Polymer Degradation and Stability 18 (1987) 247 259
The Thermal Stability of Polyetherimide Huang Farong, Wang Xueqiu & Li Shijin Department of Polymer Materials, East China University of Chemical Technology, Shanghai, China (Received 26 January 1987; accepted 9 February 1987)
A BS TRA C T A technique called bivariate .fitting has been devised for obtaining kinetic parameters of thermal decomposition of polyetherimide from TGA data. In this way an apparent order of 5"33 was obtained, an apparent activation energy of 192kJ/mol, and a pre-exponential .factor of 1"45 × 1012 min-1. Compared with other methods such as Friedman's 5 and Ozawa's, 6, this method was found to be a good approaeh to the solution of the kinetics of the decomposition. The products of the thermal degradation of the polymer were also analysed by ir, elemental analysis and pyrolysis/gas chromatography/ mass spectrometo'. A decomposition mechanism is suggested.
INTRODUCTION Polyimides have found wide application in various fields and exploitation is continuing. Polyetherimide (PEI) is a new thermoplastic polyimide and has been given a great deal of attention because of its good processing properties. It has been prepared from derivatives ofphthalic anhydride, bisphenols, and organic diamines. There have been many methods for the treatment of the kinetics of thermal decomposition and mathematical and kinetic models used to calculate kinetic parameters have been the subject of a review. 2 In the present paper, the kinetics of thermal decomposition of PEI have been studied by the use of the dynamic thermogravimetric technique under N 2. The kinetic parameters were obtained by bivariate fitting based on multiple linear regression and checked by other calculations. The degradation reactions of polyimide were studied several decades ago. 3 However, the mechanism of decomposition of PEI is unknown. We 247 Polymer Degradation and Stability 0141-3910/87/$03'50 © Elsevier Applied Science Publishers Ltd, England, 1987. Printed in Great Britain
248
Huang Farong, Wang Xueqiu, Li Shifin
have analysed the degradation reactions by means of instrumental analysis, for example, ir and pyrolysis/chromatography/mass spectrometry. A thermal decomposition mechanism is suggested. EXPERIMENTAL
Synthesis of PEI Monomers
4,4'-bis(3-nitrophthalimido)diphenyl ether (NPIE) was prepared from equivalent quantities of 3-nitrophthalic anhydride and 4,4'-diaminodiphenyl ether. 4 The other reagents were obtained commercially. Polymer
A three-necked, round-bottomed flask fitted with a mechanical stirrer, thermometer, Dean-Stark trap, reflux condenser, and nitrogen bypass was charged with Bisphenol A(BPA), 50% aqueous sodium hydroxide, DMSO and toluene. The flask was heated to reflux with stirring under nitrogen and water was collected in the Dean-Stark trap and removed. After 2 h the DeanStark trap was replaced by a Call 2 trap which allowed the condensed solvent to pass over Call2 before returning to the flask. Heating was continued for another 4 h. The toluene in the flask was then distilled and the mixture cooled to ca. 60°C. The trap and the condenser were quickly removed and NPIE was added with an equimolar amount of BPA. The system was stirred at 50°C for 4 h. The product was then cooled to room temperature (ca. 20°C) and a small amount of glacial acetic acid was added. The polymer was precipitated by addition to methanol, washed with distilled water and dried. Thereafter, it was reprecipitated by redissolving in CHCI3 and adding to methanol. Finally, it was dried in vacuo at 60°C. The PEI, in colour, was obtained. Intrinsic viscosity (DMF, 25°C); light yellow 0.33 dl/g: T~; 239°C (heating rate 20°C/rain, DSC). The chemical structure of PEI is as follows: q O O II I, ,~ |
[
/ L
x~_/
I ~_/ CH3
O
O ~
].
Analyses: calculated for the polymer C, 75'44%; H, 4"09%; N, 4.09%; O, 16.38% found, C, 75.87%; H, 3.70%; N, 4.46%; O, 16.03%.
The thermal stability of polyetherimide
249
Analytical apparatus 1.
2. 3.
A 7650 infrared spectrophotometer and an FT-IR 20SX Fourier transform infrared spectrometer were used. Samples were measured using the KBr pressed disk technique. A Dupont 1090 thermal analysis apparatus was used for the thermogravimetric tests. A Finnigan 4510 gas chromatograph/mass spectrometer and JPL-01 pyrolytic instrument (made in China) were also used.
RESULTS A N D DISCUSSION
The kinetics of thermal decomposition of PEI Representative thermograms obtained for PEI at heating rates of 2°C, 5°C, 10°C, 15°C and 20°C/min under nitrogen are shown in Fig. 1. It is seen that
,,a
2a 6C 5°
11i
. . . . . . . . . . . . . I 280
I 360
.~ . . ~ v
.. _~.~_
I 440
Temperature
I 520 ['C)
. I 600
. I 680
Fig. 1. Thermogravimetric curves for PEI at various heating rates. 1.2°C/min; 2. 5°/min; 3. 10°C/min; 4. 15°C/rain; 5. 20°C/min.
the decomposition takes place in two stages. The kinetics of thermal decomposition were studied on the basis of the first stage of the thermogram. Bivariate fitting method
The equation for the thermal decomposition rate can be expressed as follows: 2 d~ = k(1 - ~)" = Ae-E/RT(1 -- CX)n
(1)
Huang Farong, Wang Xueqiu, Li Sho'in
250
where ~. is the fractional weight loss, n is the apparent order of reaction, k is the rate constant, A is the pre-exponential factor, t is the time, E is the apparent activation energy and R is the gas constant. Taking logarithms, In ~ -
=lnA+nln(1-~)-E/RT
(2)
Therefore, if 1/T and In (1 - ~) are considered as discrete r a n d o m variables and ln(d~/dt) as a dependent variable, the kinetic parameters (A, E, n) could simultaneously be obtained by multiple linear regression.
The principle of multiple linear regression Assume that the fitted equation can take the form:
Y= alX1 + a2X2 +"" + a,X,
(3)
where al, a2 .... , a, are the fitted parameters; X 1, )(2,..., X, are independent variables. There are experimental d a t u m groups: (YI,XI~,XI2 ..... X~,), (Y2, X21, X22, ..., X2,) .... , (I'm, X,.1, X,,2 ..... X.,,). The deviation function is:
S= Z (Yk-- Ydk)2
(4)
k =1
in which Ydk is a quantity determined by experiment and Yk is a predicted value. Substituting eqn (3) into eqn (4), we obtain eqn (5):
S = ~ (alXkl + azXk2 + ' " + a.Xk.-- Ydk)z
(5)
k=l
In order to make the error minimum, there is an essential condition; namely (3S/t3al) = 0. So we can obtain the following equations: rdkXki= k=l
(i,j= 1,2 .... ,n)
~2 aj ~, Xkj~,~ki j=l
(6)
k=l
Equation (6) can be transformed into eqn (7): Slla
--1- $ 1 2 a 2 d- " ' " '1-
s21a 1 q-
Slna n = Say
$ 2 2 a 2 q- • . . _1_ S2na n
S2y
......
(7)
Snla 1 --[-Sn2a 2 --[-....q_ Snnan
Shy
where: m
rdkXk,; k=l
xk,xkj
s,j= k=l
The thermal stability of polyetherimide
251
Only when m is greater than n does eqn (7) have a solution. We have solved the system of linear equations by the Gauss's pivot element elimination. The program chart
The program chart for calculating the kinetic parameters of thermal decomposition by the bivariate fitting method is shown in Fig. 2.
No
//Enter original data
Process the data unique
J
I augmented matrix ]
Calculate the normal equation
Eliminate the elements
coefficients
L_augmented matrix
I Estimate the error
Fig. 2.
(T) Program chart of the bivariate fitting, (I) The main program. (II) The subprogram.
252
Huang Farong, Wang Xueqiu, Li Shijin
The results and check o f the bivariate fitting
The thermogram of PEI was treated by bivariate fitting with a PC-8080 personal computer. The results obtained are presented in Table 1. The data were compared with those obtained using other methods, for instance, Friedman's, 50zawa's, 6 and Kissinger's. 7 As shown in Table 1, the kinetic parameters obtained by bivariate fitting are nearly the same as that obtained by the other methods. Consequently, we are convinced that the bivariate fitting method for calculating the kinetic parameters Of thermal decomposition is efficient. The bivariate fitting method only requires one thermogravimetric curve, while the other methods require at least three TG curves at various nominal heating rates to obtain the kinetic parameters. Our method is simpler and more convenient than the others, and the kinetics can be studied over the whole range of temperature. In addition, our method appears to be more useful for studying kinetics quickly and where the kinetic parameters vary with heating rate or the degree of degradation, i.e. weight loss.
Investigation of the thermal decomposition mechanism of PEI PEI was decomposed in the pyrolytic instrument. The mixture of gases produced in the degradation of the PEI was carried into the gas chromatograph/mass spectrometer by the inert carrier gas helium. After the various components were separated by the gas chromatograph, they were separately analysed by the mass spectrometer. The gas chromatogram (GC) and mass spectra (MS) are shown in Figs 3 and 4-1 to 4-6 respectively. Figure 3 shows that there are a number of degradation fragments of PEI. The analytical results are tabulated in Table 2. The fragments are mainly
TABLE 1 The Kinetic Parameters of Thermal Decomposition of PEI Methods
Bivariate fitting° Ozawa's Friedman's Kissinger's
Weight l o s s range (%)
13-5-19'5 11"0-21"0 11"0-21"0
° Heating rate: 15°C/min.
Apparent activation energy
192 184 192 178
Apparent order
Pre-exponential factor (rain- 1)
5"33
1"45 x
1012
6.09
1'45 x
1012
The thermal stability of polyetherimide
253
1 3
2
1.00
1.20
1.40
2"00 Time
2-20
2.40
3'00
3.20
Fig. 3. Gas chromatogram of the pyrolytic fragments of PEI.
carbon dioxide, phenol, toluene, 2-methyl-1-heptene, 4-isopropenyl phenol, and 3-hydroxybenzoyl aniline and/or its isomers. Figure 5 shows ir spectra of the residues from the degradation of PEI at lower and higher degrees of degradation. In Fig. 5(a) there are peaks which can be assigned to --OH, --NH, --CN groups and many peaks overlap to form broad peaks, but the absorptions due to the groups in PEI disappear as the degree of degradation rises. Finally, the residues from thermally decomposed PEI were analysed as follows: PEI (calculated): residue (found):
C, 75.44%; H, 4.09%; N, 4.09%; O, 16.38%. C, 79.23%; H, 2.60%; N, 2-75%; O, 15.42%.
TABLE 2
MS Analysis Results of the Fragments from PEI
Figure 4-1 4-2 4-3 4-4 4-5 4-6
Main products Carbon dioxide Toluene Phenol 2-Methyl- 1-heptene 4-lsopropenyl phenol 2 and/or 3-hydroxy benzoyl aniline, N-benzoyl-p-amino-phenol
Minor products Propane 2-Methyl-l-heptene Benzonitrile Phenol, benzonitrile
44
43 45 II
I 50
40
I
I
I
60 70 Mass (m/e)
Fig. 4-1.
80
MS of peak No. 1 in Fig. 3. 91
.~
92
55 70
IIi
4~J 51
JT 8589 ,79,L.~
4O
112 I
,
80 100 Mass (m/e)
£~
i
,
140
120
Fig. 4-2. MS of peak No. 2 in Fig. 3. 94
•
66
95 103
/ I.s5 65 40
60
80
IJ
I
100 120 Mass (m/e)
I
140
Fig. 4-3. MS of peak No. 3 in Fig. 3.
I
160
55
70
94
78
40
J
Ji .
83
103
,lm,l 1,
lJ ,
60
80
Fig. 4-4.
"i
120 128
h,
I
100 120 Mass ( m / e )
140
160
MS of peak No. 4 in Fig. 3. 134
119
44 91 ,v 107 51
h, 40
~3~o77Jl 891111o~lJ 115 ,I. I .
6O
,I II J,
,..
80
Fig. 4-5.
100 120 Mass ( m / e )
I
I
140
160
MS of peak No. 5 in Fig. 3. 94
66
r~
55
s.~l63 77 83
I , 40
60
134
91
80 Mass
103
112 !
,IJ.. 100 (m/e)
120
l[l i
Fig. 4-6. MS of peak No. 6 in Fig. 3.
140
169
256
Huang Farong, Wang Xueqiu, Li Sho'in
75
67 o o
g 59
Ca)
/I
~ 60 I,..,-
40 20 (b) 4OO0
I I 2000 1600 Wavenumber$
I t 200 ( c m "1)
i
BOO
Fig. 5. Infra-red spectra of the degradation residues of PEI. (a) Lower degree of decomposition. (b) Higher degree of decomposition.
The quantity of C in the residue increases, whereas H, N and O all decrease. If a PEI chain was broken in the following two ways: (I) O
/
~
~-~
O
~ 7 ~ ~ o ~ ~ ~ /
4o_,/c~_d_,/c~;_o,~? ~ / ~ / I' [_
x_.~/
f x_~/ CH3
o
?-~ 0
J J.
(II) t
CH3
0 (f"h"~l C~N ~ " ~
o_/-3-~UL-~o~?, ~"~ k~/
J X~/ CH3
o
,-x~
0 ]3 N ~C ~-~f"h'~/
~~-~/- , ~ < ~ ! o
I Jn
the percentages of the various elements in the residues would be as follows: (I)
C, 73.40% H, 3.10%; N, 6.10%; O, 17"50%.
(II) C, 81.80%; H, 4.90%; N, 4.90%; O, 8'40%. There are obviously differences between the experimental value and the calculated value. If the degradation proceeded by the first route, the calculated percentages of H, N, O would be higher, while the calculated percentage of C would be lower, than the experimental values. On the other hand, the calculated percentages of C, H, N would be higher and the
257
The thermal stability of polyetherimide
calculated percentage of O would be lower than the experimental values if the decomposition proceeded by the second route. Thus, the degradation reactions probably proceed by the two routes. As far as the percentage of N is concerned, it is probable that N compounds escape in the degradation. On the basis of results and reasonable analysis, we suggest that PEI decomposes in two ways; namely, an ether-bond breaking mechanism and a carboxyl induced chain breaking mechanism. A. The ether-bond breaking mechanism is shown on p. 258. B. The carboxyl induced chain breaking mechanism is shown on p. 258. 8 The reaction is induced by a trace of water which is always present in the polymer sample: O II ~ ~ ~ - - ~
O 1.
--~ ~ C
O II -OOH NH~
O II --~ ~ - - C - - N H - - ~+
H20
CO 2
, O
---~N=C-~N~ 2.
+ CO2
O O ,-~F'~'X~--NH , ~ ( ) >
0
II
© I 3
' ~ C N
O .
258
Huang Farong, Wang Xueqiu, Li Sh@n
z
o=$+=o+go 0
I
P E-8 0
t
+
m
c 0
m
/
L 3
u-2
+
1 0 n 0 ____
E-~-ur ---
----
-
0
m
m
8-u” / -8 ____
---
c) 0'
O=Y li’=O z
m
0*
-_-w-l
0
m
8-E-E n
o0
m
8
O=yY=O 7
The thermal stability of polyetherimide
259
CONCLUSION The kinetic parameters of thermal decomposition of PEI were obtained by bivariate fitting and checked by Friedman's, Ozawa's and Kissinger's methods. It appears that the bivariate fitting method provides a satisfactory mathematical approach to obtaining the kinetic parameters. Meanwhile, the products of thermal degradation of the polymer were analysed by gas c h r o m a t o g r a p h y / m a s s spectrometry, ir and elemental analysis. We tentatively believe that PEI is thermally decomposed by two routes, an ether-bond breaking mechanism and a carboxyl-induced chain breaking mechanism.
ACKNOWLEDGEMENTS Our special thanks are due to Mrs Xi Yuda and M r Xu Xingrong, who helped to make it possible for us to carry out experimental work with the apparatus.
REFERENCES 1. R. O. Johnson and H. S. Burlhis, J. Polym. Sci., Polymer Symposium, 70, 129 (1983). 2. J. D. Cooney, M. Day and D. M. Wiles, J. Appl. Polym. Sci., 28, 2887 (1983). 3. S. D. Bruck, Polymer, 6, 49 (1965). 4. D. M. White, T. Takekoshi, F. J. Williams, H. M. Relles, P. E. Donhue, H. J. KIopfer, G. R. Loucks, J. S. Manello, R. O. Matthews and R. W. Schluenz, J. Polym. Sci.: Polym. Chem. Ed., 19, 1635 (1981). 5. H. L. Friedman, J. Polym. Sci.: Part-C, 6, 183 (1964). 6. T. Ozawa, Bull. Chem. Soc. Jpn, 38, 1881 (1965). 7. H. E. Kissinger, Anal. Chem., 21, 1702 (1957). 8. R. T. Conley and A. Gaudiana, in: Thermal stability of polymers, Vol. 1, (R. T. Conley (Ed)), Marcel Dekker, New York, 347 (1970).