The effect of disorder on the chemical reactivity of an organic solid, tetraglycine methyl ester: Change of the reaction mechanism

Article abstract of DOI:10.1002/jps.10066

Many drugs undergo chemical changes in the solid state, and understanding chemical reactivity of organic crystals is a critical factor in the drug development process. In this report, the impact of milling on the thermal chemical reactivity of an organic solid, tetraglycine methyl ester, was studied using DSC, isothermal calorimetry, chemical analysis (HPLC and insoluble residue determination), and powder X-ray diffraction. Significant changes in both X-ray diffraction patterns and DSC curves were detected after very brief milling (5 s). The changes were interpreted as the formation of a disordered phase. The disordered phase was tentatively identified as a crystal mesophase that combines properties of both crystalline (i.e., long-range order) and amorphous (i.e., glass transition) states. In the disordered material, the reaction mechanism changed from the methyl transfer reaction, which was observed in the intact crystal, to a polycondensation reaction when the reaction was performed at 165°C. Such changes in the reaction mechanism occurred in materials milled for > 30 s.

Full text of DOI:10.1002/jps.10066

The Effect of Disorder on the Chemical Reactivity of an
Organic Solid, Tetraglycine Methyl Ester: Change of the
Reaction Mechanism
EVGENYISHALAEV, MARINA SHALAEVA, GEORGE ZOGRAFI
School of Pharmacy, University of Wisconsin±Madison, 425 N. Charter St., Madison, Wisconsin 53706
Received 4 June 2001; revised 21 September 2001; accepted 6 November 2001
ABSTRACT: Many drugs undergo chemical changes in the solid state, and under-
standing chemical reactivity of organic crystals is a critical factor in the drug
development process. In this report, the impact of milling on the thermal chemical
reactivity of an organic solid, tetraglycine methyl ester, was studied using DSC,
isothermal calorimetry, chemical analysis ꢀHPLC and insoluble residue determination),
and powder X-ray diffraction. Signi®cant changes in both X-ray diffraction patterns and
DSC curves were detected after very brief milling ꢀ5 s). The changes were interpreted as
the formation of a disordered phase. The disordered phase was tentatively identi®ed as
a crystal mesophase that combines properties of both crystalline ꢀi.e., long-range order)
and amorphous ꢀi.e., glass transition) states. In the disordered material, the reaction
mechanism changed from the methyl transfer reaction, which was observed in the intact
crystal, to a polycondensation reaction when the reaction was performed at 1658C.
Such changes in the reaction mechanism occurred in materials milled for > 30 s. ß 2002
Wiley-Liss, Inc. and the American Pharmaceutical Association J Pharm Sci 91:584±593, 2002
Keywords: disorder; chemical reactivity; organic solid; tetraglycine methyl ester
INTRODUCTION
shown, for example, that decomposition reactions
usually start on crystalline defects.8 In addition,
such disorder has been shown to lead to the
formation of products that are ``prohibited'' in a
normal crystal.8 Common pharmaceutical opera-
tions such as milling, compaction, and freeze
drying may create a partially or completely dis-
ordered material with an increased chemical
reactivity. Note also that chemical changes them-
selves produce disorder in an initially crystalline
material, thus accelerating and facilitating
further chemical transformation.9,10
Many pharmaceuticals undergo chemical changes
in the solid state, and understanding the chemical
reactivity of organic crystals is an important
factor in the drug development process. One of
the critical factors in solid-state reactivity is the
extent of disorder in the crystal structure.1±3 As a
rule, chemical degradation is accelerated in a
disordered state because of higher energy and
increased molecular mobility.4±7 It has been
In this report, the impact of milling on the
thermal reactivity of an organic solid, tetraglycine
methyl ester ꢀTGME), has been studied. Thermal
reactivity of TGME, in general, has been studied
in some detail.11±15 In particular, it has been
shown that reaction pathways in the solid state
and in solution are different from each other,11±13
thus meeting a critical criterion for a typical solid-
Evgenyi Shalaev's present address is Groton Laboratories,
P®zer Inc., Groton, CT 06340.
Marina Shalaeva's present address is Groton Laboratories,
P®zer Inc., Groton, CT 06340.
Correspondence to: Evgenyi Shalaev ꢀTelephone: 860-441-
5909; Fax: 860-441-0467;
E-mail: evgenyi_y_shalaev@groton.p®zer.com)
Journal of Pharmaceutical Sciences, Vol. 91, 584±593 ꢀ2002)
ß 2002 Wiley-Liss, Inc. and the American Pharmaceutical Association
584
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 2, FEBRUARY 2002

CHEMICALREACTIVITY OF AN ORGANIC SOLID
585
state reaction. In our previous studies of TGME it
has been shown that, in the highly crystalline
material, the reaction starts as a homogeneous
reaction with the formation of a solid solution of
product in the reactant crystalline lattice. This
then converts into a heterophase process ꢀwith
crystallization of a product phase) after approxi-
mately 30% of conversion.14 When crystalline
material was disordered by either milling or
freeze drying, the reaction rate increased signi®-
cantly.15 Moreover, the kinetic curve changed
from an acceleratory to a deceleratory process, and
crystallization of a product phase was observed at
a much higher extent of reaction or not at all.
However, the chemical mechanism was the same
in both the disordered and crystalline states in the
temperature range from 50 to 1158C.15
In the present work, a more systematic study of
the impact of milling on the thermal properties
and crystal structure of TGME has been per-
formed using DSC, powder X-ray diffraction,
isothermal calorimetry, and chemical analysis.
In particular, evidence is presented to show that
the disordered phase, which was produced by
milling, is a crystalline mesophase. A crystalline
mesophase can be described as a phase that
combines the properties of the crystalline state
with long-range order, as evidenced by wide-angle
X-ray diffraction, and the amorphous state, with a
glass transition as detected by DSC. Examples of
mesophases are liquid crystals, plastic crystals,
and conformationally disordered crystals.16 It has
been reported in this work that the mechanism of
thermal reactivity in the disordered material and
in the initial crystal of TGME appear to be di-
fferent at a higher reaction temperature ꢀ1658C),
i.e., methyl transfer occurs in the crystalline
state, and polycondensation occurs in the dis-
ordered material.
Sample Preparation
Milled samples of TGME were prepared with a
WIG-L-BUG Heavy Duty Laboratory mill ꢀSpec-
traTech Inc.) using a stainless steel vial and ball.
The mass of the ball was 1.0229 Á 10À3 kg, the
vibration frequency was u ˆ 66 sÀ1, and amplitude
of vibration was h ˆ 1.875 Á 10À2 m. Milling
time varied from 5 to 900 s. The amount of
material placed into the vial was generally 80 to
100 mg. On several occasions, 160 to 360 mg were
milled.
Differential Scanning Calorimetry
All DSC measurements were carried out with a
Seiko SSC/5200 differential scanning calorimeter
at a heating rate of 108C/min. The instrument was
calibrated using the melting points of indium,
gallium, tin, water, and mercury as temperature
standards. Samples of TGME, from about 3 to
13 mg, were placed into sealed aluminum pans
with pin holes. Before any analysis, material was
stored into a refrigerator in tight containers
containing desiccant.
Isothermal experiments were carried out on
the same DSC instrument under nitrogen ¯ow.
Samples were heated to a speci®ed reaction tem-
perature ꢀ1658C in the majority of cases) at 108C/
min, and data were collected at this temperature.
Several isothermal experiments were also per-
formed at 115 and 1258C. Reacted samples were
analyzed using HPLC, gravimetric analysis ꢀto
determine any insoluble residue), and powder
X-ray diffraction.
X-ray Diffraction
Powder diffraction patterns were obtained on a
Scintag PADV instrument under the following
conditions: CuKꢀ radiation, 30 mA, 45 kV, scan
rate 58/min. A material was placed on a quartz
sample holder and measurements were carried
out at room temperature. The procedure for
quantitative X-ray diffraction analysis using
LiF as an internal standard has been described
elsewhere.14
EXPERIMENTAL
Materials
TGME was obtained from Professor Stephen Byrn
of Purdue University, and is the same material
reported previously.14,15 Water used for the HPLC
mobile phase was treated with a Barnstead water
puri®cation system. Acetonitrile and tri¯uoroace-
tic acid, used in the HPLC mobile phase, were
HPLC grade solvents. LiF, used as an internal
standard in the X-ray diffraction measurements,
was obtained from Aldrich at reagent grade.
Chemical Analysis
Extent of chemical reaction was determined using
HPLC and determination of insoluble material.
The fraction of insoluble material was used to
estimate the extent of the polycondensation
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 2, FEBRUARY 2002

586
SHALAEV, SHALAEVA, AND ZOGRAFI
reaction.12 Samples were dispersed in the HPLC
solvent after isothermal calorimetry experiments.
The suspension was ®ltered through a ProWeigh
47-mm glass ®ber ®lter to obtain total suspended
solids. After that, ®lters with insoluble residue
were dried, and the fraction of insoluble material
was determined as the difference between the
initial ®lter weight and the ®lter weight after
®ltration. In addition, some of the samples were
analyzed by HPLC after the isothermal calorime-
try experiments. The HPLC analysis was carried
out on an Altech Econosil C18, 5m column using an
isocratic solvent system consisting of 1% ꢀv/v) of
acetonitrile and 0.1% ꢀv/v) tri¯uoroacetic acid in
water. The conditions for the HPLC analysis and
identi®cation of the degradation products have
been described elsewhere.14 Structures of the
reactant and products are shown in Scheme 1.
RESULTS
Figure 1. Representative X-ray powder diffraction
patterns of TGME milled for different time periods.
Numbers show milling time. Peaks that are assigned to
Characterization of Milled Material
Representative X-ray diffraction patterns of
TGME samples milled for different time periods
are given in Figure 1. A very brief milling ꢀ5 to
30 s) caused a signi®cant decrease in the intensity
of all main peaks. In addition, a peak at 2y 21.48
signi®cantly increased, and several minor new
peaks at 2y 14.1, 23.6, 24.1, 25.38 were observed.
To determine if the new peaks could be
assigned to degradation products, milled samples
were analyzed by HPLC. Figure 2 shows the
TGME content and the content of the two main
$
the new crystalline mesophase are marked with .
impurities, tetraglycine and
a nonidenti®ed
impurity, as a function of milling time. It can be
seen that brief milling ꢀless than 60 s) did not
cause any noticeable changes in the chemical
composition of the material, whereas it did result
in major changes in the X-ray diffraction patterns.
Hence, we conclude that the new X-ray diffraction
peaks are associated with the formation of a new
crystalline phase of TGME ꢀTGME2) rather than
with degradation products. Here and below we
refer to TGME2 as a new crystal phase created as
a result of milling, and TGME1 as the initial
crystalline form. TGME2 is characterized by a
peak at 2y 21.48.
Quantitative X-ray diffraction data are shown
in Figure 3, where the fraction of the initial
crystalline phase ꢀFigure 3A), and the relative
fraction of a new phase ꢀFigure 3B) are shown as a
function of milling time. The fraction of the initial
crystalline form was determined using the peak
at 2y 5.78.14 The relative fraction of the new
phase has been calculated as Si/S0 where S0
and Si represent areas of the peak at 2y 21.48 for
the nonmilled sample and milled sample, respec-
tively. Peak areas were normalized using LiF as
an internal standard. Very brief milling ꢀ5 s)
caused a signi®cant decrease in the fraction of
TGME1. Increase in the milling time up to 600 s
Scheme 1.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 2, FEBRUARY 2002

CHEMICALREACTIVITY OF AN ORGANIC SOLID
587
Figure 3. Relative amount of two crystalline poly-
morphs of TGME as a function of milling time. ꢀA)
Iinitial crystal, TGME1; ꢀB) crystal mesophase, TGME2.
Inset shows fraction of TGME1 as a function of
logꢀmilling time). Straight line in the inset is a linear
regression that is shown as a visual aid.
Figure 2. Chemical composition of TGME as a func-
tion of milling time. ꢀB) Magni®ed portion of the upper
graph showing concentrations of two main impurities
after brief milling ꢀup to 60 s). *: TGME; &: tetra-
glycine; !: nonidenti®ed impurity. Lines are given as a
visual aid.
caused some additional decrease in the amount of
TGME1, as can be seen in Figure 3A ꢀinset). How-
ever, the changes associated with the increase in
milling time from 5 to 600 s are much less pro-
nounced than the change caused by the initial 5 s
of milling. The intensity of the peak at 2y 21.48
increased signi®cantly after 5 s of milling, and
remained approximately constant at longer mil-
ling times ꢀup to 600 s), indicating that the
amount of the TGME2 phase did not change
very much with longer periods of milling. Hence,
major changes were caused by very brief milling
ꢀ5 s), while extended milling did not cause any
further signi®cant changes in the phase composi-
tion of the sample as detected by X-ray powder
diffraction.
Representative DSC heating curves of samples
milled for different time periods are shown in
Figure 4. The DSC curve for an initially crystal-
line sample ꢀTGME1 crystal phase) has a melting
endotherm with an onset temperature of ꢀ 1908C,
whereas in a sample milled for 30 s melting was
observed at a much lower temperature. Lower
melting temperature is another indication of dis-
order created as a result of milling. Such depres-
sion of the melting temperature cannot be
attributed to milling-induced degradation, as
milling did not induce any noticeable chemical
changes ꢀFigure 2). Longer milling produced
further changes in the appearance of the DSC
curves, i.e., an exotherm was observed at a lower
temperature at ꢀ 1708C overlapping with the
melting event. This exothermic event corresponds
to thermal degradation of the sample, most likely
to the polycondensation reaction, as discussed
below. Magni®ed lower temperature portions of
these DSC heating curves are given in Figure 4B.
In the milled samples, an endothermic event was
observed with the onset temperature of about
308C, as reported elsewhere.15 The magnitude of
this effect increased with milling time, but the
temperature remained essentially the same.
The magnitude of the thermal event decreased
signi®cantly with the second heating scan,15
which allowed us to assign this thermal event to
enthalpy recovery at the glass transition. It
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 2, FEBRUARY 2002

588
SHALAEV, SHALAEVA, AND ZOGRAFI
Both the X-ray diffraction data ꢀi.e., a sharp
decrease in intensity of a majority of peaks), and
the DSC results ꢀi.e., observation of enthalpy
recovery) suggest that milling produced a dis-
ordered phase. In addition, milling induced the
appearance of some new crystalline lines in the X-
ray diffraction pattern indicative of a new crystal
phase. Considering the X-ray diffraction and DSC
data together, we can suggest two different inter-
pretations: ꢀ1) milling produced two new phases,
i.e., an amorphous phase ꢀwith Tg at ꢀ 308C) and a
new crystalline phase; or ꢀ2) milling produced one
new phase, i.e., a crystalline mesophase. A
crystalline mesophase can be described as a phase
that combines the properties of the crystalline
state ꢀi.e., long-range order as evidenced by wide-
angle X-ray diffraction) and the amorphous state
ꢀa glass transition as detected by DSC).16 We
suggest that the second explanation ꢀi.e., a
crystalline mesophase) is more likely. Indeed, at
T >> Tg ꢀup to 1658C), no viscous ¯ow was
observed, which indicates that a major portion of
the material was not amorphous. Quite the
opposite, the appearance of the milled materials
did not change even at temperatures as high as
1658C. In addition, very few peaks were observed
on the X-ray diffraction patterns of the milled
materials, which also supports the presence of
such a mesophase. Mesophases ꢀin particular,
liquid crystals and plastic crystals) give rather
diffuse X-ray diffraction patterns, having a re-
latively small number of re¯ections.19 Hence,
although the formation of a partially amorphous
material in samples that were milled for longer
time periods cannot be completely ruled out, we
consider that it is most likely that the disordered
phase in milled samples is a crystal mesophase.
Such a crystal mesophase can be either liquid
crystal, plastic crystal, or a conformationally dis-
ordered crystal ꢀcondis crystal). The lack of
viscous ¯ow, which was observed even at tem-
peratures well above the glass transition tem-
perature, points out that this is neither a liquid
crystal nor a plastic crystal. Most likely, milled
TGME forms a conformationally disordered crys-
talline mesophase with a more or less random
distribution of molecular conformations. Note
that in this case Tg likely corresponds to a
transition from a ``glassy crystalline state'' with
frozen conformational disorder to a conforma-
tionally mobile state. However, the data obtained
to date do not allow us to identify the type
of disorder in the crystalline mesophase more
precisely.
Figure 4. Representative DSC heating curves of
TGME samples milled for different time periods.
Numbers correspond to milling time. Heating rates:
108C/min. ꢀA) Entire DSC scans. ꢀB) Magni®ed portions
of the DSC scans which show an endothermic thermal
event that was assigned to enthalpy recovery corre-
sponding to a glass transition.
should be mentioned that the transition is
broader than usual, and, therefore, it is a
possibility that this thermal event is associated
with an artifact, for example, water evaporation.
However, the consistent growth of the transition
with milling time, and observation that the en-
dothermic event albeit of a lower magnitude
occurred during the second scan15 would seem to
indicate that water evaporation is an unlikely
explanation. It is important to note that a re-
latively broad endotherm that resembles enthalpy
recovery has also been observed for proteins
annealed under different conditions.17 In addi-
tion, a similar increase in enthalpy recovery with
milling time was observed in amorphous starch.18
Hence, we suggest that the endothermic event
observed is due to enthalpy recovery associated
with a glass transition. The glass transition
temperature for the milled samples was deter-
mined as the onset temperature. The Tg appears
to be independent of the milling time and equal to
30.3 Æ 5.78C.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 2, FEBRUARY 2002

CHEMICALREACTIVITY OF AN ORGANIC SOLID
589
Chemical Reactivity
calorimetry. The extent of the reaction ꢀꢀ)
obtained from the DSC data was determined as
Because isothermal calorimetry is a common
method to study solid-state reactions,20±22 it was
used to follow chemical reactivity of TGME.
Representative isothermal calorimetric curves of
samples, which were milled for different time
periods before making the calorimetric measure-
ments are shown in Figure 5 ꢀreaction tempera-
ture 1658C). For a nonmilled material, a strong
exothermic peak ꢀH1) was observed after a notice-
able induction period. The magnitude of the peak
decreased with milling time, whereas the induc-
tion time did not change much and was between
34 and 38 min in samples milled for 0 to 60 s. The
H1 peak was not observed in samples milled for
longer time periods ꢀ> 60 s). Instead, in these
samples an exothermic peak without any induc-
tion period ꢀH2) was observed.
To identify the nature of the H1 exothermic
peak in the isothermal calorimetric curves, sev-
eral calorimetric samples were analyzed by HPLC
before completion of the reaction. In this case, the
reaction was performed at 1258C with crystalline
ꢀnonmilled) samples. The extent of the reaction
was determined by both HPLC and isothermal
ꢀ ˆ ÁH ꢁt†=ÁH1
where DHꢀt) is the area under the curve for a
particular isothermal calorimetric run that was
interrupted before the reaction was completed,
and DH1 is an average reaction enthalpy ob-
tained from several calorimetric experiments that
were run to the completion of the reaction at the
same temperature. After the calorimetric ex-
periments were completed, the samples were
analyzed by HPLC to determine the extent of
chemical conversion.
The extent of the reaction calculated from the
calorimetric and HPLC data is shown in Figure 6.
The ®gure shows that the extent of reaction
determined from calorimetric data was less than
that determined from the HPLC data. Moreover,
signi®cant chemical reactivity ꢀi.e., ꢀ ˆ 0.17) oc-
curred before any thermal event was detected. In
addition, it follows from Figure 6 that the ``induc-
tion period'' on the calorimetric curve corresponds
to ꢀ 30% of chemical conversion, which is in a
good agreement with an earlier study in which
crystallization of a product of the methyl transfer
reaction was observed after ꢀ 25% of chemical
conversion.14 These results suggest that the
exothermic event H1 on the isothermal calorime-
try curves is not solely due to the chemical
Figure 5. Representative isothermal calorimetric
curves of TGME samples milled for different time
periods. Reaction temperature: 1658C. H1 and H2
peaks were assigned to the crystallization of the methyl
transfer product and to the polycondensation reaction,
respectively ꢀsee text for the discussion).
Figure 6. Relationship between the extent of reac-
tion determined from HPLC ꢀ*) and isothermal
calorimetry ꢀ&) data. Reaction temperature: 1258C.
Lines are given as a visual aid.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 2, FEBRUARY 2002

590
SHALAEV, SHALAEVA, AND ZOGRAFI
reaction itself but rather it is associated with
either crystallization of a product phase, or with
combined enthalpy changes from crystallization
and the methyl transfer reaction. Hence, the
induction period as observed on the isothermal
calorimetric curves appears to correspond to a
homogeneous chemical reaction, whereas the exo-
thermic peak itself is associated with the hetero-
phase portion of the chemical reaction.
Figure 7 gives the magnitude of the two exo-
therms ꢀH1 and H2) observed during isothermal
calorimetry as a function of milling time. The
crystallization exotherm of the methyl transfer
product, H1, decreased with milling time indicat-
ing that the amount of crystalline product de-
creased with milling. The H2 thermal event
ꢀassigned to a polycondensation reaction, see
below) was observed in samples milled for more
than 30 s, and remained approximately constant
after longer milling.
Changes in the appearance of calorimetric
curves in milled TGME samples ꢀi.e., disappear-
ance of the H1 event and appearance of the H2
thermal event) implies that there are changes in
the reaction mechanism as the result of milling.
Figure 8. Representative X-ray powder diffraction
patterns of TGME after isothermal calorimetric experi-
ments that were performed at 1658C. Numbers show
milling time. Peaks that correspond to the product of
$
the methyl transfer are marked as .
X-ray diffraction analysis and determination of an
insoluble fraction in the reacted samples support
this conclusion.
Reacted samples were analyzed by X-ray dif-
fraction after the isothermal calorimetry experi-
ments. Figure 8 shows X-ray diffraction patterns
of samples after reaction at 1658C as a function of
milling time. For the nonmilled sample, the X-ray
diffraction pattern of the reacted sample is identi-
cal to patterns observed at 93 to 1158C with the
crystallization of a methyl transfer product.14
The amount of the crystalline product was much
smaller in a sample milled for 15 s, and the cry-
stalline product was not observed at all in samples
milled for more than 60 s. This is consistent with
the isothermal calorimetry results where the H1
crystallization exotherm decreased and ꢀafter a
longer milling) disappeared in milled samples.
Reacted calorimetric samples were dissolved
in a mobile phase for HPLC analysis. Nonmilled
samples were completely soluble, and reaction
products produced at 1658C were very similar to
these obtained at 93 to 1158C ꢀTable 1). However,
milled samples appeared to be partially insoluble
in the HPLC mobile phase. Note that the forma-
tion of an insoluble product in TGME was ob-
served when the crystalline sample was heated at
Figure 7. Magnitude of exothermal thermal events
on isothermal calorimetry curves as a function of
milling time. Reaction temperature: 1658C. Crystal-
lization of the product of methyl transfer, H1: &;
Polycondensation reaction, H2: ~. Inset shows H1 as a
function of milling time in semilogarithmic coordinates.
Lines are given as a visual aid.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 2, FEBRUARY 2002

CHEMICALREACTIVITY OF AN ORGANIC SOLID
591
Table 1. Final Composition of Products of the Thermal Degradation of Tetraglycine
Methyl Ester in Nonmilled Samples ꢀsee Scheme 1 for Chemical Structures)
Reaction Products ꢀwt %)
Reaction
Temperature ꢀ8C) TGME ꢀwt %)
TG
Me2TG
MeTG
Source
14
14
14
93
100
115
115
165
7.4
2.1
0
14.6
14.9
15.6
12.1
13.9
12.3
63.6
65.1
67.9
4.4 Æ 4.0
16.0 Æ 1.6 12.7 Æ 1.2 62.1 Æ 2.5 This work
1.9
15.7 Æ 0.1 14.2 Æ 0.4 65.0 Æ 0.6 This work
higher temperatures ꢀ1858C), or in methanol
solution at 1008C.12 Formation of the insoluble
product appears to be associated with the poly-
condensation reaction.12 Polycondensation results
in the formation of a polyglycine polymer, which
contain both ester and sarcosyl end groups and is
accompanied by a release of methanol. The
chemical structure of the polyglycine polymer
was con®rmed by hydrolysis of the polymer
followed by chromatography, and chemical ana-
lysis of the end groups.12 As a measure of chemical
reactivity in the milled samples, we determined
the insoluble fraction by gravimetric analysis.
The results are presented in Table 2, which
demonstrates that polycondensation becomes a
major reaction in materials milled for ꢂ 230 s.
reported as well.24 When a crystal is milled for a
relatively long time it is also possible for a crystal-
to-amorphous state transition to occur, i.e., in
these cases milling induces a transformation
from a stable to a thermodynamically unstable
form.25,26 The abrupt change of the TGME crystal
seen in this study after very brief ꢀ5 s) milling
supports the conclusion that milling induces a
crystalline phase transition ꢀi.e., formation of the
crystal mesophase) rather than formation of an
amorphous phase. Further milling resulted in
change of the reaction mechanism as described in
more detail below.
Initially, after relatively short milling times of
5 to 15 s, the crystal mesophase was formed, and
depression of crystallization of a methyl transfer
product was observed ꢀFigure 5). However, the
chemical reaction mechanism did not change be-
cause very little insoluble material was found
ꢀTable 2), and the H2 thermal event ꢀassigned to
the polycondensation reaction) was not detected
in these samples. The chemical reaction pathway
changed when the milling time increased to 30 to
230 s, as evidenced by the appearance of the H2
polycondensation exotherm in isothermal calori-
metry curves, and the generation of a signi®cant
ꢀ> 50%) fraction of insoluble material.
It should be noted that the polycondensation of
TGME was reported in an earlier study when
crystalline TGME was reacted at high tempera-
ture ꢀ1858C), or in solution at 1008C.12 It has
been suggested12 that polycondensation, which
requires more signi®cant displacement of mole-
cules from their position than methyl transfer, is
prevented by the rigidity of the crystal, and that
this restriction is eliminated in either solution or
in a crystal when the reaction temperature is
close to the melting point. Our observation that
polycondensation occurred in the crystal meso-
phase but not in the initial crystal supports
this conclusion. Indeed, molecules in a crystal
DISCUSSION
We have observed in this study that very brief
milling ꢀ5 s) appears to induce a solid-solid phase
transition from a crystalline state to a crystal
mesophase. It has been reported in the literature
that milling induces a transformation from meta-
stable to stable crystal forms.23 However, for-
mation of a metastable crystalline phase by
application of a compressive force has been
Table 2. Percent of Insoluble Residue Determined
After Isothermal Calorimetry Experiments at 1658C
Milling Time ꢀs)
Insoluble Residue ꢀwt %)
0
15
3
12
30
6
60
7
230
300
600
900
67 Æ 1
54 Æ 2
46
70
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 2, FEBRUARY 2002

592
SHALAEV, SHALAEVA, AND ZOGRAFI
mesophase possess signi®cant molecular mobility
that should facilitate interaction between two
TGME molecules and/or removal of methanol,
which is the product of the polycondensation
reaction. Extending this consideration, we sug-
gest that although both the methyl transfer and
polycondensation reactions start with an interac-
tion between two TGME molecules, the activated
complexes are very likely different. In the methyl
transfer reaction, the activated complex is formed
through interaction of a carbon of the methyl ester
group "having a partial positive charge) with the
unbound electron pair of the amine nitrogen
"Scheme 2a, hydrogen atoms are hidden). In the
polycondensation reaction, an activated complex
with a more complex structure would be expected
to form "Scheme 2b). We suggest that the forma-
tion of such an activated complex in the crystal-
line state at relatively low temperatures is
severely restricted because of the rigidity of the
crystalline lattice, and therefore, the polyconden-
sation reaction is hindered. In the crystal meso-
phase, however, more mobility is expected, and
the polycondensation reaction became dominant.
In particular, conformational mobility of the
molecules is expected to play a major role in
facilitating formation of the activated complex for
the polycondensation reaction "Scheme 2b). To
test the correctness of the analysis, i.e., to
demonstrate that the crystal structure would
hinder formation of the activated complex for
the polycondensation reaction, obviously a single
crystal structure for TGME would need to b
solved. At this point, however, we have not been
able to grow a single crystal of enough size to
perform such a single-crystal analysis. In addi-
tion, methods that would allow us to follow the
chemical reaction and molecular conformations in
the solid state directly, for example, solid-state
NMR or a spectroscopic method such as FTIR,
might provide valuable information for determin-
ing the structure of the activated complex.
CONCLUSION
Isothermal calorimetry with a DSC instrument
has proved to be a useful tool to study a model
solid-state thermal reaction. However, such cal-
orimetry measurements had to be combined with
other techniques "i.e., X-ray powder diffraction
and chemical analysis) to ensure that a possible
contribution from phase transitions was taken
into consideration. Milling caused the formation
of a crystalline mesophase and corresponding
changes in the reaction mechanism. Such possible
changes in a reaction mechanism should be taken
into consideration when carrying out stability
studies of solid drug formulations, when there is
evidence that the crystalline lattice of a drug may
be disturbed during the compression or milling
process.
ACKNOWLEDGMENTS
This research was supported by the joint Purdue/
Wisconsin Industrial Program on Molecular
Mobility in Solids. The authors express their
gratitude to Dr. S. Byrn for providing crystalline
TGME, Dr. K.A. Connors for help in carrying out
the HPLC measurements, and Ms. J.M. Milkaupt
for technical assistance.
REFERENCES
1. Saleki-Gerhardt A, Ahlneck C, Zogra® G. 1994.
Assessment of disorder in crystalline solids. Int J
Pharm 101:237±247.
2. Huttenrauch R, Fricke S, Zielke P. 1985. Mechan-
ical activation of pharmaceutical systems. Pharm
Res 2:302±306.
Scheme 2.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 2, FEBRUARY 2002

CHEMICALREACTIVITY OF AN ORGANIC SOLID
593
3. Byrn SR, Pfeiffer RR, Stowell JG. 1999. Solid-state
chemistry of drugs, 2nd ed. West Lafayette, IN:
SSCI, Inc. p. 482.
4. Otsuka M, Kaneniwa N. 1990. Effect of grinding on
the crystallinity and chemical stability in the solid
state of cephalothin sodium. Int J Pharm 62:65±73.
5. Takahashi Y, Nakashima K, Nakagawa H, Sugi-
moto I. 1984. Effects of grinding and drying on the
solid-state stability of ampicillin trihydrate. Chem
Pharm Bull 32:4963±4970.
transfer of tetraglycine methyl ester. Int J Pharm
152:75±88.
16. Wunderlich B. 1999. A classi®cation of molecules,
phases, and transitions as recognized by thermal
analysis. Thermochim Acta 340±341:37±52.
17. Takizawa T, Nakata Y. 2000. New small endother-
mic peaks with hysteresis commonly observed in
the differential scanning calorimetric study of
biopolymer±water systems. Thermochim Acta
352±353:223±231.
6. Pikal MJ, Lukes AL, Lang JE, Gaines K. 1978.
Quantitative crystallinity determinations for b-
Lactam antibiotics by solution calorimetry: Corre-
lations with stability. J Pharm Sci 67:767±773.
7. Kitamura S, Miyamae A, Koda S, Morimoto Y.
1989. Effect of grinding on the solid-state stability
of ce®xime trihydrate. Int J Pharm 56:125±134.
8. Thomas JM. 1979. Organic reactions in the solid
state: Accident and design. Pure Appl Chem 51:
1065±1082.
9. Luty T, Eckhardt CJ. 1995. General theoretical
concepts for solid state reactions: Quantitative
formulation of the reaction cavity, steric compres-
sion, and reaction-induced stress using an elastic
multipole representation of chemical pressure. J
Am Chem Soc 117:2441±2452.
18. Kim YJ, Suzuki T, Hagiwara T, Yamaji I, Rakai R.
2000. Enthalpy relaxation and glass to rubber
transition of amorphous potato starch formed by
ball-milling. Carbohydr Polym, to appear.
19. Parsonage NG, Staveley LAK. 1978. Disorder in
crystals. New York: Oxford University Press,
p. 514.
20. Pikal MJ, Dellerman KM. 1989. Stability testing of
pharmaceuticals by high-sensitivity isothermal
calorimetry at 258C: Cephalosporins in the solid
and aqueous solution states. Int J Pharm 50:233±
252.
21. Koenigbauer MJ, Brooks SH, Rullo G, Couch RA.
1992. Solid-state stability testing of drugs by
isothermal calorimetry. Pharm Res 9:939±944.
22. Beckmann JW, Wilkes JS, McGuire RR. 1977.
10. Shalaev EY, Zogra® G. 1996. Interrelationships
between phase transformations and organic che-
mical reactivity in the solid-state. J Phys Organic
Chem 9:729±736.
2,4,6-Trinitrotoluene
thermal
decomposition:
Kinetic parameters determined by the isothermal
differential scanning calorimetry technique. Ther-
mochim Acta 19:111±118.
11. Sluyterman LAá, Veenendaal HJ. 1952. Reactions
of polypeptide esters in the solid state, I. Migration
of methyl group. Recueil 71:137±253.
23. Chan HK, Doelker E. 1985. Polymorphic transfor-
mation of some drugs under compression. Drug Dev
Ind Pharm 11:315±332.
12. Sluyterman LAá, Kooistra M. 1952. Reactions of
polypeptide esters in the solid state, II. A few cases
of prevailing polycondensation. Remarks concern-
ing the reaction mechanism. Recueil 71:277±284.
13. Huang KS, Johnson P, Byrn SR. 1996. Solid-state
reactionsoftetrapeptideesters. Unpublishedresults.
14. Shalaev EY, Byrn SR, Zogra® G. 1997. Single-
phase and heterophase solid-state chemical
kinetics of thermally induced methyl transfer in
24. Parkinson GM, Goringe MJ, Ramdas S,
Williams JO, Thomas JM. 1978. A new phase of
crystalline anthracene: Its importance in solid-
state photodimerization. JCS Chem Commun
134±135.
25. Tong P, Zogra® G. 1999. Solid-state characteristics
of amorphous sodium indomethacin relative to its
free acid. Pharm Res 16:1186±1192.
26. Tsukushi I, Yamamuro O, Suga H. 1994. Heat
capacities and glass transitions of ground amor-
phous solid and liquid-quenched glass of tri-O-
methyl-b-cyclodextrin. J Non-Cryst Solids 175:
187±194.
tetraglycine methyl ester. Int
29:339±348.
J Chem Kinet
15. Shalaev EY, Shalaeva M, Byrn SR, Zogra® G. 1997.
Effects of processing on the solid-state methyl
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 2, FEBRUARY 2002