Insight into the Mechanisms of Cocrystallization of Pharmaceuticals in Supercritical Solvents

Article abstract of DOI:10.1021/acs.cgd.5b00200

Carbon dioxide has been extensively used as a green solvent medium for the crystallization of active pharmaceutical ingredients (APIs) by replacing harmful organic solvents. This work explores the mechanisms underlying a novel recrystallization method - cocrystallization with supercritical solvent (CSS) - which enables APIs cocrystallization by suspending powders in pure CO₂. Six well-known APIs that form cocrystals with saccharin (SAC) were processed by CSS, namely, theophylline (TPL), indomethacin (IND), carbamazepine (CBZ), caffeine (CAF), sulfamethazine (SFZ), and acetylsalicylic acid (ASA). Pure cocrystals were obtained for TPL, IND, and CBZ (with SAC) after 2 h of CSS processing. Convection was revealed to be a determining parameter for successful cocrystallization with high-yield levels. TPL-SAC was selected as a model system to study the cocrystallization kinetics in the gas, supercritical, and liquid phases under different conditions of pressure (8-20 MPa), temperature (30 to 70°C), and convection regimes. The solubility of each substance in CO₂ was measured at the selected working conditions. TPL-SAC showed a cocrystallization rate of 2.9% min^-1, two times higher than that of IND-SAC, due to the higher solubility of TPL in CO₂. The cocrystallization kinetics was also improved by increasing the CO₂ density, showing that cocrystallization was limited by the dissolution of cocrystal formers. Overall, the CSS process has a potential for scale-up as a novel, simple, solvent-free batch process whenever the cocrystal phase is formed in the CO₂ media. (Graph Presented).

Full text of DOI:10.1021/acs.cgd.5b00200

Article
pubs.acs.org/crystal
Insight into the Mechanisms of Cocrystallization of Pharmaceuticals
in Supercritical Solvents
Luis Padrela,§ Miguel A. Rodrigues,*,§ Joao Tiago,§ Sitaram P. Velaga,⊥ Henrique A. Matos,§
̃
and Edmundo Gomes de Azevedo§
§Centro de Química Estrutural and Department of Chemical Engineering, Instituto Superior Tec
́
nico, Universidade de Lisboa, Lisboa,
Portugal
⊥Department of Health Sciences, Lulea University of Technology, 971 87 Lulea, Sweden
̊
̊
S
* Supporting Information
ABSTRACT: Carbon dioxide has been extensively used as a
green solvent medium for the crystallization of active
pharmaceutical ingredients (APIs) by replacing harmful
organic solvents. This work explores the mechanisms under-
lying a novel recrystallization methodcocrystallization with
supercritical solvent (CSS)which enables APIs cocrystalliza-
tion by suspending powders in pure CO2. Six well-known APIs
that form cocrystals with saccharin (SAC) were processed by
CSS, namely, theophylline (TPL), indomethacin (IND),
carbamazepine (CBZ), caffeine (CAF), sulfamethazine
(SFZ), and acetylsalicylic acid (ASA). Pure cocrystals were obtained for TPL, IND, and CBZ (with SAC) after 2 h of CSS
processing. Convection was revealed to be a determining parameter for successful cocrystallization with high-yield levels. TPL−
SAC was selected as a model system to study the cocrystallization kinetics in the gas, supercritical, and liquid phases under
different conditions of pressure (8−20 MPa), temperature (30 to 70 °C), and convection regimes. The solubility of each
substance in CO2 was measured at the selected working conditions. TPL−SAC showed a cocrystallization rate of 2.9% min−1,
two times higher than that of IND−SAC, due to the higher solubility of TPL in CO2. The cocrystallization kinetics was also
improved by increasing the CO2 density, showing that cocrystallization was limited by the dissolution of cocrystal formers.
Overall, the CSS process has a potential for scale-up as a novel, simple, solvent-free batch process whenever the cocrystal phase is
formed in the CO2 media.
cin−saccharin cocrystals could be produced either by super-
critical antisolvent or by supercritical atomization as straightfor-
ward methods, converting crystal clear solutions into micro- to
nanosized particles. Additionally, supercritical atomization has
shown potential for the generation of novel cocrystal
stoichiometries which had not been produced by classical
INTRODUCTION
■
Much work has been published highlighting pharmaceutical
cocrystals production and characterization and their potential
capacity to tune relevant physical and kinetic properties of
drugs.1−9 Pharmaceutical cocrystals can be prepared by several
(classical) methods, namely, mechanochemical (e.g., neat and
liquid-assisted grinding) or traditional solution crystallization
approaches such as solvent evaporation and slurry techni-
ques.10,11 However, the use of these traditional techniques may
bring undesired problems in the generation of the cocrystal
products, such as the generation of amorphous materials when
using mechanochemical methods or hydrate/solvate formation,
when using solution crystallization methods.
Because of obvious disadvantages presented by classical
cocrystal production methods, the development of novel and
better methods for screening and production of cocrystals is
highly desirable. Supercritical carbon dioxide has been used as
an alternative media for crystallization reactions, acting either as
a solvent or an antisolvent.12,13 Supercritical carbon dioxide
(SC-CO2) has been used to induce novel polymorphs, to
enable the control of the crystal habit,14−19 and has more
recently also been described as a novel tool for cocrystal
engineering.20−22 Padrela et al.20 have shown that indometha-
20
methods.21 Most interestingly, Padrela et al. observed that a
small amount of indomethacin (IND)−saccharin (SAC)
cocrystals could be formed when both pure components
(IND and SAC) were placed in contact with SC-CO2. This
effect is herein explored as a novel cocrystallization with
supercritical solvent (CSS) method that uses the solvent power
and the molecular mobility enhancement provided by super-
critical fluids, in particular in CO2, to promote the
cocrystallization of an API with a coformer. Previous results
suggest that SC-CO2 has the potential to be a simple (single
step), scalable, organic solvent-free alternative for cocrystalliza-
tion of APIs. To better understand the mechanism of CO2-
mediated cocrystallization and its full potential for industrial
Received: February 9, 2015
Revised: May 6, 2015
© XXXX American Chemical Society
A
DOI: 10.1021/acs.cgd.5b00200
Cryst. Growth Des. XXXX, XXX, XXX−XXX

Crystal Growth & Design
Article
Figure 1. Schematic diagram of the equilibrium solubility apparatus. PI − pressure indicator; V − valve; BPR − back pressure regulator; C −
compressor; TI − temperature indicator; TIC − temperature indicator and controller; VC − variable volume high-pressure visual cell; M − magnetic
stirrer; S − strainer; DV − dead volume.
calorimeter NETZSCH DSC 200 F3 Maia. The samples (6−7 mg)
were weighed directly into pierced aluminum pans, on an analytical
balance. During the experiments the sample holder was continuously
purged with a dry nitrogen flow of 50 mL/min. Nitrogen was used as a
purge gas and also as a protective gas. Measurements in the DSC were
done using a heating/cooling rate of 10 °C/min.
Determination of the Solubility of APIs and Coformers in
CO2. The APIs and coformers’ solubilities in CO2 were measured
using a visual, variable volume high-pressure cell, shown schematically
in Figure 1. The visual cell consists of a stainless steel cylinder (with an
internal diameter of 24 mm, external diameter of 60 mm, and length of
168 mm) with a high-pressure glass window. This cell is divided into
two chambers by a movable stainless steel piston provided with two O-
ring seals. The back side of the piston contains the pressuring fluid,
water in this case, responsible for moving the piston forward or
backward.
Each experimental run started by loading the cell (in the front side
of the piston) with 200 mg of the corresponding solid powder and a
magnetic stirrer set to 300 rpm. After the cell was closed, the system
was pressurized with CO2, and the air chamber was heated to the
desired conditions (see Figure 1). The CO2 was injected into the cell
through valve V2 passing by pressure indicator PI3 where the CO2
pressure was measured. The water pressure was controlled by moving
the piston by changing the pressure of the water in the back of the
piston, using the manual compressor C2 (HIP, model 86-6-5). The
temperature inside the air chamber was maintained by a digital
temperature controller (BTC-9200, ROC). Temperatures were
measured with thermocouples (Omega, type T), and the estimated
temperature variability was 0.1 °C. Pressures in the CO2 line and in
the visual cell were measured with pressure transducers (Setra, model
206) and digital indicators (Setra, model 300D) with an estimated
pressure variability of 0.01 MPa.
application, key issues must be clarified, namely: Is it possible to
obtain pure cocrystals by CO2-mediated cocrystallization?
Which systems cocrystallize in CO2 under what conditions?
What is the role of the CO2 in the cocrystallization?
This work focuses on the action of CO2 to induce the
cocrystallization of six well-known APIs that form cocrystals
with SAC: theophylline (TPL), IND, carbamazepine (CBZ),
caffeine (CAF), sulfamethazine (SFZ), and acetylsalicylic acid
(ASA). We aim to identify limiting parameters (thermodynamic
and kinetic) influencing the formation of cocrystals in SC-CO2.
Moreover, some comparative studies were carried out between
the use of SC-CO2, SC-N2 (which has much smaller solvent
power) and supercritical mixtures of CO2 and ethanol (which
have higher solvation capability). We have also measured the
cocrystallization rates of the systems IND−SAC and TPL−
SAC and the solubilities of the cocrystal formers in CO2 at
corresponding pressure and temperature conditions. Overall,
this work should provide some answers for the questions
formulated above, thus clarifying the mechanisms underlying
cocrystallization in CO2 and discussing its feasibility as a
cocrystallization media.
EXPERIMENTAL SECTION
■
Materials. Theophylline, IND (γ-form), carbamazepine, acetylsa-
licylic acid, and saccharin were purchased from Sigma Aldrich (purity
of these chemicals was >99.9%). Caffeine was purchased from Fluka
(purity > 99.0%) and sulfamethazine from Acros Organics (purity
>99.0%). Carbon dioxide (99.98%) was supplied by Air Liquide
(Portugal).
Powder X-ray Diffraction (PXRD). X-ray powder data were
collected in a D8 Advance Bruker AXS θ−2θ diffractometer, with
copper radiation (Cu Kα, λ = 1.5406 Å) and a secondary
monochromator. The tube voltage and amperage were 40 kV and
40 mA, respectively. Each sample was scanned with 2θ between 5° and
35° with a step size of 0.02° and 0.5 s at each step. The instrument had
previously been calibrated using a silicon standard.
Each test was performed overnight to ensure the equilibrium
solubility of solid components, and each point was repeated at least in
triplicate and standard deviations were calculated.
At the end of each run, the 14.0 cm3 ( 0.05 cm3) DV loop was
closed, and the visual cell was depressurized. The DV loop tubing was
immersed in a liquid nitrogen bath for approximately 30 min to freeze
the CO2. Afterward, the valves that kept the DV loop closed were
opened allowing the CO2 to bubble slowly inside a flask with a known
volume of ethanol (for the solubility measurements of IND) or
Thermal Analysis. Differential scanning calorimetry (DSC) of the
TPL-SAC samples was performed using a differential scanning
B
DOI: 10.1021/acs.cgd.5b00200
Cryst. Growth Des. XXXX, XXX, XXX−XXX

Crystal Growth & Design
Article
phosphate buffer pH 7.4 (for the solubility measurements of
theophylline and saccharin). The solution was used to rinse the DV
loop several times and was then analyzed by HPLC or UV
spectroscopy. The CO2 quantity in the DV loop was calculated
using its density for the corresponding P, T conditions.23
High-Performance Liquid Chromatography (HPLC). The
HPLC procedure for the solubility determination of APIs in CO2
(presented in detail in the Supporting Information) uses a Jasco
HPLC-system apparatus provided with a 3.9 × 150 mm Nova-Pak C18
column (Waters) and equipped with a variable UV detector (set at 254
nm). The column was kept at room temperature. The mobile phase
consisted of acetate buffer, pH 4: acetonitrile (95:5), delivered at a
flow rate of 1.2 mL min−1. It was filtered through a 0.45 μm membrane
and degassed by sonication prior to the use. Elution was isocratic, with
a flow rate of 1 mL/min. The injection volume was 20 μL and for the
detection a PDA detector was used. An elution time of 12 min sufficed
for the analysis of the samples.
(Omega model PX603), respectively. A Newport Compressor (model
46-13421-2) was used to load the CO2 into a vessel storage cylinder
(VSC) before being introduced into the high-pressure vessel. These
are placed inside a temperature-controlled air chamber. A total mass of
400 mg of one API (TPL, IND, CBZ, CAF, SFZ, or ASA) and
coformer (saccharin) in equimolar amounts (1:1 molar ratio) was
inserted in the high-pressure vessel. The two powders were then mixed
with CO2 at 20.0 MPa and 50 °C.
During each experiment, the solid mixture was subjected to
magnetic stirring at 300 rpm. After 2 h of reaction, the high-pressure
vessel was slowly depressurized and the resulting material was
collected. In order to study the effect of convection in the
cocrystallization by the CSS technique, some experiments were carried
out without magnetic stirring. Negative control experiments were
carried without CO2 (using air at atmospheric pressure) at 50 °C in
the CSS apparatus with magnetic stirring. This was important to check
the effect of the magnetic stirring alone (without the presence of
CO2).
A UV spectrophotometer was used wherever HPLC analysis was
not feasible. Each sample of IND solution was passed through a 0.20
μm filter and analyzed using an UV spectrophotometer method using
microplates. A Spectra max Plus 384 Spectrophotometer was used at a
wavelength of 319 nm.
To evaluate the effect of a different solvent medium in the
cocrystallization by the CSS technique, two approaches were used (see
Table 1); a total of 1 μL of ethanol per mg of powder was introduced
into the high-pressure vessel before the pressurization with CO2;
second, nitrogen (N2) was also used instead of CO2 (no ethanol at the
conditions mentioned above (20.0 MPa and 50 °C). Note that N2 is in
the supercritical state above 3.40 MPa and −147 °C.).
CSS Process Description and Cocrystallization. Figure 2 schemati-
cally shows the CSS setup. It consists essentially of an 8 cm3 stainless
The material recovered from the experiments was stored in a closed
desiccator prior to its characterization.
Cocrystal Formation Kinetics in CSS. Kinetics of TPL−SAC
Cocrystal Formation. Cocrystallization of TPL−SAC in the CSS
process was monitored using a PXRD quantification method presented
in the Supporting Information. As described above, TPL and SAC
were introduced in the CSS apparatus, in a 1:1 molar ratio, prior to the
pressurization with CO2. The kinetics of the cocrystallization were
analyzed by running the experiments using different reaction times: 0,
5, 20, 30, 45, 60, 90, and 120 min. This was performed under three
different conditions: at high density supercritical CO2 (20.0 MPa, 50
°C), low density supercritical CO2 (8.0 MPa, 50 °C), and liquid CO2
(8.0 MPa, 30 °C). The effect of the temperature and CO2 pressure was
also analyzed. It is worth mentioning that a significant amount of
experimental work was required to generate all the results:
approximately 200 experiments were operated by the CSS technique
to complete this task.
Kinetics of IND−SAC Cocrystal Formation. The kinetics of the
cocrystallization was studied by running the experiments using
different reaction times: 0, 5, 20, 30, 45, 60, 90, and 120 min. This
analysis was accomplished using high density supercritical CO2 (20.0
MPa, 50 °C). The results obtained for the kinetics of IND−SAC
cocrystal formation in the CSS technique were analyzed and compared
with the kinetics of TPL−SAC cocrystal formation (both processed
under the same conditions in supercritical CO2 for comparison
purposes).
Figure 2. Schematic description of the cocrystallization with
supercritical solvent (CSS) setup. VSC − vessel storage cylinder;
HPV − high-pressure vessel.
steel high-pressure vessel (HPV) with monitorized temperature and
pressure using a T-type thermocouple and a pressure transducer
Table 1. Evidence from PXRD Analysis for Cocrystallization of Several APIs with SAC after 2 h of Reaction by CSS Technique
Using Different Solvent Media at Supercritical Conditions: SC-CO2, SC-N2, and SC-CO2 with ethanol as a Cosolvent (1 μL of
a
EtOH per mg of powder)
+ saccharin
air (0.1 MPa, 50 °C) SC-CO2 (20 MPa, 50 °C) SC-CO2 (20 MPa, 50 °C) SC-(CO2 + EtOH) (20 MPa, 50 °C) SC-N2 (20 MPa, 50 °C)
API
300 rpm
not stirred
300 rpm
300 rpm
300 rpm
TPL
CBZ
IND
CAF
SFZ
ASA
− −
− −
− −
− −
− −
− −
+ −
+ −
+ −
− −
− −
− −
+ +
+ +
+ +
− −
− −
− −
+ +
+ +
+ +
− −
+ +
− −
+ −
+ −
+ −
+ −
+ −
− −
a
Control runs were carried for the stirring effect using static SC-CO2 (not stirring) and stirring only (without CO2) in air at normal pressure. − −:
mixture of API and saccharin; + −: mixture of cocrystal, API and saccharin; ++: pure cocrystal.
C
DOI: 10.1021/acs.cgd.5b00200
Cryst. Growth Des. XXXX, XXX, XXX−XXX

Crystal Growth & Design
Article
RESULTS
Table 2. Solubilities of TPL, SAC, and IND in CO2 for the
Processing Conditions Studied
■
a
Cocrystal Screening by CSS. Table 1 shows the systems
studied using the CSS technique. PXRD measurements of the
runs using CAF, SFZ, and ASA did not reveal any cocrystal
peaks when only CO2 was used. Instead, the diffractograms
obtained matched the mixture of the crystalline forms of both
pure components. None of the drugs cocrystallized in air at
atmospheric pressure and temperature of 50 °C using only the
stirring facility.
Table 1 also shows that cocrystallization was very limited
when using CO2 without stirring, only partial cocrystal
formation was observed for the systems TPL−SAC, IND−
SAC, and CBZ−SAC, and no cocrystal formation was observed
at all (by PXRD) for the remaining systems. However, when
using CO2 with stirring, full cocrystal formation was achieved
for TPL−SAC, IND−SAC, and CBZ−SAC. Table 1 also shows
that stirring by itself with atmospheric air at 50 °C does not
cause cocrystallization of these APIs with SAC.
Figures S1−S6 in the Supporting Information show the
PXRD diffractograms for the systems processed in the CSS
technique using CO2 with stirring. The d values and 2θ values
of the characteristic PXRD peaks corresponded well with those
reported elsewhere.2,20,21,24,25
The addition of ethanol as a cosolvent enabled the formation
of pure cocrystals of SFZ−SAC, which were never obtained
using CO2 only. Using N2 instead, the cocrystal was produced
with the pure components for all the systems studied except for
ASA−SAC.
pressure
(MPa)
temperature
solubility × 102
substance
TPL
CO2 state
(°C)
mol m−3
(CO2)
supercritical
supercritical
liquid
20
8
50
50
30
50
50
30
50
50
30
9.8 1.0
1.7 0.2
2.7 0.2
59.2 4.3
12.2 2.0
22.9 2.7
6.0 0.5
1.5 0.1
1.1 0.2
8
SAC
IND
supercritical
supercritical
liquid
20
8
8
supercritical
supercritical
liquid
20
8
8
a
TPL − theophylline; SAC − saccharin; IND − indomethacin.
liquid phase. In supercritical CO2 (20.0 MPa, 50 °C), SAC is 6
times more soluble than TPL and 10 times more soluble than
IND. Table 2 shows that TPL solubility in SC-CO2 is 9.8 ×
10−2 mol m−3 (20.0 MPa, 50 °C), which is in agreement with
the solubility data reported elsewhere.26
Kinetics of Cocrystal Formation by CSS. The PXRD
quantification method developed for TPL−SAC (section SI2 in
the Supporting Information) was used to study the cocrystal
formation kinetics of TPL−SAC in CO2. This was prepared for
different conditions: SC-CO2 at high density (20.0 MPa, 50
°C), SC-CO2 at low density (8.0 MPa, 50 °C), and liquid CO2
(8.0 MPa, 30 °C).
Figure 4 shows the cocrystallization kinetics for the system
TPL−SAC. A two-step linear relation of cocrystallization with
The effect of stirring is also observed in the DSC
thermograms. With magnetic stirring a single melting transition
point at 207.2 °C is observed, and not the eutectic melting
between 180 and 190 °C (typical when traces of TPL and SAC
are present in the mixture). For the experimental run with CO2
without stirring, a lower eutectic melting at 181.7 °C appears in
the DSC curve (Figure 3c), and an additional melting peak at
199 °C is observed. The impurity can also be inferred from the
PXRD diffractogram (Figure S1 in the Supporting Informa-
tion).
Solubility of APIs and Coformers in CO2. Table 2 shows
the measured solubilities of the pure substances TPL, SAC, and
IND. As expected, the solubility of each component increases
with the CO2 pressure and temperature. However, compared to
TPL and IND, the solubility of SAC is still relatively high in the
Figure 4. Formation kinetics of TPL−SAC cocrystals in (a) SC-CO2
at 50 °C and 20.0 MPa, (b) liquid CO2 at 30 °C and 8.0 MPa, and (c)
SC-CO2 at 50 °C and 8.0 MPa.
time is observed for most of the cocrystallization reaction. An
abrupt shift in the slope of the cocrystallization rate is observed
at approximately 45 min in liquid CO2 (8.0 MPa, 30 °C; Figure
4b) and at 30 min in SC-CO2 (20.0 MPa, 50 °C; Figure 4a).
This shift occurred at approximately 80% cocrystallization
either for liquid or SC-CO2. Beyond this point the percentage
increased slowly to almost 100% after 2 h of reaction. In the run
at 8.0 MPa and 50 °C (Figure 4c), full cocrystallization was still
not achieved after 2 h of reaction, and the amount of cocrystal
is rather small compared to the other conditions.
Cocrystallization rates were calculated for the linear period.
Table 3 presents data generated from the linear regression of
the curves from Figure 4a (in the range t = 0−30 min), 3b (t =
0−45 min), and 3c (t = 0−90 min).
Figure 3. DSC thermograms for (a) pure theophylline, (b) pure
saccharin, (c) TPL−SAC cocrystals produced by the CSS process
without stirring, and (d) TPL−SAC cocrystals produced by the CSS
process with stirring.
D
DOI: 10.1021/acs.cgd.5b00200
Cryst. Growth Des. XXXX, XXX, XXX−XXX

Crystal Growth & Design
Article
Table 3. Cocrystallization Rates Observed in Figure 4 and Corresponding Experimental Conditions Operated in the CSS High-
Pressure Vessel
cocrystallization rate × 103
23
pressure (MPa) temperature (°C) CO2 density × 105 g m−3 (state) TPL solubility × 102 mol m−3(CO2) g min−1 mol·min−1 m−3
curves
(CO2)
Figure 4a
Figure 4b
Figure 4c
20
8
50
30
50
7.8 (SC)
7.0 (liq)
2.2 (SC)
9.8 1.0
2.7 0.2
1.7 0.2
11.5
7.6
0.40
0.26
0.08
8
2.3
Table 3 shows that cocrystallization in SC-CO2 at 20.0 MPa
and 50 °C (Figure 4a) was nearly 5 times faster than that at 8.0
MPa and 50 °C (Figure 4c). The amount of cocrystal formation
per minute was obtained from the slope of each curve
represented in Figure 4. The specific cocrystallization rate in
SC-CO2 at 8.0 MPa and 50 °C (Figure 4c) was the smallest.
Temperature, pressure, and density all contribute to the
cocrystallization rate as will be discussed below.
Pressure and Temperature Effect on TPL−SAC
Formation. Several runs were carried on the CSS apparatus
for TPL−SAC system at 50 °C and different pressures for 2 h.
The results of the pressure effect on cocrystallization are shown
in Figure 5.
slightly decreases the cocrystallization yield. However, it should
be borne in mind that a higher temperature at constant
pressure implies a smaller quantity of CO2 due to the decrease
in its density.
Figure 6 shows that increasing temperature has a negative
impact on the crystallization rate, which could also be explained
by the decreasing of CO2 density.
Kinetics of TPL−SAC versus IND−SAC. The IND−SAC
cocrystal formation was performed by the CSS technique at
20.0 MPa and 50 °C and compared with the same conditions
used for TPL−SAC kinetic study.
Figure 7 presents the cocrystallization kinetics of TPL−SAC
and IND−SAC cocrystals in the CSS technique at 20.0 MPa
Figure 5. Effect of pressure in the cocrystallization yield (%) of TPL−
SAC at 50 °C after 2 h in CO2.
Figure 7. Formation kinetics of TPL−SAC and IND−SAC cocrystals
in supercritical CO2 at 50 °C and 20.0 MPa.
Above the critical pressure of CO2 (Pc = 7.38 MPa), there is
a noticeable increase in the cocrystal yield after 2 h compared
to the runs that used gaseous carbon dioxide (below Pc).
CSS runs (TPL−SAC system and 2 h duration) were carried
out for two constant pressure values (at 20.0 and 8.0 MPa) and
at different temperatures. Figure 6 shows that the temperature
and 50 °C. This shows that IND−SAC formation kinetics is
lower than TPL−SAC in supercritical CO2 at the same
conditions of pressure and temperature (20.0 MPa, 50 °C).
Table 4 presents data for the cocrystallization rates calculated
for the first linear period of the curves of Figure 7 for TPL−
Table 4. Cocrystallization Rates of TPL−SAC and IND−
SAC Cocrystals in the CSS Technique Calculated from
Curves of Figure 7
linear regression
cocrystallization rate
103
104 mol min−1
cocrystal
%
min−1
R2
g min−1
m−3
(cocrystal)
(CO2)
TPL−SAC
IND−SAC
2.9
1.5
0.995
0.978
11.5
6.0
4.0
1.4
SAC (between t = 0 min and t = 30 min) and IND−SAC (t = 5
min and t = 60 min). The specific cocrystallization rate of
TPL−SAC (mol min−1 m−3(CO )) in supercritical carbon
2
dioxide at 20.0 MPa and 50 °C was about 3 times higher
than that of IND−SAC.
Figure 6. Effect of temperature in the cocrystallization yield (%) of
TPL−SAC after 2 h in CO2 at (a) 20.0 MPa and (b) 8.0 MPa.
E
DOI: 10.1021/acs.cgd.5b00200
Cryst. Growth Des. XXXX, XXX, XXX−XXX

Crystal Growth & Design
Article
(1) Some impurities of both TPL and SAC may be deposited
in the walls of the apparatus; therefore these impurities are not
flowing in the CO2 with the remaining powder. After most
cocrystals are formed (∼80% cocrystallization), the absence of
TPL and SAC flowing freely in the CO2 slows down the
reaction rate.
(2) Possible differences in the kinetics of dissolution and
cocrystallization at the surface and core of the particles may
slow down the cocrystallization reaction.
DISCUSSION
■
Overall, the results show CO2 acting as a solvent, capable of
mediating molecular recognition events, thus enabling or
enhancing cocrystallization. Good yields were obtained in half
of the cases after 2 h of reaction (Table 1). Albeit the reaction
time was relatively short, it proved to be long enough to
produce pure cocrystals of the TPL−SAC (Figure 3), IND−
SAC, and CBZ−SAC, suggesting that CO2 may be considered
as a serious and very convenient alternative solvent for the
production of pure pharmaceutical cocrystals.
Nonetheless, not all systems cocrystallized in CO2. For
example, although CAF and TPL are both xanthines relatively
soluble in CO2 compared to most APIs, only TPL cocrystal-
lized with SAC. This solvent sensitivity also occurs in the
cocrystallization of many APIs with organic solvents. In fact, it
is common practice in cocrystallization screenings to use several
solvents to increase the chances of successfully obtaining the
desired cocrystal. For example, Basavoju et al.2 cocrystallized
IND with SAC in several organic solvents (using solvent
evaporation), but when using methanol or acetone a mixture of
pure components and cocrystal was obtained instead. The
different ability to cocrystallize different APIs using SC-CO2,
SC-CO2 with ethanol, and N2 shows that phase equilibria
thermodynamics plays an important role as in cocrystallization
in liquid organic solvents. This may be related to the dissolved
stoichiometric ratio, or to a transition coordination state,
required for cocrystallization, which is being hindered in the
selected solvent.27 The overall balance between the affinity of
the substances for the solvent or for each other plays an
important role herein, as in other processes that are mediated
by a solvation step in the supercritical phase, for example, as in
the case of impregnation using a supercritical solvent.28
Overall, SC-CO2 showed great potential for production of
pharmaceutical cocrystals; however, while trying to control the
number of variables in this work, it was possible that only one
cocrystal former was usedsaccharinwhich has a high
solubility in CO2 (10 times higher than TPL or IND). This
means that the dissolution of the API (TPL or IND) in the
CO2 is probably the rate-limiting step for the cocrystallization
with SAC using the CSS technique. Nonetheless, other
coformers less soluble than SAC may cause cocrystallization
to be considerably slower, an issue to be clarified in a future
work.
The results suggest that cocrystallization is mediated by the
dissolution of the substances in CO2, despite low solubility of
the cocrystal components in CO2 (Table 2). However, the
dramatic increase in the cocrystallization observed for
conditions beyond the critical point of CO2 (shown in Figure
5) suggests a possible connection with the solubility issue.
Gaseous CO2 cannot dissolve substances due to its small
density, and this ability starts once its critical state is
overpassed, even if at a very nearby range. The solvent power
increases with the fluids’ density, i.e., with either increasing
pressure or decreasing temperature. Figures 4−6 show that the
kinetics of the cocrystallization follows the density of CO2. In
denser CO2, the reaction was faster, suggesting that solubility
mediates the cocrystallization reaction.
N2 shows the potential to form cocrystals that are not formed
in the presence of CO2. However, due to the lower density of
N2 (1.9 × 105 g m−3) and possibly low solubility of the pure
components, the cocrystallization was small in SC-N2. Since
agitation alone did not promote cocrystallization at atmos-
pheric pressure (Table 1), cocrystallization in N2 is probably
also mediated by a solvation step. The method developed
herein to measure solubilities relies on freezing the supercritical
phase with liquid N2, which is not possible when N2 is the
supercritical fluid. Nonetheless, we also observe some
cocrystallization when gaseous CO2 was used (see Figure 5
below critical pressure)even though its density is 2−3 times
smaller than N2 at 20 MPa and 50 °C. Therefore, considering
the relatively high cocrystallization rates observed for super-
critical CO2,, it is plausible that a significant drop in solvation
power would still enable measurable cocrystallization after 2 h,
as observed for N2 or gaseous CO2.
The results presented herein make cocrystallization in CO2
an interesting process for scale-up. In conventional reaction
crystallization (in organic solvents), the purification of the
cocrystal from the solvent and from other nonreacted pure
components could be time and energy expensive. Therefore,
the CSS technique is in line with the current process
intensification requirements since a considerable capacity for
cocrystal formation is herein achieved in a relatively small size
reactor and time. Since CSS uses supercritical suspensions
rather than solutions, the volume of the crystallization reactor is
relatively small if compared even with other SC-CO2
technologies that require handling supercritical and liquid
solutions. Therefore, the volume under pressure, which is
usually a deterring factor for scale-up, is herein minimized.
The results obtained without stirring (Table 1) show that
transport phenomena play an important role in cocrystalliza-
tion. Without stirring, cocrystallization was significantly limited,
which has also been observed previously.20 In this case,
transport from the static powder interface is essentially
determined by the solute’s diffusivity in the supercritical fluid
(on the order of 10−8 m2/s) and the concentration gradient,
which is very small due to the component’s low solubility.
Under such conditions the supercritical phase is unlikely to
homogeneously saturate in the reactants, thus hindering
cocrystallization. Conversely, stirring promotes intense mass
transfer by convection; therefore, cocrystallization becomes not
limited by the reactants transport since CO2 is homogeneously
saturated in both components. The saturation of CO2 is evident
in Figure 4, where the cocrystallization curves show a typical
zero order reaction, i.e., linear in time. Since CO2 is always
saturated in both components, the reaction velocity is constant.
However, an abrupt shift in the linear cocrystallization rate is
observed in all cases, which can be explained by two possible
reasons:
CONCLUSIONS
■
CO2 seems to be a “green” and effective cocrystallization media
for some APIs. Cocrystals of TPL−SAC, IND−SAC, and
CBZ−SAC were produced in approximately 2 h in a CO2
suspension.
F
DOI: 10.1021/acs.cgd.5b00200
Cryst. Growth Des. XXXX, XXX, XXX−XXX

Crystal Growth & Design
Article
(9) Miroshnyk, I.; Mirza, S.; Sandler, N. Adv. Drug Delivery Rev. 2009,
6, 333−341.
Homogeneity in the cocrystallization medium caused by
stirring was determinant for a successful cocrystallization most
likely due to an enhancement of the intermolecular interactions
between the pure components. The use of a cosolvent
(ethanol) with CO2 generated SFZ−SAC cocrystals which
could not be produced using CO2 alone. When using N2
instead of CO2, some SFZ−SAC, CAF−SAC cocrystals were
also formed as well.
The cocrystallization was revealed to be mediated by a
dissolution step: the higher the concentration of the API and
coformer in the CO2 phase, the faster the cocrystallization rate.
TPL−SAC showed a higher cocrystallization rate than IND−
SAC due to the higher solubility of TPL in CO2 compared to
IND. The cocrystallization followed zero order kinetics, which
is explained by the constant saturation of the CO2 in both
cocrystal components.
Overall, CO2 showed interesting potential for cocrystalliza-
tion because it enabled reasonably fast kinetics and it is easily
separated from the products unlike organic solvents. The
proposed new technique of CSS is promising because it reduces
the equipment volume and the costs in further solvent
separation steps for the production of pharmaceutical
cocrystals.
(10) Weyna, D. R.; Shattock, T.; Vishweshwar, P.; Zaworotko, M. J.
Cryst. Growth Des. 2009, 9, 1106−1123.
̌ ̌ ́
(11) Friscic, T.; Jones, W. Cryst. Growth Des. 2009, 9, 1621−1637.
(12) Uchida, H.; Manaka, A.; Matsuoka, M.; Takiyama, H. Cryst.
Growth Des. 2004, 4, 937−942.
(13) Mazzotti, M.; Fusaro, F.; Muhrer, G. Cryst. Growth Des. 2004, 4,
881−889.
(14) Bettini, R; Menabeni, R.; Tozzi, R.; Pranzo, M. B.; Pasquali, I.;
Chierotti, M. R.; Gobetto, R.; Pellegriono, L. J. Pharm. Sci. 2009, 99,
1855−1870.
(15) Bettini, R.; Bonassi, L.; Castoro, V.; Rossi, A.; Zema, L.;
Gazzaniga, A.; Giordano, F. Eur. J. Pharm. Sci. 2001, 13, 281−286.
́
(16) Martín, A.; Scholle, K.; Mattea, F.; Meterc, D.; Cocero, M. J.
Cryst. Growth Des. 2009, 9, 2504−2511.
́
(17) Bouchard, A.; Jovanovic, N.; Hofland, G. W.; Mendes, E.;
Crommelin, D. J. A.; Jiskoot, W.; Witkamp, G.-J. Cryst. Growth Des.
2007, 7, 1432−1440.
(18) Rodrigues, M.; Padrela, L.; Geraldes, V.; Santos, J.; Matos, H. A.;
Azevedo, E. G. J. Supercrit. Fluids 2011, 58, 303−312.
́
(19) Roy, C.; Vrel, D.; Veja-Gonzalez, A.; Jestin, P.; Laugier, S.;
Subra-Paternault, P. J. Supercrit. Fluids 2011, 57, 267−277.
(20) Padrela, L.; Rodrigues, M. A.; Velaga, S. P.; Matos, H. A.;
Azevedo, E. G. Eur. J. Pharm. Sci. 2009, 38, 9−17.
(21) Padrela, L.; Rodrigues, M. A.; Velaga, S. P.; Fernandes, A. C.;
Matos, H. A.; Azevedo, E. G. J. Supercrit. Fluids 2010, 53, 156−164.
(22) Tiago, J. M.; Padrela, L.; Rodrigues, M. A.; Matos, H. A.;
Almeida, A. J.; Azevedo, E. G. Cryst. Growth Des. 2013, 13, 4940−
4947.
ASSOCIATED CONTENT
* Supporting Information
■
S
High-performance liquid chromatography method for determi-
nation of solubility of the APIs in CO2, PXRD analysis of the
samples, and a PXRD method for quantification of TPL−SAC
and IND−SAC cocrystals. The Supporting Information is
available free of charge on the ACS Publications website at
DOI: 10.1021/acs.cgd.5b00200.
(23) Linstrom, P. J.,; Mallard, W. G., Eds. NIST Chemistry WebBook;
NIST Standard Reference Database Number 69; National Institute of
Standards and Technology: Gaithersburg, MD, 2005; pp 20899;
http://webbook.nist.gov/chemistry.
(24) Lu, E.; Rodríguez-Hornedo, N.; Suryanarayanan, R. CrystEng-
Comm 2008, 10, 665−668.
(25) Padrela, L.; Azevedo, E. G.; Velaga, S. P. Drug Dev. Ind. Pharm.
2012, 38, 923−929.
AUTHOR INFORMATION
Corresponding Author
*E-mail: miguelrodrigues@tecnico.ulisboa.pt.
■
(26) Johannsen, M.; Brunner, G. Fluid Phase Equilib. 1994, 95, 215−
226.
(27) Roy, L.; Lipert, M. P.; Rodríguez-Hornedo N., In Pharmaceutical
Salts and Co-Crystals; Wouters, J., Thurston, D. E., Eds.; RSC
Publishing: London, UK, 2011; Chapter 11, pp 247−276.
(28) Diankov, S.; Barth, D.; Vega-Gonzalez, A.; Pentchev, I.; Subra-
Paternault, P. J. Supercrit. Fluids 2007, 41, 164−172.
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
For financial support, the authors are grateful to Fundaca
̧
o para
̃
a Cien
̂
cia e Tecnologia (FCT), Lisbon (Grants SFRH/BD/
39836/2007, IF/01183/2013, and Projects PTDC/EQUFTT/
099912/2008 and UID/QUI/00100/2013).
REFERENCES
■
̈
(1) Almarsson, O.; Zaworotko, M. J. Chem. Commun. 2004, 17,
1889−1896.
(2) Basavoju, S.; Bostrom, D.; Velaga, S. Pharm. Res. 2008, 25, 530−
̈
541.
(3) Basavoju, S.; Bostrom, D.; Velaga, S. Cryst. Growth Des. 2006, 6,
̈
2699−2708.
(4) Min-Sook, J.; Jeong-Soo, K.; Min-Soo, K.; Alhalaweh, A.; Cho;
Wonkyung; Hwang; Sung-Joo; Velaga, S. P. J. Pharm. Pharmacol. 2010,
62, 1560−1568.
(5) 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.
(6) Wuest, J. D. Nature Chem. 2012, 4, 74−75.
(7) Shan, N.; Zaworotko, M. J. Drug Discovery Today 2008, 13, 440−
446.
(8) Schultheiss, N.; Newman, A. Cryst. Growth Des. 2009, 9, 2950−
2967.
G
DOI: 10.1021/acs.cgd.5b00200
Cryst. Growth Des. XXXX, XXX, XXX−XXX