thermochimica

acta
ELSEVIER Thermochimica Acta 285 (1996) 259 267

Estimation of nonisothermal kinetic parameters from a TG

curve by the methods of overdetermined
system and inflection point
S.G. Viswanath a,., M.C. G u p t a b
aLaxminarayan Institute of Technology, Nagpur University, Nagpur-440010, India
bDepartment of Chemistry, Nagpur University, Nagpur-440010, India
Received 6 June 1995;accepted 12 February 1996

Abstract

Methods are described to determine nonisothermal kinetic parameters from a TG curve. The
first method uses the theory of overdetermined systems while the others use the inflection points.
These inflection point methods are an alternative to those described earlier by Viswanath and
Gupta.

Keywords: Inflection point; Non-isothermal kinetic parameters; Overdetermined system: Ther-

mogravimetry

1. Introduction

A number of methods have been proposed for the evaluation of nonisothermal

kinetic parameters of solid state reactions from a TG curve involving either mass loss or
sometimes gain. All the available methods generally involve the assumption that
a given reaction follows a simple order. The Freeman and Carroll [1] method is the
only method for the estimation of the order of a reaction; however it suffers from several
disadvantages [-2-5]. In spite of its limited precision and disadvantages, it is widely
used. In the Sharp and Wentworth I-2] method, which is a derivative method, if the
correct order of reaction is assumed and substituted in the Arrhenius rate equation, the
logarithmic form of the equation results in a linear plot. Similarly, the Coats and
Redfern [3] method is an integral method and also requires the assumption of an order

* Corresponding author.

0040-6031/96/$15.00 © 1996 ElsevierScience B.V. All rights reserved

PII: S0040-6031 (96)02917-6
260 S.G. Viswanath, M.C. Gupta/Thermochimica Acta 285 (1996) 259-267

of reaction to get a linear plot. In these two methods, the pre-exponential factor and
activation energy are determined graphically.
Viswanath and Gupta [5] reported several methods for the estimation of all the
kinetic parameters by considering the inflection point of the TG curve. The accuracy of
the evaluation of the parameters by these methods depends on the ability to locate the
inflection point accurately from the curve. We have applied this method [5] to the
study of the thermal decomposition of polycarbonate bisphenol A in the presence and
absence of some metal oxides [6]. In this communication, new derivative methods are
reported for the estimation of all nonisothermal kinetic parameters. The first method
uses the overdetermined system method. The order of reaction obtained from this
method is used for the graphical method for ascertaining the correctness of the
estimates of the parameters. Other methods use the inflection point as the basis for the
determination of kinetic parameters.
To illustrate these methods, the data for calcium oxalate decomposition reported by
Schempt et al. [7], the thermal decomposition of the solid complex formed by the
ligand 2-carboxyl-2'-hydroxyl-3',5'-dimethyl azobenzene (CHDMA) and cobalt [8],
and the TG measurements on the thermal decomposition of Cu, Ni and Co sulfates
have been re-examined.

2. Experimental

Thermogravimetry was carried out on a Perkin-Elmer TG Thermal System TGS2

in a flow of nitrogen gas and at a heating rate of 10 K min- 1in a alumina crucible at the
Regional Sophisticated Instrumentation Center, Nagpur University, Nagpur. In the
kinetic analysis of the TG data, it is assumed that the reaction follows a simple order.
The fraction decomposed ~ is defined as
(Wo - w T )
= (1)
(Wo - We)

where Wo, We, and Wr, are the initial mass, final residual mass and mass of the sample at
any temperature T, respectively. The logarithmic form Arrhenius rate equation is
written as

In ~-~ = n l n ( 1 - • ) - ~ +ln (2)

where n, A fl and E~ are the order, pre-exponential factor, rate of heating and activation
energy of a reaction, respectively.

2.1. The overdetermined system method

For x data points, there should be x equations of the above type and there are three
unknowns, namely n, A and E~, to be determined; if there are more equations than
unknowns to be determined, such a set of equations forms an overdetermined system.
The theory for finding the least-square solution of an overdetermined system of
 S.G. Viswanath,M.C. Gupta/ThermochimicaActa 285 (1996) 259-267 261

equations is found in a n u m b e r of text-books on numerical analysis. Therefore, this

theory needs no discussion here. These x equations can be resolved into three equations
which can be written as

=~1 =n ln(1-~xi)--~E~; ~ 1
i= In -d-~i i=1 (gJi=l Tii + l n x (3)

In ~-~ ln(1 - •i) = n ln(1 - cq) - ln(1 - ei)

i=i i=1 i=1

+ in ~iln(1 - :q) (4)

lnfde) 1 x 1 ~E~ 1 2
-- --=n21n(1-~/) - 2 +In ~ 1
,=, (dTJ, T, i=l Tii [ R J,=, ~ ,=, ~ (5)
Eqs.(3)-(5) are solved by either an analytical or a matrix m e t h o d to obtain the
parameters; in the present case, they are solved as described earlier [4].
The correct order of reaction obtained from the analytical solution of Eqs. (3)-(5) is
substituted into Eq. (2), which can be rewritten as

d~ 1-
ln(k) = In { [ ~ 1 / [ c~]"} = - {R@} + l n { ~ } (6)

The plot of ln(k) against 1000/T results in a straight line. The results of both solutions
are presented in Table 1 and plots are shown in Fig. 1. The correlation coefficient (C.R.)
and percent average deviation (A.D.) [5] of the points from the regression line are also
presented in Table 1.

2.2. Inflection point methods

In the absolute m e t h o d suggested by Viswanath and G u p t a [5], the (EffR) value

obtained from the inflection equation is substituted into Eq. (2). F r o m the plot of this
equation, the order of reaction n and the pre-exponential factor A are evaluated. The
n value is substituted in the inflection equation to get the activation energy of the
reaction E~.

Table 1
Nonisothermal decomposition kinetic parameters from the overdetermined system method
Analytical solution Graphical solution

Compound Order of Ej A~ Ej A~ C.R. A.D.

reaction (kJmol 1) (s 1) (kJmol t) (s-l)

Co(II)CHDMA 0.07 -5.3 1.8 x 10 4 -5.3 1.8 x 10-4 0.9996 0.03

CaC20 4 1.03 302.1 1.0 × 1019 302.1 1.0 × 119 0.9996 0.34
NiSO4 0.31 244.2 2.0 × 1 0 9 244.2 2.0 × 1 0 9 0.9999 0.21
CoSO 4 0.53 244.9 1.2 × 1 0 9 244.9 1.2 × 1 0 9 0.9998 0.22
CuSO 4 0.99 245.2 4.7 × 10l° 245.2 4.7 x 101° 0.9997 0.38
262 S.G. Viswanath, M.C. Gupta/Thermochimica Acta 285 (1996) 259-267

- 4.757

t
I
-5.6~9

"-" ~ - 5 0 0
-

-7..q-92

- 8 . ~-OS J
0.890 0.99,3 1. 0 9 6 1.1Sl9 1...-302 1.4-05 ?, ~5C)~
IO00/T

Fig. 1. Plots from the solution of the overdetermined system method: ©, Co(II)CHDMA; X~
CAC204; + ,NiSO¢; 121,COSO4; and A, CuSO 4.

At the inflection point, the second derivative (d20~/d T 2) = 0 and (da/d T) is maximum.
The equation at the inflection point can be written

(d~/d T)~ = [(1 - ~s)/n] (EJR 7~) (7)

The subscript defines the quantities at the inflection point respectively and Eq. (7) is
rearranged as

n = (EJR T2) [(1 - ~)/(d~/d T)s] (8)

The kinetic parameters are obtained by the following methods.

2.2.1. An alternate absolute method

Eq. (8) is substituted into the logarithmic form of the Arrhenius equation which is
then written

ln(d~/dT) = (EJR) [ln(1 - c0/q5 - 1/T] + ln(A/fl) (9)

where q~ = [T2/(1 -- ~s)] (d,/dT)s. The plot of ln(dc~/dT) against 1000 [ln(1 - a)/
4)- 1/T] gives a linear plot with a slope equal to (EJR) and intercept (A/fl). By
substituting the value of (EJR) in Eq.(8), n is obtained. Nonisothermal kinetic
parameters estimated by this method for some compounds are presented in Table 2 and
plots are shown in Fig. 2.
 S.G. Viswanath, M.C. Gupta/Thermochimica Acta 285 (1996) 259-267 263

Table 2
Nonisothermal decomposition kinetic parameters from the alternate absolute method

Sample no. Compound Order E/kJ tool ] A/s ~ A.D. C.R.

1 CaC204 1.08 306.5 2.2 X 1019 0.31 0.9990

2 NiSO 4 0.36 250.0 4.0 x 109 0.36 0.9996
3 CoSO 4 0.39 237.2 4.9 x l0 s 0.39 0.9995
4 CuSO 4 1.07 251.6 9.7 x 10 t° 0.26 0.9994

44.5 IG -

O---'-"-'-"----

38.19~
t

31.876
Jr
k-
"t3

?-5.555

X )~ X × xXX

19. 2 3 5 I I I 4~ I.-I- I ~ I ~Lq.-+- I

-2.751 -.o.301 - 1.esa -,.4o2 -0.953 -o.so4 -0.05

z- Cz - o c ) .

Fig. 2. Plots from the alternate absolute method: o, CaC20,,; x, NiSO4; +, COSO4; and [], CuSO~.

2.2.2. An alternate standard method

In this m e t h o d , the slope (E J R ) o b t a i n e d from Eq. (9) is s u b s t i t u t e d in the logarith-
mic form of the A r r h e n i u s rate e q u a t i o n . Eq. (2) m a y be written as

In [(dct/dT) + (E J R ) ( I / T ) ] = n ln(1 - a) + ln(A/fl) (10)

In [ ( d a / d T ) + (EJR)(1/T)] p l o t t e d a g a i n s t In ( 1 - :0 results in a straight line, from
which n a n d A can be d e t e r m i n e d . T a b l e 3 lists the results of the n o n i s o t h e r m a l kinetic
p a r a m e t e r s for s o m e c o m p o u n d s a n d Fig. 3 depicts the plots.

2.2.3. An additional standard method

T h e n value o b t a i n e d from Eq. (8) is s u b s t i t u t e d in the l o g a r i t h m i c form of the
A r r h e n i u s rate e q u a t i o n which is Eq. (6). T h e p l o t of ln(k) a g a i n s t (1000/T) results in
264 S.G. Viswanath, M.C. Gupta/Thermochimica Acta 285 (1996) 259-267

Table 3
Nonisothermal decomposition kinetic parameters from the alternate standard method

Sample no. Compound n A/s i A.D. C.R.

1 CaC20 4 1.07 2.3 × 1019 0.05 0.9981

2 NiSO 4 0.32 3.9 × 109 0.04 0.9991
3 CoSO,, 0.50 5.1 x 108 0.20 0.9995
4 CuSO 4 1.07 9.7 × 101° 0.26 0.9994

-5.945

!
I -6,573

"0
"D - 7 . 20.-0
V

-7.831

-~9. z ~ 5 9 I I I I I I
-38 -1.009 -0.92
1ooo{1,~ (' I - < ) /~-1/T } .----.

Fig. 3. Plots from the alternate standard method: (3, CAC204; x , NiSO4; + , COSO4; and C3, CuSO4.

a straight line, from which E= and A are determined. Table 4 lists the results obtained
using n values from the alternate absolute method and Fig. 4 depicts the plots.
Similarly, n obtained from the alternate standard method can also be used to obtain
such plots.

3. Discussion

In Tables 2 and 3, kinetic parameters for the thermal decomposition of Co(II)

C H D M A complex are not presented because it was found to be a zero-order reaction
which gives an inflection point at infinite T only, as is evident from Eq. (7). Therefore,
the order of reaction is a function of the parameters at the inflection point, such as C~,
 S.G. Viswanath, M.C. Gupta/Thermochimica Acta 285 (1996) 259-267 265

Table 4
Nonisothermal decomposition kinetic parameters from the additional standard method

Sample no. Compound E/kJ tool- 1 A/s-~ A.D. C.R.

1 Co(II)CHDMA -7.8 l.l x 10 4 0.12 0.9955

2 CaC20 4 307.2 2.4 × 1019 0.33 0.9996
3 NiSO 4 250.4 4.2 x 109 0.28 0.9997
4 CoSO 4 226.0 1.4 x 108 0.46 0.9990
5 CuSO 4 251.6 9.6 x 10 l° 0.28 0.9997

-5.008

-5,85~

t -6.704-
.Y

-7. 5 5 3

-8.4.0 I I I
0.890 0.993 1.09~; 1.199 1.302 L~05 1.508
IO00/T

Fig. 4. Plots from the additional standard method: ©, Co(II)CHDMA; ×, CAC204; + , NiSO4; [], COSO4:
and ~, CuSO 4.

T~, E,, and (dcqdt)s. The parameters obtained for the thermal decomposition of the
compounds by all the methods are in good agreement, except for Co(II)CHDMA
complex. Therefore, the parameters calculated from the overdetermined system
method give more accurate results, as the A.D. values are lower while the C.R. values
are higher than for other methods.
The pre-exponential factor A for the Co(II)CHDMA complex is almost zero and the
activation energy is small. This may indicate that this is a weak complex and its
decomposition is kinetically favored. In the present treatment of nonisothermal
kinetics, it is assumed that (a) a reaction follows a simple order, (b) the temperature
dependence of the rate of mass loss is governed by the Arrhenius equation, and (c) the
266 S.G. Viswanath, M.C. Gupta/Thermochimica Acta 285 (1996) 259 267

reaction mechanism takes no account of the solid state decomposition reaction, the
kinetic parameters of which may vary with various factors. However, the entire analysis
is based on the criterion of the degree of fit of the data.
Furthermore, it may be noted that the physical properties and the chemical
stoichiometry change during nonisothermal analysis. Such changes may result in
a change in the activation energy. Therefore, in the method it is desirable to define
a region in the space (c~,T) for which E~ is constant in the analysis of the T G
experiments. A constant activation energy suggests that a single mechanism will
describe the kinetics in this region. Conversely, if E~ is changing, it is certain that
a change in kinetics or a physical or chemical change, or both, is taking place. Hence, in
the data analysis by these methods, the criterion of the fit reasonably exhibits the
constancy of E~ in the range of temperature. A change in kinetics of the decomposing
spectrum frequently results from a shift in a physical-diffusion-limited case to a chemi-
cal-reaction-limited case, or vice versa [9]. It can be observed that the direct fitting of
thermoanalytical data leads to more exact kinetic parameters compared to the usual
methods of linearizing. The large variation in the values of A may be attributed to
enthalpy-entropy compensation. As the E~ increases, A must increase so that the
reaction is thermodynamically (entropy) favored. This will mean that the magnitude of
n, A, and E~ provide the understanding of the reactivity of solids. In solid state
reactions, the crystal is an association of molecules/atoms/ions with long-range
interaction between them; the relative arrangement determines the physical and
chemical properties as well as the disturbances and the reactivity of solids [10].
All thermal decomposition reactions can be classified into two groups. The first
includes chemical reactions in which rupture and formation of new bonds occur locally.
The second group covers reactions in which changes are transferred within a path
length much longer than interatomic distances. The change can be carried by electrons
or ions. Reactions which belong to the first group usually start at the surface. The
second group of reactions involves the volume of the crystal [11]. Thus defects may
play an important role in solid state reactions. The creation and distribution of defects
are entropy-favored. Therefore, a large value of A indicates that the reaction is a surface
reaction. The process involves a mechanism which is due to a large disturbance in the
solids (entropy-favored), such as the nucleation and growth of active centers on the
surface or the diffusion of species (electron or ion) through the lattice. Thus the rate
constant is related to the formation of the active centers at the surface.
It has been shown by Mental and Anderson [12] that the decomposition of C a C 2 0 4
is a first-order process, that there is an electron transition from oxalate to metal (Ca)
through interstitial defects, and that the reaction is a surface reaction which occurs at
the interface separating the parent crystal and the product.
The decomposition of NiSO 4 leads to the product NiO and order is 1/3. This gives
the functional form of the rate law similar to the Arvami-Erofeev mechanism of
nucleation and growth. Koga and Tanaka [13] reported that the decomposition of
NiSO 4 proceeds by surface and random nucleation and growth near the reaction front
as well as in bulk.
The decomposition of CoSO 4 is characterized by a contracting area geometry
model and is affected by the distribution of the active surface centers. The decompo-
 S.G. Viswanath, M.C. Gupta/Thermochimica Acta 285 (1996) 259-267 267

sition of C u S O 4 is first o r d e r a n d d u r i n g d e h y d r a t i o n , r e a c t i o n zones are p r o d u c e d in

C u S O 4 [14].

4. Conclusions

T h e m e t h o d s d e s c r i b e d here are useful for the d e t e r m i n a t i o n of the a c t i v a t i o n energy

a n d o t h e r kinetic p a r a m e t e r s for the overall processes which t a k e place in the
d e c o m p o s i t i o n process. H o w e v e r , f o r m u l a t i o n of a realistic m e c h a n i s m for a solid state
d e c o m p o s i t i o n r e a c t i o n requires at least the k n o w l e d g e of the r e l a t i o n s h i p s between
the rate a n d the rate-forcing variables such as pressure, gaseous flow, gaseous c o m p o s i -
tion, physical a n d g e o m e t r i c p r o p e r t i e s of the sample, etc.

References

[1] E.S. Freeman and B. Carroll, J. Phys. Chem., 62 (1958) 394.

[2] J.H. Sharp and S.A. Wentworth, Anal. Chem., 41 (1969) 2060.
I-3] A.W. Coats and J.P. Redfern, Nature, 201 (1964) 8.
[4] S.G. Viswanath, S.S. Umare and M.C. Gupta, Thermochim. Acta, 233 (1994) 47.
[5] S.G. Viswanath and M.C. Gupta, Thermochim. Acta (1996), in press.
1-6] M.C. Gupta and S.G. Viswanath, J. Therm. Anal., (1995), in press.
I-7] J.M. Schempt, F.E. Freberg, D.J. Royer and F.M. Angelonic, Anal. Chem., 38 (1966) 520.
1-8] H.R. Choudhari, Structural Studies of Metal Complexes, Ph.D. Dissertation, Nagpur University,
Nagpur, 1981.
1-9] J.H. Flynn, in H.H.G. Jellinek (Ed.), Aspects of Degradation and Stabilization of Polymers, Elsevier
Scientific Publication, New York, 1978, p. 573.
1-10] H.S. Ray, High Temperature Chemistry, B.A.R.C., Bombay, Vol. 6, 1985.
[11] V.V. Boldyrev, Kinet. Catal., 1 (1960) 203.
[12] J. Mental and H. Anderson, J. Therm. Anal., 40 (1993) 1307.
[13] N. Koga and H. Tanaka, J. Therm. Anal., 40 (1993) 1165.
[14] V.V. Boldyrev, J. Therm. Anal., 40 (1993) 1041.