ABSTRACT

Different models are capable to correlate a set of dynamic conversion data obtained at different heating
rates.

The question that rises is: does this fact invalidates the kinetic analysis of data at different heating rates,
as the widely accepted similar argument seems to do it in the case of data obtained at a single heating
rate?.

We believe that considering data at different heating rate is better than treating them at a single heating
rate, but none of the two possibilities guarantee the validation of a kinetic model in the mechanistic
sense. We believe that the capability of a model to represent a set of data (either at a single or at multiple
heating rates) is a necessary but not sufficient condition to validate a kinetic model. In other words,
conversion curves obtained, for example, directly from weight loss data seems to be incomplete data to
asses mechanistic results. To attain such a goal, thermogravimetric data must be complemented with
other type of techniques. The analysis of the gases evolved or the analysis of the solid fraction remaining
would be considered among other possibilities. Nevertheless, thermogravimetric data fitting at different
rates is a powerful tool to discard pseudokinetic models in order to determine apparent kinetic
parameters capable of simulate the weight loss processes under conditions similar to those present in the
thermogravimetric equipment.

In this work we show different examples where we illustrate that different models, with different kinetic
parameters, could be able to reproduce the same conversion curves at different heating rates
simultaneously. Thus, using a similar reasoning, the kinetic analysis, even considering data at different
heating rates should be rejected.

Far of this, we believe that is always better considering data at different heating rate rather than at a
single heating rate, but none of the two possibilities guarantee the validation of a kinetic model in the
mechanistic sense. The capability of a model to represent a set of data (either at a single or at multiple
heating rates) is a necessary but not sufficient condition to validate a kinetic model. In other words,
conversion curves obtained, for example, directly from weight loss data seems to be incomplete data to
asses mechanistic hypothesis. To attain such a goal, thermogravimetric data must be complemented with
other type of techniques. The analysis of the gases evolved or the analysis of the solid fraction remaining
would be considered among other possibilities. Nevertheless, thermogravimetric data fitting at different
rates is a powerful tool to discard pseudokinetic models in order to determine pseudokinetic parameters
capable of simulate the weight loss processes under conditions similar to those present in the
thermogravimetric equipment.


ACKNOWLEDGEMENT
The authors of the work wish to thank financial support provided by the Spanish Comisin de Investigacin Cientfica y
Tecnolgica de la Secretara de Estado de Educacin, Universidades, Investigacin y Desarrollo and the European
Community (FEDER refunds) (CICYT CTQ2004-02187) and by the Generalitat Valenciana (project GRUPOS03/159).
RESULTS
Conversion curves have been obtained from different models and different sets of kinetic parameters, by
integration by the 4
th
order Runge-Kutta procedure the equation:
INTRODUCTION

Dr. Marek Maciejewski reported in 1999 in an excellent work the second part of a series of papers
dealing with the results of the ICTAC (International Confederation for Thermal Analysis and
Calorimetry) Kinetic Analysis Project [1-5]. We fully agree with most of the contents of the paper
that was presented in a very didactical way and wonder why still many authors keep on falling into
the same errors and mistakes pointed out in this paper and in many others.

We believe that it is worth mentioning some of Maciejewskis statements:

1. Model fitting methods based on a multi-step process can be very useful for describing the course
of solid reactions.

2. The physical meaning of the kinetic triplet (kinetic constant, activation energy and conversion
degree function) should be analyzed very carefully. Although many authors tend to analyze uniquely
one kinetic parameter (mainly Ea), it is generally incorrect, since the whole reaction pattern depends
on the kinetic model used and its corresponding conversion degree function, the characteristic
parameter of the kinetic model (for example the reaction order of the n-order reaction model), the
activation energy (Ea) and the preexpoenential factor (A). Only one of them can not be used to
compare processes.

Nevertheless, in this work Dr. Maciejewski emphasized the very limited applicability of the kinetic
methods that use single-heating rate data and stated that to obtain reliable kinetic descriptions,
computational methods that employ multi-heating rate data should be used for treating multi-step
processes.
KINETIC MODELS EMPLOYED
REFERENCES
[1] M.E. Brown, M. Maciejewski, S. Vyazovkin, R. Nomen, J. Sempere, A. Burnham, J. Opfermann, R. Strey, H.L.
Anderson, A. Kemmler, R. Keuleers, J. Janssens, H.O. Desseyn, Chao-Rui Li, Tong B. Tang, B. Roduit, J. Malek, T.
Mitsuhashi, Thermochim. Acta 355 (2000) 125-143.
[2] M. Maciejewski, Thermochim. Acta 355 (2000) 145-154.
[3] S. Vyazovkin, Thermochim. Acta 355 (2000) 155-163.
[4] A. K. Burnham, Thermochim. Acta 355 (2000) 165-170.
[5] B. Roduit, Thermochim. Acta 355 (2000) 171-180.
[6] P. Budrugeac, E. Segal, E. Int. J. Chem. Kinet. 33(10), (2001) 564-573.
CONCLUSIONS

Different models are capable to correlate a set of dynamic conversion data obtained at different heating rates.

The question that rises is: does this fact invalidates the kinetic analysis of data at different heating rates, as the
widely accepted similar argument seems to do it in the case of data obtained at a single heating rate?.

We believe that considering data at different heating rate is better than treating them at a single heating rate, but
none of the two possibilities guarantee the validation of a kinetic model in the mechanistic sense. We believe that
the capability of a model to represent a set of data (either at a single or at multiple heating rates) is a necessary
but not sufficient condition to validate a kinetic model. In other words, conversion curves obtained, for example,
directly from weight loss data seems to be incomplete data to asses mechanistic results. To attain such a goal,
thermogravimetric data must be complemented with other type of techniques. The analysis of the gases evolved
or the analysis of the solid fraction remaining would be considered among other possibilities. Nevertheless,
thermogravimetric data fitting at different rates is a powerful tool to discard pseudokinetic models in order to
determine apparent kinetic parameters capable of simulate the weight loss processes under conditions similar to
those present in the thermogravimetric equipment.
STUDY OF THE THERMAL BEHAVIOUR OF PVC IN THE
PRESENCE OF ZEOLITES AND A SEPILOITE

A.Marcilla, J.C. Garca-Quesada and R. Ruiz.
CHEMICAL ENGINEERING DEPARTMENT, UNIVERSITY OF ALICANTE -
SPAIN

Model Symbol
) (o f
Reaction order model Fn
n
) 1 ( o
SestakeBerggren (empirical kinetic
model)
SBnm
m n
) 1 ( o o
Random nucleation and growth of
nuclei (AvramiErofeev equation)
An
| |
( )
n n
1 1
) 1 ln( ) 1 (

o o


The reasons supporting this argument are presented in his
Fact 2, where it was stated that two different models
(reaction order and Jander models) represent very similar o-
T curves at one heating rate (5C/min), but at other two
different heating rates (2.5 and 10C/min) the same kinetic
parameters used at 5C/min illustrate very different curves.

2.5Cmin
5C/min
10C/min
0.0
0.2
0.4
0.6
0.8
1.0
500 600 700 800 900
Temperature (C)
o
Reaction order model
Jander model
Thus, in order to avoid parameter interactions and the possibility that different sets of kinetic parameters
could represent very similar conversion curves, these and other authors strongly recommend that kinetic
analysis of single heating rate experiments should be generally avoided. With the following examples,
we intend to illustrate that interactions among parameters are even higher than these authors described,
since there is a possibility that different models, with different kinetic parameters could be able to
reproduce very similar conversion curves at different heating rates simultaneously. Consequently, and
following the same type of reasoning, kinetic modeling could not be used as a tool of model validation,
i.e., a good fit of experimental data at different heating rates is a important, but not the unique
requirement to ask to a model.

dt f e dT
f
e
RT
E
RT
E
) ( A
) (
A
t
0
T
0
o
|
o
o
} }
+ +
= =
Fn An
Ln A (s
-1
) 19.0 17.84
Ea (kJ/mol) 125.0 119.3
n 1 1.04

CASE 1. THE NUCLEATION AND
THE REACTION ORDER MODELS
CASE 2. THE NUCLEATION AND
THE SESTAKE-BERGREN MODELS
An SBnm
Ln A (s
-1
) 22.0 20.67
Ea (kJ/mol) 125.0 115.6
n 2 0.79
m --------- 0.54

CASE 3. OVERLAPPED PROCESSES


Lets us consider another type of hypothetical situation where a product A can decompose via two
different reactions, as may occur during polymers degradation, and 50% of A follows the first route,
while 50 % the second route:

Reaction 1 and reaction 2 are described by the reaction order model. Although their corresponding
kinetic parameters are different (Table 4), both processes are not easily separated by changing the
heating rate, as possible to observe in Figure 4.













Are in consequence the mechanism and the model suggested valid (i.e. one single step with the reaction
order model)?.

In our opinion kinetic analysis can not be used uniquely to validate models or reaction mechanisms. A
good fit of experimental results is only one condition, but it should not be the unique one. Other
techniques, as for example infrared or mass spectroscopy can be useful in order to obtain more
information about processes involved. In this example, the monitoring of likely signals attributed to B
and C could have revealed that they are not released concurrently and the suggested single step
mechanism is wrong.
C A
B A

A mere analysis of the overall conversion degree curve

or alternatively its derivative could reveal that only one
process is taking place (see Figure 5), i.e.:


C B A +
The kinetic analysis of these curves can be achieved by
considering one single step and the reaction order
model. For example, there exits a set of kinetic
parameters (ln A(s-1)=10.62, Ea(kJ/mol)=98.8, n=1.38)
that allows a very good fit of the conversion degree
curves, as is possible to observe in the Figure.