Conformational, concomitant polymorphs of 4,4-diphenyl-2,5- cyclohexadienone: Conformation and lattice energy compensation in the kinetic and thermodynamic forms

Article abstract of DOI:10.1002/chem.200501417

4,4-Diphenyl-2,5-cyclohexadienone (1) crystallized as four conformational polymorphs and a record number of 19 crystallographically independent molecules have been characterized by low-temperature X-ray diffraction: form A (P2 ₁, Z′=1), form B (P1, Z′ = 4), form C (P1, Z′=12), and form D (Phca, Z′ = 2). We have now confirmed by variable-temperature powder X-ray diffraction that form A is the thermodynamic polymorph and B is the kinetic form of the enantiotropic system A-D. Differences in the packing of the molecules in these polymorphs result from different acidic C-H donors approaching the C=O acceptor in C-H...O chains and in synthons I-III, depending on the molecular con formation. The strength of the C-H...O interaction in a particular structure correlates with the number of symmetry-independent conformations (Z′) in that polymorph, that is, a short C-H...O interaction leads to a high Z′ value. Molecular conformation (E_conf) and lattice energy (U_latt) contributions compensate each other in crystal structures A, B, and D resulting in very similar total energies: E_total of the stable form A = 1.22 kcal mol¹, the metastable form B = 1.49 kcal mol^-3, and form D = 1.98 kcal mol^-1. Disappeared polymorph C is postulated as a high-Z′ high-energy precursor of kinetic form B. Thermodynamic form A matches with the third lowest energy frame based on the value of U _latt determined in the crystal structure prediction (Cerius ², COMPASS) by full-body minimization. Re-ranking the calculated frames on consideration of both E_conf (Spartan 04) and U _latt energies gives a perfect match of frame #1 with stable structure A. Diphenylquinone 1 is an experimental benchmark used to validate accurate crystal structure energies of the kinetic and thermodynamic polymorphs separated by < 0.3 kcal mol^-1 (～1.3kJ mol^-1).

Full text of DOI:10.1002/chem.200501417

FULL PAPER
DOI: 10.1002/chem.200501417
Conformational, Concomitant Polymorphs of 4,4-Diphenyl-2,5-
cyclohexadienone: Conformation and Lattice Energy Compensation in the
Kinetic and Thermodynamic Forms
[
a]
[a]
[a]
[b]
Saikat Roy, Rahul Banerjee, Ashwini Nangia,* and Gert J. Kruger*
ꢀ
1
Abstract: 4,4-Diphenyl-2,5-cyclohexa-
dienone (1) crystallized as four confor-
mational polymorphs and a record
number of 19 crystallographically inde-
pendent molecules have been charac-
terized by low-temperature X-ray dif-
formation. The strength of the CꢀH···O
B=1.49 kcalmol , and form D=
ꢀ
1
interaction in a particular structure cor-
relates with the number of symmetry-
independent conformations (Z’) in that
polymorph, that is, a short CꢀH···O in-
1.98 kcalmol . Disappeared poly-
morph C is postulated as a high-Z’,
high-energy precursor of kinetic form
B. Thermodynamic form A matches
with the third lowest energy frame
based on the value of Ulatt determined
in the crystal structure prediction
teraction leads to a high Z’ value. Mo-
lecular conformation (Econf) and lattice
energy (Ulatt) contributions compensate
each other in crystal structures A, B,
and D resulting in very similar total en-
ergies: Etotal of the stable form A=
fraction: form A (P2 , Z’=1), form B
1
¯ ¯
P1, Z’=4), form C (P1, Z’=12), and
(
2
form D (Pbca, Z’=2). We have now
confirmed by variable-temperature
powder X-ray diffraction that form A
is the thermodynamic polymorph and
B is the kinetic form of the enantio-
tropic system A–D. Differences in the
packing of the molecules in these poly-
morphs result from different acidic Cꢀ
(Cerius , COMPASS) by full-body min-
imization. Re-ranking the calculated
frames on consideration of both Econf
(Spartan 04) and Ulatt energies gives a
perfect match of frame #1 with stable
structure A. Diphenylquinone 1 is an
experimental benchmark used to vali-
date accurate crystal structure energies
of the kinetic and thermodynamic poly-
ꢀ
1
1.22 kcalmol , the metastable form
Keywords: conformation analysis ·
crystal structure prediction
·
polymorphism supramolecular
·
H donors approaching the C=O accept-
or in CꢀH···O chains and in synthons
chemistry · symmetry-independent
molecule · X-ray diffraction
ꢀ
1
morphs separated by <0.3 kcalmol
ꢀ
1
I–III, depending on the molecular con-
(~1.3 kJmol ).
Introduction
ty of at least two different arrangements of the molecules of
[2]
that compound in the solid state” is widely accepted today.
The existence of polymorphism implies that free-energy dif-
[
1]
McCroneꢀs definition of a polymorph as “a solid crystal-
line phase of a given compound resulting from the possibili-
ꢀ
1
ferences between various forms are small (<3 kcalmol )
and that kinetic factors are important during crystal nuclea-
tion and growth. Molecular conformations, hydrogen bond-
ing, packing arrangements, and lattice energies of the same
molecule in different supramolecular environments may be
[
a] S. Roy, R. Banerjee, Prof. A. Nangia
School of Chemistry, University of Hyderabad
Hyderabad 500 046 (India)
Fax : (+91)40-2301-1338
E-mail: ashwini_nangia@rediffmail.com
[3]
compared in polymorphic structures. Polymorphs are ideal
systems to study molecular structure–crystal structure–crys-
tal energy relationships with a minimum number of varia-
bles because differences arise due to different intermolecu-
[
b] Prof. G. J. Kruger
Department of Chemistry, University of Johannesburg
PO Box 524, Auckland Park, Johannesburg 2006 (South Africa)
Fax : (+27)11-489-2819
[4]
lar interactions (supramolecular synthons) and crystal
packing effects and not because they are different chemical
species. There is increasing interest in understanding poly-
morphism, growing new crystal forms, controlling the selec-
tive growth of one form, transformations between poly-
E-mail: gjk@rau.ac.za
Supporting information for this article is available on the WWW
under http://www.chemeurj.org/ or from the author. Simulated
powder XRD plots of forms A–D, DSC of diphenylquinone 1, confor-
mational overlay diagram, Rietveld fit of Polymorph Predictor frame
with form A, and tables containing crystal structure prediction
frames, energies, and cell parameters.
[5]
morphs, and the high-throughput crystallization of drugs.
Polymorphism is more widespread in pharmaceutical
Chem. Eur. J. 2006, 12, 3777 – 3788
ꢁ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
3777

[6]
[7]
solids than the estimates of 4–5% polymorphic crystals
in the Cambridge Structural Database (CSD) suggest.
Table 1. Crystallographic data for polymorphs A–D of diphenylquinone
1.
[
13]
[8]
Form A
Form B
Form C
Form D
Among organic crystal structures, there is one example of a
compound with seven polymorphs (5-methyl-2-[(2-nitrophe-
[
a]
CSD refcode
space group
Z’, Z
a [Š]
b [Š]
c [Š]
a [8]
b [8]
HEYHUO HEYHUO01 HEYHUO02HEYHUO03
[9]
¯
¯
P2
1
P1
P1
Pbca
2, 16
10.7921(6)
17.4749(12)
27.9344(19)
90
nyl)amino]-3-thiophenecarbonitrile, common name ROY),
sulfathiazole has five forms, there are 14 clusters of tetra-
1, 2
7.9170(6)
8.4455(6)
10.3086(9) 16.2921(4)
90 88.2570(10)
105.758(2) 85.3380(10)
4, 8
10.0939(2)
16.2592(3)
12, 24
18.3788(4)
19.9701(4)
24.4423(5)
95.008(1)
111.688(1)
105.218(1)
7871.8(3)
0.1120.059
[9b,10a]
morphs, and over a 100 trimorphic systems.
Polymor-
phism is of great current interest because different solid-
state forms can have different physical, chemical, and func-
tional properties, for example, melting point, stability, color,
bioavailability, pharmacological activity, and a nonlinear op-
tical response.
90
90
g [8]
V [Š ]
90
83.6450(10)
2648.00(10)
0.068
3
663.36(9)
0.050
5268.2(6)
R factor
[11]
Concomitant polymorphs
crystallize simultaneously
[
a] See ref. [8].
from the same solvent and crystallization flask under identi-
[12]
2
2
cal crystal growth conditions. Bernstein carried out early
studies on conformational polymorphs (different conforma-
tions of the same molecule in different crystal structures)
and conformational isomorphs (different conformers of the
same molecule in the same crystal structure). Herein we de-
scribe a tetramorphic cluster of conformational polymorphs
in which molecular- and lattice-energy compensation results
in very small differences in the total energy of the concomi-
tant polymorphs A–D of 4,4-diphenyl-2,5-cyclohexadienone
R (16) ring through D and D molecules. The parameters
i
ii
of the CꢀH···O geometries are listed in the legend to
Figure 1. We have not found polymorphs of bis(biphenylyl)
[
16]
ketone 2 or its substituted phenyl derivatives 3 (4-Cl/Br/
[10b]
Me)
so far.
The number of symmetry-independent or crystallographi-
cally unique molecules/ions in a crystal lattice is Z’. Alterna-
tively, Z’ may be defined as the number of formula units (Z)
divided by the number of independent general positions for
that space group. Z’ is typically 1 or 0.5 in crystal structures
(87%). A high Z’ of 12for form C is a record for poly-
(
1). Experimental conditions are described for the prepara-
tion of thermodynamic form A and kinetic form B in a rea-
sonably pure state. The implication of a molecular and crys-
tal-packing balance is to rank crystal structure prediction
frames derived from metastable rotamers by consideration
of both conformation- and lattice-energy contributions,
which gives an excellent match of the global minimum struc-
ture with stable form A.
[9b]
[8]
morph clusters
(only five hits).
(Table 2) and as such rare in the CSD
Structures with high Z’ values continue to interest crystal-
lographers but it is still not properly understood why some
categories of structures exhibit a higher frequency of Z’>1.
Steed has critically reviewed the reasons for high-Z’ crys-
tal structures. 1) The molecule has a packing problem be-
cause of its awkward shape, which is reconciled by having
[17]
[18]
Results and Discussion
two or more molecules in different conformations. 2 ) The
molecules organize in stable clusters prior to reaching the
highest symmetry arrangement in strong OꢀH···O hydrogen-
Crystal packing and multiple Z’ in polymorphs: Crystallo-
graphic data for the polymorphs A–D of diphenylbenzoqui-
bonded structures because of the enthalpic advantage de-
[
13]
[19]
none 1 are listed in Table 1.
The molecule has several
rived from s-cooperative chains, for example, as in alco-
2
acidic, activated donor hydrogen atoms of sp - and phenyl
hols, phenols, steroids, nucleotides, and nucleosides. 3) Sev-
eral low-lying molecular conformations interconvert in solu-
tion and more than one molecule may crystallize simultane-
ously for kinetic reasons. The last of these situations occurs
in the conformational polymorphs of 1, which provides a
unique opportunity to study polymorphic structures with
multiple values of Z’.
CꢀH-type whereas there is a single carbonyl acceptor. De-
pending on the molecular conformation one or more of the
[14]
several possible CꢀH···O interactions are optimized in the
crystal structure (Figure 1). For example, form A has zigzag
chains of CꢀH···O interactions between screw-axis-related
[
15]
A molecules of graph-set notation C(8). Form B has Cꢀ
i
H···O quinone dimer synthon I
and p-phenyl CꢀH···O synthon
Cholesterol (Z’=16) is a prototype example of strong Oꢀ
H···O hydrogen bonds being associated with unusually high
2
2
2
2
II [graph set R (8) and R (20)]
Z’ values. An exceptional case in the weak hydrogen bond
between B , B and B , B mol-
category is the crystal structure [ReCl
A
T
E
N
i
ii
iii
iv
2
[
20]
ecules, respectively. Form C has
the same synthons and overall
packing as B but has 12mole-
cules (C –C ) in the asymmetric
A
H
R
U
G
3
2
dense network of CꢀH···O and CꢀH···Cl interactions. We
observed an interesting trend in the polymorphs of 1,
namely that the number of conformations in a particular
structure (Z’) correlates with CꢀH···O bond strength/dis-
tance. Figure 2is a distance–angle scatter plot of C ꢀH···O
interactions in polymorphic forms A–D. CꢀH···O contacts in
i
xii
unit. Form D has C(10) chains
that connect to form the cyclic
3
4
R (32) pattern and o-phenyl
CꢀH···O synthon III with a
form C are, in general, shorter than those in forms B and D
3778
www.chemeurj.org
ꢁ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2006, 12, 3777 – 3788

Polymorphs of 4,4-Diphenyl-2,5-cyclohexadienone
FULL PAPER
Figure 1. a) Helices of CꢀH···O hydrogen bonds (2.55 Š, 163.68; 2.60 Š, 129.98) between 2
i
related molecules in form A [conformer A , graph set C(8)].
1
2
b) Centrosymmetric CꢀH···O synthon I between B
, Bii molecules of R
(8) pattern (2.33 Š, 169.28; 2.49 Š, 123.58) and synthon II between Biii, Biv mole-
i
2
2
cules of R
cules (C
related D
2
(20) pattern (2.64 Š, 138.08; 2.74 Š, 121.68) in form B. The crystal structure of form C is similar to B. Twelve symmetry independent mole-
3
–Cxii) engage in similar synthons instead of four molecules in B. c) CꢀH···O interactions of graph set R
4
(32) between translation and screw-axis-
i
2
molecules (2.36 Š, 137.98; 2.54 Š, 166.98) in form D. d) Centrosymmetric CꢀH···O synthon III between Dii molecules [R (16) pattern] and Cꢀ
i
2
H···O interaction (2.57 Š, 144.58; 2.47 Š, 165.18). Neutron-normalized distances are quoted. Note that different CꢀH donors participate in CꢀH···O inter-
actions in different crystal forms. Cyclic CꢀH···O synthons I–III are labeled.
[
a]
Table 2. Data for polymorphs (ꢁ3 forms) in organic crystal structures
with multiple molecules in the asymmetric unit.
[
b]
Entry
CSD refcode
No. of polymorphs
Highest Z’
[
a]
conformational polymorphs (ꢁ4 forms)
1
QAXMEH
5
7
1
2
3
4
5
6
7
8
SUTHAZ
2
BEWKUJ
BIXGIY
HEYHUO
KAXHAS
MABZNA
RUWYIR
4
4
4
4
4
4
2
1
12
1
4
2
[
c]
[
a]
multiple molecules in asymmetric unit (Z’>4)
[c]
9
1
1
1
1
PUBMUU
IFULUQ
DUVZOJ
3
4
3
16
8
6
0
1
2ZZZVTY
3
3
5
Figure 2. H···O distance (2.2–3.0 Š) versus CꢀH···O angle (120–1808)
scatter plot of interactions in tetramorphs A–D. A=* (Z’=1), B=
THIOUR
3
4.5
&
[
[
a] Cut-offs were made to limit the number of structures analyzed.
b] See ref. [8]. [c] Compound has a high number of polymorphs and a
(Z’=4), C=~ (Z’=12), D=^ (Z’=2). The shortest H···O distance
(marked with an arrow in the linear band) is inversely related to Z’ (the
number of symmetry-independent conformations).
high Z’ value.
Chem. Eur. J. 2006, 12, 3777 – 3788
ꢁ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
3779

A. Nangia, G. J. Kruger et al.
and longer in form A. Interestingly, the shortest linear inter-
action (q=160–1758) in a particular form is inversely related
to the Z’ value of that structure: for example, form C has
the shortest H···O distance of 2.30 Š and the highest Z’
value of 12, form B has H···O=2.33 Š, Z’=4, and forms D
and A have even longer H···O distances of 2.47 and 2.55 Š
and smaller Z’ values of 2and 1, respectively. Our finding
that the strength of CꢀH···O interactions can promote Z’>1
(~10%). The ratios were determined by the Rietveld refine-
[22]
ment
of observed powder XRD plots with simulated
peaks for each crystal structure (Powder Cell 2.3). Triclinic
forms B and C are taken together because it is not possible
to distinguish between these closely related forms from their
overlapping diffraction patterns (Figure S1, Supporting in-
formation). The mixture of forms at room temperature was
heated to study phase transformations. The peak profile is
relatively stable between 30–608C, however, we noticed
changes as the sample was heated to 708C (Figure 4): Cer-
tain peaks disappeared and the overall pattern became sig-
nificantly sharper with fewer but more intense lines. The
PXRD profile is relatively unchanged between 70–1008C,
after which the material became mostly amorphous and
then gradually turned to a semi-solid/melt mass at 105–
1158C. There are no reflections from the sample at T>
1058C except for the peak from the sample holder at 25.28.
VT-PXRD shows that heating polymorphs A–D to a pre-
melt temperature of 708C transforms the mixture to form A
(Figure 5) with good polymorph purity (>95%), based on a
match with simulated peaks of the crystal structure. Mono-
clinic polymorph A therefore is a thermodynamic modifica-
tion of the enantiotropic system of polymorphs 1 between
30–808C.
among flexible molecules means that the well-known exam-
ples of structures with values of Z’>1, that is, alcohols and
phenols, can be expanded to new categories of crystal struc-
tures stabilized by weak hydrogen bonds. We show that the
relative strengths of directional CꢀH···O interactions are im-
[21]
portant in promoting high Z’ structures
particularly as
their energies are comparable to favorable close-packing
forces. The inverse relationship between Z’ and H···O dis-
tance is observable in polymorphic cluster 1 because de-
tailed supramolecular effects can be clearly discerned
against a background of a constant molecular structure.
The occurrence of high-Z’ polymorphs and the relation-
ship between Z’ and CꢀH···O strength alludes to the impor-
tance of kinetic factors during crystallization. We therefore
wanted to identify which of the concomitant polymorphs A–
D is the kinetic form and which is the thermodynamic one
and determine the nature of the phase transitions between
these forms.
Chiral form A was prepared in high purity and shown to
have a nonlinear optical signal equal to that of urea when ir-
3
+
radiated with a Nd –YAG laser (1.06 mm). However, the
mixture of polymorphs obtained from a typical solution
crystallization does not emit light at 532nm. The prepara-
tion of form A by the above heating method is preferred
over controlled crystallization at ꢀ58C because of contami-
nation from other polymorphs over a period of time, pre-
Variable-temperature powder X-ray diffraction: We ob-
tained single crystals of all four forms in preliminary batch-
[
13]
es but subsequent experiments gave mostly forms A and
B, as determined by unit-cell checking of several crystals.
However, powder X-ray diffraction shows all four forms in
the concomitant mixture at room temperature (Figure 3). A
typical solid upon crystallization from EtOAc/n-hexane con-
tains form A (~40%), forms B+C (~50%), and form D
[23]
sumably due to accidental seeding of laboratory space,
a
term used to describe difficulties in isolating an early poly-
morph after the appearance of other forms of the same
compound. Polymorphs of 1 do not follow Ostwaldꢀs rule of
[24]
stages, with stable form A appearing first from solution
crystallization followed by metastable forms B and C.
Kinetic form B was prepared by heating the polymorphic
mixture to a melt phase in the powder X-ray diffractometer
pan at ~1158C. Cooling the sample to room temperature af-
forded reasonably pure polymorph B (Figure 6). This was
confirmed by unit-cell checking of a few randomly picked
crystals. Although polymorphs A–D have quite different ar-
rangements of molecules and unit cells, they melt at the
same temperature (T =120.458C) and there is no apparent
m
phase transition other than the melting endotherm in differ-
ential scanning calorimetry (Figure S2, Supporting informa-
tion).
There are alternative explanations for the occurrence of
[11]
concomitant polymorphs:
Simultaneous nucleation of
more than one form from solution, interconversion between
polymorphs, their appearance in order of stability, and het-
erogeneous cross nucleation. The simultaneous crystalliza-
tion of all four forms A–D from the homogeneous medium
is the most likely reason for the concomitant cluster 1. Inter-
conversion in solution is minimal at room temperature be-
Figure 3. Powder X-ray diffraction of solid 1 at room temperature:
black=experimentally observed powder pattern; red, blue, and green=
calculated powder pattern of form A (37.5%), B+C (52.0%), and D
[5e]
(
10.5%), respectively. Rietveld refinement in Powder Cell 2.3: R
wp =19.32. Percentages of polymorphic forms in different batches vary
within 5%.
p
=14.82,
R
3780
www.chemeurj.org
ꢁ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2006, 12, 3777 – 3788

Polymorphs of 4,4-Diphenyl-2,5-cyclohexadienone
FULL PAPER
Conformation
and
lattice
energy compensation: Com-
pound 1 is an excellent system
for studying how interconvert-
ing molecular conformations in
solution lead to different crystal
packing arrangements involving
several conformers in the solid
state. The energies of the mo-
lecular conformations and the
crystal lattice were calculated
to obtain a quantitative picture
of polymorphism in 1 and a
clue as to why these conforma-
tional polymorphs appear con-
comitantly. The conformation
energies (Econf) and dipole mo-
ments (m) were calculated using
[25]
the Spartan 04 package
Table 3). Each molecule was
(
extracted from the crystal struc-
ture and energy-minimized
(
HF/6-31G**) keeping the con-
formation fixed (heavy carbon
and oxygen atoms invariant)
while the hydrogen atoms were
allowed to relax to reasonable
geometries
method of Yu et al. The tor-
following
the
[
9a]
sion angles t and t that define
1
2
the rotation about the Cquinone
ꢀ
Cphenyl single bonds lie in the
range of 8–22 and 12–388, re-
spectively, along the scatter plot
diagonal (Figure 7). In general,
in the 19 rotamers t ¼
6
t2 save
1
conformer A and C (see Fig-
i
xi
ure S3, Supporting information,
for the overlay diagram). Mole-
cule B has the most stable con-
i
formation (Econf =ꢀ479813.50
ꢀ
1
kcalmol , a value that is arbi-
trarily fixed to 0); the energies
of the conformers of forms A,
B, and D (Econf) are within
Figure 4. a) Powder X-ray diffraction patterns of 1 recorded at different temperatures. The sample is a mixture
of forms A–D at room temperature, it transforms to forms A at around 708C, and becomes amorphous upon
further heating to 1058C. The wide peak at 25.28 is from the sample holder. b) Powder XRD of 1 at 28 (top)
ꢀ
1
and 698C (bottom). Peaks that disappear upon heating are marked with an arrow. Note the increase in intensi- 1.3 kcalmol
of Bi (Table 3).
ty of the peaks and the overall simplification of the profile at a higher temperature.
Conformers C –C are higher
i
xii
in
energy
(Econf =2–9 kcal
ꢀ
1
mol ) but this could be due to
cause the percentages of various forms in different batches
are within the experimental limit of 5%. As mentioned, the
system does not obey Ostwaldꢀs law of stages. Heterogene-
ous cross nucleation in the heated/melt phase is ruled out
because if this were happening then cooling the sample
from 70 and 1158C would not only give form A and B, re-
spectively, but also other polymorphs from seeded nuclea-
tion and growth.
an error in the experimental X-ray geometry as the R factor
of form C is high (11.1%). Therefore we focus on structures
A, B, and D in this discussion. The conformers of forms A,
B, and D readily interconvert in solution through geared
[13]
(correlated) rotation of the phenyl rings about t and t2
1
because the energy barrier should be accessible through the
ꢀ
1
thermal motion of atoms (RT~0.5 kcalmol at 298 K) in
the crystallization regime between ꢀ5 and 1008C.
Chem. Eur. J. 2006, 12, 3777 – 3788
ꢁ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
3781

A. Nangia, G. J. Kruger et al.
stability. Either the molecular conformer or the lattice
energy is at the minimum, but both intra- and intermolecu-
lar energies are not the lowest in any structure. For example,
ꢀ
1
form A has the lowest Ulatt value (ꢀ32.69 kcalmol ), but
conformer A is higher in energy (E =1.22 kcalmol ).
ꢀ
1
i
conf
Molecular conformations B –B are lower in energy (E =
conf
i
iv
ꢀ
1
0
.00, 0.06, 0.66, 1.12kcalmol ) but crystal lattice B is meta-
ꢀ
1
stable (Ulatt =ꢀ31.66 kcalmol ). Both intra- and intermolec-
ular energy contributions are included in the total energy
term (Table 5). In increasing order of Etotal (COMPASS):
form A (thermodynamic, most stable)<B (kinetic, inter-
mediate stable)<D (least stable). This energy order is con-
sistent with the thermodynamic stability of form A deter-
mined in the VT-PXRD experiments and the crystallization
of the kinetic form B from the melt of the polymorphic mix-
ture. On the other hand, the Ulatt (COMPASS) stability
order of A<D<B and the energies calculated by using the
DREIDING 2.21 force field (D<A<B) do not agree with
experiment. We believe that calculations using the COM-
PASS force field give an indication of the relative stability
of the polymorphs with an energy difference of 0.3–
Figure 5. Experimental powder XRD of 1 at 698C (black line) matches
with the calculated powder pattern of polymorph A (dotted line). Riet-
veld refinement in powder cell 2.3: R =14.39, Rwp=18.81. The starting
p
solid was the mixture of forms A–D shown in Figure 3.
ꢀ
1
1
.0 kcalmol for small organic molecules.
The molecular conformations and crystal lattice energies
suggest the following picture of polymorphism in 1. Depend-
ing on the geometry of a particular conformation, a different
CꢀH···O interaction and aromatic packing motif lead to a
metastable crystal structure. The reason for conformational
polymorphism in 1 is that CꢀH···O motifs in its crystal struc-
tures (Figure 1) involve one of the phenyl CꢀH donors,
except for those in dimer I which are between the quinone
groups. Therefore a change in molecular conformation will
alter the strength of weak CꢀH···O and van der Waals inter-
actions and in turn the preferred crystal-packing motif. The
energy penalty in the molecular conformation is compensat-
ed by lattice-energy gain from intermolecular interactions
and close packing, and vice versa, because their magnitudes
ꢀ
1
are comparable, DEconf ꢂDU =0–2kcalmol . Facile ro-
latt
Figure 6. Experimental powder XRD of 1 from melt crystallization at
tamer interconversion in solution and very similar crystal
energies mean that more than one molecular conformation
may crystallize out simultaneously to give concomitant con-
formational polymorphs of 1.
1
158C (black line) shows good agreement with the calculated powder
pattern of polymorph B (dotted line). Rietveld refinement in powder cell
.3: R =11.78, Rwp =15.86. The starting solid was the mixture of forms
2
p
A–D shown in Figure 3.
[29]
Price and co-workers recently examined intra- and in-
termolecular energy compensation in the conformational
polymorphs of some drugs, for example, aspirin, barbituric
acid, and piracetam. Surprisingly, the quintessential poly-
morphic compound, ROY, appears to be an exception to the
above-mentioned energy balance situation: the stable per-
pendicular conformation is present in the thermodynamic
yellow crystal form. This prompts the question: Will meta-
stable crystal forms of ROY give way to the stable poly-
morph because the system may gradually transform to the
bottom of the molecular and lattice energy well?
The crystal lattice energies, Ulatt, of polymorphs A–D
were computed using the COMPASS and DREIDING 2.21
2
[26]
force fields (Table 4, Cerius ).
We discuss COMPASS
numbers because this force field is better parametrized and
[27]
gives more accurate energies of organic molecules
that
[30]
are typically stabilized by hydrogen bonds, intermolecular
[28]
interactions, and van der Waals forces. COMPASS
is
better suited for molecule 1 because electrostatic stabiliza-
tion by CꢀH···O hydrogen bonds and edge-to-face aromatic
interactions is included in the coulombic term. Form A has
Having discussed forms A, B, and D we will briefly men-
tion the unusually high Z’ value (=12) of polymorph C. We
ꢀ
1
the most stable crystal structure (Ulatt =ꢀ32.69 kcalmol )
[13]
and forms B, C, and D are less stable by 1.03, 1.06, and
previously postulated that form C represents a snap-shot
picture of an evolving crystal nucleus on the way to form B
(Z’=4) wherein the molecules have aggregated to form the
ꢀ
1
0
.82kcalmol , respectively. However, all these crystal struc-
tures are a compromise between intra- and intermolecular
3782
www.chemeurj.org
ꢁ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2006, 12, 3777 – 3788

Polymorphs of 4,4-Diphenyl-2,5-cyclohexadienone
FULL PAPER
Table 3. Energies and dipole moments of 19 conformers of 1 calculated using Spartan 04.
terpretation is consistent with
the following observations. 1)
The higher-energy conformers
in form C participate in stron-
ger CꢀH···O interactions and
[
a]
Crystal polymorph
Molecular conformation
Ph torsion angles
[8] [8]
HF/6-31G**
[
b]
ꢀ1
t
1
t
2
E
conf [kcalmol
]
m [D]
form A
form B
A
i
12.5
12.3
14.9
19.1
11.5
12.8
11.9
18.4
12.1
15.2
10.2
11.5
18.3
20.9
14.0
14.9
20.1
18.5
8.4
12.6
16.0
23.6
31.8
17.74
16.0
16.7
32.3
22.7
22.0
21.0
17.4
28.2
30.8
16.7
15.1
31.5
36.8
16.5
22.3
1.22
0.00
0.06
0.66
1.125.2
2.81
3.94
4.16
4.76
5.19
5.51
7.99
8.14
8.15
8.31
8.54
8.90
1.08
1.25
5.15
5.22
5.20
5.34
B
B
B
B
i
ii
this alludes to the greater role
of kinetic factors in the nuclea-
tion of this polymorph. 2) The
iii
iv
form C
Ci
5.20
5.40
5.06
5.30
5.40
5.24
5.50
5.08
5.21
5.15
5.23
5.39
5.21
5.17
4.87
C
C
C
C
C
C
C
C
C
C
C
ii
close
relationship
between
iii
iv
v
forms C and B is evident from
their crystal structures: They
have the same space group and
overall packing that is sustained
vi
vii
viii
ix
x
through CꢀH···O synthons
I
and II and overlapping powder
XRD peaks. In short, high Z’
xi
xii
structures are not just a crystal-
lographic oddity but they open
a window to “see” crystal nu-
cleation and growth.
Does molecular conformation
determine crystal packing or
are favorable crystal packing ef-
form D
energy-minimized
D
D
i
ii
[
c]
22.3
ꢀ2.78
[
(
i
a] Molecules are numbered in order of increasing energy. [b] Conformer B has the most stable conformation
ꢀ479813.50 kcalmol ) and is arbitrarily fixed to 0.00 for comparison of conformation energies. [c] The mo-
ꢀ
1
lecular skeleton was minimized to the stable gas-phase rotamer.
fects able to trap metastable
molecular conformations? This
question is difficult to quantify in the present case because
Econf and Ulatt differences are of the same order of magni-
tude. We can nonetheless say that a particular conformation
is associated with a specific CꢀH···O synthon even though
that rotamer is present in different polymorphs. For exam-
ple, Biv and Cvii rotamers have similar conformations and
they engage in synthon II in the crystal structures of poly-
morphs B and C. Such a reality, namely that a less stable
conformation can lead to a more stable crystal lattice in con-
formational polymorphs, is a major challenge in the crystal
structure prediction of flexible molecules.
Computational prediction of form A: The ab initio predic-
tion of crystal structures of organic molecules from their
[31]
molecular diagram is a global research activity,
which
should give us a better understanding of the crystallization
process and even protein folding. In crystal structure predic-
tion (CSP), the input for flexible molecules is progressing
from the first phase of using stable gas-phase conformations
[29]
to metastable conformations. The main difficulties in pre-
dicting the structures of conformationally flexible molecules
are the following. 1) The stable conformation may not result
in the thermodynamic crystal structure. 2) Which metastable
conformation out of several low-energy rotamers should be
selected for simulations? 3) Both conformation- and lattice-
energy contributions to crystal structure stabilization must
be taken into account. These issues are pertinent to mole-
cule 1 and we now present a possible solution for predicting
and ranking structure frames of a conformationally flexible
molecule. Thermodynamic polymorph A is the target in
crystal structure prediction because at the present time CSP
computations are only able to predict the lowest energy
Figure 7. a) Conformations of diphenylquinone 1 defined by torsion
angles t
tallographically independent conformations of 1 lie along the diagonal.
All 19 conformers converge to the gas-phase rotamer (t =t =22.38)
after energy minimization. A=*, B=
, C=~, D=^, and gas phase=ꢃ.
1
and t
2
about the CquinoneꢀCphenyl single bonds. b) Nineteen crys-
1
2
&
crystal lattice but the periodicity is yet to reach the highest
possible crystal symmetry. The 12high-energy conformers of
crystal structure C may be viewed as metastable relics of the
relatively stable four conformers of form B. Attempts to
study the transformation of form C to B in the laboratory
were thwarted by our inability to grow crystals of form C
after the initial X-ray diffraction experiment. The above in-
Chem. Eur. J. 2006, 12, 3777 – 3788
ꢁ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
3783

A. Nangia, G. J. Kruger et al.
ꢀ
1
2
Table 4. Lattice energies [Ulatt, kcalmol ] of forms A–D computed in Cerius , corrected to per molecule of 1.
Form A Form B Form C
COMPASS DREIDING 2.21 COMPASS DREIDING 2.21 COMPASS DREIDING 2.21 COMPASS DREIDING 2.21
Form D
U
latt
total
ꢀ32.69
ꢀ28.01
ꢀ4.68
–
ꢀ42.12
ꢀ27.78
ꢀ12.50
ꢀ1.84
ꢀ31.66
ꢀ28.19
ꢀ3.47
–
ꢀ39.66
ꢀ27.22
ꢀ10.43
ꢀ2.01
ꢀ31.63
ꢀ28.18
ꢀ3.45
–
ꢀ39.71
ꢀ27.23
ꢀ10.49
ꢀ1.99
ꢀ31.87
ꢀ27.80
ꢀ4.07
–
ꢀ42.42
ꢀ27.19
ꢀ12.49
ꢀ2.74
van der Waals
coulombic
hydrogen bond
[
a]
[
a] The hydrogen bond energy is partitioned in the DREIDING 2.21 force field but it is part of the coulombic component in the COMPASS force field.
[
a]
ꢀ1
[b]
Table 5. Relative energies [per molecule, kcalmol ] of crystal forms A, B, and D.
answer, our reproduction of a
large and flexible molecule
[c]
Polymorph
U
latt
E
conf
E
total =Ulatt +Econf
Graph set symbol of CꢀH···O
interaction
(
compound 1) is quite good in
COMPASS DREIDING HF/6-
COMPASS DREIDING
2.21
relation to ongoing CSP efforts
2
.21
31G**
[31]
by other groups.
A
B
D
0.00
1.03
0.820.00
0.30
2.76
1.22
0.46
1.98
1.22
1.49
1.52
3.22
C(8)
How important is the starting
conformation in generating a
particular crystal structure of 1?
Given that the molecular con-
formation, CꢀH···O synthon
2
2
R
R
2
(8), R
2
(20)
2
3
1.16
1.16
2
(16), R (32)
4
[
a] Values taken from Tables 3 and 4. The values of Ulatt and Econf are relative energies. [b] Form C is excluded
because the errors are too large. [c] The average Econf value is estimated for multiple conformers as (ꢀEconf)/Z’.
and crystal packing are inti-
structure and those with Z’=1. The objective of our simula-
tions was to reproduce the known stable polymorph A and
to identify structures of similar energies as a guide for
future crystallization experiments. We have used Polymorph
2
[26]
Predictor (Cerius ) (PP) computations to generate several
putative crystal structures of 1 from the stable molecular
[32]
conformation derived from Gaussian 03.
In structure prediction, the conformation should be al-
lowed to vary during the energy minimization of frames (de-
fined as full-body minimization) to cover the complete
range of possible crystal structures in flexible molecules.
Rigid-body minimization (the molecular conformation is
held fixed) is about five times faster but it generates struc-
tures that correspond to local minima for a particular con-
formation, not necessarily the global minimum. Crystal
structure frames were generated in six common space
¯
groups, P2 /c, P1, C2/c, Pbca, P2 , and P2 2 2 . Ten unique
1
1
1 1 1
ꢀ
1
Figure 8. Lattice energy versus net volume (V/Z) for structures of mole-
cule 1 generated in six common space groups by full-body minimization
using the Polymorph Predictor software package. Experimental crystal
structure A matches with the third rank predicted structure based on
values of Ulatt. See Table 6 for details.
low-energy frames within 4 kcalmol of the global mini-
mum are plotted for each space group in Figure 8 (total of
6
0 structures, see Table S1, Supporting Information, for
values). The experimental structure A is the third rank
frame based on values of Ulatt. Cell parameters, torsion
angles, and the lattice energy of
CSP frame #3 match remarka-
[
a]
[b]
Table 6. Comparison of predicted and experimental structure parameters.
bly well the experimental struc-
ture A having a deviation of
within 3% (Table 6). The crys-
tal packing in the simulated
structure is identical to the ob- form A
served form and their powder rigid-body minimization
XRD profiles are in good
agreement (Figure 9 and Fig-
ure S4, Supporting Informa-
Form
a
b
[Š]
c [Š] b [8]
V/Z
[Š ]
t
1
, t
2
[8]
U
mol
latt [kcal
3
ꢀ1
[
Š]
]
full-body minimization
frame #3
7.712 8.286 10.415 104.32 322.45
7.713 8.286 10.415 104.32 322.46
14.6, 15.9 ꢀ92.201
14.6, 15.9 ꢀ92.196
frame #1
form A
crystallographic parameters from
Table 1
7.699 8.133 10.165 104.96 307.51
7.701 8.139 10.160 105.00 307.54
12.6, 12.7 ꢀ32.697
12.6, 12.7 ꢀ32.695
[c]
tion). Although the fact that form A
frame #3 matches the observed
7.917 8.445 10.308 105.75 331.68
12.5, 12.6
–
2
2
[
a] Structure predicted by Cerius (COMPASS). [b] Experimental form A minimized in Cerius . [c] The devia-
structure is short of the correct tion in cell parameters from those of the full-body minimized structure #3 is <3%.
3784
www.chemeurj.org
ꢁ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2006, 12, 3777 – 3788

Polymorphs of 4,4-Diphenyl-2,5-cyclohexadienone
FULL PAPER
is significant and so one should accurately compute both in-
tramolecular and intermolecular energy terms. Structures
predicted by full-body minimization adopt the best molecu-
lar conformation for a stable crystal structure in that space
group because the conformation is allowed to adjust during
the simulation. The lattice energy quantifies the intermolec-
ular component arising from hydrogen bonds, electrostatic
interactions, and van der Waals forces. The gain/penalty
from the molecular conformation must be added/subtracted
to accurately calculate the total crystal energy. To imple-
ment this method, the U component was calculated by
latt
33]
[
rigid-body minimization
and Econf was calculated using
Spartan 04. The starting rotamer was extracted from the
full-body minimization frame and structures were generated
by the rigid-body method. The minimum energy structure in
the rigid-body simulation matches with the full-body refer-
ence frame in all respects: Cell parameters, crystal packing,
and simulated PXRD (Table S3, Supporting Information).
Frame numbers 1–10 of the flexible-body method (Table S1)
were re-ranked based on the total energy, Etotal =Econf +Ulatt
,
of rigid-body minimized frames (Table 7). There is signifi-
cant reorganization in the rankings of predicted structures
when accurately calculated crystal lattice and molecular con-
formation energies are considered together to prioritize the
simulated frames using increasing Etotal as the criterion in-
stead of only Ulatt. Frame #3 of the flexible-body minimiza-
tion is now the global minimum frame #1 and it perfectly
matches stable form A. We take advantage of both the flexi-
2
ble-body and rigid-body minimization methods in Cerius
Polymorph Predictor to simulate the crystal structures of a
molecule with several low-energy conformations. The best
metastable conformation for generating stable crystal pack-
ing is determined by allowing torsion angles to vary. Then
Ulatt and Econf are accurately calculated for this ideal confor-
mation and their sum is taken to finalize the lowest energy
structure. The above iterative method for deriving the cor-
rect metastable conformation and then re-ranking predicted
crystal structure frames is not reported in the CSP literature
to our knowledge. It is suitable for structure prediction of
flexible drug molecules in which conformation- and lattice-
energy contributions must be properly quantified.
Figure 9. Comparison of experimental stable polymorph A with simulated
structure #3 derived from full-body minimization. a) Unit-cell packing in
predicted frame #3. b) Unit-cell packing in form A. c) Simulated powder
XRD of predicted and experimental crystal structures. See Figure S4,
Supporting Information, for Rietveld refinement of PXRD.
Among the 60 predicted structures (Table S1, Supporting
Information), the molecular conformation in global mini-
¯
mum frame #1 in the P1 space group (t , t =28.4, 28.68,
1 2
Econf =2.55 kcalmol ) is not too high in energy relative to
mately related in the polymorphs of 1, we were encouraged
to find that rigid-body minimization starting from rotamer
ꢀ
1
A gave the lowest energy structure #1 as polymorph A
i
the stable rotamer B and may be accessible if a much more
i
(
Table 6). To further attest the significance of metastable
stable Ulatt is able to compensate for the Econf penalty
through stronger intermolecular interactions and better
close packing. We are searching for new polymorphs of 1
conformations in generating structures of 1, the gas-phase
conformation, which is not observed in any known poly-
morph so far, was used in CSP. Predicted structures from
the stable Gaussian 03 rotamer (t =t =23.88) are less
stable (Ulatt =ꢀ2 8 toꢀ30 kcalmol , Table S2, Supporting
Information) than the observed crystal structures (U
[34]
through exhaustive crystallization screens.
1
2
ꢀ
1
=
Conclusion
latt
ꢀ
1
ꢀ
31 to ꢀ33 kcalmol , Table 4) with metastable conforma-
tions.
The discussions so far imply that the contribution of mo-
lecular conformation energy to crystal structure stabilization
Our experimental and computational results on tetramor-
phic cluster A–D of diphenylquinone 1 can be summarized
as follows. 1) Variable-temperature powder XRD shows that
Chem. Eur. J. 2006, 12, 3777 – 3788
ꢁ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
3785

A. Nangia, G. J. Kruger et al.
Table 7. Ulatt of the lowest energy frame determined by the rigid-body method starting from the molecular
conformation in full-body minimized frames #1–10. Econf is calculated in Spartan 04. The simulated structures
of 1 are re-ranked based on the sum of the intra- and intermolecular energies, Etotal [kcalmol ].
trans-Stilbene oxide (1.20 g, 6.0 mmol)
was stirred with BF
3
·Et
2
O
(0.5 mL,
Cl
2
ꢀ1
225 mg, 3.0 mmol) in dry CH
2
[
c]
(60 mL) for 30 min at 08C. The reac-
tion mixture was washed with water
Frame # in full-
body method
Space
group
U
latt in rigid-body
E
conf
E
total =Ulatt +Econf Re-ranked frame
[
a]
[b]
ꢀ1
ꢀ1
ꢀ1
method [kcalmol
]
[kcalmol
]
[kcalmol
]
based on Etotal
(
2”50 mL) and the solvent evaporated
to give diphenyl acetaldehyde
960 mg, 80%).
[
1
d]
3
1
6
P2
P1
ꢀ30.93
ꢀ30.83
ꢀ30.91
ꢀ30.421.82
ꢀ30.88
ꢀ30.06
ꢀ30.85
ꢀ30.01
ꢀ31.03
ꢀ29.88
1.38
2.03
2.25
ꢀ29.55
ꢀ28.80
ꢀ28.66
ꢀ28.60
ꢀ28.29
ꢀ28.24
ꢀ28.06
ꢀ27.87
ꢀ27.76
ꢀ27.51
1
2
3
4
5
6
7
8
9
¯
(
P2
1
2
1
2
1
Ethanolic KOH (3n, 0.5 mL) was
added dropwise over a period of 5 min
to a mixture of diphenyl acetaldehyde
(700 mg, 3.6 mmol) and methyl vinyl
ketone (MVK, 0.3 mL, 3.7 mmol) in
dry THF (30 mL) at 08C. The mixture
was stirred for 2h at 0 8C and then for
1
7
8
5
4
2
9
0
C2/c
¯
P1
2.59
1.82
2.79
2.14
3.27
2.37
C2/c
P2
P2
P2
1
1
1
/c
2
1
2
1
¯
P1
10
2
h at room temperature. Neutraliza-
[
a] Taken from Table S1. [b] See Table S3 for matching cell parameters. [c] Relative to the stable gas-phase ro-
tamer, Econf =ꢀ479812.98 kcalmol . [d] Unit cell in full-body/rigid-body minimized structure: 7.712/7.713 Š,
ꢀ1
tion with 20% HCl, extraction with
EtOAc, and work up gave the crude
3
8
.286/8.289 Š, 10.415/10.412 Š, 75.68/76.658, 322.45/322.44 Š .
form A is the thermodynamic polymorph and form B is the
kinetic phase. The enantiotropic relationship means that
ꢀ
1
lattice-energy differences as small as 0.3 kcalmol
ꢀ
1
(
(
~1.3 kJmol ) computed using the COMPASS force field
Cerius ) are experimentally verified. This value serves as a
2
benchmark for future work on crystal structures that lie in a
shallow potential energy well such as concomitant and/or
conformational polymorphs. 2) We have shown that weak
but directional CꢀH···O interactions promote multiple mole-
cules in the asymmetric unit and moreover that a short
H···O distance in a particular polymorph relates to a high Z’
value of that crystal structure. Z’ is 4 in kinetic polymorph B
and 1 in thermodynamic polymorph A. 3) The reason for
conformational polymorphism in 1 is the presence of several
low-energy, interconverting conformations in solution. The
crystal structures of 1 are a compromise between the mini-
mization of intramolecular (rotamer) and intermolecular
Scheme 1. Synthesis of diphenylquinone 1 from trans-stilbene.
product. Purification by column chromatography yielded pure 4,4-di-
phenyl-2-cyclohexenone (440 mg, 50%).
DDQ (908 mg, 4.0 mmol) and a catalytic amount of p-TsOH (15 mg,
0.1 mmol) was added to a solution of the above cyclohexenone (248 mg,
1.0 mmol) in 1,4-dioxane (40 mL) and the solution was refluxed for 72h.
After cooling, the reaction mixture was filtered through Celite and the
filtrate was diluted with CH
2 2
Cl and washed with 10% NaOH solution
(
interaction) energies. Crystal packing stabilizes the meta-
(
3”30 mL). Workup gave the crude product which was purified on a
stable molecular conformation of 1 in the solid state. 4)
Thermodynamic polymorph A is reproduced as frame #3 in
full-body minimized structure prediction based on Ulatt. Re-
ranking of frames by including the Econf contribution to the
crystal energy gives global minimum structure #1 which
matches form A. This exercise gives a posteriori validity to
silica gel column to give pure 4,4-diphenyl-2,5-cyclohexadienone
1
1
(100 mg, 40%). M.p. 1208C. H NMR (400 MHz, CDCl , 2 58 C, TMS):
3
d=6.38 (d, J=10 Hz, 2H, a-enone Hs), 7.25–8.15 (m, 12H, Ph+b-enone
Hs) ppm; IR (KBr): n˜ =3028, 1655 (C=O), 1620, 1487, 1446, 1400, 1267,
ꢀ
1
1
226 cm .
Polymorphs A—D: The pure solid 1 after column chromatography was
analyzed by powder XRD. The bulk material (100 mg of the solid) shows
all four forms A–D in the concomitant crystallization batch (Figure 3):
monoclinic form A 37.5%, triclinic forms B+C 52.0%; orthorhombic
form D 10.5%.
2
Cerius Polymorph Predictor for a conformationally flexible
medium-sized organic molecule. Our results on prototype
system 1 are currently being examined and evaluated in
other polymorphic clusters.
Polymorphic forms A, B, and C were crystallized by slow evaporation of
a solution of 1 in 5% EtOAc/n-hexane at ambient temperature. Three
types of morphologies were observed: needle, block, and plate. Needle-
type crystals correspond to form A and block/plate crystals correspond to
form B as confirmed by random cell-checking of different crystals. Crys-
tallization by slow evaporation at ꢀ58C in a domestic refrigerator yielded
form A whereas crystal growth from a saturated solution by fast evapora-
tion at ambient temperature yielded predominantly form B. Crystals of
Experimental Section
Synthesis: Compound
1 was prepared in four steps as shown in
[
35]
2 2
form D were obtained from a CH Cl /EtOAc/n-hexane solvent mixture.
In recent batches over the last 2–3 years we have not found a single crys-
tal corresponding to form C in several random cell-checking experiments.
Scheme 1.
trans-Stilbene (1.1 g, 6.0 mmol) in CH
to a stirred solution of m-CPBA (1.14 g, 6.6 mmol) in dry CH
2
Cl
2
(60 mL) was added dropwise
Cl
15 mL) at 08C. The reaction was continued for 30 h. The mixture was
2
2
(
Pure form A: The polymorphic mixture 1 (100 mg) was heated in a test
tube at ~708C in an oil bath for 30 min and then slowly cooled to room
temperature. The mixture converted to form A in >95% purity as con-
firmed by powder XRD (Figure 5).
3
washed with a NaHCO solution and water. The resulting epoxide was
extracted with CH
2
Cl
2
and evaporated in vacuo to give pure trans-stil-
bene oxide (1.2g, 95%).
3786
www.chemeurj.org
ꢁ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2006, 12, 3777 – 3788

Polymorphs of 4,4-Diphenyl-2,5-cyclohexadienone
FULL PAPER
ꢀ
1
Pure form B: The polymorphic mixture 1 (100 mg) was heated in the alu-
minium pan of a powder X-ray diffractometer until the compound
melted (~1208C). The cooled solid is form B (powder XRD in Figure 6),
as confirmed by unit-cell matching of a few randomly picked crystals.
ture powder X-ray diffraction data at a heating rate of 18Cmin were
collected at T=28, 39, 49, 59, 69, 79, 89, 94, 98, 102, and 1058C. The
sample was cooled to room temperature (318C) and data were recollect-
ed. Powder XRD profiles are plotted in the range of 2q=10–358 for all
samples. There are no significant peaks below 108 and only minor peaks
between 35 and 508. Wide peaks resulting from the sample holder appear
at 25.2 and 34.88.
Differential scanning calorimetry: DSC was performed with a Mettler
Toledo 822e module. Samples (4–6 mg) were placed in crimped but
ꢀ
1
vented aluminium pans and heated at 108C min from 30–2008C. The
ꢀ
1
instrument was purged with a stream of dry nitrogen at 150 mLmin
.
2
2
Spartan 04, Gaussian 03, and Cerius computations: Cerius simulations
[
26]
and crystal energy: All simulations were carried out using version 4.8
of the Cerius molecular modeling software package on a Silicon Graph-
2
Acknowledgements
ics workstation. Geometry optimization was carried out using density
functional theory (DFT) at the B3LYP/6-31G (d,p) level of theory with
Gaussian 03. The global minimized rotamer of 1 from Gaussian 03 was
We thank the Department of Science and Technology for research fund-
ing (SR/S5/OC-02/2002), for support of the CCD X-ray diffractometer
(IRPHA), and the CMSD computational facility. The UPE program of
the UGC provided infrastructure funds for UH. S.R. and R.B. thank the
UGC for fellowships. G.J.K. and Prof. Gautam R. Desiraju (UH) thank
the Indo-South Africa Program on Research Cooperation for financial
support (DST/INT/SAFR/COP/2001).
[
32]
entered as the input for the Polymorph Predictor. Crystal structure pre-
diction was carried out in six common space groups (P2
1
, P2
1
/c, C2/c,
¯
Pbca, P2 2 2 , and P1). The cell parameters of predicted frames of the
1 1 1
C2/c space group were compared with reduced cell parameters to con-
firm that they represent different structures. Reduced cell parameters are
used throughout the paper. Atom point charges were assigned using the
COMPASS force field. It was not felt necessary to calculate multipole
charges because molecule 1 does not contain strong hydrogen-bonding
groups. Multipole charges are known to give superior results but the ad-
vantage is more evident in strongly hydrogen-bonded structures and that
too at the expense of an approximate 10-fold increase in computer
[1] W. C. McCrone in Physics and Chemistry of the Organic Solid State,
Vol. 2 (Eds.: D. Fox, M. M. Labes, A. Weissberger), Wiley Inter-
science, New York, 1965, pp. 725–767.
[2] J. Bernstein, Polymorphism in Molecular Crystals, Clarendon Press,
Oxford, 2002.
[
36]
time.
Default options were used throughout with the fine search
option in Monte Carlo simulations and for the clustering of frames to
obtain unique structures. Lattice-energy minimization of predicted struc-
tures was carried out without any modifications except for the use of the
Ewald summation of van der Waals interactions at a cut off of 6.0 Š. All
calculations were carried out either by relaxing the molecular conforma-
tion during the minimization, referred to as the full-body method, or by
keeping the conformation fixed during minimization, the so-called rigid-
body method. Full-body lattice energy minimizations were carried out
even though these calculations take approximately five times more com-
puter time because this method gives more accurate results for flexible
molecules such as 1. The lattice energies of the experimental polymorphs
[3] For some recent papers on polymorphism dealing with these issues,
see: a) A. Kµlmµn, L. Fµbiµn, G. Argay, G. BernPth, Z. Gyarmati, J.
Am. Chem. Soc. 2003, 125, 34; b) I. Weissbuch, V. Y. Torbeev, L.
Leiserowitz, M. Lahav, Angew. Chem. 2005, 117, 3290; Angew.
Chem. Int. Ed. 2005, 44, 3226; c) M. Morimoto, S. Kobatake, M. Irie,
Chem. Eur. J. 2003, 9, 621; d) P. Raiteri, R. Martoòµk, M. Parrinello,
Angew. Chem. 2005, 117, 3835; Angew. Chem. Int. Ed. 2005, 44,
3
769; e) P. K. Thallapally, R. K. R. Jetti, A. K. Katz, H. L. Carrell,
K. Singh, K. Lahiri, S. Kotha, R. Boese, G. R. Desiraju, Angew.
Chem. 2004, 116, 1169; Angew. Chem. Int. Ed. 2004, 43, 1149; f) H.
Chow, P. A. W. Dean, D. C. Craig, N. T. Lucas, M. L. Scudder, I. G.
Dance, New J. Chem. 2003, 27, 704; g) R. G. Gonnade, M. M.
Bhadbhade, M. S. Shashidhar, Chem. Commun. 2004, 2 530; h) C.
Guo, M. B. Hickey, E. R. Guggenheim, V. Enkelmann, B. M.
Foxman, Chem. Commun. 2005, 2220; i) S. Aitipamula, A. Nangia,
Chem. Commun. 2005, 3159; j) W. I. F. David, K. Shankland, C. R.
Pulham, N. Bladgen, R. J. Davey, M. Song, Angew. Chem. 2005, 117,
2
of 1 were computed using the Cerius program by energy minimization of
crystal structures using DREIDING 2.21 and COMPASS force fields.
Force-field charges were assigned with COMPASS and the charge equili-
brium method was used with DREIDING 2.21. COMPASS is better par-
ametrized for structure prediction and the energy of organic molecules.
Crystal lattice energies were calibrated to account for the number of mol-
ecules in the unit cell (per molecule).
7
194; Angew. Chem. Int. Ed. 2005, 44, 7032.
[
[
4] G. R. Desiraju, Angew. Chem. 1995, 107, 2541; Angew. Chem. Int.
Ed. Engl. 1995, 34, 2311.
5] a) R. J. Davey, Chem. Commun. 2003, 1463; b) N. Blagden, R. J.
Davey, Cryst. Growth Des. 2003, 3, 873; c) P. Erk, H. Hengelsberg,
M. F. Haddow, R. van Gelder, CrystEngComm 2004, 6, 474; d) J.
Bernstein, Chem. Commun. 2005, 5007; e) L. Yu, J. Am. Chem. Soc.
Spartan 04 computations: The energies of all 19 conformers were calcu-
[
25]
lated using Spartan 04 with crystallographic coordinates as the input;
the positions of the hydrogen atoms were reoptimized at the HF/6-31G**
level of theory while keeping the heavy atoms fixed. The gas-phase con-
formation of 1 was obtained by global energy minimization. The gas-
phase rotamer of 1 calculated using Spartan 04 (t
similar to the minimized conformation derived using Gaussian 03 (t
=23.88).
1 2
=t =22.38) is very
2
003, 125, 6380; f) P. Vishweshwar, J. A. McMahon, M. Oliveira,
M. L. Peterson, M. J. Zaworotko, J. Am. Chem. Soc. 2005, 127,
6802; g) C. P. Price, A. L. Grzesiak, A. J. Matzger, J. Am. Chem.
Soc. 2005, 127, 5512.
1
=
t
2
1
Variable-temperature powder X-ray diffraction: Powder X-ray data were
collected with a PANlytical X’Pert PRO powder X-ray diffractometer
using a parallel beam of monochromated Cu-K radiation (l=1.54056 Š)
a
[6] S. R. Bryn, R. R. Pfeiffer, J. G. Stowell, Solid-State Chemistry of
Drugs, SSCI, West Lafayette IN, 1999.
and an X’celerator detector at 40 kV and 40 mA. Diffraction patterns
were collected over the 2q range of 5–508. Samples were ground to a par-
ticle size of >20 mm and loaded in an 18 mm alumina holder for varia-
ble-temperature powder X-ray diffraction data collection and in an alu-
minium sample holder with a 10 mm diameter sample cavity for the col-
lection of data at room temperature. Vigorous grinding was avoided to
minimize potential phase transitions. The program X’Pert High Score
was used for the processing and comparison of powder patterns. Powder
Cell 2.3 was used for calculating the PXRD patterns and for the profile
fitting and Rietveld refinement of unit-cell parameters, a displacement
parameter, a background polynomial function, peak shape asymmetry
terms, and an overall temperature factor using the known single-crystal
structures of polymorphs A, B, and D as the model. Variable-tempera-
[7] a) X. Zheng, B. Wang, U. Engelert, G. E. Herberich, Inorg. Chem.
2001, 40, 3117; b) J. van de Streek, S. Motherwell, Acta Crystallogr.
Sect. B 2005, 61, 504. In 2000 the values were 4.1% of organic com-
pounds and 5.5% of organometallic compounds (ref. [7a]), which in
2004 was updated to 14600 accurate crystal structures out of 325000
entries (4.5%) (ref. [7a]).
[8] F. H. Allen, R. Taylor, Chem. Soc. Rev. 2004, 33, 463. The CSD is an
archive of over 350000 small-molecule organic and organometallic
crystal structures (August 2005 release); see www.ccdc.cam.ac.uk
[9] a) L. Yu, G. A. Stephenson, C. A. Mitchell, C. A. Bunnell, S. V.
Snorek, J. J. Bowyer, T. B. Borchardt, J. G. Stowell, S. R. Byrn, J.
Am. Chem. Soc. 2000, 122, 585; b) S. Chen, I. A. Guzei, L. Yu, J.
Am. Chem. Soc. 2005, 127, 9881.
[
22]
Chem. Eur. J. 2006, 12, 3777 – 3788
ꢁ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
3787

A. Nangia, G. J. Kruger et al.
[
10] a) V. S. S. Kumar, PhD Thesis, University of Hyderabad, 2002;
b) crystal structures of 3 have been deposited with the Cambridge
Crystallographic Data Centre (see ref. [8]): CCDC 293133–293135.
11] J. Bernstein, R. J. Davey, J.-O. Henck, Angew. Chem. 1999, 111,
[31] a) W. D. S. Motherwell, H. L. Ammon, J. D. Dunitz, A. Dzyabchen-
ko, P. Erk, A. Gavezzotti, D. W. M. Hofmann, F. J. J. Leusen, J. P. M.
Lommerse, W. T. M. Mooij, S. L. Price, H. Scheraga, B. Schweizer,
M. U. Schmidt, B. P. van Eijck, P. Verwer, D. E. Williams, Acta Crys-
tallogr. Sect. B 2002, 58, 647; b) G. M. Day, W. D. S. Motherwell, H.
Ammon, S. X. M. Boerrigter, R. G. Della Valle, E. Venuta, A.
Dzyabchenko, J. D. Dunitz, B. Schweizer, B. P. van Eijck, P. Erk,
J. C. Facelli, V. E. Bazterra, M. B. Ferraro, D. W. M. Hofmann, F. J. J.
Leusen, C. Liang, C. C. Pantelides, P. G. Karamertzanis, S. L. Price,
T. C. Lewis, H. Nowell, A. Torrisi, H. A. Scheraga, Y. A. Arnautova,
M. U. Schmidt, P. Verwer, Acta Crystallogr. Sect. B 2005, 61, 511.
[32] Gaussian 03 (Revision B.05), M. J. Frisch, G. W. Trucks, H. B. Schle-
gel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman, J. A. Montgomer-
y, Jr., T. Vreven, K. N. Kudin, J. C. Burant, J. M. Millam, S. S. Iyen-
gar, J. Tomasi, V. Barone, B. Mennucci, M. Cossi, G. Scalmani, N.
Rega, G. A. Petersson, H. Nakatsuji, M. Hada, M. Ehara, K.
Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda,
O. Kitao, H. Nakai, M. Klene, X. Li, J. E. Knox, H. P. Hratchian,
J. B. Cross, C. Adamo, J. Jaramillo, R. Gomperts, R. E. Stratmann,
O. Yazyev, A. J. Austin, R. Cammi, C. Pomelli, J. W. Ochterski, P. Y.
Ayala, K. Morokuma, G. A. Voth, P. Salvador, J. J. Dannenberg,
V. G. Zakrzewski, S. Dapprich, A. D. Daniels, M. C. Strain, O.
Farkas, D. K. Malick, A. D. Rabuck, K. Raghavachari, J. B. Fores-
man, J. V. Ortiz, Q. Cui, A. G. Baboul, S. Clifford, J. Cioslowski,
B. B. Stefanov, G. Liu, A. Liashenko, P. Piskorz, I. Komaromi, R. L.
Martin, D. J. Fox, T. Keith, M. A. Al-Laham, C. Y. Peng, A. Na-
nayakkara, M. Challacombe, P. M. W. Gill, B. Johnson, W. Chen,
M. W. Wong, C. Gonzalez, J. A. Pople, Gaussian, Inc., Pittsburgh,
PA, 2003.
[
[
[
3
646; Angew. Chem. Int. Ed. 1999, 38, 3440.
12] J. Bernstein in Organic Solid State Chemistry (Ed.: G. R. Desiraju),
Elsevier, Amsterdam, 1987, pp. 471–518.
13] a) V. S. S. Kumar, A. Addlagatta, A. Nangia, W. T. Robinson, C. K.
Broder, R. Mondal, I. R. Evans, J. A. K. Howard, F. H. Allen,
Angew. Chem. 2002, 114, 4004; Angew. Chem. Int. Ed. 2002, 41,
3
848; b) A. Nangia in Nanoporous Materials: Science and Engineer-
ing (Eds.: G. Q. Lu, X. S. Zhao), Imperial College Press, London,
004, pp. 165–187.
2
[
14] G. R. Desiraju, Acc. Chem. Res. 2002, 35, 565.
[
15] J. Bernstein, R. E. Davis, L. Shimoni, N.-L. Chang, Angew. Chem.
1
995, 107, 1687; Angew. Chem. Int. Ed. Engl. 1995, 34, 1555.
[
[
[
16] V. S. S. Kumar, A. Nangia, Chem. Commun. 2001, 2392.
17] J. W. Steed, CrystEngComm 2003, 5, 169.
18] a) C. P. Brock, L. L. Duncan, Chem. Mater. 1994, 6, 1307; b) C. P.
Brock, Acta Crystallogr. Sect. B 2002, 58, 1025.
[
[
19] A. Gavezzotti, G. Fillippini, J. Phys. Chem. 1994, 98, 4831.
20] H. Jacobsen, H. W. Schmalle, A. Messmer, H. Berke, Inorg. Chim.
Acta 2000, 306, 153.
[
[
21] K. E. Plass, K. Kim, A. J. Matzger, J. Am. Chem. Soc. 2004, 126,
9
042.
22] Rietveld refinement was carried out using Powder Cell 2.3, N.
Krauss, G. Nolze, Federal Institute for Materials Research and Test-
ing, Berlin, 2000.
[
[
23] a) J. D. Dunitz, J. Bernstein, Acc. Chem. Res. 1995, 28, 193; b) J.-O.
Henck, J. Bernstein, A. Ellern, R. Boese, J. Am. Chem. Soc. 2001,
23, 1834.
24] W. Ostwald, Z. Phys. Chem. 1897, 22, 289. Ostwaldꢀs rule of stages:
When leaving a metastable state, a given chemical system does not
[33] The value of Ulatt for minimized crystal structures after full-body
ꢀ1
1
minimization is much lower (ca. ꢀ92kcalmol ) than rigid-body
ꢀ1
minimized structures (ca. ꢀ32kcalmol ). The contribution from va-
“
lence terms (bonds, angles, torsions, cross-terms) is around
ꢀ
1
seek out the most stable state, rather the nearest metastable one
that can be reached with minimum loss of free energy.”
25] Spartan 04 computational methods include Hartree–Fock, density
function theory, and Moøller–Plesset on an interactive PC interface:
see www.wavefun.com
ꢀ60 kcalmol for full-body minimization of molecule 1 but valence
terms are not considered in rigid-body minimization (fixed to zero).
This difference between the two sets of energies is not a problem
because we are comparing relative energies using the same method.
[34] For CSP and isolation of the polymorph, see: a) A. T. Hulme, S. L.
Price, D. A. Tocher, J. Am. Chem. Soc. 2005, 127, 1116; b) G. M.
Day, A. V. Trask, W. D. S. Motherwell, W. Jones, Chem. Commun.
2006, 54.
[
[
[
2
26] Crystal Packer and Polymorph Predictor of the Cerius suite of soft-
ware were used for crystal lattice energy calculation and crystal
structure prediction: see www.accelrys.com
27] A. Dey, M. T. Kirchner, V. R. Vangala, G. R. Desiraju, R. Mondal,
J. A. K. Howard, J. Am. Chem. Soc. 2005, 127, 10545.
[35] a) D. J. Reif, H. O. House, Org. Syn. 1963, Coll. Vol. IV, 860; b) D. J.
Reif, H. O. House, Org. Syn. 1963, Coll. Vol. IV, 375; c) A. C. Cope,
P. A. Trumbell, E. R. Trumbell, J. Am. Chem. Soc. 1958, 80, 2844;
d) H. E. Zimmerman, K. G. Hancock, G. C. Licke, J. Am. Chem.
Soc. 1968, 90, 4892; e) K.-B. Chai, P. Sampson, J. Org. Chem. 1993,
58, 6807.
[
28] H. Sun, J. Phys. Chem. B 1998, 102, 7338.
[
29] a) T. C. Lewis, D. A. Tocher, S. L. Price, Cryst. Growth Des. 2004, 4,
9
79; b) C. Ouvard, S. L. Price, Cryst. Growth Des. 2004, 4, 1119;
c) H. Nowell, S. L. Price, Acta Crystallogr. Sect. B 2005, 61, 558.
[
30] a) J. D. Dunitz, A. Gavezzotti, Cryst. Growth Des. 2005, 5, 2180. The
[36] G. M. Day, W. D. S. Motherwell, W. Jones, Cryst. Growth Des. 2005,
5, 1023.
ꢀ
1
DEtotal maximum is larger in ROY (5 kcalmol between the Y and
ORP forms) because Econf and Ulatt contributions are additive than
Received: November 15, 2005
Published online: March 3, 2006
ꢀ
1
in 1 (<1 kcalmol ) when the components counter one another.
3788
www.chemeurj.org
ꢁ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2006, 12, 3777 – 3788