1

Advanced Kinetics and Technology Solutions

AKTS AG TECHNOArk 1 CH-3960 Siders


To who it may concern

AKTS AG TECHNOArk 1
3960 Siders
Switzerland
Phone +41 (0) 848 800 221
Fax +41 (0) 848 800 222
E-Mail info_contact@akts.com
Internet www.akts.com

Information about advanced thermokinetic analysis for thermal aging and safety

The reaction rates can be considered as a function of only two variables, T and :
) f( T k
dt
d

) ( = (1)
The dependence of the process rate on temperature is represented by the rate constant, k(T), the
dependence on the extent of conversion by the reaction model, f(). Eq. 1 describes the rate of a
single-step process.

The temperature dependence of the process rate is typically parameterized through the Arrhenius
equation
|
|

\
|
=
RT(t)
E
exp A T k ) ( (2)
where A and E are kinetic parameters, the preexponential factor and the activation energy,
respectively, and R is the universal gas constant. The temperature mode applied during investigations
is usually isothermal, T=const, or non-isothermal, T=T(t). The most common non-isothermal mode is
the one in which the temperature changes linearly with time so that
const
dt
dT
= = where is the heating rate.

A wide variety of the reaction models f() is applied in the solid-state kinetics, some of which are
presented in Table 1.

Table 1. Some of the kinetic models used
Reaction model Abbreviation: f( ) Reaction model Abbreviation: f( )
first order F1 : 1- Avrami-Erofeev A1.5 : 1.5 (1-) [-ln(1-)]^(1/3)
second order F2 : (1-)^2 Avrami-Erofeev A2 : 2 (1-) [-ln(1-)]^(1/2)
third order F3 : (1-)^3 Avrami-Erofeev An : n (1-) [-ln(1-)]^(1-1/n)
nth order Fn : (1-)^n contracting cylinder R2 : 2 (1-)^(1/2)
power law P1 : ^0 contracting sphere R3 : 3 (1-)^(2/3)
power law P2 : 2 ^(1/2) Rn : n (1-)^(1-1/n)
power law P3 : 3 ^(2/3) 1-dimensional diffusion D1 : 1/(2)
power law P4 : 4 ^(3/4) 2-dimensional diffusion D2 : [-ln(1-)]^-1
power law Pn : n ^(1-1/n) 3-dimensional diffusion D3 : 1.5 [1-(1-)^(1/3)]^-1 (1-)^(2/3)
autocatalytic (1-)^n ^m

Although there is a significant number of various reaction models f(), they all can be reduced to three
major types when considering the dependence of the reaction progress on the time: accelerating,
decelerating and autocatalytic. Each of these types has a characteristic reaction profile or kinetic
curve, the terms frequently used to describe a dependence of or d/dt on t or T. Such profiles are
readily recognized for isothermal data. The respective vs. t reaction profiles are shown in Figure 1.

2

Fig. 1: Relationship of the reaction progress vs time for decelerating, autocatalytic and accelerating reaction
models.

(1) Models of the decelerating type represent processes whose rate has maximum at the beginning of
the process and decreases continuously as the extent of conversion increases. The most common
example here is a reaction-order model:
f()=(1-)
n


where n is the reaction order. Diffusion models (Table 1) belong also to the decelerating models.


(2) Autocatalytic models represent processes whose initial and final stages demonstrate respectively
the accelerating and decelerating behavior so that the process rate reaches its maximum at some
intermediate values of the extent of the conversion. The autocatalytic model can be described by the
following equation:
f()=(1-)
n

n



(3) Accelerating models represent processes whose rate increases continuously with increasing the
extent of conversion and reaches its maximum at the end of the process. Models of this type can be
exemplified by a power-law model:
f()= n
(1-1/n)

where n is a constant.


Only those kinetic methods that allow considering all three types of the conversion dependencies can
be recommended as reliable methods. Sestak and Berggren (SB) have introduced an empirical model:
f () =
m
(1 )
n
[ ln(1 )]
p

in which the values of the exponents m, n, and p characterize the contribution of the different reaction
models in the observed reaction rate. The SB model is generally used in truncated form (p=0) being
equivalent to the so called Prout-Tompkins autocatalytic model.

The extent of the conversion (or the reaction progress) is determined experimentally as a fraction of
the overall change in a physical property that accompanies a process as a function of the time or
temperature. If a process is accompanied by the mass loss (or gain) as in TG (thermogravimetry), the
extent of the conversion at certain time t
i
or temperature T
i
is evaluated as a ratio of the actual mass to
the total mass loss in the process. If a process is accompanied by release or absorption of heat
(Differential Scanning Calorimetry DSC or Differential Thermal Analysis DTA), the extent conversion is
3
evaluated as a ratio of the amount of the evolved or consumed heat at t
i
or T
i
to the total amount of
heat released or absorbed in the process. The value of the conversion varies from 0 to 1 when the
process progresses from initiation to completion.

Combining eqns. 1 and 2 yields:
) (

f
RT(t)
E
xp e A
dt
d
|
|

\
|
= (3)

From the computational standpoint, the purpose of kinetic analysis of thermally stimulated processes
is to establish mathematical relationship between the reaction rate d/dt, the extent of conversion ,
and the temperature or time. This can be accomplished in several ways. The most straightforward way
is determining a kinetic triplet, which is a term frequently used for a single set of A, E and f() values.

A major practical purpose of the kinetic evaluation of the experimental data is the prediction of the
process rate and reaction progress at arbitrarily chosen values of t or T under any mode of the heat
treatment. The knowledge of these parameters allows modelling of the process of the materials aging.
The predictions are reliable only when the sound kinetic analysis methods are used. The main
difficulty in the correct kinetic interpretation of the experimental data is due to the fact that even
apparently simple, one-step process, may in the reality contain multiple steps which require more
complex kinetic elaboration of the experimental results. This remark is illustrated in Fig.2 where
unknown amount of reactions can contribute to the observed, apparently simple, shape of the
experimental signal (bold line). This problem and the resulting equation (3) provide a basis for the
isoconversional kinetic methods.


Fig. 2: Thermokinetic analysis of heat flow signals: the stages and physico-chemical reaction pathways are
generally unknown.

General idea: All isoconversional methods take their origin in the isoconversional principle that states
that the reaction rate at constant extent of conversion is only a function of temperature. This can be
easily demonstrated by taking the logarithmic derivative of the reaction rate (eq. 1) at = constant:
a a a
T
d
f d
T
d
T k d
T
d
dt
d
d
(
(
(
(

+
(
(
(
(

=
(
(
(
(

1
)) ( ln(
1
)) ( ln(
1
) ln(

(4)

4
where the subscript indicates isoconversional values, i.e., the values related to a given extent of the
conversion . Because at =const, f() is also constant therefore the second term in the right hand
side of eq. 4 is zero. Thus:

R
E
T
d
dt
d
d
a
) (
1
) ln(

=
(
(
(
(

(5)

It follows from eq. 5 that the temperature dependence of the isoconversional rate can be used to
evaluate isoconversional values of the activation energy, E() without assuming or determining any
particular form of the reaction model f(). For this reason, isoconversional methods are frequently
called model-free methods. However, one should not take this term literally. Although the methods
do not need to identify the reaction model, they do assume that the conversion dependence of the rate
obeys some f() model. This is visible with the most common differential isoconversional method of
Friedman based on the equation (3). Friedman proposed to apply the logarithm of the conversion rate
d/dt as a function of the reciprocal temperature at any conversion :
) (
) (
)

f
RT(t)
E
xp e A(
dt
d
|
|

\
|
= (6)

)) ( ln(
) (
) (
)) ( ln( ) ln(

f
t RT
E
A
dt
d
+ = (7)

f() is a constant in the last term at any fixed value of and the dependence of the logarithm of the
conversion rate d/dt on 1/T shows a straight line with the slope m = -E/R and intercept equal to
ln(A()*f()).


Fig. 3: Differential isoconversional analysis


By extension
) (
) (
)) ( ' ln( ) ln(
t RT
E
A
dt
d

= (8)
with ) ( ) ( ) ( ' f A A =

So, having determined A'() and E() from the Friedman analysis we can predict the reaction rate or
reaction progress using the following expression:
5
|
|

\
|
=
RT(t)
E
xp e ( A'
dt
d ) (
)

(9)
at any temperature profile without the explicit knowledge of any reaction model f(). It follows that for
thermal aging determination, it means for calculation of the time necessary to reach a given reaction
progress one can use the following expression:

|
|

\
|

= =

0
) (
)
0
RT(t)
E
xp e ( A'
d
dt t
t
a
(10)
for any temperature profile T(t) at mg, kg and ton scales by applying the corresponding heat balance
equations (see Fig. 4) depending on the sample mass.


Fig. 4: From mg (shelf-life) to kg and tons (thermal safety)

The differential isoconversional method of Friedman does not require the assumption of the form of
the reaction models f() and additionally does not assume the constancy of the A and E values during
the course of the process. Therefore it is much more precise than commonly used method requiring
the knowledge of f() and assuming that A and E values are constant for range from 0-1.

For more information: http://www.akts.com