Article
a vacuum. The isolated yield of the 1-MA cocrystal was 99.4% in terms
of 1. Chemical purity was determined by HPLC and NMR, while
phase purity was determined by XRPD, and DSC, and was determined
to be acceptable (data not shown).
Single Crystal X-ray Diffraction Analysis of 1-BA Cocrystal.
Single crystals of 1-BA suitable for X-ray diffraction were grown by
layering heptane over a solution of compound 1 with benzoic acid in a
1:1 ratio in dichloromethane. 1-BA cocrystal (C28H33N7O2), MW =
499.608, triclinic, P1, a = 9.9966(4), b = 15.7845(5), c = 17.0972(7) Å,
̅
α = 83.436(3), β = 75.007(4), γ = 84.875(3)°, V = 2583.84(17) Å3,
Z = 4, Dx = 1.284 g cm−3, monochromatized radiation λ(Mo) = 0.7107
Å, μ = 0.08 mm−1, F(000) = 1064, T = 100 K, θ limit = 27.04° (24801
reflections). There are 11 193 unique reflections with 6049 observed at
the 2σ level. The final model was refined using 673 parameters and all
11193 data. All non-hydrogen atoms were refined with anisotropic
thermal displacements. The final agreement statistics are R = 0.041
(based on 6049 reflections with I > 2σ(I)), wR = 0.073, S = 0.81 with
(Δ/σ)max = 0.01. The maximum peak height in a final difference
Fourier map is 0.226 e Å−3 and this peak is without chemical
significance.
Single Crystal X-ray Diffraction Analysis of 1-MA Cocrystal.
Colorless single crystals of the 1-MA were prepared by evaporation of
a concentrated tetrahydrofuran (THF) solution containing 1 and
maleic acid in a 1:3 ratio. 1-MA cocrystal C25H31N7O4, MW =
493.558, monoclinic, P2/c, a = 21.5550(3), b = 15.58596(18), c =
15.0489(2) Å, β = 100.2339(13)°, V = 4975.32(11) Å3, Z = 8, Dx =
1.318 g cm−3, monochromatized radiation λ(Mo) = 0.7107 Å, μ = 0.09
mm−1, F(000) = 2096, T = 100 K, θ limit = 30.12° (62 346
reflections). There are 13 423 unique reflections with 9470 observed at
the 2σ level. The final model was refined using 653 parameters and all
13 423 data. All non-hydrogen atoms were refined with anisotropic
thermal displacements. The final agreement statistics are R = 0.041
(based on 9470 reflections with I > 2σ(I)), wR = 0.109, S = 1.00 with
(Δ/σ)max < 0.01. The maximum peak height in a final difference
Fourier map is 0.344 e Å−3, and this peak is without chemical
significance.
Figure 2. Example design and results of a 96-well-plate for a cocrystal
screen for 1 with six different carboxylic acids. Two different cocrystal
stoichiometries were investigated in each column and different solvent
slurries at 30 μL solvent volume were evaluated for each row. Six other
carboxylic acids were screened with 1 in a similar fashion (result not
shown as no conversion resulted).
shown in Figure 2. Sonication rather than mechanical grinding
was applied as this was more amenable to high throughput
screening, and the slurries were sonicated for 2 h. XRPD
analysis of solids post sonication showed that wells containing 1
and maleic acid or benzoic acid gave strong, consistent, and
reproducible new XRPD spectra in several slurry conditions,
suggesting the formation of a new cocrystal form. Wells
containing other cocrystal formers had weak and non-
reproducible new XRPD patterns, potentially due to weak
polymorphs or solvates, which were not pursued. The six
remaining potential carboxylic acid cocrystal formers were
screened in a similar format, but no significant conversion was
observed (data not shown).
The best conversion for the new crystal form of compound 1
and maleic or benzoic acid was found using polar organic
solvents such as ethanol, dichloromethane, acetone, and ethyl
acetate (Figure 2). In contrast, no conversion was observed in
DMSO, which is likely due to the high solubility of the
components resulting in full dissolution rather than slurry
formation for those samples. Limited conversion was observed
in water and generally poor to no conversion was observed for
the nonpolar solvent cyclohexane, which is likely due to the low
solubility of the API or CCF components in those solvents. As
a control experiment, we performed the same cocrystallization
sonication reaction without the addition of solvent. No
conversion was observed upon sonicating or heating the dry
solid materials after 2 h, emphasizing the importance of forming
a concentrated slurry for initiating cocrystallization under
solvent-drop conditions.
RESULTS AND DISCUSSION
■
1. Discovery of Cocrystals of API 1. Compound 1 is a
weakly basic (pKa1 = 3.3 pyrazine group, pKa2 = 1.9 pyrimidyl
group) GPR-119 agonist that was developed for the treatment
of type 2 diabetes.30 Two free form polymorphs of 1 (Forms I
and II) were identified during initial stages of its development.
The most stable crystalline form has low water solubility (<30
μg/mL at pH 8.8 at 25 °C) and is classified as a BCS Class II
compound with low solubility and high permeability. The oral
bioavailability of 1 is good across species at low dose (2 mg/kg
in 20% hydroxypropyl-β-cyclodextrin vehicle): 66% in mice,
73% in rats, and 79% in dogs. However, at higher doses 1
displays solubility-limited absorption typical of BCS II
compounds with subproportional increases in exposure with
dose. A salt screen was performed on 1 but due to its weakly
basic nature, the formation of stable salts with acids was not
feasible (data not published). Compound 1 thus provided an
ideal candidate for application of a cocrystallization approach to
improve its physicochemical properties and behavior.
Since 1 contains several hydrogen bond (H-bond) accepting
groups that could act as cocrystal synthons, we screened the
cocrystallization of 1 with 12 different carboxylic acid
compounds that are generally regarded as safe (GRAS) by
the FDA as possible cocrystal formers. Since compound 1
contains multiple polar functional groups potentially capable of
cocrystal interactions, different stoichiometries of the CCFs
were screened for cocrystal formation in small volumes of a
variety of solvents. An example of a plate design for a screen
involving compound 1 with six potential cocrystal formers is
1a. Summary of Crystal Structure Analysis of 1-BA
Cocrystal. Single crystal X-ray diffraction analysis confirmed
cocrystal formation and a 1:1 stoichiometry of the individual
components. The asymmetric unit of the 1-BA cocrystal
1256
dx.doi.org/10.1021/cg201270s | Cryst. Growth Des. 2012, 12, 1254−1262

Crystal Growth & Design
Article
contains two molecules of 1 and two molecules of benzoic acid.
Hydrogen-bonding interactions were observed between each
1-benzoic acid pair, with the benzoic acid and the (more basic)
pyrazine nitrogen group in close contact with N--O distances of
2.62 and 2.68 Å (Figure 3). Therefore, the hydrogen-bonding
Figure 3. Left: A view of a unit cell of the 1-BA cocrystal along (010).
Right: The asymmetric unit of 1-BA cocrystal showing close contacts
between O−H---N hydrogen bonds.
Figure 4. Top: A view of a unit cell of the 1-MA cocrystal along (010).
Bottom: The hydrogen bonding network present in the cocrystal 1-
MA. Two types of hydrogen bonds are shown between the two acid
groups of the maleic acid and pyrazine and pyrimidine groups of
compound 1.
motif in 1-BA crystal structure follows Etter’s hydrogen-bond
rules (general rule # 3) which states that “the best proton
donors and acceptors remaining after intramolecular hydrogen-
bond formation form intermolecular hydrogen bonds to one
another”.33 The stable carboxylic acid-pyrazine supramolecular
heterosynthon is also consistent with results found from
cocrystal engineering reported in the literature.34−40 Partic-
ularly, this result is consistent with a report from Zaworotko
and co-workers who surveyed the Cambridge Structural
Database (CSD) and demonstrated that heterosynthons
containing pyridine groups persist even in the presence of a
CN moiety that could act as an alternative synthon such as in
compound 1.40 The C−O bond distances of the benzoic
carboxylic acid groups indicate that the groups are not
deprotonated (mean CO bond distance 1.218(3) Å, mean
C−OH bond distance 1.321(3) Å), indicating that a true
cocrystal was formed rather than salt formation through
complete proton transfer. The XRPD spectrum predicted
from the single crystal structural data matches closely with the
experimental XRPD spectrum of the material (see Supporting
Information).
1b. Summary of Crystal Structure Analysis of 1-MA
Cocrystal. Single crystal X-ray diffraction analysis confirmed
cocrystal formation and a 1:1 stoichiometry of the individual
components. The asymmetric unit of the 1-MA cocrystal
contains two molecules of 1 and two molecules of maleic acid.
Two types of hydrogen-bonding interactions were observed
between 1 and maleic acid: the first between the carboxylic acid
group of maleic acid and the (more basic) pyrazine nitrogen
group of 1, and the second between the second carboxylic
acid group of maleic acid and the (less basic) pyrimidyl
nitrogen group of 1 (Figure 4). Therefore, hydrogen-bonding
motif in 1-MA crystal structure also follows Etter’s general
hydrogen-bond rules. The stable carboxylic acid-pyrazine/
pyrimidine supramolecular heterosynthon is also consistent
with results found from cocrystal engineering reported in the
literature.34−40 Particularly, this result is again consistent with a
report from Zaworotko and co-workers who surveyed the
(CSD) and demonstrated that heterosynthons containing
pyridine groups persist even in the presence of a CN moiety
that could act as an alternative synthon such as in compound
1.40 The C−O bond distances of the maleic carboxylic acid
groups also indicate that the groups are not deprotonated
(mean CO bond distance 1.209(2) Å, mean C−OH bond
distance 1.317(2) Å), indicating that a true cocrystal was
formed rather than salt formation through complete proton
transfer. The XRPD spectrum predicted from the single crystal
structural data matches closely with the experimental XRPD
spectrum of the material (see Supporting Information).
2. Scale-up of Cocrystal Hits. The scale-up of potential
cocrystal leads involves solvent selection followed by
determination of the critical region on the TPD within which
pure cocrystals can be isolated. To streamline the scale-up of
cocrystal hits identified during screening, we developed a
general method for selecting solvents for cocrystallization and a
method for rapid determination of the region of stability of
cocrystals on the TPDs of API + solvent + CCF systems. This
process enabled the scale-up to 1−2 g of a cocrystal lead for
initial physicochemical and PK evaluation.
2a. Rational Approach to Selection of Solvent for
Cocrystallization. The shape of the region of stability of
cocrystals on a TPD is intimately dependent on the nature of
the solvent. Therefore, solvent selection is a key step in
cocrystallization. However, experimental determination of
solubility is a resource-, time-, and material-intensive step, all
of which are limited during drug discovery and early
development. To address this challenge we developed a two-
pronged approach that involves: (i) solubility modeling and
prediction from minimal experimental data using a process
described elsewhere41 and, (ii) a method of classifying solvents
based on the predicted solubility. The solubility of 1, benzoic
acid, and maleic acid were experimentally determined in five
solvents. The melting point (Tm) and the enthalpy of fusion
(ΔHfus) of benzoic acid and 1 were determined experimentally
using differential scanning calorimetry (DSC) and were found
to be 122 °C, 141 J/g and 175 °C, 118 J/g, respectively.
1257
dx.doi.org/10.1021/cg201270s | Cryst. Growth Des. 2012, 12, 1254−1262

Crystal Growth & Design
Article
Table 1. Classification of Solvents for Cocrystallization of 1 with Benzoic Acid and with Maleic Acid Based on the Order of
a
Magnitude of the Predicted Solubility
common
solvents
pred sol S1
mg/mL
pred sol SBA mg/mL (ratio BA/1 pred sol SMA mg/mL (ratio MA/1
solvent class for 1-BA
solvent class for 1-MA
molar sol)
molar sol)
cocrystal
cocrystal
water
<1
1−10 (>100)
>100 (>100)
IV
II
IV
II
acetone
10−100
10−100
10−100
<1
>100 (10−100)
>100 (10−100)
>100 (10−100)
10−100 (>100)
>100 (10−100)
>100 (>100)
>100 (10−100)
>100 (10−100)
>100 (10−100)
<1 (<0.1)
ethanol
II
II
2-propanol
heptane
II
II
IV
II
IV
II
ethyl acetate
isopropyl acetate
methanol
1−10
1−10
1−10
10−100
10−100 (10−100)
1−10 (0.1−10)
>100 (>100)
III
III
II
I
>100 (>100)
III
II
methyl ethyl
ketone
>100 (10−100)
>100 (10−100)
acetonitrile
THF
1−10
10−100
1−10
>100 (>100)
>100 (>100)
>100 (10−100)
<1 (0.1−10)
III
II
III
II
>100 (10−100)
10−100 (>100)
toluene
III
IV
a
The predicted solubility is reported as <1 mg/mL, between 10−100 mg/mL, and >100 mg/mL, while predicted molar solubility ratios are reported
as <0.1, between 0.1−10, between 10−100, and >100.
However, due to decomposition concomitant with melting
observed for maleic acid, its ΔHfus could not be determined
reliably from DSC measurements. Therefore, the ΔHfus for
maleic acid was estimated from the melting point onset
(141 °C) and assuming the entropy of fusion to be 13.5 eu for
rigid molecules.42 The experimental solubility and the enthalpy
of fusion data were used to predict the solubility of the cocrystal
components in 12 commonly used solvents using the
nonrandom two-liquid segment activity coefficient model
(NRTL-SAC)43 incorporated in the commercially available
software ASPEN. Table 1 shows the predicted solubility of 1
(S1), benzoic acid (SBA), and maleic acid (SMA) in various
solvents and the ratio of the predicted molar solubility (M) of
benzoic acid to 1 (RBA‑1) and maleic acid to 1 (RMA‑1) at 25 °C.
For scale-up via solution crystallization, solubility, and its
coefficient of change with antisolvent and/or temperature are
important considerations in solvent selection.44 Previous
reports on solubility predictions of pharmaceutical compounds
using NTRL-SAC model indicated a maximum log10 squared
error of 0.151.41 Therefore, in this work we have used an order
of magnitude predicted solubility grouping as a means of
classifying solvents. Solvents in which the API and CCF are
predicted to have solubility in the 1−100 mg/mL range were
designated as preferred. Furthermore, solvents in which the API
and CCF manifest similar solubility are expected to yield a
more symmetric TPD, wherein a large excess of API or
cocrystal former would not be required to stabilize the
cocrystal. Therefore, in this work solvents in which the
predicted molar solubility ratio of API to cocrystal former is
within 0.1 to 10 were designated as preferred.
Solvents were grouped in the order of preference from class I
(ideal) to class III (not suitable) based on the magnitude of the
predicted solubility of the API and CCF and the corresponding
predicted molar solubility ratio. For example, for the 1 +
benzoic acid system, a solvent was assigned a class I if S1 and
SBA were both between 1 and 100 mg/mL and the ratio of
molar solubility RBA‑1 was between 0.1 and 10. Class II was
assigned for solvents with S1 between 1 and 100 mg/mL,
SBA >1 mg/mL, and RBA‑1 between 10 and 100, while class III
was assigned when RBA‑1 was outside the range of 0.1 and 100.
Finally, class IV was assigned to a solvent if S1 or SBA was <1
mg/mL. Solvents in class I and II were shortlisted for initial
small scale cocrystallization feasibility experiments, while
solvents in class IV were identified as potential antisolvents.
On the basis of the solvent ranking, 2-propanol (IPA) and
tetrahydrofuran (THF) were selected for small scale experi-
ments to check the feasibility of cocrystallization for 1-BA,
while isopropyl acetate (IPAc) and THF were selected to check
feasibility of cocrystallization of 1-MA. Solvents IPA and IPAc
were dropped from further consideration because crystallization
from their respective solutions resulted in an undesirable
solvated phase so binary mixtures of THF and heptane
(antisolvent) were chosen for detailed TPD determination
and subsequent scale-up of the two cocrystals.
2b. Determination of the Critical Region of the TPD.
Figure 5 shows a schematic of a typical TPD. As illustrated in
the figure, the critical region (outlined in blue) is defined by
three critical points, namely, EU1, EU2, and C. Pure cocrystal is
thermodynamically more stable than the individual components
or mixtures of pure components and cocrystal for any system
composition (referring to the ratio of API:CCF:solvent) within
the critical region. EU1 refers to the solvent composition of API
and CCF when the solid phase at equilibrium contains both
CCF and cocrystal (region 3). Similarly, EU2 refers to the
solvent composition of API and CCF when the solid phase at
equilibrium contains both API and cocrystal (region 4). Point
C is the cocrystal composition. Gibb’s phase rule,45 states that
at fixed temperature and pressure the degree of freedom, F =
number of components − number of phases. Therefore, the
solution phase composition of a ternary system at equilibrium is
fixed at EU1 (or EU2) as long as the total composition is within
region 3 (or 4), when two phases coexist at equilibrium.
Furthermore, the curve connecting EU1 and EU2 represents
the solubility of the cocrystal with the change in API:CCF ratio
in the system.
For this work we have chosen the technique of
cocrystallization by addition of antisolvent. In order to reduce
the risk of crystallization of pure components (API or CCF),
while using this technique, it was imperative to ensure that the
total system composition stayed within the critical region
during the course of cocrystallization. In order to design such a
cocrystallization process, we determined two critical points,
EU1 and EU2, in binary solvent mixtures that bracket the initial
and final system composition during crystallization. The third
critical point, C, does not change with solvent composition.
Under the assumption that the curve connecting EU1 and EU2
1258
dx.doi.org/10.1021/cg201270s | Cryst. Growth Des. 2012, 12, 1254−1262

Crystal Growth & Design
Article
Figure 6. Ternary phase diagram of 1 + BA + THF−heptane
(solvent). The solvent is a binary mixture of THF + heptane, with
vfTHF/Hep varying from 0.67 (initial) to 0.25 (final). See Figure 5 for
the definition of EU1, EU2, A and B. The label T refers to the total
system composition during cocrystallization for 1-BA. The subscript I
refers to initial solvent composition (in red), while f refers to the final
solvent composition (in blue). The solvent axis has been scaled by a
factor of 5 while the BA axis has been scaled by a factor of 10 for
clarity.
Figure 5. A schematic for ternary phase diagram (TPD) of the API +
CCF + solvent system illustrating the critical region outlined with blue
curve. Point A is the solubility of CCF in the solvent, B is the solubility
of API in the solvent, and C is the composition of the cocrystal. The
numbered points refer to the various regions on the TPD: 1, the
system is undersaturated with no solid at equilibrium; 2, pure
cocrystal(s) in equilibrium with a solution of CCF and API; 3, a
mixture of cocrystal + CCF in equilibrium with a solution of
composition EU1; 4, a mixture of cocrystal + API in equilibrium with a
solution of composition EU2; 5, CCF in equilibrium with a solution of
CCF and API; 6, API in equilibrium with a solution of CCF and API.
The solvent drop grinding technique typically samples the TPD close
to the dotted region around point C, thus increasing the likelihood of
identifying a cocrystal phase.
antisolvent was added to reduce loss of mother liquor). Figure 6
is a graphical description of the data listed in Table 2. It
shows the TPD for 1-BA with the critical points at initial
solvent composition (i), and at final solvent composition (f). In
a similar fashion, Table 3 lists the solution phase composition at
critical points and pure component solubility at 25 °C for 1,
MA, and 1-MA. These values were determined at the initial
(vfTHF/Hep = 1) and final (vfTHF/Hep = 0.4) solvent compositions
of THF-heptane encountered in the reaction vessel during
cocrystallization. Figure 7 is a graphical description of the data
listed in Table 3 and shows the TPD for 1-MA with the critical
points at initial solvent composition (i) and at final solvent
composition (f).
2c. Design of Scale-up Procedure. The method used to
scale-up 1-BA and 1-MA cocrystals is described in the
experimental section. To scale-up 1-BA, Figure 6 was used to
select the relative amounts of 1:BA:THF−heptane binary
solvent mixture with vfTHF/Hep = 0.67 (2 g: 6 g: 83.5 mL).
Under these conditions, as the volume fraction of heptane
increased from 0.33 to 0.75, the initial and final total system
composition (Ti and Tf) stayed within the critical region of the
TPD. The yield of the 1-BA cocrystal solid was 90.7% with
respect to free form equivalents according to HPLC and NMR
analysis (data not shown). Similarly, Figure 7 was used to select
the relative amounts of 1:MA:THF components (2 g: 2 g: 37 mL).
can be approximated by the dotted straight line, knowing the
critical points EU1, EU2, and C completely determines the
critical region of the TPD. This approximation not only reduces
the large number of experiments needed to fully determine the
curve EU1−EU2 to just two, but also saves time and resources
which is crucial within the tight timelines of drug discovery
and early development. However, the drawback of this
assumption is that the exact degree of supersaturation with
respect to the cocrystal phase cannot be determined during
cocrystallization.
The critical regions for 1-BA and 1-MA cocrystals were
identified as described in the experimental section. Table 2 lists
the solution phase composition at critical points and pure
component solubility at 25 °C for 1, BA, and 1-BA. These
values were determined at the initial (vfTHF/Hep = 0.67) and
final (vfTHF/Hep = 0.25) solvent compositions of THF-heptane
encountered in the reaction vessel during cocrystallization. The
initial and final ratios of THF and heptane were selected for
optimal seeding conditions (some oversaturation, but not high
enough to induce self-nucleation) and yield (additional
Table 2. Solution Phase Composition at Critical Points and Solubility of Pure Components for 1-BA Cocrystal in THF−
Heptane Mixtures with vfTHF/Hep = 0.67 and 0.25
solid phase in equilibrium = 1-BA + BA
solid phase in equilibrium = 1-BA + 1
solid phase in equilibrium = 1 or BA
solvent composition
(THF vol fraction in
heptane)
conc of 1 in solution
phase (mg/mL)
EU1
conc of BA in
solution phase
(mg/mL) EU1
conc of 1 in solution
phase (mg/mL)
EU2
conc of BA in
solution phase
(mg/mL) EU2
conc of BA in
solution phase
(mg/mL) A
conc of 1 in
solution phase
(mg/mL) B
0.67 (initial)
0.25 (final)
17.13
0.5
321.7
97.6
15.9
0.4
39.2
11.9
266.4
88.9
6.3
0.2
1259
dx.doi.org/10.1021/cg201270s | Cryst. Growth Des. 2012, 12, 1254−1262

Crystal Growth & Design
Article
Table 3. Solution Phase Composition at Critical Points and Solubility of Pure Components for 1-MA Cocrystal in THF−
Heptane Mixtures with vfTHF/Hep = 1 and 0.4
solid phase in equilibrium = 1-MA + MA
solid phase in equilibrium = 1-MA + 1
solid phase in equilibrium = 1 or MA
solvent composition
(THF vol fraction in
heptane)
conc of 1 in solution
phase (mg/mL)
EU1
conc of MA in
solution phase
(mg/mL) EU1
conc of 1 in solution
phase (mg/mL)
EU2
conc of MA in
solution phase
(mg/mL) EU2
conc of MA in
solution phase
(mg/mL) A
conc of 1 in
solution phase
(mg/mL) B
1 (initial)
0.4 (final)
32.2
0.2
400.4
36.4
41.5
1.0
23.6
1.6
344.7
35.3
34.1
1.1
Figure 8. XRPD spectra of the 1-BA cocrystal and components. Top:
compound 1. Middle: benzoic acid. Bottom: 1-BA cocrystal.
Figure 7. Ternary phase diagram of 1 + MA + THF−heptane
(solvent). The solvent is a binary mixture of THF + heptane, with
vfTHF/Hep = 1 and 0.4 (final). See Figure 5 for the definition of the
labels EU1, EU2, A and B. The label T refers to the total system
composition during cocrystallization for 1-MA. The subscript I refers
to initial solvent composition (corresponding curves in red), while f
refers to the final solvent composition (corresponding curves in blue).
The solvent and MA axes have been scaled by a factor of 5 for clarity.
Differential scanning calorimetry (DSC) analysis of the solid
products for the reaction of 1 with benzoic acid confirmed the
formation of a new crystalline form displaying a melting point
of 145.1 °C, which lies between the melting points of the
individual cocrystal component species (Figure 9, Tm benzoic
acid = 122.3 °C and Tm 1 = 173.8 °C). For an overlay of DSC
and TGA data, see Supporting Information. The FT-IR
spectrum of the solid product contained a CO stretching
frequency at 1684 cm−1, suggesting that the benzoic acid moiety
remained protonated as a result of a hydrogen-bonding
cocrystal interaction rather than formation of a deprotonated
salt form (see Supporting Information),46 consistent with the
bond lengths observed in the single crystal structure.
3b. 1-MA Cocrystal Characterization. 1-MA cocrystals were
scaled-up by the antisolvent crystallization method as described
above and characterized. The XRPD spectrum of the resulting
solid material revealed that a new crystalline form distinct from
the starting components had been generated (Figure 10). DSC
analysis of the solid products for the reaction of 1 with maleic
acid confirmed formation of a new crystalline form displaying a
melting point of 145.5 °C (followed by a small exothermic
event attributed to decomposition), which lies between the
melting points of the individual cocrystal component species
Under these conditions, as the volume fraction of heptane
increased from 0 to 0.6, the initial and final total system
composition (Ti and Tf) stayed within the critical region of the
TPD. The yield of the 1-MA cocrystal solid was 99.4% with
respect to free form equivalents according to HPLC and NMR
analysis (data not shown).
3. Solid-State Characterization of Cocrystals. Having
successfully scaled-up the preparation of gram-size quantities of
pure cocrystals 1-BA and 1-MA, the resulting cocrystalline
material was fully characterized. Table 4 shows a summary of
the solid state characterization of the pure samples of the two
cocrystals.
3a. 1-BA Cocrystal Characterization. 1-BA cocrystals were
scaled-up by the antisolvent crystallization method as described
above and characterized. The XRPD spectrum of the resulting
solid material revealed that a new crystalline form distinct from
the starting components had been generated (Figure 8).
Table 4. Summary of Physicochemical Properties of the Scaled-up Samples of Cocrystals of 1 with Benzoic Acid and with Maleic
Acid
physicochemical property
1-BA cocrystal
1-MA cocrystal
thermogravimetric
analysis
2.9% weight loss up to melting attributed to loss of residual solvent 0.7% weight loss up to melting attributed to loss of residual solvent
particle size distribution
surface area
average = 39.6 μm SD = 19.4 μm 95% < 77.1 μm
average = 22.7 μm SD = 12.0 μm 95% < 64.3 μm
0.4 m2/g
1.0 m2/g
hygroscopicity
<0.1 wt % gain between 20% to 80% RH, maximum weight gain of <0.3 wt % gain between 20% to 80% RH, maximum weight gain of
<0.1 wt % at 95% RH
<0.1 wt % at 95% RH
1260
dx.doi.org/10.1021/cg201270s | Cryst. Growth Des. 2012, 12, 1254−1262

Crystal Growth & Design
Article
consistent with the bond lengths observed in the single crystal
structure.
CONCLUSION
■
Cocrystals represent a promising new approach for improving
the crystallinity and physicochemical properties of
pharmaceutical compounds. We have outlined a new high
throughput screening approach for quickly identifying potential
cocrystal leads and have successfully discovered two new
cocrystal forms of the pharmaceutically active compound 1. In
order to follow up on these leads, we have developed a new
procedure for rapidly mapping the critical region in the TPD
and determining a procedure to scale-up pure cocrystals of 1
with benzoic and maleic acids. This provides cocrystal material
of sufficient quantity and purity suitable for physicochemical
characterization and PK evaluation within short timeframes as
required by short drug discovery and early development
timelines. These new cocrystals were also characterized through
structure determination using single crystal X-ray diffraction.
The bioperformance of these new phases of compound 1 are
currently being investigated. We have already and will in the
future apply this new screening and scale-up methodology to
the cocrystallization of other pharmaceutically active com-
pounds.
Figure 9. DSC data for 1 (blue solid line, Tm = 173.8 °C), benzoic acid
(red dashed line, Tm = 122.3 °C), and the 1-BA cocrystal (green
dashed-dotted line, Tm = 145.1 °C).
ASSOCIATED CONTENT
■
S
* Supporting Information
X-ray crystallographic information files (CIF) for the 1-BA and
1-MA cocrystals, DSC-TGA overlay for 1-BA, experimental and
predicted XRPD spectra for 1-BA and 1-MA, and FT-IR
spectra for 1-BA and 1-MA. This information is available free of
charge via the Internet at http://pubs.acs.org.
AUTHOR INFORMATION
■
Figure 10. XRPD spectra of 1-MA cocrystals and components. Top:
compound 1. Middle: maleic acid. Bottom: 1-MA cocrystal.
Corresponding Author
*(D.L.) Phone: 732-594-6982; e-mail: dennis_leung@merck.
com. (S.L.) Phone: 732-594-3509; e-mail: sachin_lohani@
merck.com.
(Figure 11, Tm maleic acid = 141.4 °C and Tm 1 = 174.2 °C).
The FT-IR spectrum of the solid product contained a CO
ACKNOWLEDGMENTS
■
The authors acknowledge Arlene McKeown, Yanfeng Zhang,
Alexander Chin, Jonathan Ashby, Narayan Variankaval, Tanise
Shattock, Davida Krueger, Xiaoling Jin, Kevin Belyk, Kathy
Lyons, and James Cuff for their assistance.
REFERENCES
■
(1) Bergese, P.; Colombob, I.; Gervasonib, D.; Deperoa, L. E. Mater.
Sci. Eng., C 2003, 23, 791−795.
(2) Lipinski, C. Am. Pharm. Rev 2002, 5, 82−85.
(3) Cooke, M. W.; Stanton, M.; Shimanovich, R.; Bak, A. Am. Pharm.
Rev. 2007, 10, 54−59.
̈
(4) Almarsson, O.; Zaworotko, M. J. Chem. Commun. 2004, 1889−
1996.
(5) Bak, A.; Gore, A.; Yanez, E.; Stanton, M.; Tufekcic, S.; Syed, R.;
Akrami, A.; Rose, M.; Surapaneni, S.; Bostick, T. J. Pharm. Sci. 2008,
97, 3942−3956.
Figure 11. DSC data for 1 (blue solid line, Tm = 173.8 °C), maleic
acid (red dashed line, Tm = 141.9 °C), and the 1-MA cocrystal (green
dashed-dotted line, Tm = 150.0 °C).
(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.
stretching frequency at 1709 cm−1, suggesting that the maleic
acid moiety remained protonated as a result of a hydrogen-
bonding cocrystal interaction rather than formation of a
deprotonated salt form (see Supporting Information),46
(7) Hickey, M. B.; Peterson, M. L.; Scoppettuolo, L. A.; Morrisette,
S. L.; Vetter, A.; Guzman, H.; Remenar, J. F.; Zhang, Z.; Tawa, M. D.;
̈
Haley, S.; Zaworotko, M. J.; Almarsson, O. Eur. J. Pharm. Biopharm.
2007, 67, 112−119.
1261
dx.doi.org/10.1021/cg201270s | Cryst. Growth Des. 2012, 12, 1254−1262

Crystal Growth & Design
Article
(8) Childs, S. L.; Chyall, L. J.; Dunlap, J. T.; Smolenskaya, V. N.;
Stahly, B. C.; Stahly, G. P. J. Am. Chem. Soc. 2004, 126, 13335−13342.
(9) Variankaval, N.; Wenslow, R.; Murry, J.; Hartman, R.; Helmy, R.;
Kwong, E.; Clas, S.-D.; Dalton, C.; Santos, I. Cryst. Growth Des. 2006,
6, 690−700.
(39) Bhogala, B. R.; Vishweshwar, P.; Nangia, A. Cryst. Growth Des.
2002, 2, 325−328.
(40) Bis, J. A.; Vishweshwar, P.; Weyna, D. R.; Zaworotko, M. J. Mol.
Pharmaceutics 2007, 4, 401−416.
(41) Tung, H.-H.; Tabora, J.; Variankaval, N.; Bakken, D.; Chen,
C.-C. J. Pharm. Sci. 2008, 97, 1813−1820.
(42) Yalkowsky, S. H. Ind. Eng. Chem. Fundam. 1979, 18, 108−111.
(43) Chen, C.-C.; Song, Y. Ind. Eng. Chem. Res. 2004, 43, 8354−
8362.
(44) Mullin, J. B. In Crystallization; Butterworth-Heinemann: 2001.
(45) Gibbs, J. W. In The Scientific Papers of J. W. Gibbs: On the
Equilibrium of Heterogeneous Substances; Dover Publications, Inc.:
New York, NY, 1961.
(46) Colthup, N.; Daly, L. H.; Wiberley, S. E. In Introduction to
Infrared and Raman Spectroscopy, 3rd ed.; Academic Press: New York,
NY, 1990.
(10) Kennedy, M.; Hu, J.; Gao, P.; Li, L.; Ali_Reynolds, A.; Chal, B.;
Gupta, V.; Ma, C.; Mahajan, N.; Akrami, A.; Surapaneni, S. Mol.
Pharmaceutics 2008, 5, 981−993.
(11) Trask, A. V.; Motherwell, W. D. S.; Jones, W. Int. J. Pharm.
2006, 320, 114−123.
(12) Aakeroy, C. B.; Forbes, S.; Desper, J. J. Am. Chem. Soc. 2009,
̈
131, 17048−17049.
(13) Aakeroy, C. B.; Fasulo, M. E.; Desper, J. Mol. Pharmaceutics
̈
2007, 4, 317−322.
(14) Byrn, S. R.; Pfeiffer, R. R.; Stowell, J. G. In Solid-State Chemistry
of Drugs, 2nd ed.; SSCI, Inc.: West Lafayette, 1999.
(15) Rodríguez-Hornedo, N.; Nehm, S. J.; Jayasankar, A. In
Encyclopedia of Pharmaceutical Technology, 3rd ed.; Swarbrick, J.
Taylor and Francis: New York, 2006. 615−635.
(16) Trask, A. V.; Motherwell, W. D. S.; Jones, W. Cryst. Growth Des.
2005, 5, 1013−1021.
(17) Good, D. J.; Rodríguez-Hornedo, N. Cryst. Growth Des. 2009, 9,
2252−2264.
(18) Stahly, G. P. Cryst. Growth Des. 2007, 7, 1007−1026.
(19) Etter, M. C. Acc. Chem. Res. 1990, 23, 120−126.
(20) Schultheiss, N.; Newman, A. Cryst. Growth Des. 2009, 9, 2950−
2967.
(21) Chiarella, R. A.; Davey, R. J.; Peterson, M. L. Cryst. Growth Des.
2007, 7, 1223−1226.
(22) Childs, S. L.; Rodríguez-Hornedo, N.; Reddy, L. S.; Jayasankar,
A.; Maheshwari, C.; McCausland, L.; Shipplett, R.; Stahly, B. C.
CrystEngComm 2008, 10, 856−864.
̌ ̌ ́
(23) Friscic, T.; Childs, S. L.; Rizvi, S. A. A.; Jones, W.
CrystEngComm 2009, 11, 418−426.
(24) Trask, A. V.; Jones, W. Top. Curr. Chem. 2005, 254, 41−70.
(25) Buca
32−33.
̌
r, D.-K.; MacGillivray, L. R. J. Am. Chem. Soc. 2007, 129,
(26) Kelly, D. R.; Harrison, S. J.; Jones, S.; Masood, M. A.; Morgan, J.
J. G. Tetrahedron Lett. 1993, 34, 2689−2690.
(27) Sheikh, A. Y.; Hammond, R. B.; Roberts, K. J. CrystEngComm
2009, 11, 501−509.
(28) Yu, Z. Q.; Chow, P. S.; Tan, R. B. H. Cryst. Growth Des. 2010,
10, 2382−2387.
(29) Lohani, S.; Chin, A.; Leung, D.; Wang, Y.; Zhang, Y.; Rhodes,
T.; Bak, A. Rapid determination of ternary phase diagrams to enable
efficient scale-up of cocrystal leads via solution crystallization;
American Association of Pharmaceutical Scientists (AAPS) - Annual
Meeting and Exposition; New Orleans, Louisiana, USA, 2010;
American Association of Pharmaceutical Scientists (AAPS) , Arlington,
VA.
(30) Szewczyk, J. W.; Acton, J.; Adams, A. D.; Chicchi, G.; Freeman,
S.; Howard, A. D.; Huang, Y.; Li, C.; Meinke, P. T.; Mosely, R.;
Murphy, E.; Samuel, R.; Santini, C.; Yang, M.; Zhang, Y.; Zhao, K.;
Wood, H. B. Bioorg. Med. Chem. Lett. 2011, 21, 2665−2669.
(31) Sheldrick, G. M. Acta Crystallogr. 1990, A46, 467−473.
(32) Sheldrick, G. M. SHELXL-97. Program for the Refinement of
Crystal Structures; University of Gottingen: Germany: 1997.
̈
(33) Etter, M. C. J. Phys. Chem. 1991, 95, 4601−4610.
(34) Etter, M. C.; Adsmond, D. A. J. Chem. Soc., Chem. Commun.
1990, 589−591.
(35) Vishweshwar, P.; Nangia, A.; Lynch, V. M. J. Org. Chem. 2002,
67, 556−565.
(36) Walsh, R. D. B.; Bradner, M. W.; Fleischman, S.; Morales, L. A.;
Moulton, B.; Rodríguez-Hornedo, N.; Zaworotko, M. J. Chem.
Commun. 2003, 186−187.
(37) Vishweshwar, P.; Nangia, A.; Lynch, V. M. Cryst. Growth Des.
2003, 3, 783−790.
(38) Aakeroy, C. B.; Beatty, A. M.; Helfrich, B. A. J. Am. Chem. Soc.
̈
2002, 124, 14425−14432.
1262
dx.doi.org/10.1021/cg201270s | Cryst. Growth Des. 2012, 12, 1254−1262