Journal of Thermal Analysis and Calorimetry, Vol.

68 (2002) 741–749

INFLUENCE OF MOISTURE ON THE GLASS

TRANSITION OF A SPRAY-DRIED COMPOUND
USING THE ISOSTEPTM METHOD

J. E. K. Schawe* and U. Hess

Mettler Toledo GmbH, Sonnenbergstrasse 74, CH-8603 Schwerzenbach, Switzerland

Abstract
The interaction among moisture content, solvent loss and glass transition temperature is relevant for
processing of spray-dried pharmaceuticals, since the glass transition temperature determines the ap-
plication range of a compound. Conventional Differential Scanning Calorimetry (DSC) does usually
not allow to separate glass transitions from common kinetic effects like evaporation or crystalliza-
tion. Based on classical DSC methods, the IsoStepTM method allows the independent determination
of heat capacities and kinetic effects, and thus, the separation of kinetic effects from effects arising
from heat capacity changes. This technique is used to separate glass transition and evaporation pro-
cesses, and to find the relation between moisture content and glass transition temperature for a phar-
maceutical sample based on a modified Gordon–Taylor equation.

Keywords: DSC, evaporation, glass transition, moisture

Introduction
Pharmaceutical substances are frequently spray-dried in order to remove solvent resi-
dues that remain from the production process. The solvent content influences the
phase behavior of the substances. In the case of amorphous or partially crystalline
materials solvent residues may act as plasticizers and may have an influence on the
glass transition. The glass transition temperature decreases with higher solvent con-
tent. This could be important for further processing since substances become often
sticky above the glass transition. Therefore, it is necessary to know the dependence of
the glass transition temperature on the solvent content.
Conventional Differential Scanning Calorimetry (DSC) is a well established tech-
nique for the investigation of glass transition. However, in the case of moist materials,
glass transitions in DSC curves are often overlapped by predominant moisture loss peaks
or annealing phenomena such as relaxation peaks. A technique which provides the capa-
bility of separation of such overlapping effects on the glass transition is Temperature

* Author for correspondence: E-mail: [email protected]

1418–2874/2002/ $ 5.00 Akadémiai Kiadó, Budapest

© 2002 Akadémiai Kiadó, Budapest Kluwer Academic Publishers, Dordrecht
742 SCHAWE, HESS: SPRAY-DRIED PHARMACEUTICAL COMPOUND

Modulated DSC® (TMDSC). Royall and co-worker used this technique for the analysis of
moisture uptake effects on the glass transition of pharmaceuticals [1].
On moist amorphous pharmaceutical compounds a separation of glass transitions
from overlapping evaporation peaks by TMDSC was performed by Schubnell and
Schawe [2] as well as a quantitative analysis of the glass transitions. Based on this
work the newly developed and straight forward IsoStepTM method was applied for heat
capacity determinations and simultaneous separations of an overlapping kinetic pro-
cesses. In the experiments described in this paper, the kinetic process is the evapora-
tion of water.
It is demonstrated that the dependence of the glass transition temperature on the
moisture content can be determined efficiently and accurately by the IsoStepTM
method.
A pharmaceutical compound was investigated. In order to avoid sticking of the
powder during further processing the dependence of the glass transition temperature
from the water content was investigated.

Experimental
The DSC experiments were performed using a Mettler-Toledo DSC 822e connected to
an intracooler and a Mettler-Toledo Thermogravimetrical Analyzer TGA/SDTA 851e.
Both systems are operated by the STARe Software where the DSC included the
IsoStepTM DSC software option. For the mass-spectrometry (MS) measurements a
Thermostar mass spectrometer (Balzers) was connected with the TG.
The IsoStepTM method combines a conventional heat capacity measurement with
a quasi-isothermal kinetics evaluation method [3]. The characteristic temperature pro-
gram is based on sequences of small temperature increases followed by isothermal seg-
ments for sample equilibration. The parameters of the temperature program (i.e. these
are the time of the isothermal, the heating rate and step high of the heating steps) are
not subject to any restrictions and can be varied during the measurement depending on
the thermal event. As described by Mraw and Naas [4], the heat capacities are given by
the heat flows of the temperature ramp segments. The isothermal segments contain the
information about kinetic processes and can be used to evaluate heat flows caused by
kinetic processes. A corresponding method, called Isothermal Stepping DTA, was de-
veloped by Doerr [5]. Because they are purely related to kinetic heat flows and do not
contain any information about heat capacity changes, these curves are very suitable for
water loss determination.
The measurements were performed with 1 K temperature increases and heating
rates of 2 K min–1 followed initially by 30 s isothermal segments. The measuring cell
was purged with nitrogen for all DSC and TG measurements.
The sample was a spray-dried mixture of two incompatible, amorphous, low mo-
lecular mass compounds with a remnant water content of about 6%. The dry mass con-
tents of the components were 65% and 35%. The sample has two glass transitions. It
becomes sticky above the first glass transition and liquid above the second one.

J. Therm. Anal. Cal., 68, 2002

 SCHAWE, HESS: SPRAY-DRIED PHARMACEUTICAL COMPOUND 743

The samples with masses of about 8 mg were placed in aluminum crucibles

sealed by standard pierced lids with 50 µm holes. Samples with different moisture
contents were prepared by preheating at 80°C for different lengths of time.

Results
The uppermost curve in Fig. 1 is a DSC curve of a sample that was not preheated be-
fore starting the IsoStepTM experiment. The curve was measured with a heating rate of
1 K min–1. The heat capacity curve and the kinetic curve resulting from the IsoStepTM
measurement are depicted below. There are several endothermic thermal events in
the DSC curve but a clear interpretation of the DSC curve is obviously difficult. A
better understanding can be obtained from the heat capacity curve: The step-like in-
creases of the specific heat capacity cp at approximately 68 and 120°C indicate two
glass transitions which are caused by the two amorphous components contained in
the sample. With this information given, the endothermic peaks of the DSC curve at
about 60 and 125°C can be identified as enthalpy relaxation peaks overlaying a huge
water loss peak. The enthalpy relaxation peak disappears after heating in the glass
transition range and reappears after a few days storage at room temperature. The ki-
netic curve at the bottom is very similar to the DSC curve. The difference is explained
by the heat flow contributions from the sample heat capacity which are not included
in the kinetic curve. Because its baseline is free of any contributions from heat capac-
ity changes, the kinetic curve was used for evaporation related evaluations.

Fig. 1 DSC curve (heating rate: 1 K min–1) and the results of the IsoStepTM experiment
of a not preheated sample. On top – DSC curve; middle – heat capacity curve;
bottom – kinetic curve; Φ – heat flow

In order to vary the moisture content the sample was preheated at 80°C for dif-
ferent lengths of time. Prior DSC, IsoStepTM and TG experiments showed the revers-
ibility of glass transition shifts and water absorption. As shown in Fig. 2 simulta-
neous mass loss measurements and evolved gas analyses proofed that the material

J. Therm. Anal. Cal., 68, 2002

744 SCHAWE, HESS: SPRAY-DRIED PHARMACEUTICAL COMPOUND

Fig. 2 TG curve of a heating experiment with 5 K min–1 (upper diagram). In the lower
diagram the related mass spectrometer curves for the mass numbers 18 and 44
are shown

does not decompose at 80°C. The TG heating curve (heating rate 5 K min–1, sample
mass 30 mg) and the evolved gas analysis of water (m/z=18) indicate maximum water
loss rates at about 80°C. At temperatures above 110°C a second specific mass signal
with m/z=44 appears and indicates a decomposition process.
In order to perform a systematic investigation of the influence of water on the
first glass transition at approximately 68°C (Fig. 1) the mass loss during preheating at
80°C is measured by TG. The measured curve is shown in Fig. 3 and the water losses
after 20, 40, 60, 80 and 160 min are indicated. These preheating periods were chosen
to prepare samples with different water contents for subsequent IsoStepTM experi-
ments. The water content of each preheated sample and before starting an IsoStepTM
experiment (Table 1) was determined from the mass losses measured in the TG ex-
periment (Fig. 2).
As shown in the kinetic curve in Fig. 4, the enthalpy relaxation peak at approxi-
mately 60°C disappears after preheating at 80°C. The sample was preheated for
20 min before the measurement. Compared to the heat capacity curve of the sample
without preheating treatment the glass transition has shifted from 68 to 72°C. From
the TG measurement in Fig. 2 the water content before the IsoStepTM measurement
was found to be 4.32%.
In order to take the additional water loss into account which occurs during the
IsoStepTM experiment itself, the percentage of the remaining water evaporated before
the glass transition was determined by evaluating the partial peak area of the evapora-
tion peak up to the glass transition temperature on the kinetic curve (Fig. 4). The nor-
malized partial peak area represents the specific evaporation enthalpy ∆he related to
the sample mass . The water loss during the IsoStepTM measurement ∆wm is given by

J. Therm. Anal. Cal., 68, 2002

 SCHAWE, HESS: SPRAY-DRIED PHARMACEUTICAL COMPOUND 745

∆he
∆wm = (1)
∆hw
where ∆hw is the specific evaporation enthalpy of water bound to the sample which
was determined as ∆hw=2500 J g–1 [2].
The water content at the glass transition temperature wTg is

Fig. 3 Isothermal TG experiment of a 8.102 mg in a 50 µm pierced crucible at 80°C

showing percentages of water loss after 20, 40, 60, 80 and 160 minutes of pre-
heating

Fig. 4 Heat capacity and kinetic curve of the IsoStepTM experiment after 20 min of pre-
heating. In the heat capacity curve the glass transition temperature is evaluated.
The determination of the evaporation enthalpy ∆he from the partial area is shown
on the kinetic curve

J. Therm. Anal. Cal., 68, 2002

746 SCHAWE, HESS: SPRAY-DRIED PHARMACEUTICAL COMPOUND

Fig. 5 Heat capacity curves obtained by the IsoStepTM method for different preheating
times. The preheating time is the parameter. The arrows indicate the glass transi-
tion temperature. Note: the moisture content at Tg is not only given by the pre-
heating time period but also by the moisture loss during the IsoStepTM
experiment

Fig. 6 Glass transition temperature of component A (Tg) vs. the overall water content
(ww) at Tg. The curve is a fit result of the Eq. (8)

wTg = wini −∆wm (2)

where wini is the initial water content after preheating and obtained from the TG mea-
surement in Fig. 3. Using Eq. (2) the water content at the glass transition temperature
was calculated.
The heat capacity curves are shown in Fig. 5 for various 80°C preheating peri-
ods. The glass transition temperature shifts to higher temperatures with increasing

J. Therm. Anal. Cal., 68, 2002

 SCHAWE, HESS: SPRAY-DRIED PHARMACEUTICAL COMPOUND 747

preheating periods and hence, decreasing moisture contents. The glass transition tem-
peratures (indicated by the arrows) are determined from these curves. For a two-
component system, containing one drug component and water, the relation between
moisture content and glass transition temperature would be given by these results and
could be displayed by a curve similarly to the one displayed in Fig. 6. Such a curve
may be used to optimize industrial processing conditions. However, the sample in-
vestigated here is a three-component system, consisting of two drug components and
water. For two incompatible drug components with a known molecular mass distribu-
tion, there will be only one additional unknown parameter describing the water distri-
bution within the two components. By using the Gordon–Taylor equation and curve
fitting, the water distribution can be determined and the relation between glass transi-
tion temperature and the water content described.

Discussion
The relationship between the glass transition temperature Tg and the fractional mois-
ture uptake ww of a two-component system can be described by the Gordon-Taylor
equation [5]:
wATgA +KwwATgw
Tg = (3)
wA +KwwA
where wA and wwA are the mass fractions of a compound A and water absorbed in A.
TgA and Tgw are the glass transition temperatures of the pure compound and water. K is
an empirical parameter. This equation is often used for the quantitative estimation of
the water dependence of the glass transition temperature of drugs [1].
Since the investigated system is a three-component system, consisting of water
and the incompatible pharmaceutical compounds A and B, the Gordon–Taylor equa-
tion has to be modified accordingly. Particularly, the water distribution has to be
taken into account as ∆ mw, this is the water portion absorbed by component B. The
∆

mass fractions of Eq. (3) are then:

mA
wA = (4)
mA +mwA −∆mw
mwA −∆mw
wwA = (5)
mA +mwA −∆mw
where mA is the dry mass of the component with the low temperature glass transition
and mwA the mass of water absorbed in A. The masses of Eqs (4) and (5) can be ex-
pressed by the measurable fractions ww, the overall water content, and wAd, the mass
fraction of the component A relative to components A and B:
mw
ww = (6)
mA +mB +mw

J. Therm. Anal. Cal., 68, 2002

748 SCHAWE, HESS: SPRAY-DRIED PHARMACEUTICAL COMPOUND

mA
wAd = (7)
mA +mB
mB and mw are the masses of the component B and of water absorbed to A and B. Rear-
ranging Eqs (6) and (7) and substituting in Eqs (4) and (5) a modified Gordon–Taylor
equation is obtained
TgA +δq w Tgw
Tg = (8)
1+δq w
where qw is a measure of the water content:
ww
qw = (9)
1−ww
The parameter δ depends on the relative amount of water in the component A:
 ∆m  K
δ= 1− w  (10)
 mw  wAd
The parameter of Eq. (8) are determined by non-linear curve fitting of the data in
Fig. 5. The glass transition temperature of water is assumed to be Tgw=136 K [6]. The
fitting procedure results in a glass transition temperature of TgA=379.5 K for com-
pound A and δ=3.65. The fit-curve is shown in Fig. 6. If the parameter K is known,
the affinities of both components for water absorption can be determined. An approx-
imation of K is
ρ ATgA
K= (11)
ρ w Tgw

where ρA and ρw are the densities of component A and water [1]. If a density of
ρA=1.2 g cm–3 for a typical organic drug is assumed, K is estimated to be 3.35. The
known ratio of the two compounds can be expressed as wAd=0.65. From Eq. (10) fol-
lows that (1–∆mw/mwA)= 0.71. This means that 71% of the water is absorbed by com-
ponent A. When the weight distribution of A and B is taken into account it follows that
component A absorbs 10% more water than component B.

Conclusions
The IsoStepTM method is a powerful tool for the separation of glass transitions and
overlapping kinetic effects. Particularly, the separation of solvent evaporation and
glass transition is an important application for process control in the production pro-
cess of pharmaceuticals. The experiments allow to relate moisture contents to glass
transition temperatures, and to thus find drying conditions that do not affect sample
integrity, are appropriate for further sample processing and economic, in terms of re-
quired heat and time for drying.

J. Therm. Anal. Cal., 68, 2002

 SCHAWE, HESS: SPRAY-DRIED PHARMACEUTICAL COMPOUND 749

The experiments have shown that the glass transition can shift from 68°C for a
moist sample to 107.5°C for an almost completely dry sample. This means, for in-
stance, that the moist sample is likely to be sticky at temperatures above 68°C. This
will possibly affect the result of subsequent processing steps.
The intention of this paper was to introduce a method for finding a relation be-
tween moisture content and glass transition temperature and thus, to provide a practi-
cal tool for process optimization. Using a modification of the Gordon–Taylor equa-
tion the dependence of the the glass transition temperature from the moisture content
can be described for a three-component system.

References
1 P. G. Royall, D. Q. M. Craig and C. Doherty, Int. J. Pharm., 192 (1999) 39.
2 M. Schubnell and J. E. K. Schawe, Int. J. Pharm., 217 (2001) 173.
3 S. C. Mraw and D. F. Naas, J. Chem. Thermodyn., 11 (1979) 567.
4 W. F. Hemminger and H. K. Cammenga, Methoden der Thermischen Analyse,
Springer-Verlag, Berlin 1989, p. 260.
5 L. A. Wood, J. Polym. Sci., 23 (1958) 319.
6 G. Widmann and R. Riesen, J. Therm. Anal. Cal., 52 (1998) 109.

J. Therm. Anal. Cal., 68, 2002