Dehydration, hydration behavior, and structural analysis of fenoprofen calcium

Article abstract of DOI:10.1002/jps.1038

Fenoprofen calcium (FC) is a nonsteroidal, anti-inflammatory, analgesic, and antipyretic agent. The dehydration behavior of FC dihydrate and the rehydration of the dried FC were investigated using differential scanning calorimetry (DSC), thermogravimetric

Full text of DOI:10.1002/jps.1038

Dehydration, Hydration Behavior, and Structural Analysis
of Fenoprofen Calcium
HAIJIAN ZHU, JIA XU, PETER VARLASHKIN, STACEY LONG, CHRIS KIDD
Pharmaceutical Development Discovery and Analytical Sciences Departments, Glaxo Wellcome Inc., Five Moore Drive,
Research Triangle Park, North Carolina 27709
Received 16 May 2000; revised 4 December 2000; accepted 8 December 2000
ABSTRACT: Fenoprofen calcium (FC) is a nonsteroidal, anti-infl ammatory, analgesic,
and antipyretic agent. The dehydration behavior of FC dihydrate and the rehydration of
the dried FC were investigated using differential scanning calorimetry (DSC),
thermogravimetric analysis (TGA), and powder X-ray diffractometry (PXRD). The
stoichiometry, the crystal packing arrangement, and water environments in FC
dihydrate were determined using single-crystal X-ray diffraction (XRD) analysis. The
Arrhenius plot (natural logarithm of the dehydration rate constant versus the reciprocal
of absolute temperature) for FC dihydrate from isothermal TGA is not linear. The
ꢀ
activation energy of dehydration was 309 kJ/mol in the 50
–
60 C range and 123 kJ/mol
ꢀ
in the 60
–
80 C range. The difference in activation energy can be explained from the
crystal structure data where one water molecule is sandwiched between repeating polar
carboxylate groups and the other water is in a slightly less polar region of the crystal.
Single-crystal XRD analysis also indicated each calcium ion is coordinated to six
oxygens. Two coordinating oxygens are provided by two water molecules and the other
four oxygens are provided by the carboxylate group of four separate fenoprofen anions.
Each fenoprofen anion, which can provide two oxygens for coordination, is associated
with two different calcium ions. Hot-stage PXRD suggested that only a loss of 1 mole of
water per mole of FC dihydrate (forming a monohydrate) was required to convert the
material to a partially crystalline state. The monohydrate is not completely disordered
as evidenced by a strong diffraction peak as well as some weaker peaks in the PXRD
pattern. The rehydration of the anhydrous form of FC follows a solution-mediated
transformation, prior to crystallizing as the dihydrate. ß 2001 Wiley-Liss, Inc. and the
American Pharmaceutical Association J Pharm Sci 90:845
– 859, 2001
Keywords: fenoprofen calcium XRD; PXRD; TGA; DSC
INTRODUCTION
result from the association of the drug with
different salt forming counterions. Furthermore,
1
Salts are usually considered alternatives when
the physicochemical characteristics of the parent
drug molecules are unsuitable or inadequate for
satisfactory formulations. Considerable variation
in solubility, dissolution rate, bioavailability, and
other pharmaceutically important properties can
because of lack of predictive relationships bet-
ween the physicochemical properties of the resul-
tant salts, the selection of an appropriate salt
form with desired combination of properties can
be a diffi cult, semi-empirical choice.
Pharmaceutical salts frequently exist as hy-
drates. Because the nature and stability of a salt
hydrate are known to vary with a change of the
This work is presented in memory of one of the authors,
Jia Xu.
Correspondence to: P. Varlashkin (Telephone: 919-483-8464;
Fax: 919-315-1091; E-mail: PGV48846@glaxowellcome.com)
2
counterion, knowledge of the stoichiometry and
nature of the binding environment of water in
the salt hydrate is essential for optimum salt
selection.
Journal of Pharmaceutical Sciences, Vol. 90, 845
– 859 (2001)
ß 2001 Wiley-Liss, Inc. and the American Pharmaceutical Association
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 90, NO. 7, JULY 2001
845

8
46
ZHU ET AL.
used for data reduction and analysis has been
6
described. UNC single-crystal structure results
were reported to the authors in UNC Single
Crystal X-ray Facility report number c99083.
The unit cell and atomic coordinate information
is provided in Table 1.
Differential Scanning Calorimetry (DSC)
Figure 1. Molecular structure of fenoprofen calcium.
A TA 2910S differential scanning calorimeter
equipped with a data station (TA Instruments,
New Castle, DE) was used to determine the ther-
mal behavior of the FC dihydrate. The tempera-
ture axis and the cell constant of the DSC cell
were calibrated with indium. Sample (ꢁ5 mg) in a
Fenoprofen calcium (calcium methyl-3-phenox-
ybenzeneacetate, Figure 1) is a nonsteroidal, anti-
in¯ammatory, analgesic, and antipyretic agent.
3
Fenoprofen calcium (FC) was reported to exist as
4,5
ꢀ
a monohydrate and a dihydrate. Some physico-
crimped aluminum pan was heated at 10 C/min
4
,5
chemical data of those hydrates are available.
under a nitrogen purge at 40 mL/min.
In this study, we describe the use of thermal
analysis and hot stage/environmentally controlled
X-ray diffraction analysis (XRD) to study the dehy-
dration and rehydration behavior of FC. The cry-
stal structure of FC dihydrate is reported here for
the ®rst time and is used to con®rm the stoichio-
metry of water to FC and to provide insight into
the dehydration behavior of FC dihydrate.
Thermogravimetric Analysis (TGA)
The TGA curves were obtained using a TA Hi-Res
2950 thermogravimetric analyzer linked to a data
station (TA Instruments, New Castle, DE). Iso-
thermal TGA was performed on samples (10 mg)
ꢀ
in open aluminum pans at 50, 55, 60, 70, and 80 C
with a nitrogen purge at 60 mL/min.
EXPERIMENTAL SECTION
Materials
Karl Fischer Titrimetry (KFT)
The amounts of water in the FC dihydrate were
determined using a Moisture Meter (Mitsubishi
CA-05, Mitsubishi Chemical Industries Ltd.,
Tokyo, Japan). Sample was weighed and quickly
transferred to the titration vessel containing
anhydrous methanol prior to titration.
FC dihydrate (MW ˆ 558.64 g/mol) was purchased
from the Sigma Chemical Company. The material
was con®rmed as the dihydrate reported pre-
4
,5
viously through physical characterization using
powder X-ray diffraction (PXRD), Karl-Fischer
titrimetry, differential scanning calorimetry
Powder X-ray Diffraction (PXRD)
(DSC), and thermogravimetric analysis (TGA).
The powder X-ray diffractometer and an ``in-
house'' built environmental sample chamber have
been described. For the dehydration study, scans
Preparation of Single Crystals and Structure
Determination of FC Dihydrate
7
were collected at 4 degrees 2y/min at a digital
resolution of 0.03 degrees 2y using copper Ka
The single crystal of FC dihydrate was prepared
by gradually cooling a saturated FC aqueous
ꢀ
radiation while drying the covered sample at 50 C
ꢀ
solution from 60 C to room temperature. Wet
under a stream of dry nitrogen. The sample cover
uses a X-ray transparent window. XRD peak inte-
gration was performed by removing background
and then integrating using the BACKGROUND
and AREA programs, respectively, of Scintag's
DMS software, v. 3.35.
Two series of experiments were conducted to
qualitatively examine the dehydration/rehydra-
tion of FC dihydrate. One series of experiments
utilized a Baxter DP-32 vacuum drying oven to
crystals were submitted to the X-ray Crystal-
lographic Facility at the Chemistry Department,
University of North Carolina (UNC), Chapel Hill,
NC. The 2y and intensity data were collected at
ꢀ
�
100 C using a Bruker SMART diffractometer.
Unit cell dimensions were obtained from 6683
re¯ections. Structure solution and re®nement
was based on 2466 unique and signi®cant re¯ec-
tions (R ˆ 0.062 and R ˆ 0.064). The program
f
w
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 90, NO. 7, JULY 2001

CHARACTERIZATION OF FENOPROFEN CALCIUM
847
a
Table 1. Single-Crystal X-ray Diffraction Unit Cell and Atomic Coordinates
Fraction Atomic Coordinate
z
b
Atom
x
y
Biso
Cal
O1
O2
C3
C4
C5
C6
C7
0.98906 (6)
0.96583 (20)
0.94168 (22)
0.9440 (3)
0.9244 (4)
0.9030 (5)
0.8723 (4)
0.8739 (4)
0.8272 (4)
0.7793 (4)
0.7776 (3)
0.8242 (4)
0.7283 (3)
0.7059 (5)
0.6371 (6)
0.6124 (10)
0.6566 (14)
0.7281 (11)
0.7531 (6)
1.10105 (21)
1.07599 (19)
1.1105 (3)
1.1789 (6)
1.2352 (5)
1.2431 (4)
1.1977 (5)
1.1421 (5)
1.1329 (6)
1.0943 (5)
1.0826 (5)
1.0604 (6)
1.0445 (7)
1.0545 (8)
1.0770 (7)
1.0907 (4)
1.03566 (21)
0.88541 (22)
1.1849
0.74938 (15)
0.5070 (5)
0.7794 (4)
0.6259 (9)
0.5832 (8)
0.7314 (10)
0.4315 (8)
0.3289 (9)
0.1897 (8)
0.1528 (9)
0.2564 (11)
0.3965 (8)
0.2352 (8)
0.0722 (14)
0.0457
� 0.116 (3)
� 0.239 (3)
� 0.2197 (19)
� 0.0584 (18)
0.7213 (5)
0.9950 (5)
0.8677 (8)
0.7275 (10)
0.6200 (11)
0.4748 (9)
0.4376 (9)
0.5490 (11)
0.6917 (11)
0.5298 (10)
0.3692 (16)
0.2359 (22)
0.078 (3)
0.0677 (22)
0.199 (3)
0.3511 (17)
0.8856 (5)
0.6075 (5)
0.8738 (15)
1.0466 (18)
0.8968 (17)
1.0173 (17)
0.542
0.02478 (6)
� 0.07293 (20)
� 0.090053 (21)
� 0.1127 (3)
� 0.1857 (4)
� 0.2280 (4)
� 0.1924 (4)
� 0.2490 (4)
� 0.2574 (4)
� 0.2081 (5)
� 0.1522 (4)
� 0.1436 (4)
� 0.1025 (3)
� 0.0863 (4)
� 0.0801 (5)
� 0.0596 (10)
� 0.0454 (11)
� 0.0499 (6)
� 0.0727 (5)
� 0.02755 (24)
� 0.03653 (21)
� 0.0536 (3)
� 0.1468 (6)
� 0.1353 (4)
� 0.1772 (5)
� 0.2284 (4)
� 0.2391 (5)
� 0.1983 (6)
� 0.2918 (5)
� 0.3228 (6)
� 0.2846 (6)
� 0.3161 (11)
� 0.3839 (12)
� 0.4238 (7)
� 0.3898 (7)
0.12680 (20)
0.0586 (3)
2.71 (4)
3.89 (19)
4.45 (22)
3.5 (3)
5.3 (4)
7.4 (4)
4.2 (3)
4.6 (3)
5.2 (4)
5.5 (4)
5.4 (4)
5.1 (4)
9.2 (4)
6.7 (5)
9.0 (7)
13.5 (14)
16.6 (18)
13.3 (11)
9.1 (6)
4.90 (23)
4.03 (20)
3.9 (3)
8.1 (6)
7.4 (5)
6.0 (4)
5.9 (4)
7.2 (5)
7.9 (6)
11.8 (6)
7.4 (6)
9.8 (7)
12.8 (12)
13.1 (12)
11.3 (10)
8.0 (7)
4.48 (20)
6.8 (3)
2.70 (22)
4.2 (3)
4.0 (3)
3.8 (3)
6.0
C8
C9
C10
C11
O12
C13
C14
C15
C16
C17
C18
O21
O22
C23
C26
C27
C28
C29
C30
C31
O32
C33
C34
C35
C36
C37
C38
O41
O42
C241
C251
C242
C252
H4
� 0.0856 (6)
� 0.1306 (7)
� 0.1195 (7)
� 0.0988 (7)
� 0.206
1.1889 (7)
1.1511 (7)
1.2149 (7)
0.967
0.893
0.861
0.939
0.908
0.828
0.748
H5a
H5b
H5c
H7
H8
H9
H11
H14
H15
H16
H17
0.694
0.779
0.818
0.353
0.121
0.056
0.467
0.139
� 0.274
8.2
8.2
8.2
5.3
5.9
6.3
5.9
9.8
� 0.209
� 0.228
� 0.283
� 0.299
� 0.213
0.822
0.605
0.563
0.637
� 0.103
� 0.090
� 0.136
� 0.057
14.3
17.5
14.1
� 0.348
� 0.032
0.759
� 0.313
� 0.037
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 90, NO. 7, JULY 2001

848
ZHU ET AL.
Table 1. (Continued)
Fraction Atomic Coordinate
z
b
Atom
x
y
Biso
H18
H27
H28
H29
H31
H34
H35
H36
0.803
1.268
1.282
1.204
1.094
1.056
1.027
1.045
1.082
1.108
1.079
1.047
0.853
0.836
1.225
1.237
1.156
1.185
1.122
1.243
1.199
1.245
� 0.040
0.644
0.398
0.339
0.768
0.250
� 0.018
� 0.043
0.188
0.450
0.857
1.006
0.604
0.633
0.856
1.054
1.046
1.145
0.956
1.045
1.126
0.960
� 0.079
� 0.099
� 0.169
� 0.258
� 0.207
� 0.236
� 0.290
� 0.405
� 0.472
� 0.414
0.151
9.9
8.3
6.8
6.8
8.8
10.7
13.6
13.8
12.1
8.8
5.2
5.2
7.7
7.7
3.7
4.7
4.7
4.7
4.8
4.9
4.9
4.9
H37
H38
H41b
H41a
H42a
H42b
H241
H251a
H251b
H251c
H242
H252a
H252b
H252c
0.121
0.096
0.053
� 0.053
� 0.149
� 0.165
� 0.100
� 0.155
� 0.138
� 0.079
� 0.065
a
1
UNC report code c99083; space group and cell dimensions: monoclinic, P2 /n;
Ê
ꢀ
a ˆ 18.9996(8), b ˆ 7.7270(3), c ˆ 19.4811(9) A; beta ˆ 91.530(1) ; unit cell volume ˆ 2859.00
Ê
(
(
21) A cubed; FW ˆ 558.64 g/mol; Z (# molecules in unit cell) ˆ 4; calculated density ˆ 1.298 g/cc.
ESDs refer to the last digit printed)
b
Biso is the mean of the principal axes of the thermal ellipsoid.
dehydrate the FC dihydrate and an Espec LHU-
drated, as con®rmed by TGA. The hot stage was
lowered to 40 C, and a RH100 humidity generator
ꢀ
ꢀ
1
12 humidity cabinet set at 40 C/75% relative
humidity (RH) to rehydrate the material. FC
dihydrate was mounted on a silicon zero back-
ground wafer, and the initial PXRD pattern was
collected. The sample was then placed in the
(VTI Corp., Hialeah, FL) was utilized to produce a
75% RH atmosphere in the chamber. The humi-
di®er on the RH100 was set at 55 C, and the
ꢀ
heated transfer line between the generator and
ꢀ
ꢀ
vacuum drying oven and dried at 70 C under
environmental chamber was set at 45 C. The ¯ow
vacuum. A second PXRD pattern was collected
of the dehydrated material. The sample was
placed in the humidity cabinet and periodically
removed and scanned. Visual observations were
also recorded.
rate of moist nitrogen from the RH100 was 350
mL/min. A heat lamp equipped with a dimmer
helped maintain the temperature inside the
chamber at 40 C. PXRD scans were collected
periodically.
ꢀ
A second series of dehydration/rehydration
experiments were performed in situ using a hot
stage/environmental chamber designed for the
Cerius2 Modeling
7
Scintag XDS2000. FC dihydrate was mounted on
a silicon wafer and placed in the hot stage/envi-
ronmental chamber. An initial PXRD pattern was
Single-crystal structure data were imported into
Cerius2 (Molecular Simulations, Inc.), v. 3.9 to
provide visualization of the crystal structure and
calculation of the simulated PXRD pattern. For
atomic charge assignments and energy minimiza-
tions, the Charge Equilibrium method and Dreid-
ing v. 2.21 and Universal v. 1.02 force ®elds were
used, respectively.
ꢀ
recorded and then the hot stage was set to 70 C
and a dry stream of nitrogen (350 mL/min) was
blown into the environmental chamber. Scans
were collected continuously until the diffraction
patterns indicated the dihydrate had fully dehy-
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 90, NO. 7, JULY 2001

CHARACTERIZATION OF FENOPROFEN CALCIUM
849
RESULTS AND DISCUSSION
Typical isothermal dehydration plots are
shown in Figure 2. In all temperatures, the shape
of the curve is sigmoidal and the dehydration rate
increases with temperature. The dehydration
curves were ®tted by various rate equations (see
Table 2). The dehydration kinetics was best des-
cribed by the three-dimensional growth of nuclei
equation (eq. A3) or the three-dimensional phase
boundary equation (eq. R3). The activation ener-
gies evaluated by graphical solutions of Arrhenius
equation using the rate constant calculated from
either equation was ꢁ309 kJ/mol for the tempera-
Dehydration Kinetics by Isothermal TGA
The weight loss, expressed as fraction dehy-
drated, x, was obtained at ®xed temperatures
ꢀ
covering the range 50 to 80 C maximum possible
weight loss of water in dehydrating the fenopro-
fen calcium. Therefore, 100% fraction dehydrated
(
x ˆ 1) corresponds to the loss of two water
molecules of hydration. Prior to each isothermal
scan, the TGA furnace was rapidly heated to the
required study temperature and the samples were
maintained at that temperature until dehydra-
tion was complete. The fraction dehydrated, x,
was plotted as a function of time, t, according to
the kinetic models of the reaction mechanisms
ꢀ
ture range 50±60 C and 123 kJ/mol for the temp-
ꢀ
erature range 60±80 C (Table 3 and Figure 3).
Because the sample presentation, surface area,
bulk density, particle size, ¯ow rate, and RH of
the drying stream, etc. may affect dehydration
rate, it is not surprising that the rate constants
obtained in this work do not agree with the results
8
±10
known to occur in the solid state.
The
2
correlation coef®cient squared, r , was deter-
mined to test the extent of conformity of the data
to the kinetic models. The activation energies of
dehydration were determined from Arrhenius
plots of the rate constant derived from the
statistically best ®tting mechanism.
4
� 1
by Sathe and Moore (i.e., 0.02 versus 0.04 h at
ꢀ
50 C, respectively). Interestingly, even though
the prior work of Sathe and Moore did not ment-
ion two possible dehydration modes, their Arrhe-
nius plot (natural logarithm of the dehydration
Figure 2. Plot of fraction dehydrated against time for the dehydration of FC
dihydrate under isothermal conditions.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 90, NO. 7, JULY 2001

8
50
ZHU ET AL.
8
Table 2. Kinetic Equations of the Most Common Mechanism of Solid-State Decompositions and Correlation
2
Coef®cients Squared (r ) of Isothermal TGA Data Fitting
2
Symbol
Equation
Rate-Controlling Process
r
P1
A2
A3
F1
R1
ln[x/(1 � x)] ˆ kt
Random nucleation (Prout±Tompkins equation)
Two-dimensional growth of nuclei (Avrami±Erofeev, n ˆ 1/2)
Three-dimensional growth of nuclei (Avrami±Erofeev, n ˆ 1/3)
Random nucleation (®rst-order mechanism)
One-dimensional phase boundary reaction (zero-order
mechanism)
Two-dimensional phase boundary reaction (zero-order
mechanism)
Three-dimensional phase boundary reaction (spherical
symmetry)
0.9620
0.9764
0.9915
0.9439
0.8102
1
/2
[ � ln(1 � x)] ˆ kt
1
/3
[ � ln(1 � x)] ˆ kt
� ln (1 � x) ˆ kt
x ˆ kt
1
/2
R2
R3
[1 � (1 � x)] ˆ kt
0.9788
0.9919
1
/3
[1 � (1 � x)] ˆ kt
2
D1
D2
D3
x ˆ kt
One-dimensional diffusion
Two-dimensional diffusion
Three-dimensional diffusion (Jander equation)
Three-dimensional diffusion (Ginstling±Brounshtein equation)
0.9388
0.9568
0.9126
0.7992
(1 � x) ln(1 � x) ‡ x ˆ kt
1
/3 2
[1 � (1 � x) ] ˆ kt
2
/3
D4
1 � (2/3)x � (1 � x)
ˆ kt
rate constant versus the inverse of the absolute
temperature) suggests a break that may indicate
two ``modes'' of dehydration process as pointed out
in this paper. The calculated activation energies
the formation of a partially crystalline phase. The
simply ordered phase is not purely amorphous
because of the presence of a strong diffraction
peak, as well as some minor peaks (Figure 4).
After the diffraction peaks of the crystalline
dihydrate were not detectable, the drying experi-
ment was stopped and the residual moisture in
the dried sample on the XRD environmental stage
was immediately determined by KFT. The sample
contained ꢁ3.8 % water, which corresponds to a
FC monohydrate.
The PXRD pattern of a crystalline powder is
characteristic of the crystal lattice of that parti-
cular compound, including a hydrate. By measur-
ing the rate of disappearance (or appearance) of a
peak unique to the hydrate (or dehydrated form),
the kinetics of the solid-state dehydration can be
determined. Consider a mixture of FC dihydrate
4
from the data by Sathe and Moore at two differ-
ꢀ
ent temperature ranges 50±60 and 60±80 C are
345 and 135 kJ/mol, respectively, and are compar-
able with the data in this work. The difference in
the activation energies for the two temperature
ranges is due to the different water environments
in FC dihydrate, as con®rmed by the single-
crystal XRD analysis.
Dehydration Kinetics by PXRD
Figure 4 shows the PXRD pattern of FC dihydrate
ꢀ
over time when exposed to 50 C. Over a period of a
few hours, the diffraction peaks for the crystalline
dihydrate are reduced in intensity followed by the
(
A) and monohydrate (B). The intensity of PXRD
ꢀ
growth of a low angle diffraction peak (5 2y).
peak of the ith diffraction line of the dihydrate (A)
component is given in eq. 1:
Qualitatively, the dehydration seems to follow the
loss of the crystalline dihydrate phase followed by
Ã
IiA ˆ ꢂk X †=ꢀ
ꢂ1†
Table 3. Dehydration Rate Constants by Fitting the
Isothermal TGA Data to the Three-Dimensional Phase
Boundary Equation
iA
A
mat
where IiA is the intensity of the ith diffraction
peak of component A, X is the weight fraction of
Dehydration Rate
Constant (k, min
A
ꢀ
�
1
Ã
Temperature ( C)
)
component A, m
mat
is the mass absorption coef®-
cient for the total matrix (A ‡ B), k is a constant
iA
5
5
6
7
8
0
5
0
0
0
0.0003
0.0013
0.0090
0.0422
0.1129
containing instrumental factors, atomic factors,
and the density of A. The value of kiA is dependent
on the chosen diffraction peak.
Similarly, the intensity of the X-ray powder
diffraction peak of the jth diffraction line of the
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 90, NO. 7, JULY 2001

CHARACTERIZATION OF FENOPROFEN CALCIUM
851
Figure 3. Arrhenius plot of the rate constants obtained for the dehydration of FC
dihydrate.
Ã
dehydrated (B) component is given in eq. 2:
errors in PXRD peak integration to consider m
mat
constant during the dehydration to the monohy-
Ã
Ã
IjB ˆ ꢂk X †=ꢀ
ꢂ2†
drate. With a constant m , the need for an
jB
B
mat
mat
internal standard is avoided. With a single
sample preparation and the dehydration per-
formed at the X-ray diffractometer, errors due to
changes in preferred orientation during drying
are eliminated, minimizing the need for any met-
hod validation.
Assuming X ‡ X ˆ 1, from the ratio of eqs. 1 and
A
B
2
, the weight fraction of the dihydrate, X , is
A
solved as in eq. 3:
X ˆ I =ꢂI ‡ I ꢂk =k ††
ꢂ3†
A
iA
iA
jB iA
jB
Using the strongest diffraction peaks for the
crystalline dihydrate (a split peak centered aro-
ꢀ
De®ne I as the intensity of phase A when X ˆ 1
ꢀ
ꢀ
A
A
und 6.2 2y) and the simply ordered monohydrate
(
i.e., the FC dihydrate main peak prior to drying)
FC (5.0 2y), the weight fraction of the dihydrate
ꢀ
and I as the intensity of phase B when X ˆ 1
B
B
was determined versus time of drying. Figure 5
(
i.e., no detectable FC dihydrate peak) and assume
shows the plot of the weight fraction of the dihy-
Ã
remains constant; then k /k ˆ IꢀA/I B,
ꢀ
that m
ꢀ
mat
iA jB
drate versus time at 50 C. The integrated peak
and eq. 3 can be represented as in eq. 4:
data used to calculate Figure 5 is provided in
Table 4.
X ˆ I =ꢂI ‡ I ꢂIꢀA=I
ꢀB††
ꢂ4†
A
iA
iA
jB
Qualitatively, Figure 4 suggests that the dehy-
ꢀ
dration of the dihydrate at 50 C produces a two-
The mass absorption coef®cient of the sample
matrix will change as the sample is dried and
unbound water is evaporated. However, the calcu-
lated mass absorption coef®cients for the dihy-
drate and the monohydrate (using the atom mass
phase system comprised of a mixture of the
crystalline dihydrate and the partially crystalline
monohydrate. If the dehydration process involved
a single-phase system where the crystalline
material uniformly became more disordered with
increased water loss, one should observe one set of
diffraction peaks gradually shifting from their
1
1
2
absorption coef®cients ), 17.6 and 17.8 cm /g,
respectively, are suf®ciently close relative to the
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 90, NO. 7, JULY 2001

8
52
ZHU ET AL.
Figure 4. The powder diffraction pattern of FC dihydrate over time when exposed to
ꢀ
5
0 C.
ꢀ
original position to their ®nal position after dry-
ing rather than observing (as we do) the separate
dihydrate and the monohydrate peaks. In addi-
and 10 2y that were not observed in the prior
work; the absence of these peaks is probably
due to sensitivity differences between the PXRD
instruments.
ꢀ
tion to the diffraction peaks in the 3±11 2y range
for the monohydrate, a weak broad ``halo'' or
hump in the baseline is observed around the 16±
Dehydration/Rehydration Study Using PXRD
ꢀ
22 2y range. The weak broad ``halo'' represents
an amorphous aspect of the monohydrate com-
pared with that observed by the strong low-angle
diffraction peak.
Duplicate dehydration/rehydration experiments
were performed using the oven and humidity
cabinet. Dehydration of the FC dihydrate to the
anhydrate was complete after 40±45 min in the
ꢀ
Further drying at 80 C produced an FC anhy-
ꢀ
drate, as con®rmed by TGA, with a similar pat-
tern as the monohydrate obtained by drying at
vacuum oven at 70 C, as indicated by the PXRD
pattern (Figure 7). The appearance of the FC was
changed little during the dehydration process.
Both the dihydrate and the anhydrate consisted of
white granular particles.
ꢀ
ꢀ
5
0 C. Additional drying at 120 C produced little
change except for a slight sharpening of the diffr-
action peaks and slight shifts to higher d-spacings
ꢀ
(
lower 2y), which were probably due to decreased
Exposing the anhydrate to 40 C/75% RH ini-
unit cell density with increased temperature (see
Figure 6). The PXRD pattern for the anhydrate in
Figure 6 is consistent with the pattern for the
tially caused the material to become more dis-
ꢀ
ordered (Figure 7). After ꢁ20±30 min at 40 C/
75% RH, the sample became a clear, liquescent
4
ꢀ
anhydrous FC presented in an earlier work.
mass. As the material aged at 40 C/75% RH, the
There are weak peaks in Figure 6 between 8
formation of the dihydrate crystalline phase was
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 90, NO. 7, JULY 2001

CHARACTERIZATION OF FENOPROFEN CALCIUM
853
Table 4. Integrated XRD Peak Intensities
ꢀ
ꢀ
Form A (6.2
Split Peak)
Form B (5.0
Peak)
Drying
Dihydrate
Monohydrate
time (min)
(Arbitrary Units)
(Arbitrary Units)
0
80830
51492
33694
20633
13285
8934
7076
3784
1727
0
0
17400
33938
46130
55331
59547
65602
66422
70896
69053
1
2
3
4
5
6
7
8
9
1
1
2
2
3
3
4
4
4
Figure 5. Plot of % crystalline FC dihydrate versus
ꢀ
time when exposed to 50 C by quantitative PXRD
analysis.
readily apparent both visually and by PXRD
A second experiment using the same methodol-
(
Figure 7). The presence of an intermediate lique-
ogy produced similar results, however the crystal-
lization of the dihydrate occurred more rapidly
than in the ®rst experiment. After 55 min at 40 C/
75% RH, crystallization of the dihydrate appeared
to be complete by PXRD.
scent phase prior to crystallizing of the dihydrate
suggests a solution-mediated rehydration process.
Crystallization of the dihydrate appeared essen-
ꢀ
ꢀ
tially complete after 190 min at 40 C/75% RH.
ꢀ
Figure 6. PXRD patterns of FC dihydrate, monohydrate obtained by drying at 50 C,
ꢀ
ꢀ
anhydrate obtained by drying at 80 C, and anhydrate with further heating at 120 C.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 90, NO. 7, JULY 2001

8
54
ZHU ET AL.
Figure 7. PXRD patterns showing the dehydration and rehydration of the FC when
exposed to different conditions.
Duplicate dehydration/rehydration experi-
Single Crystal Results
ments were also performed using the hot stage/
environmental chamber. For both experiments,
dehydration of the dihydrate was complete after
The asymmetric unit for FC dihydrate is shown in
Figure 8. The compound crystallized in a mono-
1
0 min in the drying phase. Rehydration of the FC
clinic lattice with a P2 /n space group (Table 1).
1
occurred much slower in the environmental cham-
ber than it did in the humidity cabinet. The ®rst
experiment showed no evidence of the crystalline
The crystal structure is racemic with one R-
enantiomer and one S-enantiomer of fenoprofen
anion and two molecules of water per divalent
calcium atom. In the asymmetric unit, one
fenoprofen anion exhibited disorder, which was
modeled by the crystallographer by two differ-
ent atomic coordinates at 50% occupancy each for
ꢀ
dihydrate after exposure to 40 C/75% RH for
ꢁ
14.5 h. Some crystalline dihydrate was observed
during the second experiment, but only after
ꢀ
ꢁ
30 h at 40 C/75% RH. The experiment was
carried out to 41.5 h, and crystallization of the
dihydrate never went to completion. The observed
difference in the rate of recrystallization between
the environmental cabinet and the PXRD envir-
onmental chamber may be a consequence of static
versus dynamic moisture introduction or due to
differences in sample mass, geometry, or ¯ow rate
of the gas stream.
both the CH and CH carbons. For viewing simp-
3
licity, only one set of carbon atoms is shown in
Figure 8. Unlike the case for the reported stru-
1
2
cture of cromolyn sodium hydrates, where the
disorder in one of the sodium ions may contribute
to the disorder and variable hydration nature,
neither the calcium ions nor water molecules in
our crystal structure are disordered. Except for
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 90, NO. 7, JULY 2001

CHARACTERIZATION OF FENOPROFEN CALCIUM
855
structure analysis, determining the extent of hy-
drogen bonding of water with the fenoprofen is
dif®cult. Using the ``ADD HYDROGENS'' feature
of Cerius2, hydrogens were added to each oxygen
atom corresponding to water molecules at an OH
Ê
bond distance of 0.85 A. The positions of the
inserted hydrogen atoms were then re®ned by two
approaches. One approach involved assigning
atom charges using the Charge Equilibrium
method in Cerius2 and then minimizing the
lattice energy with the Dreiding force®eld, v.
2.21, with the constraint that only the inserted
hydrogen positions could change. The other atom
positions and unit cell dimensions remained ®xed.
After minimization, the hydrogen bonds were
calculated using the ``CALCULATE H-BONDS''
feature of Cerius2. In the other approach, the
orientation of the hydrogen atoms were varied
using the ``3D SKETCHER'' tool to establish
optimum hydrogen bonding patterns (using the
``CALCULATE H-BONDS'' tool) between the
water molecules and the carboxylate oxygens.
Using the energy minimization approach, one
water molecule was involved in a hydrogen bond
between a water hydrogen and a carboxylate oxy-
gen and the other water molecule was not invol-
ved in any hydrogen bonding. Using the "manual"
approach to optimize H-bonding, both water mole-
cules appear to have similar hydrogen bonding
opportunities (i.e., the strength of the hydrogen
bonding should be similar). Thus, while the ®rst
approach suggests that one water molecule may
be more tightly bound within the crystal structure
due to limited hydrogen bonding, this suggestion
is not con®rmed by the ``manual'' approach. Beca-
use of the limitations in the Dreiding force ®eld,
the reliance on molecular mechanics calculations
for optimizing hydrogen positions can be useful
but not absolute. Ideally, the hydrogen positions
should be re®ned from single-crystal X-ray or
neutron diffraction. In addition, the distance bet-
ween the water oxygen atoms and the calcium
ions is about the same so that the extent of ele-
ctrostatic attraction between the calcium cations
and the electronegative water oxygens should be
equivalent.
Figure 8. The asymmetric unit for fenoprofen cal-
cium dihydrate. The oxygen atoms of the water mole-
cules are represented by black balls.
the hydrogens on one CHCH group, no hydro-
3
gens are shown because they were not re®ned in
the structure. Thus, the water molecules are
represented by their oxygen atoms only. Sur-
rounding each calcium ion are six oxygens with a
Ê
calcium±oxygen distance of ꢁ2.3±2.4 A. Two oxy-
gens are provided by the water molecules and the
other four oxygens come from four separate feno-
profen ions, as shown in Figure 9. Thus the two
oxygens in any one carboxylate group coordinate
to different calcium ions. In addition to the six
Ê
oxygens within ꢁ2.4 A from a central calcium ion,
Ê
there is a seventh oxygen from a FC group at 2.7 A.
A search through the Cambridge Structural
Database of calcium-containing compounds indi-
cated that calcium is coordinated with 3 to 8 or 10
neighboring oxygens or nitrogens, with coordina-
tions of 6 and particularly 8 being very common. If
one water molecule per calcium ion is removed by
drying, the stability of the hexacoordination state
is lost producing a less ordered (less crystalline)
material.
The one signi®cant observable difference bet-
ween the water molecules is their location. One
water molecule (labeled by its oxygen atom as
O42) is located between repeating carboxylate
groups. The oxygen O42 atoms are represented as
red balls in Figure 10. The other water molecule
(labeled by its oxygen atom as O41) is located
above/below the carboxylate groups, as shown as
Because the hydrogen atoms of the water mole-
cules were not re®ned during the single-crystal
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 90, NO. 7, JULY 2001

8
56
ZHU ET AL.
Figure 9. Coordination of calcium with oxygen atoms from four separate fenoprofen
anions (O1, O2, O21, O22) and two distinct water molecules (O41, O42). Ca±O distances
Ê
(
A) denoted in blue.
red balls in Figure 11. Visually, free pathways out
of the crystal seem to exist along the b and c axes
for the O41 water molecule and along the a and c
axes for the O42 water molecule. However, using
the Connolly Surfaces module in Cerius2 to cal-
culate solvent channels, no such channels large
enough to ®t water molecules are observed, sugge-
sting there is insuf®cient room for the dihydrate
crystal structure to lose water without disrupting
the crystal. At elevated temperatures, the crystal
unit cell may expand allowing egress of water.
Thus, during drying, water at one crystal-
lographically distinct position is probably prefer-
entially removed relative to water at the second
position. Based on the hydrogen bonding calcula-
tions of the energy minimized, the dihydrate
structure as well as the location of the O42 water
lying in between repeating polar carboxylate
groups, the O41 water, located in a slightly less
polar region of the crystal, is probably preferen-
tially lost during drying. However, it takes some
energy to remove this water because of the
electrostatic attraction between the calcium ions
and water's oxygen atom where the lattice favors
the hexacoordination sphere for calcium.
Based on the crystal structure analysis, the
presence of two ``modes'' of dehydration process as
suggested by PXRD and thermal analysis data is
con®rmed. One water is lost from a crystalline
matrix followed by loss of a second water from a
partially crystalline phase.
Figure 12 shows a staggered overlay of the
experimental with the simulated PXRD pattern
(the latter calculated from the crystal structure).
Indexing of the experimental PXRD pattern as
well as Rietveld re®nement (DBWS) of the crystal
structure indicated there as an approximate � 0.1
degree 2y error in the experimental PXRD pat-
tern, probably due to negative sample displace-
ment error. The experimental PXRD pattern has
been offset to compensate for the 2y error in
Figure 12. Rietveld re®nement indicated that the
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 90, NO. 7, JULY 2001

Figure 10. The O42 oxygen atom from water is represented by a red ball. Other
oxygens, carbons, and hydrogens (not all shown) are in red, gray, and white colors,
respectively. The view is down the b crystal axis.
Figure 11. The O41 oxygen atom from water is represented by a red ball. Other
oxygens, carbons, and hydrogens (not all shown) are in red, gray, and white colors,
respectively. The view is down the b crystal axis.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 90, NO. 7, JULY 2001

8
58
ZHU ET AL.
CONCLUSIONS
Arrhenius plot (natural logarithm of the dehydra-
tion rate constant versus the inverse of the
absolute temperature) from isothermal TGA
suggests a break that indicates two ``modes'' of
dehydration process for FC dihydrate. Using
quantitative PXRD, the loss of only 1 mol of water
per mole of FC dihydrate was required for the
material to convert to a partially crystalline
phase. Both water molecules are required to
maintain the original crystalline structure. Pre-
sumably the ®rst loss of water starts from a
crystalline state followed by loss of the second
water from a partially crystalline state. Single-
crystal XRD analysis con®rmed the dihydrate
stoichiometry and revealed the crystal packing
arrangement in FC. The oxygens from the feno-
profen anions and from the water molecules
served to ®ll the hexacoordination sphere for
calcium. Analysis of the structure data indicated
the two water molecules were inequivalent, with
one water located in between repeating carbox-
ylate groups. Analysis of the relative amounts of
hydrogen bonding for the two water molecules is
inconclusive but suggests that the water located
in between the carboxylate groups is in a better
position for hydrogen bonding to the carboxylate
moiety compared with the other water molecule.
The rehydration process is not the reverse of the
dehydration process. The rehydration of the
anhydrous FC goes through a solution-mediated
process prior to crystallizing as the dihydrate.
Figure 12. Overlay of experimental (top) and simu-
lated (bottom) XRD patterns.
ꢀ
unit cell dimensions at � 100 C (determined by
single-crystal XRD) and room temperature (from
PXRD) are the same within experimental error.
The crystal structure of the dihydrate was
``computationally'' dehydrated by ®rst adding
missing hydrogens and minimizing the crystal
structure while maintaining the unit cell para-
meters and other atom positions, as discussed
previously. From this ``re®ned'' structure, one
water molecule was removed from the crystal
structure (to make the monohydrate), and the
resulting structure energy was minimized with no
atom and unit cell constraints. Both Dreiding and
Universal force ®elds were tried. The process was
repeated by removing the other water molecule
from the dihydrate crystal structure to make
another monohydrate. Finally, both waters were
removed (fully dehydrated structure) and the
dehydrated structure minimized. The PXRD
patterns were calculated for all minimized struc-
tures. Because energy minimization will not
ACKNOWLEDGMENTS
The authors thank Sean Lynn, Glaxo Wellcome
Research and Development (Stevenage, UK), for
performing the coordination number search of
calcium compounds in the Cambridge Structural
Database and Peter White, X-ray Crystallogra-
phic Facility, Chemistry Department, UNC, for
providing the crystal structure of calcium feno-
profen. The authors also thank one of the revie-
wers for helpful comments regarding comparisons
of our work with that of prior authors (Sathe and
Moore) as well as all of the reviewers for com-
ments that help strengthen the manuscript.
``convert'' a crystalline structure to a partially cry-
stalline structure, it was not surprising that none
of the calculated PXRD patterns were similar to
the experimental patterns for the partially dried
monohydrate or fully dried anhydrous material
monohydrate. Also, the anhydrates are not dehy-
drated hydrates but are simply ordered structures
with possible remnants of the original crystalline
dihydrate structure.
REFERENCES
1. Berge SM, Bighley SD, Monkhouse DC. 1977.
Pharmaceutical salts. J Pharm Sci 66:1±19.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 90, NO. 7, JULY 2001

CHARACTERIZATION OF FENOPROFEN CALCIUM
859
2
. Forbes RT, York P, Fawcett V, Shields L. 1992.
Physicochemical properties of salts of p-aminosa-
licylic acid. I. Correlation of crystal structure and
hydrate stability. Pharm Res 9:1428±1435.
. Hirsch CA, Messenger RJ, Brannon JL. 1978.
Fenoprofen: Drug form selection and preformula-
tion stability studies. J Pharm Sci 67:231±236.
. Sathe R, Moore S. 1995. Dehydration and hydra-
tion kinetics, phase change, solubility and dissolu-
tion behavior of fenoprofen calcium. Drug Dev Ind
Pharm 21(16):1887±1894.
7. Buckner C, Kidd C, Long S, Varlashkin P, Zhu H.
1999. Environmentally controlled sample stage for
Scintag's six-position XRD sample changer. Powder
Diffraction 14(4):296±299.
8. Sharp JH, Brindley GW, Achar BNN. 1966. Nume-
rical data for some commonly used solid state re-
action equations. J Am Ceram Soc 49:379±382.
9. Byrn SR. 1982. Solid state chemistry of drugs. New
York: Academic Press, Inc.
10. Monkhouse DC, Van Campen L. 1984. Solid state
reactions- Theoretical and experimental aspects.
Drug Dev Ind Pharm 10:1175±1276.
11. Cullity BD. 1978. Elements of X-Ray Diffraction,
Appendix 8. Reading, MA: Addison-Wesley Pub-
lishing Company, Inc.
3
4
5
6
. Florey K, editor. 1977. Analytical Pro®les of Drug
Substances and Excipients, Vol. 6. New York: Aca-
demic Press, pp. 161±182.
. Gabe EJ, Le Page Y, Charland J-P, Lee FL, White PS.
1
989. NRCVAX an interactive program system for
12. Chen LR, Young Jr. VG, Lechuga-Ballesteros D,
Grant DJW. 1999. J Pharm Sci 88:1191±1200.
structure analysis. J Appl Crystallogr. 22:384±387.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 90, NO. 7, JULY 2001