251
Journal of Analytical and Applied Pyrolysis, 11 (1987) 251-261 Elsevier Science Publishers B.V.. Amsterdam - Printed in The Netherlands
THERMAL
BEHAVIOUR
L. COSTA and G. CAMINO
OF POLYBENZHYDROLIMIDES *
Istituto di Chimica Macromolecolare
dell’Unioersitci, V.G. Bidone 36, IO125 Torino (Italy)
M.E. QUENNESON Centre d’Etudes des Matkriaux 69390 Vernaison (France)
Organiques pour Technologies Avanckes, B.P. 3,
M. BARTHOLIN Luboratoire des Matkriaux
Organiques du CNRS, B.P. 24, 69390 Vernaison (France)
ABSTRACT Simultaneous measurements of weight loss and rate of evolution of either organic products or water using a specially designed thermogravimetry-evolved gas analysis technique have been employed to study the evolution of small amounts of residual solvent and water formed in reactions occurring on heating polybenzhydrolimides (PBHI). On heating PBHI to 400“ C, a redox reaction takes place involving the elimination of water from =CHOH groups. Diarylketone moieties were identified in the polymer as one of the products of this reaction. Evolved sis.
gas analysis;
polybenzhydrolimides;
polymers;
pyrolysis;
thermogravimetric
analy-
INTRODUCTION
Aromatic polyimides are widely used owing to their outstanding thermal stability, high dielectric strength, and high resistance to solvents. Synthesis of aromatic polyimides is generally carried out in two steps, e.g.:
0. 0’;m O@
Cfi ;: c. ‘0
COOH
HOOC +
H2N-Ar-NH2
-s-mAr-
CONH --AT-
HNOC I
-Hz0
(1)
0165-2370/87/$03.50
0 1987 Elsevier Science Publishers
B.V.
252
While the monomers and the polyamic acid precursor (I) are soluble, the polyimide (II) is generally not. Thus, the imidization step, which is often performed on setting, takes place essentially in the heterogenous phase. Under these conditions, full imidization may not be reached, giving rise to structural defects which are weak points vulnerable to thermal stress or chemical attack. The presence of the =CHOH group in polybenzhydrolimides:
III
makes them soluble in polar aprotic solvents, allowing their complete synthesis in solution. Very high imidization yield can thus be obtained with cyclisation defects below the detection limits of spectroscopic analytical methods such as IR and NMR spectroscopy [l]. The use of polyimides often means exposure of the polymer to temperatures up to 450°C. While the aromatic imide structure withstands these temperatures, release of residual solvent or absorbed water, firmly retained by polar imide groups, is detrimental to the final properties of polyimides. The polarity of polybenzhydrolimides is increased, as compared with typical polyimides, by the presence of =CHOH groups. Therefore, specific conditions for thermal release of solvent and absorbed water from polybenzhydrolimides must be studied. Furthermore, the effect of thermal treatment of benzhydrol structures must be understood for full exploitation of potential uses of polybenzhydrolimides. In this work we have studied the effect of heat on polybenzhydrolimides using a specially designed technique combining thermogravimetry with high-resolution gas chromatography and a water-sensitive probe.
EXPERIMENTAL
Polybenzhydrolimide
(PHBI)
The polymer was prepared by condensation of the bis-methyl hemiester of 3,3”-4,4’-benzhydroltetracarboxylic acid (DDB) and p,p’-diaminodiphenylmethane.in N-methyl-2-pyrrolidone (NMP).
PBHI
253
The concentration varies from 40% at the beginning of the reaction to 20% at the end. In the meantime the temperature is increased from 160 o C, the temperature at which the elimination of water and methanol begins, up to 200 o C for complete imidization. The solution is precipitated in water, the precipitate crushed, washed with methanol and then dried under vacuum at 60” C. The polymer is characterised by IR and NMR spectroscopy. Its weight average molecular weight, determined by low angle light scattering, is 97,000. The polydispersity obtained by size exclusion chromatography (SEC) is 4.5. More details are given in refs. 1 and 2. Model compounds A series of model compounds, for the structural unit of PBHI or eventual products formed by thermal treatment, were prepared:
Iv:
X=CH,
V:
x = CHOH
VI:
x=co
(BHI)
IV is prepared by refluxing 3 g (0.01 mol) of 4,4’-methylene diphthalic anhydride with 2 g (0.02 mol) of aniline in 20 cm3 pyridine for 4 h. The solution is cooled at 20°C and shaken for 24 h. Crystals are filtered off, washed with pyridine then methanol and finally dried under vacuum at 130°C. V is obtained by reacting 91 g (0.24 mol) of DDB with 47 g (0.50 mol) of aniline in 200 cm3 of pyridine. The mixture is heated to 150°C until the elimination of methanol and water is complete. Upon cooling, the solution is precipitated in water, washed with dilute HCl, then more water, and finally vacuum dried at 120 OC. VI is prepared according to the two-step classical synthesis of imides: 32.2 g (0.1 mol) of benzophenonetetracarboxylic acid dianhydride (BTDA) and 23.6 g (0.25 mol) of aniline are dissolved in 100 cm3 of dioxane and heated at 40 OC. The resulting amic acid is shaken at room temperature with 113 g of acetic anhydride and 113 g of pyridine under argon. The solid is filtered off, treated with diethyl ether, then methanol and dried under vacuum at 40°C. Thermogravimetry
(TG)
Thermogravimetry was carried out under a nitrogen flow of 70 cm3/min in a Du Pont 951 thermobalance coupled to a Du Pont 1090 thermal
254
analyser, using a cylindrical silica sample holder (diameter 7 mm, height 7 mm). Samples of polymer were films lo-20 pm thick cast from NMP which, prior to TG, were heated at lO”C/min to 150” C to eliminate absorbed water. Simultaneous thermogravimetry-evolved
gas analysis (TG-EGA)
The basic features of the technique are those of an automated combined thermogravimetry-high-resolution gas chromatography system, previously described in detail [3], which was modified to allow determination of evolved water. A thermobalance is coupled to a gas chromatograph and to a water-sensitive probe by means of a sampling valve, as shown in Fig. 1. Typical operating conditions were as follows: TG heating rate, 2.6 o C/mm; purge gas, nitrogen dried over molecular sieves (4 A) at a flow of 70 cm3/min; interface oven temperature, 300°C; transfer line temperature, 300” C; sampling loop, 10 ~1; capillary column 20 m x 0.3 mm I.D., SE-52 (coating thickness 0.2 pm); flame ionisation detector, 330 o C; carrier gas, helium at 3 cm3/min; column oven temperature, 200 OC; gas chromatographic separation time, 3 min; loop loading time, 1 min. The thermobalance and the water probe respectively supply continuous recording of the weight of the sample and of instantaneous water concentration in the exiting purge gas as a function of TG temperature. Gas chromatography gives the area of peaks corresponding to the separated
Carrier gas Fig. 1. Thermogravimetry-evolved gas analysis (TG-EGA). TG = thermobalance, Du Pont 951. HRGC = high-resolution gas chromatograph, Fractovap 4160, Carlo Erba. 1= TG furnace tube; 2 = sample holder; 3 = TG furnace; 4 = interface oven; 5 = sampling valve, A2C6WT, Valco Instruments, (- - -) loop loading position, (. . . . . .) loop injection position; 6 = sampling loop; 7 = water probe, Hygrometer III, Panametrics; 8 = heated aluminium block; 9 = HRGC column oven.
255
organic volatile products evolved on heating as measured every 4 min, that is at 10.4” C intervals. The rate of evolution of organic volatile products, identified by classical analytical method (IR, gas chromatography-mass spectrometry, etc.) in separate experiments, is calculated using a previously described calibration technique relating the rate of volatilisation of pure reference compounds in TG to the gas chromatographic peak area [3]. The rate of water evolution as a function of temperature was calculated by multiplying concentration by purge-gas flow-rate which was stabilised by means of a flow controller (Milliflow, Varian). Although the probe is calibrated by the manufacturer, we checked its response in the TG-EGA system. For this, we compared the rate of evolution of water, as measured by the probe, with that measured from weight loss at room temperature of a glass capillary filled with water and placed on the sample pan of the thermobalance. Under these conditions, a constant rate of water evaporation (weight loss) is obtained which depends on the capillary diameter. The results obtained were in agreement with supplied calibration data and within the accuracy claimed by the manufacturer: +5% in terms of dew point or ~15% in terms of rate of water evolution over the range used here. Although apparently poor, the accuracy of the probe is satisfactory for characterisation of the evolution of very small amounts of water as in the present study. The quantities of either water or organic products evolved during heating can be obtained by graphical integration of differential curves of the rate of evolution versus temperature (TG-EGA), or versus time in experiments carried out at constant temperature. Infrared spectroscopy
IR spectra were measured on films or by dilution in KBr pellets using a 20 SX Nicolet Fourier Transform Infrared instrument.
RESULTS AND DISCUSSION
Overall thermal behaviour
The TG curve of PBHI heated at 10” C/mm shows a small weight loss (ca. 4%) between 250 and 400°C (Fig. 2). The resulting product is stable to 500” C; above this it undergoes extensive degradation to 700” C, leaving 60% of the original weight of PBHI as a charred residue. Although the process occurring at 250-400” C involves limited weight loss, important properties of PBHI such as solubility, glass transition temperature and dielectric characteristics vary significantly above 250 OC. In
256 100
s :: ol
‘a, 5c 3
C
Temperature,
Fig. 2. TG curves of PBHI (PBHI, 24 mg; V, 16 mg.
“C
-
-)
and of model compound
V (-).
Sample size:
particular, PBHI becomes insoluble and in the present study chemical modifications induced by heating were mostly investigated by IR spectroscopy. In order to obtain better evidence of IR modifications, a parallel study was carried out on model compound V (BHI) which is characterised by the same benzbydrolimide structure as the polymer. BHI shows the same behaviour as PBHI in TG up to 400 o C (Fig. 2). However, above this temperature, it continues to lose weight up to 700” C leaving 16% residue. First degradation step (T < 400 “C)
The TG-EGA curves of Fig. 3a show that, at a slow heating rate (2.6 o C/mm), PBHI eliminates residual solvent NMP between 210-340 OC with a maximum rate at 270 OC. Evolution of water begins at 250 OC with a
300 Temperature.“C
400
Fig. 3. TG-EGA curves of (a) PBHI, 12 mg and pyrrolidone; P = pyridine; A = aniline; W = water.
(b) BHI, 39 w;
NMP = : N-methyl-2-
257
maximum rate at 345 o C. Since absorbed water is eliminated in the pretreatment to 150°C, water formed in this step should be a product of chemical modification of PBHI. The first conclusion that can be drawn from these results is that residual NMP is strongly linked to the polymer, very probably to the OH groups. The maximum rate of evolution of NMP is observed well above its boiling point (202°C). This suggests that NMP elimination may be related to the transformation of the OH group shown below. A strong reduction in OH absorption at 3480 cm-’ is the most evident modification of the IR spectrum of PBHI upon heating to 315” C for 7 h (Fig. 4), at which temperature water is eliminated at a relatively high rate (Fig. 3a). In addition, a shoulder appears at 1675 cm-’ that is in the region of C=O stretching in aromatic ketones (1660-1670 cm-‘) [4]. In fact, benzophenone shows this absorption at 1660 cm-‘. Furthermore, in the IR spectrum of heated PBHI there is an increase or decrease in intensity of several medium intensity absorptions in the region 1100-1300 cm-’ as indicated in Fig. 4. However, the imide structure of PBHI is not affected by heating at this stage since the most intense absorptions of PBHI below 3200 cm-’ and, in particular, typical imide absorptions at 720, 1380, 1725 and 1780 cm-‘, are not affected by the water-elimination reaction.
b
3500
1
3000
I
I
1000
1500 Wavenumber,
cm’
Fig. 4. IR spectra of PBHI film (1) before and (2) after heating at 315°C nitrogen. Arrows indicate absorption increase or decrease.
for 7 h under
258 r
3500
3000
1500
I
I
1000 Wavenumber,
500 cm-’
Fig. 5. IR spectra of BHI, (1) original and heated at 340 o C under nitrogen 50% overall weight loss.
to (2) 10% and (3)
As far as the model compound, BHI, is concerned, TG-EGA of Fig. 3b shows that it eliminates water with the same kinetic characteristics as PBHI. Pyridine and aniline, also evolved between 160 and 4OO”C, are residual solvent and reagent used in the synthesis, respectively. Upon elimination of water, the IR spectrum of BHI, which is closely similar to that of PBHI, undergoes essentially the same modifications listed above in the case of the polymer (Fig. 5). In particular, BHI clearly shows an increased absorption at 1660 cm-l which appears as a shoulder at 1675 cm-’ in PBHI. A slight difference between heated PBHI and BHI occurs in the region 1100-1300 cm-‘, where a weak band at 1190 cm-’ tends to disappear on heating BHI, whereas in the case of the polymer it is not affected. Minor differences can also be detected in the region below 900 cm- ‘. During water elimination from BHI, condensation of a solid product as long needles was observed on the section of the tube, enclosing the thermobalance, where it emerges from the furnace. This explains the extensive weight loss overlapping water elimination in the case of BHI (Figs. 2 and 3). The IR spectrum of the condensed product is essentially that of the ketone-imide model compound VI (Fig. 6, trace 2) which is characterised by
259
: 1 ”
-
1
1500
I
I
1000
500
Wavenumber,
Fig. 6. IR spectra of model compounds: (1) V, (2) indicate typical absorptions of the model compound.
cm-’
VI and (3) IV. Underlined
wavenumbers
the presence of the absorptions shown to appear or increase in residues after elimination of water from either PBHI or BHI. These data indicate beyond doubt that water is formed in a reaction involving the =CHOH group of the benzhydrol structure, which is at least in part oxidised to the corresponding ketone. In the case of BHI, the resulting product is the model compound VI which is volatile at the temperatures reached for water elimination. Indeed, the model compound VI, heated in TG at 10 o C/mm, mostly evaporates unaltered between 390 and 530 o C. Mechanism
of reaction
Since the thermal treatments were carried out under an inert atmosphere, elimination of water from PBHI and from BHI should imply an internal redox mechanism. Stoichiometry of water elimination from PBHI and from BHI was obtained from plots of the type shown in Fig. 7 of the rate of water evolution,
260
Time.min
Fig. 7. Rate of water evolution
from BHI at 360 o C under nitrogen.
measured at constant temperature, versus time. Although the rate of water formation was expected to decrease continuously with increasing time, an initial increase from zero to a maximum value is shown in Fig. 7 which corresponds to the time required by the sample to reach the selected temperature (8 min), after which the expected trend is observed. From integration of the area below the rate-time curve, it was calculated that the total amount of water evolved corresponds, within experimental error, to about 50% of the total original number of moles of OH groups in both BHI and PBHI. Thus the following complete redox reaction can be tentatively written for the =CHOH group of benzhydrolimide structures: 2’
7/
‘i
i)H
-
‘c’
+ II
'CH;'
+
H,O
(2)
6
While the presence of C=O groups in the dehydration products of PBHI was clearly established, that of methylene-bridged aromatic imide structures is not easy to detect by IR spectroscopy, owing to the overlapping of absorption with those of the same structures bridged by C=O or =CHOH groups, as shown by comparison of the IR spectra of corresponding model compounds in Fig. 6. Furthermore, PBHI contains diphenylmethane moieties in its molecule, whose methylene groups should give rise to IR absorptions similar to those of the methylene groups bridging the imide structures. On the other hand, the model compound BHI is of little help in this respect since the model compound IV, characterised by the presence of the methylene bridge, which would be formed in reaction (2), was shown in a separate TG to volatilise between 315 and 350°C upon heating at a rate of 10 c,C/mm. Therefore, completion of reaction (2) in the case of BHI, should lead to almost complete volatilisation of a 1 : 1 mixture of model compounds IV and VI, which should condense in the cold section of the thermobalance. However, in this case too, the presence of compound IV would not be easily detectable by IR spectroscopy. As a matter of fact, upon heating BHI to constant weight at 390 o C, only a 75% total weight loss was attained after 17 h. The IR spectrum of the solid condensed from the purge gas was indeed
2611
close to that of the ketone-imide model VI. As far as the involatile residue is concerned, its IR spectrum, apart from the disappearance of OH absorption at 3480 cm-‘, and the presence of a very weak shoulder at 1660 cm-‘, is closely similar to that of original BHI. This suggests that condensation reactions of benzhydrolimide structures, through OH elimination, may also occur, leading to an involatile product. Thus, the redox process involving benzhydrolimide structures, which occurs on heating PBHI, seems more complex than that summarised by reaction (2). Work is in progress to elucidate the detailed mechanism of this process.
REFERENCES 1 2 3 4
M.E. Quenneson, J. Garapon, M. Bartholin and B. Sillion, in preparation. M.E. Quenneson, M. Bartholin and B. Sillion, in preparation. L. Costa, G. Camino and L. Trossarelli, J. Chromatogr., 279 (1983) 125. L.J. Bellamy, The Infra-red Spectra of Complex Molecules, Chapman and Hall, London, 1975.