Infrared Studies of The Thermal Conversion of Mefenamic Acid Between Polymorphic States

You might also like

Download as docx, pdf, or txt
Download as docx, pdf, or txt
You are on page 1of 9

Vibrational Spectroscopy 37 (2005) 53–59

www.elsevier.com/locate/vibspec

Infrared studies of the thermal conversion of mefenamic


acid between polymorphic states
R.K. Gilpin*, W. Zhou
Brehm Research Laboratories, College of Science and Mathematics, Wright State University, Dayton, OH 45435, USA

Received 7 April 2004; received in revised form 2 June 2004; accepted 3 June 2004
Available online 10 December 2004

Abstract

An infrared method has been developed and used to study the thermal conversion of mefenamic acid from its polymorphic I form to the
polymorphic II form. Rates of conversion for the crystal to crystal transition have been measured at temperatures of 150, 155, and 160 8C
and subsequently used to calculate the activation energy for the process. The value of 71.6 kcal/mol obtained using the current infrared
method is
significantly smaller than a previously reported value of 86.4 kcal/mol that was determined by differential scanning calorimetry. In order to
explain this difference, additional HPLC experiments were carried out to evaluate the potential loss of analyte due to sublimation when
samples of mefenamic acid are maintained at elevated temperatures in the DSC pans prior to measuring the exothermic event indicative of
the polymorph I content of a sample. Based on the current study, the rate of anaylte loss due to sublimation is the likely mechanism
explaining the larger reported value for the activation energy of the polymorphic conversion of mefenamic acid obtained by DSC.
# 2004 Elsevier B.V. All rights reserved.

Keywords: Infrared spectroscopy; Polymorph; Thermal conversion; Mefenamic acid; HPLC

1. Introduction determining bioavailability. Likewise, crystallographic


measurements of fenamates, including mefenamic acid,
Mefenamic acid is a potent prostaglandin synthetase and their complexes have shown that they share a common
inhibitor that is used widely as a non-steroidal anti- and invariant feature [10–15]. The carboxyl group, the ring
inflammatory and analgesic–antipyretic drug [1–3]. Since containing it, and the bridging amino group are all coplanar
its introduction, there have been numerous manuscripts resulting from resonance interactions and internal hydrogen
published that discuss various aspects of the compound bonding between the NH and the carboxyl group (Fig. 1a).
including important structural and physical properties [4– The conformational properties of fenamates also have been
6], as well as, numerous individual analytical studies and considered in terms of semiempirical structural
quantitative methods. An overview of this work may be considerations using AM1 and PM3 molecular orbital
found elsewhere [7–9] and in other similar past biannual methods and the computational methods produce structures
reviews in the series. consistent with Fig. 1 [5].
Mefenamic acid is a highly insoluble compound in water Another of the important physical properties of
with about 40 and 80 mg/ml dissolving at 25 and 37 8C, mefenamic acid is that it exists in two polymorphic forms
respectively at pH 7.1 [4]. Likewise, it is only slightly that differ significantly in solubility and conversion
soluble in ethanol. As such, the required liquid to dissolve between the two forms is measurable by differential
a typical 250 mg dose of the compound is over 6 l [6].
scanning calorimetry as an exothermic event in the 170–
Because
190 8C range depending on the solvent used for
of this, it has been suggested that solubility is a key factor in
crystallization [4]. In
addition to the DSC and extensive solubility
* Corresponding author. Tel.: +1 937 775 2691; fax: +1 937 775 3122. measurements, infrared measurements also were carried
E-mail address: roger.gilpin@wright.edu (R.K. Gilpin).
out in this same study. Although it was noted that the NH
0924-2031/$ – see front matter # 2004 Elsevier B.V. All rights reserved. stretching band in
doi:10.1016/j.vibspec.2004.06.001
54 R.K. Gilpin, W. Zhou / Vibrational Spectroscopy 37 (2005) 53–59

HO O
H
CH3
OOH
HCH3
Grade potassium bromide for preparing the infrared pellets
N
N CH3
were from Fisher Scientific (Pittsburgh, PA). The
CH3
mefenamic acid was obtained from Sigma (St. Louis, MO).
(a)
(b)
The deionized water was prepared in-house with a
Labconco water deionization system (Kansas, MO).
Fig. 1. Conformational changes in mefenamic acid between polymorph I the IR
and II.

the infrared shifted from 3313 cm —1 for polymorph I (i.e.,


the initially crystallized form) to 3347 cm —1 for polymorph
II (i.e., produced via heating) little additional information
was provided in terms of discussing these spectral
differences as they relate to structural changes between
the two polymorphic forms of mefenamic acid. Infrared
spectroscopy also has been used to study the Fe(III)
complexes [16] and the sodium and calcium complexes
[17] of mefenamic acid and in the latter instance, it was
noted that the NH bands did not change, which is
inconsistent with the spectra that appear in the manuscript.
Moreover, this drug is not stable and products of its
decomposition can enhance undesirable effects. However,
in formulations, cyclodextrins can be used to increase both
the stability and solubility of mefenamic acid [18].
In another account, the kinetics and activation energy
for conversion of mefenamic acid between polymorphic
forms was studied by differential scanning calorimetry
[19]. After maintaining samples at temperatures between
140 and 150 8C in the DSC pans for varying periods of
time, scans
were carried out to quantitate the amount of remaining
polymorph I. Using this experimental approach, the
kinetics of the crystal to crystal transition were reported to
proceed to half completion in less than an hour for the
sample maintained at 140 8C to less than 10 min for the
150 8C sample. Further, the activation energy for the
polymorphic
conversion was reported to be 86.4 kcal/mol.
Based on the observations discussed above, an infrared
study was carried out to examine the feasibility of using
changes in the infrared band frequency and shape to
elucidate structural differences between the polymorph I
and II forms of mefenamic acid. During the course of the
work, it was noted that the kinetics of conversion were
significantly slower than previously reported [19] and that
the activation energy determined by the current infrared
method was about 15 kcal/mol smaller than that determined
by DSC. In order to reconcile these differences, HPLC
experiments were carried out to evaluate sublimation losses
as a possible cause of the previously reported, more rapid
conversion kinetics and higher activation energy measured
by DSC.

2. Experimental

2.1. Chemical and reagents

The HPLC grade acetonitrile and methanol, as well as


the ACS grade reagents used for buffer preparation and
R.K. Gilpin, W. Zhou / Vibrational Spectroscopy 37 (2005) 53–59 55

2.2. Infrared equipment and procedures

All infrared measurements were carried out on a


Bomem (Quebec City, Que., Canada) model DA-8 high
resolution FT-IR spectrometer, equipped with a globar
source, KBr beam splitter and MCT detector. Samples of
mefenamic acid were ground with IR Grade potassium
bromide at weight ratios of 1:20, pressed at 6000 psi into
pellets and mounted between two 13 mm 2 mm KBr ×
windows (ThermoSpec- tra-Tech, Madison, WI). Spectra
were an average of either 64 or 128 scans and were
collected in the vacuum mode. Interferograms were
collected at a 5.0 mm aperture,
1.00 cm/s mirror speed, 0.5 cm—1 resolution, and trans-
formed using boxcar appodization. Band shape analysis
was carried out using the Bomem-Grams software package
which includes baseline correction, spectral smoothing,
and curve fitting algorithms.
The thermal conversion studies of mefenamic acid
from polymorph I to II were carried out at 150, 155, and
160 8C using the following procedure. Samples of
mefenamic acid (300 mg) were weighed into small glass
bottles, the openings of the bottles covered loosely with
aluminum
foil, and then placed into a constant temperature oven (i.e.,
a GC oven). They were maintained at elevated
temperatures for varying periods up to three days. During
this time, bottles were removed from the oven at specified
times, cooled to room temperature, and the mefenamic
acid blended with KBr as described above.
The standard addition experiments were carried out as
follows. Pure polymorph II was prepared by heating the
original as received mefenamic acid for 48 h at 160 8C and
then verifying the identity and purity of the product that
formed by infrared spectroscopy (i.e., the absence of any
traces of an NH stretch band at 3312 cm—1 and only a band
at 3346 cm—1). Additional details about this process are
presented in the Results and Discussion section. Subse-
quently, varying amounts of pure polymorph II and
mefenamic acid, as it was received from Sigma (i.e., about
90% polymorph I), were weighed out and blended
together with KBr as described above to produce pellets
with accurately known standard addition compositions.
In curve fitting the individual bands within a data set,
an iterative process was used that involved initially
allowing the instrument’s software to pick both the
frequency and width of the individual spectral
components. Subsequently, the average values for the
frequency and width were calculated and these values held
constant and new fits calculated across all data within an
individual temperature. In most cases the r2 values of the
spectral fits were 0.997 or better and in all cases they were
at least 0.993.
2.3. Chromatographic equipment and procedures

The liquid chromatographic experiments were carried


out using a Varian (Walnut Creek, CA) model 9002
isocratic pump, Spectra Physics (San Jose, CA) model
Spectra Focus detector set at 254 nm, and a model 4400
Chromjet integrator. The samples were injected using a
Rheodyne (Berkely, CA) model 7125 injection valve
equipped with a 20 ml sample loop. The separations were
performed on a 4.6 mm
i.d. ×250 mm Hypersil ODS 5 mm column (Supelco,
Bellefonte, PA) using acetonitrile-ammonium acetate buffer
(pH 4.0; 10 mM) (65:35, v/v). Quantitation was carried out
using six point calibration curves that were linear (i.e.,
r2 > 0.99) over the range studied. All samples were
injected at least three times for an individual assay point.
Temperature studies to evaluate loss of mefenamic acid
through sublimation and to evaluate its stability at these
same temperatures were carried out as follows. Samples of
mefenamic were carefully weighted into glass bottles
(10 mg), the bottles loosely covered with aluminum foil,
placed into a constant temperature oven (i.e., a GC oven),
and maintained at elevated temperatures for varying
periods up to four days. Duplicate sampling was made at Fig. 3. Infrared spectra of the NH stretch region of mefenamic acid.
the different time points by removing two samples from the Samples: (a) as received, and after heating at 160 8C for (b) 12 h, (c) 24
oven, cooling them to room temperature, carefully adding a h, (d) 36 h, and (e) 48 h. Bands: Polymorph I at 3311–3312 cm—1,
polymorph II at 3346–3347 cm—1 and unidentified trace component at
known volume of methanol, and analyzing them by HPLC. 3353 cm—1.
The compound’s stability and loss rate via sublimation
were studied at three different temperatures, 140, 160, Except for the presence of a small amount of polymorph II
and 180 8C. This range of temperatures was selected such (i.e., about 10%) the spectrum is consistent with that
that it previously published for the polymorphic I form [4]. An
covered thermal conditions studied previously by differ-
important spectral feature that can be used to distinguish
ential scanning calorimetry [19].
these two crystal forms is the NH stretching band that
occurs between 3300 and 3350 cm—1. This is illustrated in
Fig. 3 by the series of partial spectra over this region for
3. Results and discussion mefenamic acid as it is thermally converted (i.e.,
maintained at 160 8C
Shown in Fig. 2 is a representative spectrum of a sample for varying periods of time) from the as received sample
of mefenamic acid as it was received from the supplier.
Absorbance

500 1000 1500 2000 2500 3000 3500 4000


Wavenumbers, cm-1

Fig. 2. Infrared spectrum of mefenamic acid as received.


(spectrum a) to pure polymorph II form of mefenamic acid In order to study the rate of conversion between
(spectrum e). The band at about 3311–3312 cm —1 arises polymorph I and II, changes in the NH stretch region were
from the amino group internally hydrogen bonding with the monitored as a function of time at temperatures of 150,
carbonyl group and the band at about 3346–3347 cm—1 is 155, and 160 8C. Two additional temperatures, 140 and 180
from the NH stretch in the polymorphic II form. These 8C, also were studied, however results from these
structural differences are illustrated in Fig. 1. measurements are not included because of extremely slow
As shown by the dashed vertical line in Fig. 3, the band conversion kinetics in the case of the lower temperature
for polymorph II in the initially unheated sample (spectrum and analyte
a) appears to be shifted to slightly higher wavenumber degradation at the higher temperature. In the latter case, the
compared to the heat treated samples shown in spectra b–e. mechanisms and kinetics of thermal degradation of
By employing standard curve fitting procedures, the shifted mefenamic acid at elevated temperatures is being reported
band could be resolved into two spectral components, the in a separate study. In the 150 to 160 8C temperature range
3346–3347 cm—1 band arising from polymorph II and an mefenamic was found to be thermally stable (i.e., no
unidentified component appearing at 3353 cm—1. Both of measurable degradation) using a HPLC stability indicating
these features are shown as dashed lines in spectrum a. In assay over the time period needed to monitor the
the initial sample the unidentified 3353 cm —1 band was polymorphic changes.
approximately 4%. However, it was not detectable in any Because the molar absorptivity of the NH stretches are
of the heated samples. Two possible causes for these different between the two polymorphs and only one of the
observations are traces of solvent or crystal defects in the polymorphs could be obtained in pure form (i.e., produced
unheated sample. In any case, the thermal conversion data by thermally converting polymorphic I completely to
discussed below are corrected for this trace unidentified polymorphic II by heating at 160 8C for 2 days) a standard
component as well as differences in infrared band addition method was used to correct for these differences.
absorptivity as discussed below using a standard addition Shown in Fig. 4 is a plot of the area % of polymorph II
approach. compared to the total area of the two bands plotted against
There are two important spectral features that are shown the percentage of polymorph II added to the as received
in Fig. 3. One is the frequency shift to high wavenumbers mefenamic sample (i.e, polymorph I with about 10%
as mefenamic acid changes from polymorph I to polymorph II). In carrying out the standard addition
polymorph II and the second is the change in band width 1/2 calibration experiment, multiple measurements were made
from about 30 wavenumbers when the NH group is at the individual data points and their average values used
hydrogen bonding with the carbonyl group to about 16–17 to construct the calibration plot. As in the case of the
wavenumbers in the polymorphic II form. Both of these thermal measurements, the r2 values for the statistical fits
spectral changes are consistent with crystallographic data of the infrared bands were in almost all cases better than
[10–15] and quantum mechanical considers of the structure 0.997. Subsequently, the points shown in Fig. 4 were fitted
of mefenamic acid and related fenamic compounds [5]. by a second order polynomial which is the solid line with
Likewise, they are indicative of rotational changes in the an intercept value of 11.8% (i.e., the amount of polymorph
carboxyl group that influences the internal hydrogen II in the original as received sample of mefenamic acid
bonding interactions as the crystal to crystal transition used to prepare the calibration mixtures). Also, shown in
occurs. Fig. 4 are the coefficients for the second order
regression fit.

1.0
Area Polymorph II/Area Polymorph I & II

0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1 0.1168 + 0.003196*x +
5.729E-005*x2
0.0
0 10 20 30 40 50 60 70 80 90 100
% Polymorph II Added

Fig. 4. Standard addition calibration curve for the initial as received mefenamic acid sample with varying amounts of pure polymorph II added.
1.0
0.9
0.8

Fraction Polymorph II
0.7
0.6
0.5
0.4
0.3
0.2
0.1

0 20 40 60 80 100 120 140 160


Time, hours

Fig. 5. Thermal conversion of mefenamic acid to polymorph II for temperatures of: solid lines (a) 150 8C, (b) 155 8C, and (c) 160 8C. Also, shown as a dashed
line is the150 8C data not corrected for differences in molar absorptivity of the different NH bands.

Subsequently, this equation was used to correct the area the subsequent calculations are based on standard addition
ratio values (i.e., amino stretch band response differences) measurements.
for the thermal conversion experiments carried out at The slopes from plots a–c in Fig. 5 were used to
temperatures of 150, 155, and 160 8C. determine the activation energy for the crystal to crystal
The relative rates of thermal conversion of polymorph
I– transition of mefenamic acid. This is shown in Fig. 6 where
II are shown in Fig. 5 as solid lines for the three the relative rate of change is plotted against the inverse of
temperatures studied. The experimental points are shown the temperature in K. The slope of this plot yielded an
as the solid circles and were obtained using the standard activation energy of 71.6 kcal/mol. Interestingly, the
addition procedure discussed above. Also, included for thermal conver- sion of mefenamic acid from polymorph I
illustrative purposes, is the linear fit (i.e., the dashed line) to II has been suited previously by differential scanning
for one set of data obtained using curve fitting and simple calorimetry and reported to have an activation energy of
area normalization to calculate the amount of polymorph 86.4 kcal/mol [19]. Further in this same study, significantly
I and II present in a given sample. In using this latter faster rates of conversion were noted at the individual
procedure, differences in the NH band absorptivity temperatures. In carrying out the previously reported DSC
between polymorph I and II are ignored. On a relative work, apparently small amounts of sample (1–2 mg) were
basis, the overall error in employing the simpler area placed in the sample pans, heated to a given elevated
normalization procedure, over the more accurate standard temperature between 140 and 150 8C, and maintained for
addition procedure is in the 20–25% range. As such, all of varying periods of time.
Following this, scans were completed through the thermal

-3.6

-4.0
ln Conversion Rate

-4.4

-4.8

-5.2

-5.6

-6.0

0.00230 0.00232 0.00234 0.00236


1/T in K

Fig. 6. Plot of the natural logarithm of the rate of conversion vs. reciprocal temperature in K.
1.0
c
0.9

Relative Sublimation Loss


0.8 b
0.7
0.6
0.5
0.4
0.3
0.2
0.1 a
0.0

0 10 20 30 40 50 60
Time, hrs

Fig. 7. Loss of analyte due to sublimation measured using a stability indicating reversed-phase HPLC procedure. Curves: (a) 140 8C, (b) 160 8C, and (c) 180
8C.

range where the crystal to crystal transition of mefenamic changes through sublimation losses do not influence the
acid could be observed as an exothermic event. Subse-
quently, the area of the thermal transition was plotted
against
time for hold temperatures of 140, 145, and 150 8C and the
isothermal conversion rates used to construct a van’t Hoff
plot that yielded the 86.4 kcal/mol value for the activation
energy.
During the course of the current study, as well as
parallel investigations being carried to examine the thermal
stability and mechanisms of degradation of mefenamic
acid, it was observed that appreciable sample loss occurred
when the samples were maintained at elevated
temperatures due to sublimation. Based on this observation,
it was believed that sample loss might be a possible
explanation for differences between the current infrared
activation energy and that obtained by DSC. As such,
controlled sublimation studies
were carried out at temperatures of 140, 160, and 180 8C.
The results from these experiments are summarized in
Fig. 7, which shows plots of sample loss vs. time at the
three different temperatures. These data were obtained
using a stability indicating HPLC assay and all of the
sample losses
are due to simple sublimation and not analyte degradation.
In the case of the 180 8C plot and over the time period
shown a small amount of analyte degradation occurred,
however over 90% of the loss was still attributable to
sublimation. These measured loss profiles (i.e., the 140 and
160 8C data) were used to calculate the relative sample loss
profiles for temperatures used previously in the DSC
studies [19]. Although the results cannot be applied
quantitatively to correct the previous data, they clearly
demonstrate that
sublimation loss from the DSC pan (i.e., even though the
authors note that no change in weight was observed, they
do not explain how they verified this) would easily account
for the 15 kcal/mol activation energy difference as well as
the much faster apparent isothermal conversion profiles. In
the case of the current infrared method, sample weight
data since they are based on NH stretching area ratio
measurements.

4. Conclusion

The current work clearly demonstrates that infrared


spectroscopy provides a better measure of the relative rate
of conversion of mefenimic acid from polymorph I–II than
the previously described DSC approach that failed to
consider potential sample loss at elevated temperatures
due to sublimation when heating was carried out directly
in the DSC pans prior to measuring the exothermic
transition. In the case of DSC, this problem likely could
have been avoided by accurately weighing a known
amount of the analyte into the sample pans following prior
heating, assuming appropriate precautions are taken to
obtain a homogeneously representative sample. Work is in
progress to apply the current infrared approach to study
thermally induced crystal to crystal transitions in other
structurally related fenamates.

References

[1] J. Sato, E. Owada, K. Ito, Y. Niida, A. Wakamatsu, M. Umetsu,


Simple, rapid and sensitive reversed-phase high-performance
liquid-chromatographic method for the determination of mefe-
namic-acid in plasma, J. Chromatogr.-Biomed. Appl. 493 (1989)
239–243.
[2] F.A. Aly, S.A. Al-Tamimi, A. Alwarthan, Determination of
flufenamic acid and mefenamic acid in pharmaceutical preparations
and biolo- gical fluids using flow injection analysis with tris(2,20-
bipyridyl)- ruthenium(II) chemiluminescence detection, Anal.
Chim. Acta 416 (2000) 87–96.
[3] P.C. Ioannou, N.V. Rusakova, D.A. Andrikopoulou, K.M. Glynou,
G.M. Tzompanaki, Spectrofluorimetric determination of anthranilic
acid derivatives based on terbium sensitized fluorescence, Analyst
123 (1998) 2839–2843.
[4] S. Romero, B. Escalera, P. Bustamante, Solubility behavior of poly- 2-([3-(trifluoromethyl)phenyl]amino)benzoic acid (flufenamic acid),
morphs I and II of mefenamic acid in solvent mixtures, Internat. J Acta Crystallogr. Sec. B 38 (1982) 315–317.
Pharm. 178 (1999) 193–202. [13] V. Dhanaraj, M. Vijayan, Structural studies of analgesics and
[5] J.J. Lozano, R. Pouplana, M. Lopez, J. Ruiz, Conformational their interactions. 12. Structure and interactions of anti-inflammatory
analysis of the anti-inflammatory fenamates: a molecular mechanism fenamates – a concerted crystallographic and theoretical conforma-
and semiempirical molecular orbital study, J. Mol. Struct. 335 (1995) tional study, Acta Crystallogr. Sec. B 44 (1988) 406–412.
215–227. [14] J.F. McConnell, 30 -Trifluoromethyldiphenylamine-2-carboxylic acid,
[6] C.N. TenHoor, V. Bakatselou, J. Dressman, Solubility of mefenamic C14H10F3NO2, flufenamic acid, Cryst. Struct. Commun. 2 (1973) 459–
acid under simulated fed- and fasted-state conditions, Pharm. Res. 8 461.
(1991) 1203–1205. [15] J.F. McConnell, F.Z. Company, N-(2,3-Xylyl)anthranilic acid,
[7] R.K. Gilpin, L.A. Pachla, Pharm. Relat. Drugs, Anal. Chem. 71 C15H15NO2, mefenamic acid, Cryst. Struct. Commun. 5 (1976) 861–
(1999) 217R–233R. 864.
[8] R.K. Gilpin, L.A. Pachla, Pharma. Relat. Drugs, Anal. Chem. 73 [16] H. Bojarowicz, Z. Kokot, A. Surdykowski, Complexes of Fe(III)
(2001) 2805–2816. ions with mefenamic acid, J. Pharm. Biomed. Anal. 15 (1996) 339–
[9] R.K. Gilpin, L.A. Pachla, Pharma. Relat. Drugs, Anal. Chem. 75 342.
(2003) 2907–2918. [17] A. Topacli, S. Ide, Molecular structures of metal complexes with
[10] H.M. Krishna Murthy, M. Vijayan, Structural studies of analgesics mefenamic acid, J. Pharm. Biomed. Anal. 21 (1999) 975–982.
and their interactions. VI. 2-{[3-(trifluoromethyl)phenyl]amino}-3- [18] T. Hladon, J. Pawlaczyk, B. Szafran, Stability of mefenamic acid in
pyridi- necarboxylic acid (niflumic acid), Acta Crystallogr. Sec. B the inclusion complex with b-cyclodextrin in the solid phase, J. Incl.
35 (1979) 262–263. Phenom. Macrocycl. Chem. 35 (1999) 497–506.
[11] H.M. Krishna Murthy, M. Vijayan, Structural studies of analgesics [19] T. Umeda, N. Ohnishi, T. Yokoyama, T. Kuroda, Y. Kita, K. Kuroda,
and their interactions. 8. Rotational-isomerism and disorder in the E. Tatsumi, Y. Matsuda, A kinetic study on the isothermal transition
crystal- structure of meclofenamic acid, Acta Crystallogr. Sec. B 37 of polymorphic forms of tolbutamide and mefenamic acid in the
(1981) 1102–1105. solid state at high temperatures, Chem. Pharm. Bull. 33 (1985) 2073–
[12] H.M. Krishna Murthy, T.N. Bhat, M. Vijayan, Structural studies of 2078.
analgesics and their interactions. 9. Structure of a new crystal form
of

You might also like