Conformational Color Polymorphism and Control of crystallization of 5-Methyl-2-[(4-methyl-2-nitrophenyl)amino]-3-thiophenecarbonitrile

Article abstract of DOI:10.1002/1520-6017(200103)90:3<371::AID-JPS12>3.0.CO;2-N

5-Methyl-2-[(4-methyl-2-nitrophenyl)amino]-3-thiophenecarbonitrile is an example of conformational and color polymorphism. The compound crystallizes in red (R), dark red (DR), light red (LR), and orange (O) modifications. There are two specific goals for this study. One is to characterize the complex thermodynamic relationship among these four known forms, and the other is to use the knowledge of the thermodynamic relationship to control the crystallization of these forms. The different forms were characterized by X-ray powder diffractometry as well as Fourier-transform infrared (FT-IR) and Raman spectroscopy; their complex thermodynamic relationships were determined by thermal analysis, solubility measurements, and slurry conversion studies. According to the solubility results, all forms are enantiotropically related: R is the thermodynamically most stable form above 60 deg C, O is the most stable form between room temperature and 60 deg C, LR is the most stable form below - 15 deg C, and DR is metastable throughout the entire temperature range. DR, LR, and O have very similar free energy at ambient temperature, which is the reason for the complex transition behavior. Finally, a schematic energy - temperature diagram was constructed that combines all experimental data in a comprehensive thermodynamic picture and provides insights into how to control the crystallization of the individual forms.

Full text of DOI:10.1002/1520-6017(200103)90:3<371::AID-JPS12>3.0.CO;2-N

Conformational Color Polymorphism and Control of
Crystallization of 5-Methyl-2-[(4-methyl-2-
nitrophenyl)amino]-3-thiophenecarbonitrile
XIAORONG HE,1* ULRICH J. GRIESSER,2 JOSEPH G. STOWELL,1 THOMAS B. BORCHARDT,3 STEPHEN R. BYRN1
1Department of Industrial & Physical Pharmacy, Purdue University, West Lafayette, Indiana 47907-1336
2Institute of Pharmacognosy, University of Innsbruck, Innrain 52, A-6020 Innsbruck, Austria
3G.D. Searle Pharmaceuticals, PDAD Department, 4901 Searle Parkway, Skokie, Illinois 60077-2980
Received 13 July 2000; revised 1 October 2000; accepted 9 October 2000
ABSTRACT: 5-Methyl-2-[(4-methyl-2-nitrophenyl)amino]-3-thiophenecarbonitrile is an
example of conformational and color polymorphism. The compound crystallizes in red
(R), dark red (DR), light red (LR), and orange (O) modi®cations. There are two speci®c
goals for this study. One is to characterize the complex thermodynamic relationship
among these four known forms, and the other is to use the knowledge of the thermo-
dynamic relationship to control the crystallization of these forms. The different forms
were characterized by X-ray powder diffractometry as well as Fourier-transform
infrared (FT-IR) and Raman spectroscopy; their complex thermodynamic relationships
were determined by thermal analysis, solubility measurements, and slurry conversion
studies. According to the solubility results, all forms are enantiotropically related: R is
the thermodynamically most stable form above 60ꢀC, O is the most stable form between
room temperature and 60ꢀC, LR is the most stable form below 15ꢀC, and DR is metas-
table throughout the entire temperature range. DR, LR, and O have very similar free
energy at ambient temperature, which is the reason for the complex transition behavior.
Finally, a schematic energy±temperature diagram was constructed that combines all
experimental data in a comprehensive thermodynamic picture and provides insights
into how to control the crystallization of the individual forms. ß 2001 Wiley-Liss, Inc. and
the American Pharmaceutical Association J Pharm Sci 90:371±388, 2001
Keywords: color polymorphism; solid state; crystallization; phase transition; confor-
mational polymorphism; solubility; energy±temperature diagram; thermodynamics;
crystal forms
INTRODUCTION
tures.2 The structural differences often have a
profound effect on the physicochemical properties
of a given compound. For example, polymorphs
can have different melting points, crystal shapes,
colors, chemical reactivities, dissolution rates, and
solubilities.3 Controlling polymorphism is very
important in such ®elds as the food and pigment
industries as well as the pharmaceutical industry.
For drug substances, the structural differences
About one third of organic molecules crystallize
into more than one solid form (polymorphs).1 By
de®nition, polymorphs share the same chemical
composition but have different crystal struc-
Present address: Pharmacia Inc., 7000 Portage Road,
Kalamazoo, Michigan 49001-0199.
are known to signi®cantly affect their bioavail-
Correspondence to: S. R. Byrn (Telephone: 765-494-1460;
Fax: 765-494-6545; E-mail: sbyrn@pharmacy.purdue.edu) or
X. He (Telephone: 616-833-6999; Fax: 616-833-7290; E-mail:
xiaorong.he@am.pnu.com)
6
ability and processability.4
Because crystal-
lization is the major process for drug substance
puri®cation, it is important to control the crystal-
lization to obtain a pure polymorph consistently.
Journal of Pharmaceutical Sciences, Vol. 90, 371±388 (2001)
ß 2001 Wiley-Liss, Inc. and the American Pharmaceutical Association
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 90, NO. 3, MARCH 2001
371

372
HE ET AL.
A subclass of polymorphism is conformational
polymorphism, where a ¯exible molecule can
adopt different conformations in different crystal
forms. The Gibbs free energy difference between
conformational polymorphs is often small, espe-
cially when van der Waals interactions dominate
the structure.7 This small difference presents a
potential challenge to characterize the thermo-
dynamic relationships of conformational poly-
morphs. In addition, the transition temperatures
of conformational polymorphs are often close to
room temperature or within the crystallization
temperature range. Hence, it is challenging to
control the crystallization of each form because
any of the polymorphs may be isolated under
similar crystallization conditions.
and co-workers.8;9 Stephenson et al.9 reported
that the parent compound exhibits dramatic color
polymorphism, crystallizing in red, orange, and
yellow modi®cations (nicknamed ``ROY,'' accord-
ing to the initials of the colors) of which there
are now a total of six polymorphs with solved
crystal structures. Similarly, 40-Me crystallizes
at least four solvent-free crystal forms, red (R),
dark red (DR), light red (LR), and orange
(O). The four forms were characterized using a
variety of techniques. Based on these studies, an
energy±temperature diagram was constructed.
The diagram provides information of the best
temperature region to obtain each polymorph
(Figure 11).
A common approach to control the crystal-
lization of a given compound is trial-and-error
where the crystallization conditions are varied
randomly to ®nd an optimum set of conditions to
crystallize a polymorph consistently. However,
this approach can be very time-consuming be-
cause many factors (e.g., solvent, supersatura-
tion, impurity, cooling rates, stirring rates, etc.)
can in¯uence the crystallization outcomes. Very
often, seeding is also used to control the crystal-
lization of polymorphs. However, seeding does not
always work. Sometimes, the seeded form has a
very slow crystal growth, and other undesirable
polymorphs may grow in the presence of seeds;
other times, seeding may result in the crystal
growth of the desired form, yet, the desired form,
if metastable, may transform quickly to the more
stable form. In this study, we present an approach
to control the crystallization of conformational
polymorphs. The objectives of this study are two
fold: (1) to characterize the complex thermody-
namic relationship of conformational polymorphs,
and (2) to control the crystallization of these
polymorphs based on their thermodynamic
relationships. The compound investigated is 5-
methyl-2-[(4-methyl-2-nitrophenyl)amino]-3-thio-
phenecarbonitrile (C13H11O2N3S; see Figure 1)
or 40-Me, a derivative of the parent compound
5-methyl-2-[(2-nitrophenyl)amino]-3-thiophene-
carbonitrile previously studied by Stephenson
EXPERIMENTAL SECTION
Materials
The synthetic procedure for ROY was modi®ed to
prepare the 40-Me derivative.10 The synthetic
path consisted of two step, as shown in Scheme 1.
2-Amino-5-methyl-3-thiophenecarbonitrile
Sulfur (64 g, 2 mol, Aldrich), propanol (170 mL, 2
mol, Aldrich), and dimethylformamide (DMF)
(430 mL, Fisher) were transferred to a three-
necked round-bottomed ¯ask equipped with a
dropping funnel, mechanical stirrer, and conden-
ser. The resulting mixture was cooled to 0ꢀC, and
triethylamine (170 mL, Aldrich) was added in a
dropwise manner via the dropping funnel over a
period of 1 h. The resulting dark solution was then
warmed to room temperature over a period of 1 h.
A solution of malonitrile (132 g, 2 mol, Aldrich) in
DMF (240 mL, Fisher) was transferred to the
addition funnel and added in a dropwise manner
over a period of 2 h. The resulting brownish
mixture was stirred at room temperature for 15 h
(overnight). At the end of this time, the mixture
was poured over 6 L of ice and water to yield an
orange precipitate. The solid was ®ltered, washed
with chilled water, and dried in vacuo; yield, 176.1
g (64%); melting point, 99±101ꢀC.
40-Me
To 200 mL of tetrahydrofuran (THF) was added
NaH (70.3 g, 3 equivalents, 60% dispersion in oil).
2-Amino-5-methyl-3-thiophenecarbonitrile (139 g,
1 mol) and 4-¯uoro-3-nitrotoluene (124.5 mL,
1 mol, Aldrich) were dissolved in THF (300 mL,
Figure 1. Molecular structure of 40-Me.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 90, NO. 3, MARCH 2001

CONFORMATIONAL COLOR POLYMORPHISM AND CONTROL OF CRYSTALLIZATION
373
Scheme 1. The synthesis procedure of 40-Me.
Fisher) and added in a dropwise manner to the
suspension while the temperature was main-
tained below 30ꢀC. The reaction mixture was
allowed to stir overnight under N2 purge. The
mixture was then poured into 3 L of ice water,
neutralized with concentrated HCl, and extracted
with 10 L of dichloromethane. The product
was puri®ed in portions by passing 2.5 L of the
dichloromethane extracts through a silica column
(230±400 mesh). The eluate was evaporated in
vacuo to give a red solid by rotary evaporation.
The solid was recrystallized from ethanol to give
red crystals of 40-Me: yield, 83 g (30.4%); melting
point, 127ꢀC.
Company, Weston, MO). R was crystallized at
65±70ꢀC with an initial concentration > 40 mg/
mL. A small amount of the DR form was crystal-
lized from a 20-mg/mL ethanol solution at room
temperature. These DR crystals were used as
seeds and mixed with R, and the mixture was
suspended in ethanol. The suspension was stirred
at 0ꢀC until R converted to DR (within days).
LR was crystallized from an ethanol solution
(25±30 mg/mL) by fast cooling from 60ꢀC to room
temperature. The pure orange form was produced
by stirring a suspension of R in ethanol at 25ꢀC for
1 day.
X-Ray Powder Diffractometry (XRPD)
Purity AnaIysis
Samples were analyzed with a D-500 Diffrakt-
ometer (Siemens AG, D-Karlsruhe) using Cu-Ka
radiation, a voltage of 40 kV, and a current of
20 mA. A Kevex Psi peltier cooled silicon [Si(Li)]
detector was used. The samples were scanned at a
The chemical purity of 40-Me was analyzed by
reversed-phase high-performance liquid chroma-
tography (HPLC) using a C18 column (5 mm, 25 cm
long, i.d. 4.6 mm). The mobile phase initially
consisted of 55% acetonitrile and 45% water, and
then ramped to 70% acetonitrile and 30% water.
The ¯ow rate was 1 mL/min. The injection volume
was 10 mL, and the detector wavelength was set at
215 nm. The retention time of 40-Me was ꢁ 12 min
and the chemical purity is > 99.6%. Elemental
analysis showed less <0.2% difference for C, H, N
between the calculated and experimental values
(calculated: C 57.14%, H 4.03%, N 15.38%; experi-
mental: C 57.23%, H 3.92%, and N 15.41%.).
ꢀ
rate of 6ꢀ/min from 4±40 2y with a step size of
0.04ꢀ2y.
Variable Temperature X-Ray Powder Diffractometry
(VT XRPD)
VT XRPD measurements were recorded with
a XRD-6000 Diffractometer (Shimadzu, Japan)
equipped with a temperature controller (TTK,
Anton Paar AG, Graz, Austria) using Cu-Ka radia-
tion, a voltage of 30 kV, and a current of 30 mA.
DR, LR, and O were heated from 35ꢀC to
temperatures slightly below melting points of
the respective forms (113ꢀC for LR, 115ꢀC for DR
and O) at a heating rate of 5 K Á min 1. The
Preparation of Polymorphs
The four forms studied were prepared by crys-
tallization from ethanol (McCormick Distilling
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 90, NO. 3, MARCH 2001

374
HE ET AL.
temperature was held at appropriate tempera-
ture (113 or 115ꢀC) for 1 min, then increased to
spectra were measured using a Bruker RFS 100
FT-Raman Spectrometer (Bruker Analytische
Messtechnik, Karisruhe, Germany) equipped with
a diode pumped Nd:YAG (1064 nm) as the exci-
tation source and a liquid-nitrogen cooled, high-
sensitivity detector. The powder samples were
packed into small aluminum cups and the spectra
were recorded at an output power of 200 mW
(64 scans per spectrum, instrument resolution:
1
125ꢀC at 5 K Á min , held at 125ꢀC for 1 min,
followed by cooling to 35ꢀC. The samples were
scanned from 14 to 17.5 ꢀ2y at the isothermal
temperature steps of 35ꢀC, 113 (for LR) or 115ꢀC
(for O or DR), 125ꢀC, and back to 35ꢀC. The
scanning rate was 2ꢀ/min, with a step size of 0.02.
The temperature at the surface of the sample was
checked by measuring the melting points of benzil
and the 40-Me R form. According to hot-stage
microscopy, benzil melts at 94.0ꢀC (Tm95ꢀC,
Merck Index) and the 40-Me R form melts at
126.5ꢀC. The two compounds melted at 97.0 and
129.5ꢀC, respectively, according to the tempera-
ture controller for the XRPD hot-stage, indicating
that the sample temperature is 3ꢀC below the
programmed temperature displayed on the XRPD
temperature controller.
2 cm 1). The scanned ranges for IR and Raman
1
spectra were 4000 to 600 cm
cm 1, respectively.
and 3500 to 50
Solubility Studies
The solubility of each polymorph was measured at
10, 20, 25, 35, and 45ꢀC in ethanol, and the
temperature was controlled by a programmable
temperature bath (Fischer Scienti®c, model 9510,
Pittsburg, PA). Ten milliliters of ethanol was
transferred to a 25-mL scintillation vial and
equilibrated at the test temperature for 30 min.
Then, an excess amount of one polymorph was
added, the vial was capped, and the suspension
was stirred continuously with a magnetic stirrer.
Prior to sampling, the stirring was stopped and
the suspension was allowed to settle for 30 min.
The supernatant was then withdrawn and ®ltered
either through a 0.45-mm membrane ®lter (Var-
ian, Nylon-66, Walnut Creek, CA) or through a
pledget of glass wool in a disposable Pasteur
pipette. No signi®cant difference between glass-
wool ®lters and membrane ®lters could be detec-
ted. The concentration was measured at 410 nm
by a Beckmann DU 640 ultraviolet±visible (UV±
vis) spectrophotometer (Beckmann, Fullerton,
CA). A standard curve was constructed at this
Thermal Analysis
Thermomicroscopic analysis was performed with
a Zeiss (Jena, Germany) polarized-light micro-
scope and a Mettler FP5 hotstage with a FP5
temperature controller (Mettler Inc., Greifensee,
Switzerland). Samples were heated at various
heating rates. The ``equilibrium melting point''
was determined by subsequently heating and
cooling the partly melted material. The differen-
tial scanning calorimetry (DSC) thermograms
were recorded with a DSC 2920 (TA Instruments,
New Castle, DE) and a DSC 7 (Perkin-Elmer,
Norwalk,CT). The DSC 2920 was temperature
calibrated with mercury (mp,
38:83ꢀC) and
indium (mp, 156.6ꢀC); the DSC 7 was calibrated
with pure benzophenone (mp, 48.0ꢀC) and caffeine
(mp, 236.2ꢀC). Indium was used for the enthalpy
calibration of both instruments (ÁHf ˆ 28:45 J/g).
About 2 to 4.5 mg of each sample was heated in
a sealed aluminum pan. The heating rate varied
from 1 to 20 K Á min 1. The ¯ow rates of the
helium (DSC 2920) and nitrogen (DSC 7) purges
were 25 and 20 mL/min, respectively.
wavelength, with a correlation coef®cient > 0.999.
1
The absorption coef®cient is 5:3 Á 106 cm 1 M
.
Samples were withdrawn at different time inter-
vals to ensure that the solubility of each poly-
morph was reached. The solution concentration
plateaus within 2 h. Except for R, which trans-
forms to O after 3 h, no solution-mediated trans-
formations were observed during the time scale of
the experiments.
Fourier-Transform Infrared (FT-IR) and Fourier-
Transform Raman Spectroscopy (FT-Raman)
Slurry Conversion Studies
IR spectra were recorded with a Bruker IFS 25
FT-IR spectrometer (Bruker Analytische Mes-
stechnik, Karlsruhe, Germany). Samples were
scanned as potassium bromide disks at an instru-
ment resolution of 2 cm 1. One hundred scans
were recorded for each spectrum. The Raman
Five milliliters of saturated 40-Me in ethanol was
transferred to a 25-mL scintillation vial. The
solution was equilibrated at the tested tempera-
ture before a pair of polymorphs (1:1 w/w) was
added. The suspension was stirred continuously
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 90, NO. 3, MARCH 2001

CONFORMATIONAL COLOR POLYMORPHISM AND CONTROL OF CRYSTALLIZATION
375
until one form transformed to the other. After-
wards, the suspension was ®ltered, air dried, and
analyzed by XRPD. The slurry-conversion studies
were performed at different temperatures in 5ꢀC
intervals. The temperature was controlled by a
programmable temperature bath (model 9510,
Fischer Scienti®c, Pittsburgh, PA).
patterns of the four modi®cations are distinctive
in the 14.5±17.5 ꢀ2y region, this range was used to
monitor the solid-state transformation of LR, O,
and DR to R in the VT XRPD studies.
Thermal Analysis
Figure 4 shows the DSC curves of the four poly-
morphs reported and the values are listed Table 1.
The DSC curve of R shows a single endothermic
peak at 127ꢀC, corresponding to the melting point
of R. At a heating rate of 5 K Á min 1, the DR, LR,
and O forms show small endothermic peaks at
ꢁ 100, 110, and 115ꢀC, respectively, followed by
an endothermic peak at 127ꢀC. This result
indicates that these forms undergo an endother-
mic solid-state transformation to R on heating.
As shown in Figure 5 for the DR form, the
endothermic transformation peak is shifted to
lower temperatures with decreasing heating rate.
This behavior is characteristic of solid-solid trans-
formations and shows the kinetic control of the
reaction. Similar behavior was observed for the
other forms. When observed under the hot-stage
microscope at a heating rat of 1 K Á min 1, the
majority of DR transforms to R above 100ꢀC and
then completely melts at 127ꢀC. LR and O behave
in the same way. However, a small portion of the
samples of the DR, LR, or O forms do not exhibit a
solid±solid transformation to R on heating but
melt and recrystallize to R or stay liquid up to the
melting temperature of R. The presence of such
crystals, which do not contain or form nuclei of R
until the melting temperature of R is attained,
made it possible to determine the ``equilibrium
melting temperature'' of each individual form
(Table 1). This also indicates a low nucleation
rate of R in the supercooled melt, which allowed
the investigation of the solid±liquid equilibrium
of a lower melting form without recrystallization
to the higher melting form (R). The melting tem-
peratures and transition enthalpies of DR, LR,
and O are extremely close to each other, which is
consistent with the general observation that the
free energy differences between conformational
polymorphs are often very small.7
Density Measurements
The true density of 40-Me was measured with a
AccuPyc 1330 (Micromeritics Instruments Corp.,
Norcross, GA) helium pycnometer using a 12 cm3
cell and 2±4 g of materials for each determina-
tion. The reported densities are based on at least
three replicate measurements.
Crystallization
A known amount of 40-Me was dissolved in 400 mL
of ethanol and ®ltered through a 0.45-mm Milli-
pore ®lter membrane (Varian, Nylon-66, Walnut
Creek, CA). The ®ltrate was added to a 1000-mL
crystallization ¯ask. The ¯ask temperature was
controlled as already described in the Preparation
of Polymorphs section. At the beginning of the
experiment, the temperature was maintained at
70ꢀC for 15 min (well above the saturation point of
40-Me) to ensure that all possible seeds were
dissolved. The temperature was then lowered
rapidly to 5ꢀC above the saturation point and
cooled at a variety of rates to 20ꢀC. Seeds of 1% of
the total 40-Me used (w/w) were added to the
solution 0.5ꢀC above the saturation point. The
stirring rate was controlled at 200 rpm with a
Lightnin mixer (lightnin Inc., Chargrin Falls,
OH) with an 8-cm impeller (A100) and the convex
face of the blades up. Ten minutes after the bath
temperature reached 20ꢀC, the crystals were ®l-
tered and air-dried.
RESULTS
The color differences of the four forms are shown
in Figure 2. The results of various studies of the
color polymorphs of 40-Me are discussed next.
Fourier-Transform Infrared (FT-IR) and Fourier-
Transform Raman Spectroscopy (FT-Raman)
X-Ray Powder Diffractometry (XRPD)
Figure 3 shows the XRPD patterns for the four
modi®cations reported. Each of the XRPD pat-
terns is unique, although the diffractograms for O
and LR are strikingly similar. Because XRPD
The IR and Raman spectra are shown in Figures 6
and 7, respectively, and selected vibrational
frequencies are summarized in Table 2. Absor-
ption differences between the spectra of the
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 90, NO. 3, MARCH 2001

376
HE ET AL.
Figure 2. Photographs of the four 40-Me polymorphs.
modi®cations can be observed in the entire spec-
tral region. The differences between the crystal
lattices of the individual forms are clearly ex-
pressed by the differences of the lattice vibrations
below 400 nm (Raman). Like the X-ray powder
patterns, the IR and Raman spectra also show the
structural similarity of LR and O. The NÐH
stretching frequency ꢂnNH† is shifted to lower
wavenumbers in the order O, LR, R, and DR (i.e.,
with increasing color depth). The nNH frequencies
of O, LR, and R (Table 2) are in the range of the six
known forms of ROY,8;9 which is 3307 to 3284
cm 1. The nNH of DR is distinctly lower (3251
cm 1), which indicates a stronger intramolecular
hydrogen bond between the amino and the nitro
group. The nCN frequency decreases by 15 cm 1 in
the order R, O, LR, and DR. In the structurally
similar compound ROY, there is an interesting
9
relationship between polymorph color and nCN
.
As the color changes from red to orange to yellow,
nCN increases. This change is because the degree
of p conjugation between the amino and nitrile
groups decreases from the more planar red form
to the ``perpendicular'' yellow form. Thus, the
electron-donor perturbation decreases and nCN
increases from the red to the yellow form. Being
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 90, NO. 3, MARCH 2001

CONFORMATIONAL COLOR POLYMORPHISM AND CONTROL OF CRYSTALLIZATION
377
Figure 3. X-Ray powder diffraction patterns of the four 40-Me polymorphs.
1
Figure 4. DSC curves of the four 40-Me crystal forms at a heating rate of 5 K Á min
.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 90, NO. 3, MARCH 2001

378
HE ET AL.
Table 1. Physicochemical Data of the Four 40-Me Polymorphs
Parameter
R
DR
LR
Needles
O
Morphology
Plates
127
Plates
115
Needles
a
TfUSꢂꢀC†
113
115.5
ÁHfus (kJ/mol)b
24:57 Æ 0:13
ÁHtrs (to R) (kJ/mol)c
3:77 Æ 0:1
2:97 Æ 0:04
3:53 Æ 0:14
d
TtrsꢂꢀC†
52 Æ 2 (to DR)
57 Æ 3 (to O)
1:3803 Æ 0:0004
5:56 Æ 0:02
34.75
15 to 15 (to LR)
Density (g/cc)
1:3909 Æ 0:0014
1:4008 Æ 0:0003
1:3854 Æ 0:0005
4:88 Æ 0:007
36.47
Solubility at 25ꢀC (mg/mL)
ÁHsol (kJ/mol)e
4:99 Æ 0:02
5:14 Æ 0:08
37.49
2
38.02
3
Order of stability at 25ꢀC
4
1 (stable)
aMelting point.
bHeat of fusion (obtained from DSC).
cHeat of transition to R (obtained from DSC investigations).
dTransition temperature to R (obtained from slurry conversion data).
eHeat of solution (calculated from solubility data).
structurally similar, 40-Me may have a compar-
able relationship among the four modi®cations.
The torsional angle between the nitrophenyl ring
and the thiophene ring of DR is ꢁ 8ꢀ,9;10 which is
smaller than that of the ROY red form (24ꢀ).8;9
Based on the conformation±color relationship
observed with ROY, we can assume that the other
modi®cations of 40-Me adopt conformations with
larger angles between the planes of the nitrophe-
nyl and thiophene rings.
DR has the lowest nitrile stretching frequency,
whereas R, LR, and O have very similar nCN. If the
relationship between color and nCN still holds for
40-Me, the angular differences between the nitro-
phenyl ring and the thiophene ring should be
small for R, LR, and O. Because the crystal struc-
Figure 5. DSC curves of DR at different heating rates.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 90, NO. 3, MARCH 2001

CONFORMATIONAL COLOR POLYMORPHISM AND CONTROL OF CRYSTALLIZATION
379
Figure 6. IR spectra of the four 40-Me polymorphs.
Figure 7. Raman spectra of the four 40-Me polymorphs.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 90, NO. 3, MARCH 2001

380
HE ET AL.
1
Table 2. Selected Vibrational Frequencies of 40-Me Polymorphs (cm
)
Group
R
DR
LR
O
nNHa(cm
nCNab(cm
)
)
3282
2219
1626/1578
1531
3251
2204
1625/1587
1545
3291
2214
1629/1579
1537
3293
2217
1630/1579
1534
1
1
Benzene ringa(cm
)
1
Thiopheneb(cm
)
l
1
nasNO2a(cm
nsNO2a(cm
)
1519
1249
1521
1253
1531
1252
1530
1251
1
)
aIR spectra.
bRaman spectra.
tures of R, O, and LR are not known, further
correlation concerning the conformational diffe-
rences based on the IR and Raman data are
speculative.
the regression coef®cients listed in Table 4. All
forms have very similar solubility in the investi-
gated temperature range. Particularly, the solu-
bilities of DR, LR, and O are almost identical at
ambient temperatures, indicating they have very
similar free energies in this temperature range.
These similarities agree well with the thermoana-
lyticalresults. Thefree-energydifferences between
polymorphic pairs can be calculated using eq. 2:12
Solubility Analysis
The results of the solubility experiments are
summarized in Table 3. When a linear van't Hoff
plot was used to analyze the experimental data, it
gave correlation coef®cients ꢂr2† > 0.99. The sim-
pli®ed van't Hoff equation used to ®t the data is
written as:11
ÁG1;2 ˆ RT ln ꢂCS2=CS1
†
ꢂ2†
where ÁG1;2 is the difference in the Gibbs free
energies of two forms, R is the gas constant in
kJ/lmol, T is the temperature in K, and CS1 and
CS2 are the solubilities of the forms. At the ther-
modynamic transition temperature ꢂTtrs†, the
solubility of two enantiotropically related forms
1
ln CS ˆ a ‡ b Á T
ꢂ1†
where CS is the solubility at temperature T in
mg/mL, T the temperature in K, and a and b are
Table 3. Solubility of R, DR, LR, and O in Ethanol at Different Temperatures
Temperature
Form
R
Parameter
10ꢀC
10.5ꢀC
20ꢀC
25ꢀC
35ꢀC
44.5ꢀC
a
Cs;m
Cs;c
2.68 (0.21)
2.64
Ð
Ð
Ð
4.47 (0.04)
5.56 (0.02)
8.59 (0.05)
8.77
13.35 (0.11)
13.16
b
4.38
2
5.56
0
Dc
1.5
2:1
1.4
a
DR
LR
O
Cs;m
Cs;m
Dc
Ð
Ð
Ð
2.35 (0.03)
2.33
3.88 (0.04)
3.89
4.99 (0.02)
5.03
8.05 (0.05)
8.22
12.94 (0.15)
12.73
b
0.8
0:3
0:8
2:1
1.6
a
Cs;m
Cs;c
Dc
Ð
Ð
Ð
2.35 (0.02)
2.34
3.98 (0.04)
3.95
5.14 (0.08)
5.13
8.23 (0.04)
8.44
13.37 (0.07)
13.16
b
0.4
0.8
0.6
2:5
1.6
a
Cs;m
Cs;c
Dc
2.33 (0.02)
2.28
Ð
Ð
Ð
3.85 (0.04)
3.88
4.88 (0.01)
7.75 (0.06)
8.04
12.72 (0.04)
12.32
b
4.98
2
2.1
0:7
3:7
3:1
aMeasured values in mg/mL ÆSD.
bValues (in mg/mL) calculated with eq. 1 and the regression coef®cients in Table 4.
cDifference of measured and calculated solubility values in %.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 90, NO. 3, MARCH 2001

CONFORMATIONAL COLOR POLYMORPHISM AND CONTROL OF CRYSTALLIZATION
381
Table 4. Regression Coef®cients for the Solubility
Curves of R, DR, LR, and O Calculated from Equation 1
even <0.1 kJ/mol, which is within the experi-
mental error in the low-temperature range.
Because the transition temperature is de®ned
as the crossing point in the free energy curves of
the two forms, the range of transition tempera-
ture can be read directly from a free energy plot
similar to Figure 8, but plotted with triplicate
results for each form. As listed in Table 5, the
transition temperature between DR and LR is
Form
a
b
r2
R
15.739
16.746
16.979
16.328
4:179
4:509
4:573
4:387
0.998
0.999
0.999
0.998
DR
LR
O
calculated to lie in the range between
17 and
24ꢀC, and that of LR and O is in the range
between 25 and 17ꢀC. The transition tempera-
ture between DR and O is even less accurate,
being 87 to 24ꢀC.
The heat of solution ꢂÁHsol† given by eq. 3 can
be estimated from the slope of the linear van't
Hoff plot:
ÁHsol
ˆ
R Á b
ꢂ3†
where R is the gas constant and b is the slope
listed in Table 4. The heat of solution of each form
is listed in Table 1. The heats of transition
between the different forms can be estimated by
the differences in their heats of solution. The
calculated differences for the transitions DR to R
(2.74 kJ/mol) and O to R (1.72 kJ/mol) are lower
than the heats of transition measured by DSC
(Table 1), but show the same order. However,
the calculated value for the transition LR to R
(3.27 kJ/mol) is inconsistent with the correspond-
ing heat of transition measured by DSC, which
will be discussed later.
The solubility data was analyzed according to
a nonlinear regression analysis described else-
where.13;14 Because the solubility±temperature
functions of all forms provide good linear plots,
the results of the nonlinear regression analysis
are not signi®cantly different and thus are not
discussed further.
Figure 8. Relatlve energy±temperature diagram of
the four 40-Me polymorphs (relative to O).
are equal. Thus, their free energy difference at
this temperature is zero (i.e., these two forms are
in equilibrium with each other). Consequently,
from the solubility data it is possible to calcu-
late the thermodynamic transition temperatures
between the forms. Triplicate studies were car-
ried out at each temperature for each form, and
the range of transition temperature is calculated
from the triplicate solubility data. However, be-
cause DR, LR, and O have very similar solubi-
lities, the predicted transition temperatures are
not very accurate. As shown in Figure 8, the free
energy difference between any of the forms at
20ꢀC is <0.4 kJ/mol. The free energy difference
between DR, LR, and O at this temperature is
Table 5. Transition Temperatures Calculated from the Solubility Data and Slurry Conversion
ꢀ
ꢀ
a
b
Pair of forms
Ttrs ꢂ C†
Ttrs ꢂ C†
R/DR
R/LR
R/O
DR/LR
DR/O
LR/O
51 to 61
42 to 48
64 to 81
17 to 24
87 to 24
25 to 17
50 to 54
Not determined because LR and R transformed to O directly
54 to 60
15 to 15 (based on visual observation and XRPD)
unknown
unknown
aTransition temperature, Ttrs, from solubility data.
bTransition temperature, Ttrs, from slurry conversion studies.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 90, NO. 3, MARCH 2001

382
HE ET AL.
Figure 9. X-Ray powder patterns of slurry conversion experiments in ethanol with (a)
pure DR, (b) mixture of DR and R (1:1 w/w) after stirring at 50ꢀC for 2 d, (c) pure R, and
(d) mixture of DR and R (1:1 w/w) after stirring at 54ꢀC for 2 d.
Slurry Conversion Studies
at this temperature. Alternatively, after equili-
bration for 48 h at 54ꢀC, the XRPD of the DR:R
(1:1) sample resembles that of R. The results
indicate that the transition temperature between
DR and R is between 50 and 54ꢀC.
In addition to the solubility investigations, slurry-
conversion studies in different temperature
regions were performed to check the thermody-
namic stability of the individual forms. The
solution-mediated polymorphic transformation
was observed to be fast for 50/50 mixtures of the
40-Me polymorphs except for the LR/O and LR/DR
pairs. Thus, it was possible to estimate the tran-
sition temperature range between three out of six
pairs of polymorphs (R/DR, R/O, DR/LR). The
results of the slurry conversion studies are sum-
marized in Table 5.
Similarly, the transition temperature between
R and O is between 54 and 60ꢀC. When a mixture
of LR and R is slurried at 50ꢀC or above, LR
converts to R. However, at a temperature of 45ꢀC
or below, the mixture directly converts to O.
The transformation between LR and DR could
not be veri®ed by XRPD alone. DR did not
completely transform to LR at low temperature
(
15ꢀC) within 15 d, although peaks of DR
The preliminary temperature range in the
slurry conversion studies was decided from the
transition temperature predicted by solubility
data. If the transition temperature could not be
located in the preliminary range, the range was
expanded to locate the transition temperature.
Figure 9 shows the XRPD diffraction pattern of
the solids from a DR/R (1:1) slurry collected after
48 h at each temperature. After equilibration for
48 h at 50ꢀC, the XRPD of the sample is identical
to DR, indicating that R completely converts to DR
diminish with respect to the original mixture. In
addition, the color of the mixture becomes lighter
with time, indicating DR may transform to LR
during the conversion.
The slurry experiments illustrate the potential
application and limitations of simple slurry-
conversion experiments to the investigation of
the thermodynamic relationship between poly-
morphs. For example, the experiments helped to
verify and narrow the transition temperature
range for the R/DR and R/O enantiotropic pairs.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 90, NO. 3, MARCH 2001

CONFORMATIONAL COLOR POLYMORPHISM AND CONTROL OF CRYSTALLIZATION
383
The slurry experiments were less helpful in nar-
rowing the transition temperature range for the
DR and LR polymorphs, but helped verify their
thermodynamic relationship. Unfortunately, we
were not able to determine the transition tem-
peratures for the DR/O, R/LR, and LR/O pairs
of polymorphs, either because of the slow trans-
formation kinetics or because the mixture con-
verted to another more stable form directly.
bonding is not likely to be a dominant force in the
crystal structure; thus, 40-Me, can be expected to
follow the density rule. The density values of the
individual forms (Table 1) are very close to each
other (the relative density differences are within
1.5%). However, as discussed later, density is very
helpful and important information for the deter-
mination of the relative thermodynamic stability
of the forms.
Density Measurements
Variable-Temperature X-Ray
Diffractometry (VT X-Ray)
According to Kitaigorodsky's close packing prin-
ciple15 and the density rule,16 a form with a
higher density is more stable at absolute zero.
Thus, thermodynamic stability information can
be deduced from the density of each form. Based
on the chemical structure of 40-Me, hydrogen
Variable-temperature X-ray diffractometry was
used to study the high-temperature transforma-
tion processes of DR, LR, and O to R. The VT
X-ray results for DR are shown in Figure 10a.
Prior to the melting point of DR (115ꢀC), DR has
Figure 10. Series of variable temperature X-ray powder diffraction pattern of three
40-Me polymorphs.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 90, NO. 3, MARCH 2001

384
HE ET AL.
completely transformed to R. As mentioned ear-
lier, a temperature gradient exists between the
surface and the bottom of the sample, and the
surface temperature is 3ꢀC lower than that of
the bottom. Thus, the transformation of DR is
complete at 112ꢀC. Shown in Figure 10b, there is
still LR present in the sample prior to the melting
point of LR (113ꢀC). This result indicates that the
solid-state transformation of LR to R is slower
than that of DR to R. The VT X-ray patterns for O
(Figure 10c) indicate that a small amount of O
remains in the heated sample even at a tempera-
ture of 125ꢀC, which is much higher than the
melting point of O (115.5ꢀC). At the end of the
experiment, a small amount of O is visually
observed on the surface of the sample bulk. When
O is held at 125ꢀC for 1 min, the peaks from O
diminish with time, whereas, the peak from R
increases; when O is held at 125ꢀC for 10 min, the
peaks from O still do not disappear completely.
This behavior agrees with the fact that a distinct
temperature gradient exists from the bottom of
the sample holder to the top of the sample bulk. In
addition, when the tiny needles of O transform to
R, the small particles may form a dense barrier
that hinders the heat transfer from the bottom to
the surface of the sample. The VT X-ray experi-
ments verify the DSC results with regards to
the order of the transition temperatures of the
different forms (DR<LR<O). In addition, the VT
X-ray results con®rm the transformation of DR,
LR, and O to pure R on heating.
Figure 11. Schematic energy/temperature diagram
of 40-Me modications. Key: (L) liquid phase (melt; (H)
enthalpy; (G) Gibbs free energy; ꢂÁtrsH† heat of transi-
tion; ꢂÁfusH† heat of fusion; ꢂTtrs† transition point; ꢂTfus
†
melting point.
tion rule for the solid±solid transition endotherms
observed in the DSC thermograms for each form.16
In addition, the higher transition enthalpy of DR
compared with O indicates that the H isobar of O
is above that of DR and thus DR is more stable
than O at absolute zero (consistent with the
density rule). Because DR has the lower melting
point, the G isobars of DR and O must intersect
(consistent with enantiotropism).
DISCUSSION
The complex relative thermodynamic relation-
ships between the different forms are visualized
by the schematic energy±temperature diagram,17
shown in Figure 11. This diagram can be derived
from thermal analysis and solubility and density
data, as well as the slurry conversion studies
based on the following considerations. Because the
melting points are in the order LR<DR<O ꢃ R
and in inverse order to the densities (LR > DR >
O > R), it follows that the G isobars of all forms
intersect. In other words, all forms are enantio-
tropically related, with R being the most stable
form at high temperature. Consequently, there
are six transition points between the polymorphic
pairs (LR/DR, LR/O, LR/R, DR/O, DR/R, and O/R)
in this polymorphic system.
Solubility studies and slurry-conversion stu-
dies also clearly demonstrate that there is an
enantiotropic relationship between O, DR, and
R. The thermodynamic transition temperatures
ꢂTtrs† of O/R and DR/R have been veri®ed to be
57 Æ 3 and 52 Æ 2ꢀC, respectively, which matches
the predicted order in the diagram. The Ttrs of
O/DR lies close to or below room temperature. As
already stated, the Ttrs could not be determined
more precisely for this polymorphic pair because
of the slow transition kinetics. However, the
results clearly demonstrate that O is the most
stable form between RT and 54ꢀC and R is stable
above 60ꢀC. DR is thermodynamically stable
within a small temperature window of 10 to 20ꢀC,
just below room temperature, but metastable at
all other temperatures. Although all data show
very small and not very signi®cant differences,
The enantiotropic relationship between R and
the other forms is con®rmed by the heat of transi-
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 90, NO. 3, MARCH 2001

CONFORMATIONAL COLOR POLYMORPHISM AND CONTROL OF CRYSTALLIZATION
385
they are consistent in terms of the enantiotropic
relationships between DR, O, and R.
stances in the pharmaceutical industry. It is
undesirable to have such uncontrolled crystalliza-
tion because polymorphs may differ in their
physicochemical properties and bioavailability.
Thus, the ability to control the crystallization of
polymorphs is extremely important in the phar-
maceutical industry.
This is not the case for LR, where the enthalpy
of transition is contradictory to the measured
density and the solubility data. Because the LR
was found to have the lowest heat of transition as
measured by DSC, monotropism between LR/DR
and LR/O is expected. However, LR shows the
highest density, the highest heat of solution, and
from the solubility curves, the lowest solubility at
low temperatures. These results suggest that LR
is the most stable form at absolute zero. Also,
because the melting point is lower than that of DR
and O, the G isobars of LR/DR and LR/O must
intersect at higher temperatures. Unfortunately,
the interconversion studies are not very illumi-
nating because of kinetic reasons and very close
transition temperatures. Because a peak splitting
of the melting endotherm of R in the DSC curves
is sometimes observed, we also supposed the
existence of a ®fth form that might be responsible
for the discussed inconsistencies. The DSC curves
in Figure 4 show that the melting peak of pure R
is broader than that of the corresponding melting
endotherms of the other forms. Additionally, the
peak maximum is slightly shifted to higher tem-
peratures. However, because the heats of fusion of
the melting endotherms are not different from
that of pure R or the melting endotherm of O
and DR after transformation, no direct evidence
to support the existence of another ``state'' is
observed. Also, the VT X-ray studies did not
indicate a new form. Thus, conclusive evidence for
the thermodynamic stability of LR and an expla-
nation of the inconsistency between DSC and
the solubility data cannot be given at this time.
Solid±solid transformations are an interesting but
poorly understood area18 and additional studies of
the transformation kinetics of 40-Me are planned.
In an industrial setting, seeding is most com-
monly used to control the crystallization of poly-
morphs. However, seeding is not always reliable
and does not always work, at least not with 40-Me.
For example, when a solution is seeded with R at
high temperature, R grows. However, when a
solution is seeded with R at room temperature, LR
grows. Seeding the solution with O always results
in other forms, mostly LR. Thus, seeding alone is
not suf®cient to control the crystallization of all
four forms. Two reasons might account for the
failure of seeding: one is kinetic, the other is
thermodynamic. When a solution is seeded with a
desired form, the nucleation barrier for that form
is by passed. However, whether the desired form
will grow depends on the crystal growth rate of
that form relative to the other forms and whether
solution-mediated transformation is an issue. The
relative crystal growth rate is a kinetic aspect,
whereas the solution-mediated transformation is
both thermodynamic and kinetic. For example,
DR is thermodynamically unstable at 60ꢀC; seed-
ing the solution with DR at 60ꢀC results in the
most stable form R, probably because of a fast
solution-mediated transformation (Table 7). On
the other hand, seeding the solution with the most
stable form may not always result in the most
stable form. For example, when the solution is
seeded with O (the most stable form) at room
temperature, LR always grows because the crystal
growth rate of O is much slower than that of
LR, which is coincident with the Ostwald's step
rule. Thus, it is important to understand both
the thermodynamic and kinetic effects to control
the crystallization of polymorphs. In the follow-
ing paragraphs, 40-Me will be used as a model
system to demonstrate how to control such a
complicated polymorphic system with a thermo-
dynamic approach.
Crystallization Studies
The crystallization of 40-Me is generally out of
control. Under similar conditions, mixtures of
forms or pure forms could crystallize out of solu-
tion. For example, in a 40-Me ethanol solution
(25 mg/mL) placed at room temperature, LR crys-
tallizes out of solution in many cases, DR crystal-
lizes out in some cases, and mixtures of LR and R
crystallize out in a few cases. If the solution is
placed at room temperature for a longer period of
time, all four forms are found in the crystal-
lization vessel. This variability might also be
present in the crystallization process of drug sub-
In this study, an energy±temperature diagram
is used as a key for the estimation of successful
crystallization parameters for a certain poly-
morph. According to the diagram (Figure 11), R
is the most stable form above 60ꢀC. From the
results of the solubility studies (Figure 8), we
found that the free energy difference between R
and the other three form is <0.4 kJ/mol at 10ꢀC.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 90, NO. 3, MARCH 2001

386
HE ET AL.
Table 6. Crystallization Outcome of R, DR, and LR
(stirring rate ˆ 200 rpm)
Table 7. Seeding without Knowing Thermodynamic
Relationship (stirring rate ˆ 200 rpm)
Conditions
R
DR
LR
TaꢂꢀC†
TaꢂꢀC†
63
20
20
0.33
1.5
Parameter
63
30
1
CRb ꢂK Á min
†
0.33
1.5
1
0.33
1.5
1
1
sc
CRb ꢂK Á min
sc
†
0.33
1.5
0.33
1.5
Seeding (%)
1% or no seeds
Seeding (%)
Final forms
1% DR
R
1% R
LR
aT, crystallization temperature.
bCR, cooling rate.
cs, ratio between initial concentration and the solubility at
aCrystallization temperature.
the crystallization temperature.
bCR, cooling rate.
cs, ratio between initial concentration and the solubility at
the crystallization temperature.
This energy difference becomes less and less with
increasing temperature. Thus, it is easy to control
the crystallization of R above 60ꢀC. As for LR, DR,
and O, the energy difference among them is very
small (<0.1 KJ/mol) from 10 to 40ꢀC. From the
close transition temperatures (Figure 11) and the
small free energy differences, we can assume that
the transition kinetics of one form to a more stable
form is slow in this temperature range (almost at
equilibrium). Indeed, as observed in the slurry
conversion studies, the transformation kinetics is
slow between LR/DR, LR/O, and DR/O. In the
preliminary studies, both DR and LR could grow
under similar conditions, indicating that the
growth rates of these two forms are relatively
close. Thus, it is very likely, in this case, that the
crystallization of LR and DR in the ambient tem-
perature region can be successfully controlled just
by seeding.
The results of the crystallization experiments
agree with the derived thermodynamic picture of
the system (Table 6). R can be crystallized above
60ꢀC with 0.5% R seeds. LR can also be crystal-
lized repeatedly at 20ꢀC by seeding. Being a
more stable form at room temperature, O can be
obtained by slurry conversion of R or DR. The
crystallization of DR was successfully controlled
at 20ꢀC by seeding with as little as 1% DR.
Table 7 again indicates the importance of
energy±temperature diagrams. Without knowing
the best temperature region, seeding alone does
not produce the desired form. For example, seed-
ing DR at high temperature results in the most
stable form R at this temperature; seeding R at
room temperature results in the fastest growing
metastable form LR.
transformation renders R undesirable. LR and O
crystallize into needles. The needle morphology
is generally undesirable because it makes the
material hard to ®lter and reduces the ¯owability.
DR is obtained as chunky crystals that are excel-
lent for processing and manufacturing. Although
metastable, the free energy difference between
DR and the most stable form is very small and the
transformation of DR to more stable forms is slow
in both the solid and slurry states. Therefore, DR
is the best form to formulate. Once the appro-
priate form is identi®ed, it is important to opti-
mize the crystallization conditions so that the
form can be produced consistently and ef®ciently.
Various crystallization condition were tried
to obtain the best quality DR crystals. These
conditions included two different cooling rates
1
(0:17 K Á min or 0:33 K Á min 1) and ®ve differ-
ent supersaturations S (1.5, 1.75, 2, 3, or 4). The
term S is de®ned as the ratio between the initial
concentration and the solubility of the crystal-
lizing temperature T. The data in Table 8 show
that the cooling rate does not have a signi®cant
impact on the yield, whereas the yield increases
signi®cantly as S increases. Pure DR (100% yield)
was obtained with S ˆ 4. However, crystals began
to have more defects on the surface with higher
supersaturation ꢂS ꢄ 3†. The bottom line is that
the desirable form DR can be produced consis-
tently with a high yield.
CONCLUSIONS
If 40-Me was a drug to be marketed, the next
step would be to choose a desirable polymorph to
formulate. R is the least stable form at room
temperature. Although R is kinetically stable in
the solid state, the very fast solution-mediated
5-Methyl-2-[(4-methyl-2-nitrophenyl)amino]-3-
thiophenecarbonitrile is an example of a com-
pound with conformational color polymorphism; it
crystallizes in red (R), dark red (DR), light red
(LR), and orange (O) forms. Although only the
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 90, NO. 3, MARCH 2001

CONFORMATIONAL COLOR POLYMORPHISM AND CONTROL OF CRYSTALLIZATION
387
Table 8. Crystallization Outcome of DR a Different Conditions (stirring rate ˆ 200 rpm at crystallization
temperature 20ꢀC)
Supersaturationa
Condition
1.75
2
3
4
Seeding temperature (ꢀC)b
Final form
31.5
DR
34.5
DR
43.5
DR
50.0
DR
30 Æ 2:5%
37:2 Æ 0:3%
ꢂCR ˆ 0:33†
35:0 Æ 1%
ꢂCR ˆ 0:17†
58:8 Æ 0:1%
72:2 Æ 0%
d
ꢂCR ˆ 0:33†
ꢂCR ˆ 0:33†
ꢂCR ˆ 0:33†
Yield (%)c
33:5 Æ 4:1%
ꢂCR ˆ 0:17†
aRatio between initial concentration and solubility of the crystallizing temperature, T.
bTemperature reported is the temperature for the crystallization ¯ask.
cThe yield as the weight of crystal product divided by the weight before the crystallization, multiplied by 100.
dCR, cooling rate.
crystal structure of DR is known, we can assume
that the color differences arise from the dihedral
angle between the two aromatic rings. This
assumption can be derived from previous work
with structural derivatives.8,9 Because of very
small energetic differences among the forms and
several thermodynamic transition temperatures
close to room temperature, the characterization
of the system is complex and requires the appli-
cation of a variety of methods. Furthermore, this
study demonstrates the importance of energy±
temperature diagrams in understanding complex
polymorphic systems. One goal of a polymorphism
study should be to combine various experimental
data (i.e., solubilities, densities, melting and
transition points, heats of transition, and heats
of fusion) to form a consistent thermodynamic
picture that can be expressed by a schematic
energy±temperature diagram. Based on the dia-
gram, important practical consequences can be
derived and predicted for polymorphic compounds.
It should be stressed that an energy±temperature
diagram only shows the thermodynamic relation-
ship between the different forms and does not
provide kinetic information. However, knowledge
of the thermodynamic aspects is the prerequisite
to kinetic control of any reaction. These aspects
have been demonstrated in this study by the
control of the crystallization of the individual
crystal forms, particularly the DR form, which is
a metastable modi®cation throughout the entire
temperature range.
ing research funding for this project. Ulrich
J. Griesser gratefully acknowledges an Erwin-
SchroÈdinger grant from the Austrian Science
Fund (FWF).
REFERENCES
1. Threlfall TL. 1995. Analysis of organic polymorphs
± A review. Analyst 120:2435±2460.
2. Byrn SR, Pfeiffer RR. Stowell JG. 1995. Poly-
morphs. In: Solid-state chemistry of drugs, 2nd
edition. West Lafayette, IN: SSCI, p. 139.
3. Haleblian J, McCrone W. 1969. Pharmaceutical
applications of polymorphs. J Pharm Sci 58:911±
929.
4. Joiris E, Martino PD, Berneron C, Guyot-Hermann
AM, Guyot JC. 1998. Compression behavior of
orthorhombic paracetamol. Pharm Res 15:1122±
1130.
5. Kachi S, Terada M, Hashimoto H. 1998. Effects of
amorphous and polymorphs of PF1022A, a new
antinematode drug, on angiost Rongylus Costar-
icensis in mice. Jpn J Pharmacol 77:235±245.
6. Shah JC, Chen JR, Chow D. 1999. Metastable
polymorph of etoposide with higher dissolution
rate. Drug Dev Ind Pharm 25:63±67.
7. Berstein J. 1987. Conformational polymorphism.
In: Organic solid state chemistry, Desiraju GR,
editor. Amsterdam: Elsevier pp. 471±518.
8. Stephenson GA, Borchardt TB, Byrn SR, Bowyer J,
Bunnell CA, Snorek SV, Yu L. 1995. Confor-
mational and color polymorphism of 5-methyl-2-
((2-nitrophenyl)amino)-3-thiophenecarbonitrile. J
Pharm Sci 84:1385±1386.
9. Yu L, Stephenson GA, Mitchell CA, Bunnell CA,
Snorek SV, Bowyer JJ, Borchardt TB, Stowell JG,
Byrn SR. 2000. Thermochemistry and confor-
mational polymorphism of a hexamorphic crystal
system. J Am Chem Soc 122 (4):585±591.
ACKNOWLEDGMENTS
Xiaorong He gratefully acknowledges the Purdue-
Wisconsin Water Mobility Consortium for provid-
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 90, NO. 3, MARCH 2001

388
HE ET AL.
10. Borchardt TB. 1997. The derivatization, solid-state
characterization, and crystallization of a pharma-
ceutical precursor that expresses color polymorph-
ism in the solid-state. Ph.D. Dissertation. Purdue
University, pp. 115±122.
namic data of a polymorphic drug substance
[Sulfanilamide]. Sci Pharm 41:303±314.
15. Kitaigorodsky AI. 1973. Structure of crystals. In:
Molecular crystals and molecules. New York:
Academic pp. 1±33.
11. Burger A. 1975. Dissolution and polymorphism of
metolazone. Drug Res 25:24±27.
12. Grant DJW, Brittain HG. 1995. Solubility of phar-
maceutical solids. In: Physical characterization of
pharmaceutical solids. New York: Marcel Dekker
pp. 322±326.
13. Grant DJW, Mehdizadeh M, Chow AHL, Fair-
brother JE. 1984. Nonlinear van't Hoff solubility-
temperature plots and their pharmaceutical impli-
cations. Int J Pharm 18:25±38.
14. Burger A. 1973. LoÈslichkeitsuntersuchungen zur
Ermittlung thermodynamischer Daten eines poly-
morphen Arzneimittels (Sulfanilamid). Solubility
investigations for the determination of thermody-
16. Burger A, Ramberger R. 1979. On the polymor-
phism of pharmaceuticals and other molecular
crystals. I. Theory and thermodynamic rules.
Mikrochim Acta II 259±271.
17. Grunenberg A, Henck JO, Siesler HW. 1996.
Theoretical derivation and practical application of
energy/temperature diagrams as an instrument in
preformulation studies of polymorphic drug sub-
stances. Int J Pharm 129:147±158.
18. Desiraju GR, Paul IC, Curtin DY. 1977. Conversion
in the solid state of the yellow to the red form of 2-
(40-methoxyphenyl)-1,2-benzoquinone. X-ray crys-
tal structural and anisotropy of the rearrangement.
J Am Chem Soc 99:1594±1601.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 90, NO. 3, MARCH 2001