Thermogravimetric analysis during the decomposition of
cotton fabrics in an inert and air environment
Julia Molto ´*, Rafael Font, Juan A. Conesa, Ignacio Martı ´n-Gullo ´n
Chemical Engineering Department, University of Alicante, P.O. Box 99, E-03080 Alicante, Spain
Received 29 November 2004; accepted 4 September 2005
The thermal degradation of samples of used cotton fabrics has been investigated using thermogravimetric analysis (TGA) between room
temperature and 700 8C. Experiments were carried out with about 5 mg of sample in three different atmospheres: helium, 20% oxygen in helium
and 10% oxygen in helium. Three different heating rates were used at each atmosphere condition. A kinetic model for the decomposition of used
cotton fabrics explaining the behavior of all the runs performed has been proposed and tested. For the pyrolysis of the cotton, the model comprises
two parallel reactions. For the combustion process, one competitive reaction was added to each parallel reaction of the pyrolysis model and four
combustion reactions of the different solid fractions to obtain volatiles. One single set of parameters can explain all the experiments (pyrolysis,
oxidative pyrolysis and combustion) at the three different heating rates used.
# 2005 Published by Elsevier B.V.
Keywords: Pyrolysis; Combustion; Cotton fabrics; Kinetics; Thermogravimetry
The growing interest in renewable energies is accompanied
by intensified research and development of technical processes
for the thermal conversion of biomass and wastes. Used cotton
fabrics could be used as biomass, and in this way, offer a valid
alternative to disposal in landfills. Although cotton fabrics are
usually recycled in other ways, thermal decomposition only of
the wastes is interesting for some industries that mainly focus
on obtaining the potential energy by combustion.
The development ofpyrolysis–combustion
requires an optimization of the operating conditions in order
to assure both acceptable gas outlet composition and an energy
recovery, which makes the process economically satisfactory
. A good knowledge of the kinetics of the process is
fundamental for the plant design and scale-up bases on process
Cotton is mainly comprised of cellulose. The pyrolytic
degradationof cellulose hasbeen the subject of extensivestudy,
although in many instances, knowledge of the exact nature of
degradation and decomposition remains incomplete. Chatterjee
and Conrad  studied the kinetics of cellulose decomposition
in the temperature range of 270–310 8C with absorbent cotton,
and proposed two series reactions for the pyrolysis process.
Dollimore and Hoath  used differential thermal analysis
(DTA) to follow the degradation of cellulose in air products and
obtained two and sometimes three exothermic peaks. Antal and
Va ´rhegyi  reviewed the literature of cellulose pyrolysis and
concluded that the pyrolysis of a small sample of pure cellulose
is characterized by an endothermic reaction governed by a first
order rate law with a high activation energy. Vo ¨lker and
Rieckmann  investigated the influence of the final mass on
modelling results and evaluated the applicability of established
kinetic models for engineering purposes.
Cotton fabrics, which have a major share of the textile
market, are highly inflammable and the development of
successful flame retardant systems for cotton is of major
interest. For this reason, many authors have studied the
cotton fabrics. For instance, Faroq et al.  carried out the
thermogravimetric analysis of the mechanism of pyrolysis of
untreated cotton fabrics and cotton fabrics finished with various
flame retardant, considering the fraction decomposed as
between 0.1 and 0.9. These authors evaluated the activation
J. Anal. Appl. Pyrolysis xxx (2005) xxx–xxx
* Corresponding author. Tel.: +34 96 590 34 00x3003; fax: +34 96 590 38 26.
E-mail address: firstname.lastname@example.org (J. Molto ´).
0165-2370/$ – see front matter # 2005 Published by Elsevier B.V.
energies obtaining values of 155–158 kJ mol?1for the thermal
decomposition of untreated cotton. In addition, they proposed a
second order decomposition mechanism as the most suitable.
Wanna and Powell  studied the thermal decomposition of
untreated and treated cotton fabric with selected salts in
oxidative and inert atmospheres using TGA-FTIR. On the other
hand, no papers considering the overall decomposition kinetics
of cotton cellulose or cellulose have been found.
It is clear that the n-th order kinetic model can be correct for
homogeneous gas phase kinetics, in accordance with the
collision theory and the transition state theory developed for
elemental reactions and normally with n = 1, 2 or 3. The
pyrolysis of polymers it is a non-homogeneous reaction and
consequently the n-th model should not be valid. In literature,
we can find mechanisms for the pyrolysis of different polymers
with parallel and series reactions. Considering only one
elemental reaction for the decomposition of a solid, the
proposal ofakineticexpressionbased onthe elementalreaction
is not easy because the decomposition takes place in a solid
phase. Font and Garcı ´a  proposed the application of the
transition state theory to the pyrolysis of biomass considering
the similarity between the pyrolitic reaction that take places in
the outer surface and the first order uni-molecular catalytic
surface reactions. In accordance with the model developed, it is
possible to obtain n-th models as a consequence of the increase
or decrease of the surface with active centres where the
decomposition can take place, and with fractional values of the
reaction order n.
The n-th order reaction has been used extensively by
different researchers when studying the mechanisms of
decomposition of several polymers, no matter the mechanisms
of reactions comprises parallel or series reactions. Never-
theless, in spite of the proposal of a model that could explain
fractional reaction orders, the models obtained from the TG
runs must be considered as correlation ones, and from the
analysis of the activation energy, it can be deduced if the model
is related satisfactorily to a controlling decomposition
elemental step or must be only considered valid for correlation.
The present work, which is included in a wider project
whose objective is to study the combustion of different
industrial and municipal wastes, studies the thermal decom-
position of used and waste cotton fabrics from the thermo-
gravimetric point of view, including the kinetic analysis in a
thermobalance in inert atmosphere and with different amounts
of oxygen, proposing a kinetic model.
2.1. Raw material
Used cotton fabrics were simulated by using a used blue T-
shirt made of 100% cotton. Prior to the runs, the T-shirt was cut
into small pieces with an average size of 1 cm ? 1 cm.
Table 1 shows some characteristics of the material studied.
Elemental analysis of the major components was carried out in
a Perkin-Elmer 2400. The moisture was determined by the
weight loss at 105 8C for 12 h. The calorific valuewas obtained
in an AC-350 calorimetric bomb from Leco Corporation.
Chlorine was measured using an automatic sequential spectro-
meter X-ray Fluorescence model TW 1480. Ash residue was
obtained by calcination at 850 8C.
The thermogravimetric experiments were carried out in a
Setaram thermobalance model DSC92 controlled by a PC
system. The atmosphere used for pyrolysis was helium with a
flowrateof60 ml min?1(STP),accordingtothespecificationsof
the equipment. In the combustion runs, two mixtures of helium
with the same total flow rate. The sample temperature was
measured with a thermocouple directly at the crucible, i.e., very
close to the sample. Because a water-cooled microfurnace was
used, the temperature could be lowered rapidly.
Before the runs with used cotton fabrics, an experiment with
a heating rate of 5 8C min?1usingAvicel PH-105 microcrystal-
line cellulose was done to check the good performance of the
equipment. The results obtained showed good agreement with
the kinetic evaluation of Avicel Cellulose TG curves at this
heating rate presented by Grønli et al.  in their round-robin
study of cellulose pyrolysis kinetics by thermogravimetry.
The experiments were carried out with heating rates of 5, 10
and 20 8C min?1over a variety of temperatures that included
the entire range of solid decomposition, 80–700 8C. Experi-
ments without a sample were carried out, and used as
background in order to subtract the buoyancy effect. The mass
of the samples used was approximately 5 mg, and under these
conditions the heat transfer limitations can be neglected.
3. Results and discussion
3.1. Thermogravimetric study
Figs. 1–3 show in detail the experimental curves for used
cotton fabrics pyrolysis (helium) and combustion (helium:oxy-
gen, 4:1 and 9:1) at different heating rates. The calculated
curves of the kinetic models are also shown. In the figures, w is
defined as the residual mass fraction of the solid (including
residue formed and non-reacted initial solid), i.e., the ratio
(m0). In all the processes, we can be observed the general shift
to higher temperatures when the heating rate is increased.
J. Molto ´ et al./J. Anal. Appl. Pyrolysis xxx (2005) xxx–xxx2
Characteristics of the material used
O% by difference (wt.%)
Ash content (wt.%)
Net calorific value (kJ kg?1)
Fig. 4 compares the results obtained in the different
atmospheres at a constant heating rate. As can be seen, the
at lower temperatures) and a final combustion process is
observed. On the other hand, there is not a great difference
between the runs performed under 10% oxygen and those at
3.2. Kinetics of the process
3.2.1. Pyrolysis model
Most of the papers published corresponding to the
pyrolysis of cellulose and cotton consider only one fraction
when correlating the experimental data of the primary
decomposition, in spite of the parallel and series mechanisms
proposed in literature . In the cotton fabrics used in this
work, and considering that we have extended the tempe-
rature range until high temperatures, a second decompo-
sition process has been considered in order to improve the
The kinetic model proposed for the decomposition of the
used cotton fabrics in an inert atmosphere could be interpreted
considering this waste formed by two independent parts, each
one following an independent reaction, as follows:
where C1and C2refer to different parts of the solid material to
be decomposed (cotton in this case). ‘‘Vi’’ are the gases + vo-
latiles evolved and ‘‘Si’’ are the solid residue formed in the
decomposition. The uncapitalized variables ‘‘cn’’ are the
amount of each material that is in the sample.
It is very useful to introduce the concept of the conversion
degree for each reaction:
ai¼ 1 ?Cn
i ¼ n ¼ 1; 2
(Two different subscripts have been used, because in the
combustion model presented also in this model there are
competitive reactions for the same solid.)
In the previous equations, Vi1represents the maximum
obtainable amount of volatiles via reaction ‘i’ at time infinity,
J. Molto ´ et al./J. Anal. Appl. Pyrolysis xxx (2005) xxx–xxx3
Fig. 1. Cotton pyrolysis at several heating rates: 5, 10 and 20 8C min?1.
Experimental and calculated curves.
Fig. 2. Cotton combustion at several heating rates and 20% oxygen. Experi-
mental and calculated curves.
Fig. 3. Cotton combustion at several heating rates and 10% oxygen. Experi-
mental and calculated curves.
Fig. 4. Experimental TG plots for pyrolysis and combustion with 20 and 10%
oxygen, all carried out at 20 8C min?1heating rate.
Cnrefers to the non-decomposed material at each time, and Cno
is the initial contribution offraction ‘n’to the total weight.Note
that at t = 0 thevalue of aiis zero, and that V11equals the yield
coefficient v1and that V21equals the yield coefficient v2.
The kinetic equations associated with the parallel reaction
for the pyrolysis runs, taking into account the mass balance
between products and reactants and the degreeconversions, can
be expressed as:
By integration of these reactions it is possible to calculate a1
and a2at each time; the relationship between these two values
and the weight fraction measured in the thermobalance ðwÞ is:
w ¼ 1 ? V ¼ 1 ? ðV1þ V2Þ ¼ 1 ? ðV11a1þ V21a2Þ
The values of V11 and V21 are related with the total
volatiles at time infinity (V1), that is a known amount:
V1¼ V11þ V21
The pyrolysis data and the combustion data have been
correlated together, in order to obtain a single set of parameters
atmospheres: pyrolysis, oxidative pyrolysis and combustion.
3.2.2. Combustion model
Similarly to the kinetic models found in literature to explain
pyrolytic processes, different authors propose different models
to explain decomposition mechanisms under oxidative atmo-
spheres. Obviously, if the model proposed satisfactorily fits
experimental data generated under a wide selection of
conditions, the model can be considered representative of
the process analyzed . However, literature that includes
detailed kinetic studies, fitting experimental curves at different
heating rates and oxygen content is extremely sparse. Different
models have been found in literature, concerning the
decomposition of tannery waste under oxygenated atmosphere
, polycoated materials such as milk cartons ,
polytetrafluoroethylene  and also of tire wastes . No
models for combustion of cotton or used cotton fabrics have
been found, as commented previously.
The kinetic model proposed for the combustion runs could
be interpreted considering, that the presence of the oxygen
introduces a new competitive process for the decomposition of
each fraction and in this way this model explains the fact
observed in Fig. 4, where the first weight loss in combustion
runs reaches a lower value than in pyrolysis runs.
Figs. 5 and 6 show the TG and DSC plots for pyrolysis
and combustion at 5 8C min?1heating rate. As seen in Fig. 5,
an endothermic peak appears for the pyrolysis process
and in the combustion process besides two exothermic peaks
Taking into account the behavior commented above, the
oxygen to be included in the decomposition law has been
considered, as has been done with other materials [15–16].
The following scheme represents the combustion model
proposed to explain the behaviour obtained at two different
helium: oxygen atmospheres and three heating rates:
c1C1þ O2? !
c2C2þ O2? !
J. Molto ´ et al./J. Anal. Appl. Pyrolysis xxx (2005) xxx–xxx4
Fig. 6. Experimental TG and DSC plots for combustion with 20% oxygen at
5 8C min?1heating rate.
Fig. 5. Experimental TG and DSC plots for pyrolysis at 5 8C min?1heating
ki¼ kioexp ?Ei
¼ kið1 ? aiÞni
n;i ¼ 1;2
Reactions 1 and 2 are the same as in pyrolysis. Furthermore,
as a final new process appears in combustion runs, the
combustion reactions to the chars formed has been considered:
siSiþ O2? !
Vcirefers to the volatiles produced through the combustion
reaction with solid Si.
Conversion degrees are considered for each reaction:
i ¼ 1;2;3;4
ai¼ 1 ?ðCnÞreactedbyreactioni
i ¼ 1; 2; 3; 4; n ¼ 1ðfori ¼ 1; 3Þ; n ¼ 2ðfori ¼ 2; 4Þ
i ¼ 1; 2; 3; 4; n ¼ 1ðfori ¼ 1; 3Þ; n ¼ 2ðfori ¼ 2; 4Þ
where Vi1 and Vci1 represent the maximum amounts of
volatiles evolved if the whole solid fraction decomposes only
through the reactions that lead to the corresponding volatiles,
without competitive reaction. Note that in this case Vi1equals
the yield coefficient viand that Vci1equals the yield coefficient
Reactions 1 and 3 are competitive with respect to the same
solid C1, so it is possible that none of the two values of degree
conversion a1and a2can reach the value 1 at time infinity,
although the sum a1and a2logically must be equal to 1, when
the reactant C1is exhausted. In this case, the ratio between the
non-reacted solid C1 and the initial solid C1o, taking into
account the degree conversions a1and a3, can be expressed as:
C1o¼ 1 ? a1? a3
and consequently the kinetic laws for the decomposition of the
solid fraction C1can be written as
¼ kið1 ? a1? a3Þni;
i ¼ 1;3
Similarly for reactions 2 and 4, the following expression can
¼ kið1 ? a2? a4Þni;
i ¼ 2;4
For the combustion of the residue S1(formed by the first
reaction) in accordance with the scheme:
s1S1þ O2? !
The degree conversion ac1is
The kinetic law for the combustion of S1can be written as:
In this way, the kinetic constant of this second decomposi-
tion does not depend on the initial mass fraction, and
consequently the kinetic constant k5 with distinct mass
On the other hand, it can be deduced that:
Consequently, it can be written that:
¼ a1? ac1
¼ k5ða1? ac1Þn5
Another way of obtaining similar expressions can be found
elsewhere . This procedure can be applied to the other three
combustion reactions. Consequently, the reaction model can be
solved considering the following equations:
dt¼ kið1 ? a1? a3Þni;
i ¼ 1;3
dt¼ kið1 ? a2? a4Þni;
i ¼ 2;4
¼ k5ðai? a1iÞn5;
i ¼ 1;2;3;4
v1þ s1¼ v3þ s3
v2þ s2¼ v4þ s4
Note that the same kinetic constant and reaction order are
considered for the decomposition of Si.
The total weight fraction is related to the other variables by:
w ¼ 1 ? V
¼ 1 ? ðV1þ V2þ V3þ V4þ V11þ V12þ V13þ V14Þ
w ¼ 1
? ða1V11þ a2V21þ a3V31þ a4V41 þ ac1Vc11
þ ac2Vc21þ ac3Vc31þ ac4Vc4aÞ
To take into account the effect of the partial pressure of
oxygen (which equals 0.10 and 0.20 atm for helium:oxygen 9:1
and 4:1, respectively), since different behaviour is observed
when comparing corresponding thermograms, the pre-expo-
nential factors for reactions with oxygen have been considered
to consist of two terms, one a typical pre-exponential factor k0
and the other one the partial pressure of oxygen PO2raised to
the power of an order bio:
i ¼ 3;4;5
J. Molto ´ et al./J. Anal. Appl. Pyrolysis xxx (2005) xxx–xxx5
297 298 299
The parameters optimized were: five orders of reaction, five
pre-exponential factors, five activation energies and v1, v2, v3,
v4, s1, s2, and the values of b3, b4, b5. A simplification made in
the model is to assume the same dependency of the constants
with the partial pressure of oxygen, in the combustion of the
The objective function (OF) to minimize was the sum of the
square differences between experimental and calculated weight
m; heatingrates; j; points
The model validity has been tested calculating the variation
where N and P are the number of data and parameters fitted,
respectively, and wexp is the average of the experimental
weights. According to the procedure suggested by Martı ´n-
Gullo ´n et al. , the great interrelation existing among the
pre-exponential factor, the apparent activation energy and the
reaction order can be decreased when optimization is per-
formed in terms of a ‘comparable kinetic constant’ K?
of optimizing koi. This constant is calculated in a temperature
around the maximum decomposition rate (Tmax). Since K?
and niare optimised the pre-exponential factor koiis calculated
using the following expression:
Conesa et al. [19,20] came to a conclusion in their study of
TG-DTG curves that at least three TG curves with different
heating rates must be adjusted simultaneously in order for a
kinetic model to be considered as potentially correct.
The kinetic parameters have been optimized in order to
minimize the differences between experimental and calculated
weight loss. Table 2 presents the values of the optimised
parameters. Note that all the data of the three atmospheres
(pyrolysis, oxidative pyrolysis, and combustion) and the three
heating rates have been simultaneously fitted.
As can be seen in Figs. 1–4, the model is able to explain
all the experimental data collected, at all the heating rates
studied and in the presence and absence of oxygen. Bear in
mind that all the runs are fitted with no variation of the
On analysing the values of the parameters in Table 2, some
interesting conclusions can be obtained:
OF=ðN ? PÞ
i¼ kið0:64Þni¼ koiexp
1. Considering the pyrolysis parameters, it can be observed that
fraction 1 has a value of v1, that equals a V11, around 0.75,
indicating that this fraction is the most important, although
there is a small but significant second fraction, with a value
v2 around 0.11. Analysing the TG data of some papers
concerning the pyrolysis of cellulose, this second fraction is
also present [21,5].
2. Concerning the kinetic parameters of the main fraction in the
161 kJ mol?1and reaction order around 0.57 are obtained,
that are similar to the values proposed by Antal et al. (1980)
when considering all the runs. The activation energy E1
et al.  using two different methods: the iso-conversional
technique and the most probable mechanism.
3. Withrespecttotheminorfraction inpyrolysis,alowvalueof
activation energy around 65 kJ mol?1and a reaction order
equal to 2.77 are obtained, as a consequence of the slow
decomposition of this fraction at high temperatures, and
consequently these values can be considered as correlation
parameters with no physical meaning.
4. The oxidative pyrolysis of the main fraction has activation
energy(159 kJ mol?1)similar
(161 kJ mol?1), and the reaction order (0.29) is somewhat
lower than in pyrolysis (0.57). An order reaction less than
unity can be due to an increase of the active surface with the
reaction extension , so it is possible that in the oxidative
process, the active surface increases more intensively than in
pyrolysis. The value of v3(0.569), that equals V31, is less
than the value (V11= 0.755) corresponding to the main
fraction in pyrolysis. This fact could be explained as a
tothat of pyrolysis
J. Molto ´ et al./J. Anal. Appl. Pyrolysis xxx (2005) xxx–xxx6
391 392 392
Kinetic parameters obtained for the thermal decomposition of used cotton
ioin s?1atm?bi, Eiin kJ mol?1)
9.76 ? 1010
1.32 ? 103
2.70 ? 1011
1.33 ? 1023
Combustion of char
4.47 ? 106
consequence of the partial oxidation of the molecular organic
chains to ketones, aldehydes, carboxylic acids, etc., causing a
weightloss lesser than in the casewithout oxygen. Therefore,
all these parameters can be considered as acceptable and
logical from a physical–chemical point of view.
energy is around 150 kJ mol?1, the reaction order is 0.97 and
there isno formation ofresidue solid. These parametersmust
only be accepted for correlation of the data.
6. In the combustion of the chars, assuming the same kinetics
for the three residue solids formed (S1, S2 and S3), the
apparent activation energy is around 126 kJ mol?1. The
apparent activation energy obtained for the combustion of
the char formed (126 kJ mol?1) is similar to that obtained by
other researchers. Haji-Sulaiman and Aroua  proposed
values around 135 kJ mol?1for the oxidation of Malaysian
coal chars. Henrich et al. , also obtained values of
apparent activation energy around 140 kJ mol?1are for
oxidation in pure oxygen of soot, graphite, activated
charcoals and chars of municipal waste and electronic
scrap. Nevertheless, Walker et al.  suggested apparent
activation energy around 209–242 kJ mol?1for the reaction:
C + 1/2 O2! CO. A reaction order (0.57) less than unity is
obtained. This latter aspect could also be explained as a
consequence of the increase of the active surface with the
7. The reaction orders with respect to the oxygen for the
the smaller fraction, indicatinga lowdependence ofthis with
respect to the oxygen. For the combustion of char the
reaction order with respect to oxygen is 0.31, that is a low
value in comparison with respect to the reaction order
between 0.7 and 1 considered in the gasification of the
carbon . The experimental data shown in Fig. 4 shows
the small dependence of the oxygen on the TG runs. On the
other hand, it is possible that the simplification of the model,
considering that the three combustion reactions follow the
same kinetic law causes the correlation parameters to be
somewhat different to those expected.
A study of the main by-products formed in the pyrolysis and
combustion of used cotton fabrics, was carried out in a very
recent paper . More than 90 compounds, including carbon
oxides, light hydrocarbons and PAHs, have been identified and
quantified. In the gas phase some of the main components
obtained were methane, ethene and benzene. The main
semivolatile compounds detected were styrene, phenol,
naphthalene, acenaphthylene and phenanthrene.
The evolution of the different fractions throughout the
heating process can be observed in Figs. 7 and 8, that show the
experimental and calculated mass fraction of volatiles v and
conversion factors a at 5 8C min?1in He:O24:1. The kinetics
parameters for reaction 2 show that there is a second order
reaction with a low activation energy, and this could explain
that the decomposition of the fraction C2occurs in a wide range
of temperatures, as seen in the big separation between the
curves of a2and a4(Fig. 8).
It must be emphasised that the parameters obtained must be
considered as correlation ones, some of them with physical–
chemical significance. There is a great interrelation between
them, so other sets of parameters could be also valid for the
correlation of the data with a very small increase of the
objective function. In spite of this great interrelation of the
parameters, it must be indicated that the objective function has
been minimized reducing the great interrelation that exits
between the pre-exponential factor, the activation energy and
the reaction order.
The correlation of the three pyrolysis runs and six
combustion runs have been done simultaneously, so 26
parameters were optimized for minimizing the objective
function. No papers have been found in literature where
pyrolysis and combustion runs were optimized simultaneously,
so the correlation work presented in this paper can be
considered as original. As a result, an acceptable set of
parameters is obtained, that correlates the nine TG runs
J. Molto ´ et al./J. Anal. Appl. Pyrolysis xxx (2005) xxx–xxx7
Fig. 7. Calculated conversion factor (ai, aij) at 5 8C min?1in He:O24:1.
Fig. 8. Calculated mass fraction of volatiles (vij) at 5 8C min?1in He:O24:1.
UNCORRECTED PROOF Download full-text
satisfactorily, as can be seen in Figs. 1–3 that show the
conversion degrees versus temperature for a run, observing a
logical variation in accordance with the parameters deduced.
Support for this work was provided by the Ministerio de
Educacio ´n y Ciencia of Spain, research projects PPQ2002-
00567 and PPQ2002-10548-E.
 J.A. Caballero, R. Font, M.M. Esperanza, J. Anal. Appl. Pyrol. 47 (1998)
 P.K. Chatterjee, C.M. Conrad, Text. Res. J. 36 (1966) 487.
 D. Dollimore, J.M. Hoath, Thermochim. Acta 45 (1981) 87.
 M.J. Antal, G. Va ´rhegyi, Ind. Eng. Chem. Res. 34 (1995) 703.
 S. Vo ¨lker, Th. Rieckmann, J. Anal. Appl. Pyrol. 62 (2002) 165.
 A.A. Faroq, D. Price, G.J. Milnes, Polym. Degrad. Stab. 44 (1994) 323.
 J.T. Wanna, J.E. Powell, Thermochim. Acta 226 (1993) 257.
 R. Font, A.N. Garcı ´a, J. Anal. Appl. Pyrol. 35 (1995) 249.
 M. Grønli, M.J. Antal, G. Va ´rhegyi, Ind. Eng. Chem. Res. 38 (1999) 2238.
59 (2001) 619.
 J.A. Caballero, R. Font, M.M. Esperanza, J. Anal. Appl. Pyrol. 47 (1998)
 J.A. Reyes, J.A. Conesa, A. Marcilla, J. Anal. Appl. Pyrol. 58–59 (2001)
 J.A. Conesa, R. Font, Polym. Eng. Sci. 41 (2001) 2137.
 J.A. Conesa, R. Font, A. Fullana, J.A. Caballero, Fuel 77 (1998) 1469.
 S. Vossoughi, Y.M. El-Shoubary, Thermochim. Acta 157 (1995) 37.
 I. Martı ´n-Gullo ´n, M. Esperanza, R. Font, J. Anal. Appl. Pyrol. 58–59
 R. Font, I. Martı ´n-Gullo ´n, M. Esperanza, A. Fullana, J. Anal. Appl. Pyrol.
58–59 (2001) 703.
 I. Martı ´n-Gullo ´n, M.F. Go ´mez-Rico, A. Fullana, R. Font, J. Anal. Appl.
Pyrol. 68–69 (2003) 645.
 J.A. Conesa, J.A. Caballero, A. Marcilla, R. Font, Thermochim. Acta 254
 J.A. Conesa, A. Marcilla, R. Font, J.A. Caballero, J. Anal. Appl. Pyrol. 36
 M.J. Antal, G. Va ´rhegyi, E. Jakab, Ind. Eng. Chem. Res. 37 (1998) 1267.
 M.Z. Haji-Sulaiman, M.K. Aroua, J. Inst. Energy 70 (1997) 52.
 E. Henrich,S. Bu ¨rkle,Z.I. Meza-Renken, S. Rumpel,J. Anal. Appl.Pyrol.
9 (1999) 133.
 P.L. Walker, F. Rusinko, L.G. Austin, Adv. Catal. 9 (1959) 133.
 Z. Du, A.F. Sarofim, J.P. Longwell, in: J. Lahay, P. Ehrburger (Eds.),
Fundamental Issues in Control of Carbon Gasification, 1992, p. 91.
 J.Molto ´,J.A.Conesa,R.Font,I.Martı ´n-Gullo ´n,Environ.Sci.Technol.39
(14) (2005) 5141.
J. Molto ´ et al./J. Anal. Appl. Pyrolysis xxx (2005) xxx–xxx8