Chemical Stability of lndomethacin in the Solid Amorphous and
Molten States
J. T. CARSTENSEN*' AND TOMMY MORRIS**
Received September 4, 1991, from the *School of Pharmac University of Wisconsin, Madison, WI 53706.
November 5, 1992. *Present address: Mail Drop 4923, &i Lilly and Company, Indianapolis, IN 46285.
Abstract 0 It is shown that the decomposition of an amorphous sub-
stance at temperatures close to the melting point of the stable,crystalline
modificationequals in order and activation parameters that of the molten
substance.
The physical stability of drugs in the solid amorphous form
has been extensively investigated.I-6 On the other hand, the
chemical stability of drugs in the amorphous form has not
been extensively examined.7-9 Only a few of these studies
have dealt with the mechanistic interpretation of the chem-
ical stability of a drug in the amorphous form as opposed to the
crystalline modification of the drug.
In a n amorphous substance, the order (periodicity or align-
ment regularity) is not as perfect as it is in a crystalline solid.
The phrase "solid amorphous state" is used in this paper to
indicate a state of matter in which particulate integrity exists
but in which there is no crystallinity. For high-molecular-
weight compounds, the state is denoted as rubbery in tem-
perature ranges close to the melting point of the crystalline
compound (and in ranges in which the heat capacity is close
to that of the liquid molten state) and glassy in lower
temperature ranges, in which the solid properties are more
like those of the crystalline compound. An intermediate
temperature (range) is then referred to as the glass transition
temperature. This paper does not address the question of
whether such phenomena apply to smaller molecules, but if a
solid amorphous state, as defined above, is produced by
melting a compound and cooling it through its melting point
without crystallization occurring, there may be a connection
between the stability of the drug in a molten state and that
of the drug in an amorphous state.
One might expect that in intramolecular reactions, the
amorphous entity will be less stable than the crystalline
entity. In fact, such a case has been reported in the litera-
ture7.6: cephalosporin hydrates are more stable than the
anhydrates, which are largely amorphous.
One would expect, on the basis of the information given
above, that the stability parameters of the decomposition of
an amorphous substance close to the melting point could be
extrapolated to those for a molten substance, whereas at low
temperatures, at which properties (such as heat capacity and
density) might more closely approximate those of the crys-
talline solid, the stability parameters might approximate
those of the solid crystalline material.
An excellent article by Pikal et al.10 addresses the question
of the temperature dependence of the chemical stability of
amorphous cephalosporins. However, cephalosporins decom-
pose too rapidly close to their melting points to allow for
meaningful kinetic analyses at such temperatures. Hence,
such systems do not allow experimental probing of whether
the decomposition-temperature curve of a n amorphous sub-
stance extends to that of a liquid molten substance.
0022-3549/93/0600--07$02.50/0
Q 1993, American Pharmaceutical Association
Accepted for publication
For study of this point, it is necessary to select a substance
that is of intermediate stability; i.e., rates must be such that
they can be determined in a range of temperatures close to the
melting point. The substance must also have a reasonable
tendency towards supercooling, so that an amorphous state
can be produced. Furthermore, the compound should not be
(or should not form) a hydrate, because in such a case the
stability of the crystalline solid would differ structurally from
that of the amorphous solid (which has no coordinated water).
A compound that fits these criteria is indomethacin.
It would be ideal if a compound could be studied over such
a wide temperature range that the glass transition temper-
ature and the melting point (of the most stable crystalline
modification) could both be included. Indomethacin does not
meet this criterion. In this study, amorphous indomethacin
was formed by slow cooling of molten indomethacin through
the point at which crystalline indomethacin melted and a
particulate supercooled liquid was produced. However, crys-
tallization occurred at temperatures below 137 "C, so deter-
mination of the glass transition temperature (if it existed) of
the amorphous compound made in the manner used here was
not possible. The present study was therefore confined to the
question of whether a rubbery amorphous compound exhibits
the same stability characteristics as the parent molten com-
pound.
Experimental Section
Indomethacin was used as received from the supplier (Merck & Co.,
Inc., West Point, PA). It is a crystalline modification that melts at
162 "C. Samples (50mg) were placed in ampoules and flame sealed.
They were then placed in a oil bath at 165 e 1 "C and kept there for
8 min, after which they were cooled to the tem erature of study. If the
3
study consisted of n datum points, then n random samples were
removed and tested for amorphism by examination of the samples
under cross-Nichols and occasionally X-ray diffraction.
Samples were placed in oil baths at the temperatures desired,
retrieved at various times, and assayed by the USP HPLC assay
method. Temperatures of 145, 150, 155,165,175,and 185 "C were
0.2 -J
Time (hours)
Figure 1-Decomposition of amorphous indornethacin. Data for 150 "C
are not shown for graphical clarity. Key: (0)145 "C; (A) 155 "C.
Journal of Pharmaceutical SciencesI 657
VOI. 62,NO. 6, June 1993
 0.8 1 A n
0

" 0
n

.-
c
0
m
t
0.2 1 "
4 0 0

-3 ! I
\
I 0.0 n
V . l . 1 . I

0 10 20 0 100 200 300 400 500 600

Time (hours) Time (hours)

Flgure 2-Decomposition of molten indomethacin at temperatures Figure 4-Decomposition of crystalline indomethacin at various tem-
above the melting point of the most stable crystalline modification. Key: peratures. Key: (Large circle) 125 "C; (Small circle) 135"C; (0)140 "C;
(0)165°C; (0) 175°C;(A) 185°C. (A) 145 "C.

Table I-First-Order Kinetlc Equation Parameters for Amorphous

and Molten Indomethacln'
Temperature, "C Intercept Slope Correlation Coefficient
145 -0.06 -0.10150 0.990
150 -0.04 -0.0202 0.995
155 -0.017 -0.0362 0.999
165 -0.025 -0.0492 0.997
175 0.051 -0.133 0.993
185 -0.064 -0.194 0.995
" r
In (fraction retained) = (intercept - slope) x time. 0 100 200 300
Time (hrs)
-1 - Figure 5-Decomposition data from Figure 4 treated by Bawn kinetics.
Key: (0)145 "C; (0)135 "C.

-2 -

-3 -

-2 I
1 I

2.1 2.2 2.3 2.4 2.35 2.40 2.45 2.50 2.55

1000/T
Figure &Data from Figures 1 and 2 analyzed by Arrhenius plotting.
Key: (0)amorphous; (A) molten. k, Rate constant, T. absolute temper-
ature.
used. Of these, the three former are for solid amorphous specimens
and the three latter are for molten specimens.
Crystalline indomethacin (mp, 162 "C)was tested in a like manner
at 125,135,140,and 145 "C.
For treatment of data, the decrease in the main HPLC peak was the
parameter tested.
Results and Discussion
Typical decomposition curves for amorphous indomethacin
are shown in Figure 1. The decomposition was pseudo-first
order; i.e. (asin the findings of Pikal et al.10), the pseudoliquid
amorphous state was similar to the liquid state in the manner
(order) in which decomposition occurred. Molten indometh-
acin also exhibited first-order behavior (Figure 2). The per-
tinent kinetic parameters and goodness-of-fitvalues are given
1ooorr
Figure GArrhenius plot of rate constants from Figure 5. k, Rate
constant, T, absolute temperature.
in Table I.
The rates of decomposition of the pseudoliquid amorphous
compound and of the molten compound are shown on one
Arrhenius plot in Figure 3; the curves merge into each other.
As expected, it can be concluded that the decomposition of
a pseudoliquid amorphous compound (at least of this one) is
an extension of the decomposition of the liquid molten
compound.At temperatures well below the melting point, this
conclusion may not hold true.10
The decomposition of crystalline indomethacin, as is dis-
cussed below, was of a different order. The magnitude of the
initial rate of decomposition of crystalline indomethacin in
the temperature range of 140-155 "C was much smaller for
the crystalline modification. At 145 "C, for instance, only 1%
of the material was decomposed after 48 h, whereas for the
amorphous material, such a degree of decomposition was
reached within a few hours. Hence, the crystalline modifica-

658 I Journal of Pharmaceutical Sciences

Vol. 62, No. 6, June 7993
tion is more stable than the amorphous form. At temperatures
close to the melting point, low levels of decomposition cause
sufficient melting point depression of the crystdline material
that the sample melts; hence, above 150 "C, the crystalline
solid rapidly starts behaving like the amorphous or molten
form.
Crystalline indomethacin at the temperatures in question
does not decay by a first-order pattern but rather decomposes
in a Bawn12.13 decomposition mode (Figure 4). In Bawn
kinetics, the decomposition equation is In 11 + Ax] = -(Ak&,
in which x is the fraction decomposed, A is an iterant
parameter that imposes both linearity and a zero intercept on
the data, k, is the decomposition rate in the solid state, and
t is time. For the four temperatures tested, this equation is
applicable, and the k, values fit a n Arrhenius equation, as
shown in Figures 5 and 6. Adirect comparison with the decay
of the amorphous form is not possible (e.g., on a common
graph) because the mechanism is different.
In summary, it is shown that amorphous indomethacin at
temperatures close to the melting point for the crystalline
modification exhibits first-order decomposition in the solid
(rubbery) state, that indomethacin in molten form, i.e., at
temperatures higher than the melting point for the crystal-
line modification, exhibits first-order decomposition, and that
the decomposition rates, when analyzed by Arrhenius plot-
ting, lie on a straight line, so that, kinetically (aswell as in
other respects), the rubbery state may be considered a con-
tinuation of the molten state.
References and Notes
1. Klech, C. M.; Pari, J. H. Pharm. Res. 1989,6,564-570.
2. Mathews, A. G.; Schram, C. J.; Minty, D. Nature (London) 1966,
21 1 , 959.
3. Karel, M. CRC Crit. Rev. Food Technol. 1973,3,329-339.
4. VanScoik, K., Ph.D. Thesis; School of Pharmacy, University of
Wisconsin, Madison, WI, 1987.
5. VanScoik, K.;Carstensen, J. T. Znt. J.Pharm. 1990,58,185-194.
6. Carstensen, J.T.; VanScoik, K. Pharm. Res. 1990,7,1278-1282.
7. Pfeiffer, R.R.; Engel, G. L.; Coleman, D. Antimicrob. Agents
Chemother. 1976,9, 848-851.
8. Oberholtzer, E. R.; Brenner, G. S. J. P h r m . Sci. 1979, 68,
863-867.
9. Pothisiri, P.; Carstensen, J. T. J. Pharm. Sci. 1975,64, 1931-
1934.
10. Pikal, M. J.;Lukes, A. L.; Jang, J. E. J. Pharm. Sci. 1977,66,
1312-1316.
11. Morris, T.; Ph.D Thesis; School of Pharmacy, University of
Wisconsin, Madison, WI, 1990.
12. Bawn, C. E. H. In Chemistry of the Solid State; Garner, W. E.,
Ed.; Butterworths: London, 1955,p 254.
13. Carstensen, J. T.; Kothari, R. J.Phurm. Sci. 1981,70,1095-1100.
Acknowledgments
This work was supported by a small starter grant from Merck &
Co., Inc., West Point, PA, and a substantial grant from Smith N i n e
Beecham, King of Prussia, PA, Sandoz Pharmaceutical Co., East
Hanover, NJ, and Marion-Merrill Dow, Kansas City, MO.
Journal of Pharmaceutjcal Sciences I 659
Vol. 82, No. 6, June 1993