Cocrystals of nutraceutical p-coumaric acid with caffeine and theophylline: Polymorphism and solid-state stability explored in detail using their crystal graphs

Article abstract of DOI:10.1039/c0ce00214c

Cocrystals constructed with p-coumaric acid (a phytochemical and nutraceutical compound) are investigated with xanthine compounds, caffeine and theophylline. Four cocrystals of p-coumaric acid with caffeine (1:1 and 1:2 stoichiometric ratios) and theophylline (two 1:1 polymorphs, Form I and Form II) were generated and their structures determined by single-crystal X-ray crystallography. The two theophylline cocrystals display synthon polymorphism, where both structures possess a carboxylic acid-imidazole heteromeric synthon; however, one polymorph also has a hydroxyl-carbonyl synthon (Form I), while in the other a hydroxyl-imidazole synthon (Form II) is present. Furthermore, the solid-state stability of the two p-coumaric acid:theophylline polymorphs was explored experimentally and computationally.

Full text of DOI:10.1039/c0ce00214c

View Online / Journal Homepage / Table of Contents for this issue
PAPER
www.rsc.org/crystengcomm | CrystEngComm
Cocrystals of nutraceutical p-coumaric acid with caffeine and theophylline:
polymorphism and solid-state stability explored in detail using their crystal
graphs†
Nate Schultheiss*, Melanie Roe and Stephan X. M. Boerrigter
Received 18th May 2010, Accepted 14th August 2010
DOI: 10.1039/c0ce00214c
Cocrystals constructed with p-coumaric acid (a phytochemical and nutraceutical compound) are
investigated with xanthine compounds, caffeine and theophylline. Four cocrystals of p-coumaric acid
with caffeine (1 : 1 and 1 : 2 stoichiometric ratios) and theophylline (two 1 : 1 polymorphs, Form I and
Form II) were generated and their structures determined by single-crystal X-ray crystallography. The
two theophylline cocrystals display synthon polymorphism, where both structures possess a carboxylic
acid–imidazole heteromeric synthon; however, one polymorph also has a hydroxyl–carbonyl synthon
(Form I), while in the other a hydroxyl–imidazole synthon (Form II) is present. Furthermore, the solid-
state stability of the two p-coumaric acid : theophylline polymorphs was explored experimentally and
computationally.
Coumaric acid is a hydroxy derivative of cinnamic acid and
Introduction
naturally occurs in three isomers (ortho-, meta-, and para-); p-
Cocrystals are multi-component crystals in which the individual,
coumaric acid is the most commonly occurring isomer in nature.
neutral molecules are held together by non-covalent interactions,
most often hydrogen bonds.1 Cocrystalline materials containing
Classified as a phytochemical and nutraceutical, p-coumaric
acid is found in various edible plants, such as peanuts, carrots,
active pharmaceutical ingredients (APIs) are continuing to gain
and tomatoes.9 A number of promising pharmacokinetic studies
interest within the pharmaceutical industry, due to their ability to
have been conducted on p-coumaric acid showing a positive
alter the physicochemical properties of solid dosage forms, while
response in protection against colon cancer on cultured
providing additional opportunities for intellectual property and
mammalian cells,10 as well as antioxidant and anti-inflammatory
drug product repositioning.2 Generating cocrystals of an API can
properties in rats and rabbits.11 Caffeine and theophylline are
be advantageous because (1) it allows for modifications to be
both classified as xanthine alkaloids, with the former often used
introduced to the crystal structure of a drug (which in turn can
as a food additive and central nervous system stimulant, while
alter its physical and chemical properties) without compromising
the latter has shown health benefits in the treatment of asthma.12
its intended bioactivity, and (2) it gives rise to the possibility of
Herein we report the synthesis and characterization of four
generating combination drugs where two or more active ingre-
cocrystals of p-coumaric acid with caffeine and theophylline,
dients are brought into one crystal lattice.
including single crystal X-ray structure determination for each.
When engineering a cocrystal, major criteria for selecting
Additionally, the solid-state stability of the two polymorphic p-
a cocrystal former (coformer), especially if the cocrystal is
coumaric acid : theophylline cocrystals is examined in-depth, in
intended as a final drug product, are its pharmacological and
particular how the synthons relate to the intermolecular inter-
toxicological properties. A number of different sources are often
action energies. The chemical structures of the compounds used
used for selecting a suitable coformer, including the pharma-
in this study are shown in Scheme 1.
ceutically acceptable salt formers list,3 Generally Regarded as
Safe (GRAS) list,4 and Everything Added to Food in the United
States (EAFUS) list.5 In addition to the possible coformers on
these lists, formulation excipients,6 as well as complementary
drug molecules and nutraceutical compounds7 (resulting in
possible combination drugs)8 could also be potentially incorpo-
rated into the construction of a cocrystal.
SSCI, a Division of Aptuit, 3065 Kent Avenue, West Lafayette, IN, 47906,
USA. E-mail: Nathan.schultheiss@aptuit.com; Fax: +1-765-463-4722;
Tel: +1-765-463-0112
† Electronic supplementary information (ESI) available: XRPD patterns,
DSC and TGA thermograms of 1–4, and COMPASS calculations.
CCDC reference numbers 778271–778272. For ESI and
crystallographic data in CIF or other electronic format see DOI:
10.1039/c0ce00214c.
Scheme
1 Chemical structures of p-coumaric acid, caffeine, and
theophylline (L–R).
This journal is ª The Royal Society of Chemistry 2011
CrystEngComm, 2011, 13, 611–619 | 611

View Online
X-Ray powder diffraction (XRPD)
Experimental
Patterns were collected using a PANalytical X’Pert Pro diffrac-
tometer. An incident beam of Cu Ka radiation was produced
using an Optix long, fine-focus source. PANalytical data were
collected and analysed using X’Pert Pro Data Collector software
(v. 2.2b). Prior to the analysis, a silicon specimen (NIST SRM
640c) was analyzed to verify the Si 111 peak position. PAN-
alytical diffraction patterns were collected using a scanning
position-sensitive detector (X’Celerator) located 240 mm from
the specimen.
Reagents
p-Coumaric acid, caffeine, and theophylline (anhydrous) were
purchased from Sigma-Aldrich and used as received. All other
chemicals were purchased from various suppliers and used
without further purification.
p-Coumaric acid : caffeine (1 : 1), 1
Single crystals were grown from a slow cool experiment where
a 1 : 1 mixture of p-coumaric acid (57.3 mg, 0.35 mmol) and
caffeine (67.7 mg, 0.35 mmol) was dissolved in acetonitrile (3.8
mL) with stirring at approximately 70 ꢀC. Upon cooling to room
temperature and standing for 1 day, single crystals were har-
vested. The material was scaled up following the same reaction
procedure resulting in 702 mg, 80% yield.
Thermogravimetric analysis (TGA)
Thermogravimetric analyses were performed using
a TA
Instruments 2950 thermogravimetric analyzer. The sample was
placed in an aluminium sampleꢀpan and inserted into the TG
furnace. Analyses began at ꢁ20 C, and the furnace was heated
under nitrogen at a rate of 5 or 10 K minÀ1, up to a final
p-Coumaric acid : caffeine (1 : 2), 2
ꢀ
temperature of 350 C. Nickel and Alumelꢀ were used as cali-
bration standards.
Single crystals were grown from a slow evaporation experiment.
A bulk solution of p-coumaric acid (194.0 mg, 1.18 mmol) in
acetone (4.0 mL) was initially prepared. A small amount of this
solution (600 mL, containing 0.18 mmol p-coumaric acid) was
added to a vial, and a solution of caffeine (35.7 mg, 0.18 mmol) in
acetone (5.0 mL) was added with stirring. The solution was
allowed to slowly evaporate, upon which single crystals were
culled. The material was scaled up following the same reaction
procedure resulting in 1.03 g, 64% yield. No attempt was made to
improve the yield by reacting the components in a 1 : 2 stoi-
chiometric ratio.
Differential scanning calorimetry (DSC)
DSC was performed using a TA Instruments 2920 differential
scanning calorimeter. Temperature calibration was performed
using NIST traceable indium metal. The sample was placed into
an aluminium DSC pan, and the weight was accurately recorded.
The pan was covered with a lid perforated with a laser pinhole,
and the lid was crimped. A weighed, crimped aluminium pan was
placed on the reference side of the cell. The sample cell was
equilibrated at 25 ꢀC and heated under a nitrogen purꢀge at a rate
of 5 or 10 K minÀ1, up to a final temperature of 250 C.
p-Coumaric acid : theophylline (1 : 1), 3 (Form 1) and 4 (Form
II)
Single crystal X-ray diffraction (SCXRD)
Single crystals were grown from a slow evaporation experiment. A
bulk solution of p-coumaric acid (194.0 mg, 1.18 mmol) in acetone
(4.0 mL) was initially prepared. A small amount of this solution
(600 mL, containing 0.18 mmol p-coumaric acid) was added to
a vial, and a solution of theophylline (32.7 mg, 0.18 mmol) in
methanol (5.0 mL) was added with stirring in a 1 : 1 molar ratio.
The solution was allowed to slowly evaporate, upon which single
crystals of 3 and 4 were culled. Larger quantities of pure 3 (Form I)
and 4 (Form II) were generated by adding solid theophylline to
a nearly saturated solution of p-coumaric acid in ethanol for
(Form I) or p-dioxane for (Form II) and stirring for ꢁ24 hours
before filtering.
Datasets were collected on a Bruker SMART APEX II diffrac-
tometer (1, 2 and 4) or a Bruker Kappa APEX II diffractometer
(3) using Mo Ka radiation. Data were collected using APEX II
software.13 Initial cell constants were found by small widely
separated ‘‘matrix’’ runs. Data collection strategies were deter-
mined using COSMO. Scan speed and scan width were chosen
based on scattering power and peak rocking curves. Temperature
control was provided with an Oxford Cryostream low-tempera-
ture device.
Unit cell constants and orientation matrix were improved by
least-squares refinement of reflections thresholded from the
entire dataset. Integration was performed with SAINT,14 using
this improved unit cell as a starting point. Precise unit cell
constants were calculated in SAINT from the final merged
dataset. Lorenz and polarization corrections were applied.
Absorption corrections were not applied (m  d < 0.05 in all
cases).
Data were reduced with SHELXTL.15 The structures were
solved in all cases by direct methods without incident. Unless
otherwise noted, coordinates of all –OH and –NH hydrogens
were refined. All other hydrogens were riding in idealized posi-
tions. Unless otherwise noted, all non-hydrogen atoms were
refined using anisotropic thermal parameters. Refinement details
for each cocrystal 1–4 are listed below.
Interconversion slurries
Saturated solutions of p-coumaric acid and theophylline were
prepared by adding solids of each component to three different
solvents (acetonitrile, ethyl acetate, and isopropanol) until
undissolved solids were present. Equal amounts (approximately
5 mg) of each p-coumaric acid : theophylline cocrystal, 3 (Form
I) and 4 (Form II), were then added to a filtered portion of the
liquid phase from each saturated solution, and the resulting
slurries were allowed to stir at ambient conditions for 1 day.
Solids were collected by vacuum filtration and analyzed by
XRPD.
612 | CrystEngComm, 2011, 13, 611–619
This journal is ª The Royal Society of Chemistry 2011

View Online
1 Coordinates for the acid hydrogen atom H29, and the phenol
hydrogen atom H24 were refined.
stronger than À2.0 kcal molÀ1 are shown. The full set of inter-
actions up to the default cut-off of À0.596 kcal molÀ1 are tabu-
lated in the ESI†.
2 One of the two unique caffeines existed in two different
orientations. The ratio for the two orientations was refined. A
water of hydration was located in the difference Fourier map. Its
occupancy and isotropic thermal parameter were refined;
because of the low occupancy, hydrogen atoms for this water
molecule were neither located nor included. Coordinates for the
acid hydrogen atom H29, and the phenol hydrogen atom H24
were refined.
AM1 calculations
Molecular structures were constructed using Spartan ’08
(Wavefunction, Inc. Irvine, CA), and their geometries were
optimized using AM1, with the maxima and minima in the
electrostatic potential surface (0.002 e auÀ1 isosurface) deter-
mined using a positive point charge in vacuum as a probe.
3 There were two independent acid : theophylline pairs in the
asymmetric unit. Each pair was included in a SHELXL ‘‘RESI’’
residue. Coordinates for the two acid hydrogen atoms, both
labelled H29, and both phenol hydrogen atoms H24 were refined.
4 Coordinates of amine hydrogen atom H17, phenol hydrogen
atom H24, and carboxylic acid hydrogen atom H29 were all
refined.
Results and discussion
In efforts to generate cocrystals with p-coumaric acid and
caffeine, two crystal structures with different stoichiometries
were obtained. Cocrystal 1 was synthesized from slow cooling
a 1 : 1 mixture of components, resulting in crystals with the
monoclinic space group P21/c, Table 1. The asymmetric unit of 1
contains p-coumaric acid and caffeine in a 1 : 1 stoichiometric
ratio, Fig. 1. In the crystal structure of 1, the molecules assemble
into one-dimensional chains through a series of heteromeric O–
H/N, C–H/O and O–H/O hydrogen bonds. The carboxylic
acid moiety of p-coumaric acid forms hydrogen bonds with the
imidazole of caffeine through an R22(7) heterosynthon, while the
hydroxyl moiety hydrogen bonds with the carbonyl oxygen of
Molecular modelling
Crystal structures as obtained from SCXRD were imported into
Materials Studio 5.0.0.0 (Accelrys Software Inc., San Diego) and
appropriate bond orders were assigned. Full unit cell geometry
optimizations were performed using the COMPASS 2.7 force
field16 using the force field’s charge parameters. Lattice energies
are reported as the final energies including all intra-molecular
interaction energy contributions. Molecular interaction energies
were calculated using all default settings using the ‘‘crystal
graph’’ method as implemented in the morphology module. For
the purpose of clarity in this presentation, only the interactions
ꢀ
caffeine (O24/O16, 2.6984 A), Fig. 2 and Table 2.
The second structure, cocrystal 2, crystallizes in the triclinic
ꢁ
space group P1, Table 1. Although the crystals were formed from
a 1 : 1 mixture of components in acetone, the asymmetric unit of
2 contains one molecule of p-coumaric acid and two molecules of
Table 1 Crystallographic data for 1–4
1
2
3 (Form I)
4 (Form II)
Empirical formula
Mw
C17H18N4O5
358.35
C25H28N8O7.20
555.75
C16H16N4O5
344.33
C16H16N4O5
344.33
Color, habit
Colorless, prism
0.35 Â 0.25 Â 0.15
Monoclinic
P2(1)/c, 4
6.5225(4)
10.6474(6)
23.2444(14)
90
93.983(2)
90
1610.37(17)
1.478
120(2)
Colorless, prism
0.24 Â 0.20 Â 0.16
Triclinic
Colorless, plate
0.22 Â 0.14 Â 0.08
Triclinic
Colorless, plate
0.36 Â 0.20 Â 0.08
Monoclinic
P2(1)/n, 4
6.9196(11)
26.801(4)
8.6984(12)
90
108.027(6)
90
1542.28(19)
1.491
120(2)
Crystal size/mm3
Crystal system
Space group, Z
ꢁ
ꢁ
P1, 2
P1, 4
ꢀ
a/A
8.1475(7)
11.2176(9)
14.8517(13)
75.956(2)
84.712(2)
73.340(2)
1261.16(18)
1.464
7.7073(6)
14.2718(10)
14.7035(10)
74.327(4)
82.294(5)
83.154(5)
1542.28(19)
1.483
ꢀ
b/A
ꢀ
c/A
a/ꢀ
b/ꢀ
g/ꢀ
3
ꢀ
Volume/A
Density/g cmÀ3
Temperature/K
X-Ray wavelength
m/mmÀ1
120(2)
120(2)
0.71073
0.111
752
3.13
1.54178
0.110
583
1.95
32.59
0.71073
0.113
720
1.44
31.50
0.71073
0.113
720
2.58
F(000)
ꢀ
Qmin
Qmax
/
/
ꢀ
32.14
31.00
Reflections
Collected
Independent
Observed
Rsigma
18 365
5512
4533
0.025
0.024
>2s(I)
0.0447
0.1367
26 213
8113
6433
0.026
0.023
>2s(I)
0.0538
0.1512
35 014
10 196
5212
0.086
0.086
>2s(I)
0.0665
0.1793
15 724
4826
2868
0.0935
0.0838
>2s(I)
0.0955
0.2906
Rint
Threshold expression
R1 (observed)
wR2 (all)
This journal is ª The Royal Society of Chemistry 2011
CrystEngComm, 2011, 13, 611–619 | 613

View Online
Fig. 1 Thermal ellipsoid plots and labelling schemes for 1–4. Thermal ellipsoids are displayed at 50% probability level.
ꢀ
2.72 A. The refinement of the disorder of the caffeine molecule
linked through the hydroxyl moiety resulted in site occupancies
of 0.628 : 0.372. Additionally, the water molecule within the
structure was assigned an occupancy of 0.20.
Interestingly, a 2 : 1 crystal structure between caffeine and
4-hydroxybenzoic acid (a structural analogue of p-coumaric
acid) was recently reported, and it too displayed disorder around
one caffeine molecule, resulting also in a discrete three-compo-
nent assembly.18
Fig. 2 Periodic one-dimensional chain formed through carboxylic acid–
imidazole and hydroxyl–carbonyl heteromeric synthons in 1.17
Two polymorphic 1 : 1 p-coumaric acid : theophylline coc-
rystals resulted from a 10 : 1 methanol/acetone slow evaporation
experiment. Cocrystal 3 (Form I) crystallizes in the triclinic space
caffeine, Fig. 1. In the crystal structure of 2, a discrete three-
component supramolecular assembly is generated (Fig. 3) in
contrast to the 1 : 1 one-dimensional chain formed in 1. Once
again, the carboxylic acid moiety of p-coumaric acid forms an O–
ꢁ
group P1 (Table 1) with two independent 1 : 1 cocrystals of p-
coumaric acid and theophylline in the asymmetric unit (due to
structural similarity only one will be described), Fig. 1. The
molecules are linked through heteromeric, self-complementary
N–H/O and O–H/O hydrogen bonds from the carboxylic acid
ꢀ
H/N (O29–N19, 2.7129 A) hydrogen bond with the imidazole
nitrogen atom of caffeine, while the hydroxyl moiety forms an
ꢀ
ꢀ
O–H/O (O24–O12A, 2.679 A or O24–O12B, 2.719 A) hydrogen
bond with the carbonyl group of the second caffeine molecule.
To be noted, the self-complementary C–H/O hydrogen bond
from the imidazole of caffeine to the carbonyl of p-coumaric acid
does not form even though the two molecules lie coplanar to one
ꢀ
moiety to the imidazole hydrogen (N17/O41, 2.792 A) and
carbonyl moiety of the theophylline molecule (O39/O16, 2.618
ꢀ
A). As observed in structure 1, one-dimensional hydrogen
bonded chains are formed through the hydroxyl moiety of p-
coumaric acid to the carbonyl of theophylline (O34/O12#1,
ꢀ
2.721 A). The one-dimensional rows are further linked into
ꢀ
another. The C–H/O distance is 3.07 A, which is well beyond
the van der Waals distance for a hydrogen and oxygen contact,
614 | CrystEngComm, 2011, 13, 611–619
This journal is ª The Royal Society of Chemistry 2011

View Online
Table 2 Hydrogen-bond geometries for 1–4
Cocrystal
D–H/A
D(D–H)
d(H/A)
d(D/A)
<(DHA)
1
2
O(29)–H(29)/N(19)
O(24)–H(24)/O(16)#1a
O(24)–H(24)/O(12A)
O(24)–H(24)/O(12B)
O(29)–H(29)/N(19)
N(17)–H(17)/O(40)
N(17)–H(17)/O(40)
O(39)–H(39)/O(16)
O(39)–H(39)/O(16)
O(34)–H(34)/O(12)#1b
O(34)–H(34)/O(12)#1b
O(29)–H(29)/O(16)
N(17)–H(17)/O(30)
O(24)–H(24)/N(19)#1c
0.925(16)
0.879(18)
0.87(2)
0.87(2)
0.88(2)
1.01(2)
0.90(2)
1.02(3)
1.01(3)
0.84(3)
0.99(3)
0.86(5)
0.97(5)
0.77(6)
1.783(16)
1.822(19)
1.81(2)
1.86(3)
1.84(2)
1.79(2)
1.91(2)
1.63(3)
1.61(3)
1.88(3)
1.72(3)
1.76(5)
1.75(5)
2.02(6)
2.7021(12)
2.6984(12)
2.679(9)
2.719(16)
2.7129(15)
2.792(2)
2.806(2)
2.618(2)
2.602(2)
2.721(2)
2.707(2)
2.608(4)
2.728(4)
2.748(4)
172.3(15)
174.6(17)
174(2)
171(2)
172.2(19)
173(2)
179(2)
162(2)
167(2)
178(2)
176(2)
168(5)
178(4)
159(5)
3 (Form I)
4 (Form II)
a
#1 x + 2, y + 1, z. b #1 x À 1, y, z À 1. c #1 Àx + 1/2, y À 1/2, Àz + 1/2.
(Table 1). Within the asymmetric unit, one molecule of p-cou-
maric acid and one molecule of theophylline reside. Once again
the same heteromeric synthon forms between the carboxylic acid
moiety of p-coumaric acid and the imidazole proton and
ꢀ
carbonyl moiety of theophylline (N17/O30, 2.728 A, O29/
ꢀ
O16, 2.608 A). However, the hydroxyl moiety hydrogen bonds to
ꢀ
the imidazole nitrogen atom (O24/N19, 2.748 A) and not the
carbonyl moiety of theophylline as displayed in 3 (Form I). The
molecules assemble into one-dimensional, wave-like hydrogen-
bonded chains, Fig. 5. Once again, the 2-D sheets are stacked
along the a-axis, into three-dimensions, with distances as short as
Fig. 3 Hydrogen bonding motifs between caffeine and p-coumaric acid
in the discrete assembly 2. The disorder and water molecule have been
removed for clarity.
ꢀ
3.40 A between sheets.
Fig. 4 Top view of the two-dimensional sheet composed of p-coumaric
acid and theophylline 3 (Form I) (top) and side-view of the planar,
stacked two-dimensional sheets (bottom).
two-dimensional sheets, Fig. 4. The 2-D sheets are stacked along
the a-axis, into three dimensions, with distances as short as 3.30
ꢀ
A between the layers.
Fig. 5 Top view of the two-dimensional sheet composed of p-coumaric
acid and theophylline 4 (Form II) (top) and side-view of the planar,
stacked two-dimensional sheets (bottom).
The second 1 : 1 p-coumaric acid : theophylline cocrystal, 4
(Form II), crystallizes in the monoclinic space group P21/n
This journal is ª The Royal Society of Chemistry 2011
CrystEngComm, 2011, 13, 611–619 | 615

View Online
By comparison, a 1 : 1 cocrystal between 4-hydroxybenzoic
acid and theophylline has also been found, displaying the same
hydrogen bonding patterns as seen in 4 (Form II), which includes
acid–theophylline and hydroxyl–imidazole synthons.19
unobtainable. Furthermore, the calculated densities from the
single crystal structures are very close in value (3, Form I 1.483 g
cmÀ3 and 4, Form II 1.491 g cmÀ3).22 Additionally, both forms
crystallized concomitantly upon slowly evaporating a 1 : 1 ratio
of p-coumaric acid and theophylline, suggesting both forms are
nearly identical in energy.23 Nonetheless, equal amounts of each
cocrystal (3, Form I and 4, Form II) were slurried in saturated
solutions of the individual components in three different solvents
at ambient temperature. XRPD analysis on the remaining solids
resulted in Form I, proposing it is the most stable form, of each
sample, under these conditions. Along with the experimental
results,24 a computational modelling approach was also under-
taken to examine the energy relationship between the poly-
morphs.
Polymorphism of single-component crystals is well docu-
mented and often occurs;20 however, cases of cocrystal poly-
morphism have been reported much less frequently.21 This is
potentially due to the infancy of cocrystal screening and
searching for polymorphs therein; not because it is less likely to
occur. In fact, one could make the argument that it may even be
more likely to occur, given the multifaceted nature of the typical
coformers used for screening purposes.
Therefore, as for single-component systems, determining the
thermodynamic relationship between polymorphs of cocrystals is
important, especially throughout the drug development process,
because changes in crystal form can lead to differences in phys-
ical and chemical properties.
The overall packing arrangement between forms is very
similar, in which both form one-dimensional chains that result in
relatively planar two-dimensional sheets. Not surprising, the
two-dimensional sheets stack into three dimensions. From
a hydrogen-bonding perspective, the major difference between
the two forms resides primarily in the interactions of the
hydroxyl moiety of p-coumaric acid. In 3, Form I, the hydroxyl
group hydrogen bonds to the carbonyl of theophylline, while in
4, Form II, the hydrogen bond forms between the hydroxyl
group and imidazole nitrogen atom. It has previously been
shown that, based upon semi-empirical AM1 calculations, the
highest point on the molecular electrostatic potential (MEP)
surface can be utilized to estimate the propensity for forming
electrostatic interactions (including hydrogen bonds).25 Accord-
ingly, this point for the imidazole N-atom on theophylline was
found at about À240 kJ molÀ1, whereas for the carbonyl moiety it
was found at about À270 kJ molÀ1. Thus, based on this method
and following Etter’s rules,1a the carbonyl oxygen would be
slightly preferred over the imidazole nitrogen atom as the
primary hydrogen bond acceptor (see Scheme 2).
Relative polymorph stabilities can usually be determined by
experimental approaches. Typically, the onsets of melting and
heats of fusion from differential scanning calorimetry (DSC)
experiments can give some direction to the solid-form stability of
polymorphic compounds;20 however, cocrystals 3 and 4 decom-
pose upon heating (see ESI†), and, thus, these parameters are
Scheme 2 Molecular structure of theophylline with calculated electro-
static charges for the imidazole N-atom and carbonyl moiety (both
highlighted in blue).
The COMPASS calculations on the two polymorphic struc-
tures show a lattice energy difference of 1.32 kcal molÀ1 per 1 : 1
Fig. 6 Diagrams of the 1 : 1 p-coumaric acid : theopylline polymorphs displaying the interaction strength (kcal molÀ1) between molecules. Hydrogen
bonds are represented as dashed-yellow lines and the strength of interaction follows dashed-lines: blue > purple > dark red > red.
616 | CrystEngComm, 2011, 13, 611–619
This journal is ª The Royal Society of Chemistry 2011

View Online
stoichiometric unit of each cocrystal (3, Form I À232.09 kcal
molÀ1 and 4, Form II À230.77 kcal molÀ1), favouring Form I as
the more stable form. Force field calculations do not produce
error bars, so the exact significance of this difference is unknown.
Furthermore, it should be noted that the minimized structures
represent the structures at minimal temperature, which excludes
any consideration of entropy. Therefore, we cannot conclude
with certainty, based on these calculations, which of the two
forms is the most stable at room temperature; however, at the
very least, the calculations show that the difference in enthalpy is
very similar, which is consistent with our experimental results.
The computational studies sparked an interest in how the
synthons and resulting three-dimensional structures differ
between the two polymorphs. Fig. 6 shows the molecular inter-
actions within the 2-D sheets. The strongest interactions (blue-
dashed lines) are formed in both polymorphs by the hydrogen
bonding pairs between the acid moiety of p-coumaric acid and
exist between the hydrogen bonded chains. Although the energies
within the hydrogen bonded chains are quite similar between the
two structures, the interactions between the chains are quite
different. Form I shows an alternating pattern on one side of the
chain of interactions as strong as À4.134 and À4.103 kcal molÀ1
,
.
whereas on the other side they are À2.139 and À2.186 kcal molÀ1
Form II shows a much more isotropic distribution of interactions
between the chains, all being in the range À2.033 and À2.759 kcal
molÀ1 (Fig. 6). Each of the p-coumaric acid molecules forms two
of these interactions, whereas the theophylline forms four, for
a total of six interactions per stoichiometric, asymmetric unit in
Form II. It is interesting that the total of six interactions adds up
to À14.014 kcal molÀ1, whereas the total (four interactions) adds
up to À12.562 kcal molÀ1 in Form I.26 Thus, what Form II lacks
in hydrogen bonding energy, it mostly makes up for with its
inter-chain interaction energies, where the total network energies
add up to À27.701 kcal molÀ1 for Form I and À27.228 kcal molÀ1
for Form II.
theophylline. In Form II this interaction is À13.105 kcal molÀ1
,
whereas, due to the lower symmetry, two values were computed
for Form I: À12.790 and À13.048 kcal molÀ1. A second set of
interactions is found for the molecules interacting through
a single hydrogen bond (purple-dashed lines). These are formed
between the hydroxyl moiety of p-coumaric acid and either the
carbonyl (À8.548 and À8.453 kcal molÀ1) or imidazole N-atom
(À7.116 kcal molÀ1) of theophylline. These results correlate
nicely to the MEP calculations, which also suggested the
formation of the hydroxyl–carbonyl synthon (3, Form I) as
energetically favoured over the hydroxyl–imidazole synthon (4,
Form II). Weaker interactions (dark red to red-dashed lines)
Initially, it was believed that the inter-layer interaction ener-
gies would be quite uniform and similar between the two poly-
morphs, since the molecules possess the same surface area and,
therefore, should roughly have the same van der Waals attrac-
tion. However, when the interaction energies are decomposed
into molecular interactions, a much more intricate picture
emerges. Fig. 7 shows the crystal graphs projected perpendicular
to the layer orientation. In Form II the molecules interact mostly
diagonally, producing a set of first order interactions in the range
of À6.127 to À7.048 kcal molÀ1. A second order interaction is
formed at À4.135 kcal molÀ1 followed by a set of interactions in
Fig. 7 Diagrams of the 1 : 1 p-coumaric acid : theopylline polymorphs displaying the inter-layer interaction energies (kcal molÀ1) between molecules.
All the crystal graph interactions are shown, but, for clarity, only the inter-layer interactions are labelled. The top view is a perpendicular projection with
respect to the layers. For Form I, the structure is projected along the periodic hydrogen bonded chains. The strongest inter-layer interactions show up in
pairs because of how the chains cross into the plane of projection, which is shown more clearly in the slightly tilted bottom view projection. For Form II
some weaker bonds in the order of À2 kcal molÀ1 are obscured in the top projection due to how the interactions zigzag along the projection vector. The
bottom view is slightly tilted to reveal those interactions (red lines, only labelled in the bottom view).
This journal is ª The Royal Society of Chemistry 2011
CrystEngComm, 2011, 13, 611–619 | 617

View Online
the range of À2.075 to À2.154 kcal molÀ1. Please note that an
arbitrary cut-off of À2 kcal molÀ1 was applied for the interac-
tions considered for this discussion and that many interactions
exist below this level of energy. Form I shows a number of
interactions that are much stronger between the layers than those
found in Form II, À9.374 and even À10.529 kcal molÀ1. This is
due to a larger spatial overlap; the interactions thereby align in
a more vertical orientation. Every molecule interacts with each
layer beneath and above it with at least À7.280 kcal molÀ1. This
shows that the energies of the interactions between the layers are
comparable to those of hydrogen bonds. An in-depth analysis of
these interactions revealed that they are mostly dominated by the
van der Waals attraction, where the strongest interactions lower
their energy by having atoms of opposite partial charge in close
proximity (see ESI† for a more detailed explanation). In addition
to these van der Waals interactions, Form I forms more diago-
nally oriented interactions that quickly dwindle to lower and
lower interaction strengths. Interestingly, the initial description
of cocrystals 3 and 4, as structures of 2-D sheets that stack to
form into a 3-D crystal is significantly changed. The stacking
interactions by far outweigh the lateral interactions between the
hydrogen bonded chains.
computational analyses suggest that Form I is favoured as the
low temperature form, albeit by a small difference in enthalpy,
while slurry interconversion experiments show Form I as the
most stable at room temperature. We have shown in great detail
that the crystal graphs of Forms I and II are distinctly different.
This full decomposition of the lattice energy can be a great asset
in the evaluation of how the molecules are actually interacting in
the crystal structure, which is an extremely useful tool in the
design of cocrystals and their resulting materials properties. We
have shown that the application of cocrystals is viable, at least
for these systems, and that polymorphism of these systems may
lead to opportunities in form selection to optimize the perfor-
mance of these crystals.
Acknowledgements
We would like to thank Dr John Desper (Kansas State Univer-
sity) for single-crystal data collection and structure solution of 1–
4 and SSCI’s analytical services for data collection. We also
thank Mr Eyal Barash, Dr Sarah Bethune, and Dr Brett Cowans
for their careful review of this manuscript.
It is instructive to see how these polymorphs distribute their
interaction energies and what constitutes the structural driving
forces (hydrogen bonding, van der Waals forces,.), to ulti-
mately assess the relative stabilities of the two forms. All in all, it
is the total (free) lattice energy that decides which form is most
stable at a given set of environmental conditions. In the case of
the p-coumaric acid : theophylline cocrystals, Form I (cocrystal
3) appears most stable at low temperature, based on the
computational analysis performed.
Notes and references
1 (a) M. C. Etter, Acc. Chem. Res., 1990, 23, 120–126; (b) C. B. Aakeroy
and D. J. Salmon, CrystEngComm, 2005, 7, 439–448; (c)
L. MacGillivray, CrystEngComm, 2004, 6, 77–78; (d)
G. R. Desiraju, Acc. Chem. Res., 2002, 35, 565–573; (e) B. Moulton
and M. Zaworotko, Chem. Rev., 2001, 101, 1629–1658; (f) J.-
M. Lehn, Science, 2002, 295, 2400–2403; (g) F. Grepioni and
D. Braga, Making Crystals by Design—from Molecules to Molecular
Materials, Methods, Techniques, Applications, Wiley-VCH,
Weinheim, Germany, 2007, ch. 2.5 and references therein.
2 (a) A. V. Trask, W. D. S. Motherwell and W. Jones, Cryst. Growth
Des., 2005, 5, 1013–1021; (b) A. V. Trask, W. D. S. Motherwell and
W. Jones, Int. J. Pharm., 2006, 320, 114–123; (c) D. J. Good and
This kind of exercise also affords information regarding how
the crystals may behave physically. In general, a more isotropic
distribution of energies, for instance, leads to a more isotropic
crystal morphology, which can be beneficial (due to better
flowability and filterability) when processing these materials. The
existence of planes of low energy within the structure observed in
Form I can lead to low shear strength which influences plastic
versus brittle behaviour, in turn affecting the processing behav-
iour and drug product quality such as compactibility, friability,
and compressibility. When the total lattice energies are suffi-
ciently close, as is indicated both by our experimental and
computational efforts, these properties may lead to a decisive
factor in solid form selection for single component or multi-
component product applications.
ꢂ
N. Rodrıguez-Hornedo, Cryst. Growth Des., 2009, 9, 2252–2264; (d)
S. L. Childs, N. Rodrıguez-Hornedo, L. S. Reddy, A. Jayasankar,
ꢂ
C. Maheshwari, L. McCausland, R. Shipplett and B. C. Stahly,
CrystEngComm, 2008, 10, 856–864; (e) M. B. Hickey,
M. L. Peterson, L. A. Scoppettuolo, S. L. Morisette, A. Vetter,
H. Guzman, J. F. Remenar, Z. Zhang, M. D. Tawa, S. Haley,
€
M. J. Zaworotko and O. Almarsson, Eur. J. Pharm. Biopharm.,
2007, 67, 112–119; (f) D. McNamara, S. L. Childs, J. Giordano,
A. Iarriccio, J. Cassidy, M. S. Shet, R. Mannion, E. O’Donnell and
A. Park, Pharmacol. Res., 2006, 23, 1888–1897; (g) N. Schultheiss
and A. Newman, Cryst. Growth Des., 2009, 9, 2950–2967 and
references therein; (h) N. Blagden, M. De Matas, P. T. Gavan and
P. York, Adv. Drug Delivery Rev., 2007, 59, 617–630; (i)
S. L. Childs, L. J. Chyall, J. T. Dunlap, V. N. Smolenskaya,
B. C. Stahly and G. P. Stahly, J. Am. Chem. Soc., 2004, 126,
13335–13342; (j) K. Shiraki, N. Takata, R. Takano, Y. Hayashi and
ꢃꢃ ꢂ
K. Terada, Pharmacol. Res., 2008, 25, 2581–2592; (k) T. Friscic and
W. Jones, Cryst. Growth Des., 2009, 9, 1621–1637; (l)
P. Vishweshwar, J. A. McMahon, J. A. Bis and M. J. Zaworotko,
€
J. Pharm. Sci., 2006, 95, 499–516; (m) O. Almarsson and
M. J. Zaworotko, Chem. Commun., 2004, 1889–1896; (n) W. Jones,
W. D. Samuel Motherwell and A. V. Trask, MRS Bull.,2006, 31,
875–879; (o) Crystal growth and Design ‘virtual issue’ on cocrystals,
http://pubs.acs.org/page/cgdefu/vi/1.
3 Handbook of Pharmaceutical Salts: Properties, Selection and Use, ed.
P. H. Stahl and C. G. Wermuth, Verlag Helvetica Chimica Acta,
Conclusions
We have successfully shown that the nutraceutical compound p-
coumaric acid does cocrystallize with caffeine and theophylline,
resulting in four cocrystalline materials. Single crystal structures
were determined for each cocrystal, which allowed for a better
understanding of the hydrogen bonding interactions, especially
between the two 1 : 1 p-coumaric acid : theophylline poly-
morphs.
€
Zurich, 2002.
4 Generally Regarded as Safe, http://www.cfsan.fda.gov/ꢁrdb/
opa-gras.html and http://www.cfsan.fda.gov/ꢁdms/grasguid.html.
5 Everything Added to Food Stuff in the United States, http://
vm.cfsan.fda.gov/ꢁdms/eafus.html.
Determining the relative stability of the 1 : 1 p-coumaric
acid : theophylline cocrystal polymorphs proved very difficult at
room temperature. Both the experimental and computational
methods suggest that the difference in energy is very small. The
6 Handbookof Pharmaceutical Excipients, ed. R. C. Rowe, P. J.
Sheskey, and S. C. Owen, APhA Publications, 5th edn, 2005.
618 | CrystEngComm, 2011, 13, 611–619
This journal is ª The Royal Society of Chemistry 2011

View Online
7 A ‘nutraceutical compound’ can be defined as ‘a food (or part of
a food) that provides medical or health benefits, including the
20 (a) S. R. Byrn, R. R. Pfeiffer and J. G. Stowell, Solid-State Chemistry
of Drugs, SSCI, Inc., West Lafayette, 2nd edn, 1999; (b) H. Brittain,
Polymorphism in Pharmaceutical Solids, Marcel Dekker, New York,
1999; (c) J. Bernstein, Polymorphism in Molecular Crystals,
Clarendon, Oxford, 2002; (d) R. Hilfiker, Polymorphism: in the
Pharmaceutical Industry, Wiley-VCH, New York, 2006; (e)
G. P. Stahly, Cryst. Growth Des., 2007, 7, 1007–1026.
21 (a) S. Aitipamula, P. S. Chow and R. B. H. Tan, CrystEngComm,
2009, 11, 889–895; (b) S. Aitipamula, P. S. Chow and R. B. H. Tan,
CrystEngComm, 2009, 11, 1823–1827; (c) N. J. Badu, L. S. Reddy,
S. Aitipamula and A. Nangia, Chem.–Asian J., 2008, 3, 1122–1133;
(d) D. Braga, G. Palladino, M. Polito, K. Rubini, F. Grepioni,
M. R. Chierotti and R. Gobetto, Chem.–Eur. J., 2008, 14, 10149–
10159; (e) A. N. Sokolov, D. L. Swenson and L. R. MacGillivray,
Proc. Natl. Acad. Sci. U. S. A., 2008, 105, 1794–1797.
22 The difference in calculated crystal density is less than 1% between
polymorphic forms. It has been noted that differences less than 1%
should be excluded from the density rule (ref. 20c, p. 41).
23 J. Bernstein, R. J. Davey and J.-O. Henck, Angew. Chem., Int. Ed.,
1999, 38, 3440–3461.
24 Solution calorimetry (to determine heats of solution) and/or hyper-
DSC (to obtain heat of fusion values) could have also been
introduced to aid in determining thermodynamic stabilities.
25 (a) C. A. Hunter, Angew. Chem., Int. Ed., 2004, 43, 5310–5324; (b)
prevention and/or treatment of
a
disease and possesses
a physiological benefit or reduces the risk of chronic disease’,
E. K. Kalra, AAPS PharmSci., 2003, 3, article 25, pp. 1–2.
8 A few combination formulations; Berodual (fenoterol and
ipratropium); Kaletra (Ritonavir and Lipinavir); Advair
(Fluticasone and Salmeterol); Epzicom (Abacavir and Lamivudine).
9 (a) S. T. Talcott, C. E. Duncan, D. D. Pazo-Insfran and
€
€
D. W. Gorbet, Food Chem., 2005, 89, 77–84; (b) M. P. Kahkonen,
A. I. Hopia, H. J. Vuorela, J. P. Rauha, K. Pihlaja, T. S. Kujala
and M. Heinonen, J. Agric. Food Chem., 1999, 47, 3954–3962.
10 L. R. Ferguson, Z. Shou-tun and P. J. Harris, Mol. Nutr. Food Res.,
2005, 49, 585–693.
11 (a) C. Luceri, L. Giannini, M. Lodovici, E. Antonucci, R. Abbate,
E. Masini and P. Dolara, Br. J. Nutr., 2007, 97, 458–463; (b)
F. Guglielmi, C. Luceri, L. Giovannelli, P. Dolara and
M. Lodovici, Br. J. Nutr., 2003, 89, 581–587; (c) C. Luceri,
F. Guglielmi, F. Meoni and P. Dolara, Scand. J. Gastroenterol.,
2004, 39, 1128–1133.
12 http://www.medicinenet.com/theophylline/article.htm.
13 APEXII v. 2009.11–0,
Systems, Madison, WI.
ª 2005–2009, Bruker Analytical X-ray
14 SAINT v7.68A, ª 1997–2009, Bruker Analytical X-ray Systems,
Madison, WI.
15 SHELXTL v. 2008/4, ª 2008, Bruker Analytical X-ray Systems,
Madison, WI.
€
M. Kamieth, F.-G. Klarner and F. Diederich, Angew. Chem., Int.
Ed., 1998, 37, 3303–3305; (c) C. B. Aakeroy, M. Fasulo,
€
N. Schultheiss, J. Desper and C. Moore, J. Am. Chem. Soc., 2007,
€
and J. Desper, Cryst. Growth Des., 2006, 6, 1033–1042; (e)
N. Schultheiss, K. Lorimer, S. Wolfe and J. Desper,
CrystEngComm, 2010, 12, 742–749.
16 (a) H. Sun, J. Phys. Chem. B, 1998, 102, 7338–7364; (b) H. Sun, P. Ren
and J. R. Fried, Comput. Theor. Polym. Sci., 1998, 8, 229–246; (c)
D. Rigby, H. Sun and B. E. Eichinger, Polym. Int., 1997, 44, 311–330.
17 Pictures created using Mercury CSD 2.2.
129, 13772–13773; (d) C. B. Aakeroy, D. J. Salmon, M. M. Smith
ꢃ
18 D.- K. Bucar, R. F. Henry, X. Lou, R. W. Duerst, L. R. MacGillivray
26 Notice that the numbers need to be divided by two when calculating
the energies as a portion of the lattice energy due to double counting
of interactions.
and G. G. Z. Zhang, Cryst. Growth Des., 2009, 9, 1932–1943.
19 S. L. Childs, G. P. Stahly and A. Park, Mol. Pharmacol., 2006, 3, 328–
338.
This journal is ª The Royal Society of Chemistry 2011
CrystEngComm, 2011, 13, 611–619 | 619