Aerican Coposites Manufacturers Association January 15-17, 29 Tapa, FL USA Abstract THERMAL ENDURANCE OF UNREINFORCED UNSATURATED POLYESTERS AND VINYL ESTER RESINS by Thore M. Klaveness, Reichhold AS In addition to axiu service teperature as deterined fro heat deflection teperature (HDT) or glass transition teperature (Tg), theral endurance is iportant for coposites exposed to high teperatures. Typical applications for such are for electrical coponents and under the hood autootive. It is beneficial to have accelerated ethods for the deterination of theral endurance. Therogravietric analysis (TGA) is one such ethod. Standards exist for estiating life tie at elevated teperatures by TGA. ASTM standards E 1641 and E 1877 consider the decoposition kinetics and the estiation of theral endurance respectively. Deterination of relative theral indexes can then be done according to ISO 2578. In this study selected unsaturated polyester and vinyl ester resins have been studied by TGA. Attepts to interpret results by the procedure of ASTM E 1641 was possible for isophthalic resins, but not for DCPD and vinyl ester resins due to ore coplex decoposition echaniss. By an alternative isotheral TGA ethod and interpretation by basic cheical kinetics we were able to estiate theral endurance of these resins. The results suggested better theral endurance at teperatures around and below 2 C for the DCPD resin tested than the isophthalic resin, a fact that could not easily be detected by the ASTM procedures. This behaviour was confired in an oven weight loss test at 22 C for 3 hours duration. Introduction In addition to axiu service teperature as deterined fro heat deflection teperature (HDT) or glass transition teperature (Tg), theral endurance is iportant for coposites exposed high teperatures. Typical applications for such are for electrical coponents and under the hood autootive. It is of course benefical to have accelerated ethods for the deterination of theral endurance. Therogravietric analysis (TGA) is one such ethod. Standards exists for estiating life tie at elevated teperatures by TGA. ASTM standards E 1641 [1] and E 1877 [2] consider the decoposition kinetics and the estiation of theral endurance therefro respectively. Deterination of relative theral indexes can then be done according to ISO 2578 [3]. A restriction on the use of the ASTM E 1641 ethod to deterine kinetic paraeters is that it applies to first order reaction kinetics where the decoposition follows what is described as a sooth continous ass change with a single axiu rate [1]. In the siple case of a decoposition reaction following first order reaction kinetics one can describe the ass loss by the first order rate equation: d dt = (1) k And with a rate constant, k, described by an Arrhenius relationship: E a RT k = Ae (2) Operating the TGA in a scanning ode, heating the saple at a preset rate, β [K/in]: T= T +β t (3) Cobining these and attepting to solve we will get: A T Ea RT ln = β e dt (4) T ASTM E 1641 calls for running four or ore different heating rates all between 1 and 1 K/in. The analysis is not straightforward, as can be seen fro equation 4 above. This is because the integral is not easily solved. Thus there is an iterative ethod involving tabulated nuerical integration constants described that will give the activation energy. An alternative would be to study the decoposition by isotheral experients. In the case of isotheral decoposition data obeying first order kinetics the analysis is uch sipler, as the rate constant now actually is constant during the experient: ln = k t (5) This will give the rate constant as the slope of a sei-log plot of ln() vs. tie. The ajor disadvantage is that the isotheral experient will generally take longer to run. Also several experients at different teperatures are needed to give the teperature dependence of the decoposition rate according to equation 2. Another disadvantage is the heating fro abient to test teperature necessary before the isotheral TGA starts. This will hide any deviations fro the siple first order odel of equation 5 if they take place at very low conversions. 1
Experiental Cured unsaturated polyester saples were prepared. An isophthalic resin and a dicyclo pentadiene (DCPD) resin were tested. Saples were diluted with styrene to a viscosity of 3 Pas to give acceptable castings. The curing syste used was the coonly applied redox initiation syste with cobalt octoate and MEK peroxide. Saples were cast between prepared glass plates at 2 thickness and allowed to cure 24 hours at roo teperature followed by postcuring 24 hours at 6 C and 3 hours at 15 C. Siilarly a vinyl ester resin casting was also prepared with cobalt octoate and cuene hydroperoxide as curing syste. Saples, all in one piece, approxiately 2 g each and preferably not taken fro the edges of the castings were used for the TGA analysis. For the scans at different rates according to the ASTM ethod, 3, 5, 7 and 1 K/in heating rates were used. Isotheral experients were perfored at teperatures 3, 32, 34 and 36 C, with initial 4 K/in raping rate fro abient teperature. The decoposition was followed in a nitrogen atosphere. Results: Unsaturated Polyesters Inital experients were run according to the ASTM ethod [1]. A striking difference between the isophthalic resin and the DCPD resin was observed. This is presented in Figure 1. While it shows a sooth continous ass change with a single axiu rate and following first order kinetics probably for the isophthalic resin, it certainly does not for the DCPD resin. The shape of the latter curve suggest ultiple reactions taking place. For this reason isotheral experients were perfored for both resins. The isophthalic resin was tested first as represented by the sei-log plots of Figure 2. First order rate constants for the four different teperatures have been deterined by linear regression of the sei-log data plotted above. Activation energy, Ea, in accordance with the Arrhenius equation, equation 2, have been deterined as shown in Figure 3. For all the resins tested the values obtained are listed in Table 1. Coing back to the coparison between isophthalic and DCPD resin, Figure 4 shows an exaple taken at one of the teperatures used in the isotheral easureents. This clearly shows the difference between the two resins, and in the case of the DCPD resin the deviation fro the behaviour of a single first order reaction. The DCPD resin gives a faster initial decoposition than the isophthalic resin but later it levels out with a slower rate. The isotheral experients on the DCPD resin are shown in figure 5. The sae kind of behaviour can be observed at all teperatures. This behaviour can be described by a biphasic first order rate equation: k It kiit = ae + be (6) Typically this is the behaviour of reaction echaniss involving consecutive or parallell first order reactions. Based on equation 6 the kinetic paraeters for the weight loss data was deterined by an iterative least squares ethod. The experiental data seeed to be very well described by equation 6, this is shown in Figure 6. Figure 7 shows Arrhenius plots for rate constants, ki and kii deterined for the DCPD resin. It has also been attepted for the noralized preexponential factor a. Preexponentials, a and b were noralized to a+b=1, i.e. starting the biphasic curve with no weight loss at tie equals inutes. Experiental data have tie lags due to the initial teperature rap. Good linearity in the Arrhenius plot is observed for ki. For kii it is ore uncertain. Also the linearity is not ipressing, and it should be noted that reaction echaniss including ultiple first order reactions will give linearity of this plot only in certain teperature regions. Thus the linear regression shown above, and especially extrapolations of this, should be treated with great care. The slope of the curve for the preexponential factor, a in Figure 7 corresponds to an "activation energy" of 38 kj/ol. This is expected be a function of the activation energies of different rate constants in a ultiple first order reaction echanis. Oven weight loss data at 22 C for the DCPD resin was also obtained, it is of interest to see how the kinetic odel is capable of describing these results. This is shown below in Figure 8. This was obtained using the extrapolations of the above Arrhenius plots for the kinetic odel. The result is not conclusive, but the odel sees to pick up the general trend and the agnitude of the weight loss. This is notable because uch of this effect is really that the crossover between fast initial decoposition and the subsequent slower decoposition takes place at lower weight loss at lower teperatures. This is reflected in the teperature dependence of the preexponential factors of equation 6. Note that the oven weight loss was not really in an inert gas atosphere. Still it had larger saples so the specific surface for oxygen to have an effect was less than what it would have been for TGA saples at the g scale. Theral Endurance: Unsaturated Polyesters The DCPD resin does not show theral decoposition according to a single first order reaction. Especially since the decoposition "levels out" at different levels at 2
different teperatures the ethod described in ASTM E 1877 [2] and ISO 2578 [3] for deterination of theral endurance will not be applicable. These ethods use the extrapolation of data in an Arrhenius type plot, fro easureents at high teperatures to estiate the teperature that will give a certain lifetie. To visualize the effect of the observed kinetic odel on the lifetie at lower teperatures, the sets of data are extrapolated, even though the extrapolation of the preexponential factors especially is questionable. The resulting lifetie chart for weight losses 5, 1, 15 and 2% is shown in Figure 9. And indeed it can be seen fro Figure 9 that the observed decoposition kinetics of the DCPD resin does not give tie-teperature plots that can be extrapolated in the anner described in the standards entioned. A reasonable linear relationship is achieved for the high weight loss data only. Regarding the tie necessary to reach a specified weight loss, when going fro high to low teperatures a shift towards longer tie is observed. And again that reflects the crossover between fast initial and later a slower decoposition. The siple first order odel found to fit for the isophthalic resin, on the other hand, will of course lead to tie-teperature Arrhenius type plots as assued in the standards. Vinyl Esters A standard bisphenol A based vinyl ester resin was also tested in the sae way. Figure 1 shows the isotheral TGA weight losses at the sae teperatures as for the isophthalic and DCPD resins. Biphasic kinetics can be observed, but it appears that the faster reactions observed especially at 3 and 32 C has a lower activation energy than what was the case for the DCPD resin. At higher teperatures the biphasic nature disappears possibly as the second rate constant doinates. At 34 C there is a biphasic nature, but at 36 C it can not be observed at all. Note that the scale of figure 1 is very different fro figure 5 indicating a better theral endurance in a vinyl ester. The crossovers are at very low weight losses, and the curves also indicates that this will shift further towards lower weight losses as teperature is reduced. So probably the theral endurance of this vinyl ester can be estiated assuing the first step observed at 3 and 32 C is neglitible. A single rate constant was deterined for the decoposition at 36 C. This was used together with the kii values for the other teperatures in Figure 11. This sees to fit well with the trend observed, giving good linearity. Values obtained for rate constants and activation energy is at the level as for kii of the DCPD resin shown in Figure 7. Theral Index Table 1 gives the rate constants deterined and the theral endurance of the resins investigated. The TI used here was the teperature giving 5% weight loss after 2 hours. Rearkably, a higher theral index was found for the DCPD resin, even than the vinyl ester. Again this reflects that kii values and teperature dependencies found are quite siilar for DCPD and vinyl ester resins. And also that the faster initial decoposition which is uch higher on the DCPD resin at 3-36 C becoes less iportant at lower teperatures, i.e. in the range of 2 C. At 196 C the extrapolated preexponential factor value is only.25. A weakness of this coparison is that the data for the DCPD resin depend very uch on values for kii and a for which the Arrhenius plots, Figure 7, showed poorer linearity. Conclusions Biphasic first order decopositions was found for the DCPD and vinyl ester resins tested. This was deterined in isotheral TGA easureents, which also provided the necessary kinetic paraeters for the deterination of theral endurance by extrapolation of tieteperature data. The isophthalic resin studied had a decoposition following the siple first order required in ASTM E 1641 A coparison of the resins theral endurance was still deterined by the isotheral easureents. Relative perforance of the resins in these test showed DCPD and vinyl ester resins being approxiately equal, and both outperfors the isophthalic resin. The use of the ASTM E 1641 ethod to deterine kinetic paraeters for theral decopositions of the DCPD and vinyl ester resins was not possible due to the ore coplex reaction echanis. References 1) ASTM E 1641 (1998), "Decoposition Kinetics by Therogravietry". 2) ASTM E 1877 (1997), "Calculating Theral Endurance of Materials fro Therogravietric Decoposition Data". 3) ISO 2578 (1993), "Plastics - Deterination of Tie- Teperature Liits after Prolonged Exposure to Heat". 3
Figures 1 95 9 Weight% 85 8 75 7 65 Isophthalic DCPD 6 3 31 32 33 34 35 36 37 38 39 4 41 Teperature ( C) Figure 1. TGA scan at 7 K/in for the isophthalic and the DCPD resin. -.1 -.2 3 C ln(/ ) -.3 -.4 34 C 32 C -.5 -.6 36 C 2 4 6 8 1 Tie (in) Figure 2. Isotheral weight loss data for isophthalic resin. 4
lnk -2-2.5-3 -3.5-4 -4.5-5 -5.5-6 -6.5.157.16.163.166.169.172.175 1/T (K -1 ) Figure 3. Arrhenius plot of first order rate constant for isophthalic resin decoposition. -.5 -.1 ln(/ ) -.15 -.2 -.25 -.3 -.35 -.4 DCPD resin Isophthalic resin 2 4 6 8 1 12 14 Tie (in) Figure 4. Isotheral data at 32 C on the two different unsaturated polyesters. 2
-.1 -.2 3 C ln(/ ) -.3 -.4 -.5 36 C 34 C 32 C -.6 5 1 15 2 Tie (in) Figure 5. Isotheral weight loss data for DCPD resin. 1.95 32 C data LS curve fit.9 /.85.8.75.7 5 1 15 2 25 Tie (in) Figure 6. Least squares ethod curve fit on 32 C weight loss data for DCPD resin. 3
-.7-1 -.9-2 k I -1.1-3 -1.3 lnk -4-5 a -1.5-1.7 lna -6-1.9-7 k II -2.1-8 -2.3-9 -2.5.157.16.163.166.169.172.175 1/T (K -1 ) Figure 7. Arrhenius plot of k I ( ), k II ( ) and preexponential a ( ). 5 4 Weight loss (%) 3 2 1 odel prediction oven weight loss 1 2 3 Tie (h) Figure 8. Oven weight loss for DCPD resin. 4
8 7 6 log 1 (tie) 5 4 3 2 1 2% 15% 1% 5% 2 hrs. -1.17.18.19.2.21.22.23.24 1/T (K -1 ) Figure 9. Lifetie chart for the weight loss of the DCPD resin. -.2 ln(/ ) -.4 -.6 -.8 -.1 32 C 3 C -.12 -.14 36 C 34 C -.16 5 1 15 2 Tie (in) Figure 1. Isotheral weight loss data for vinyl ester resin. 5
ln(k II ) -4.5-5 -5.5-6 -6.5-7 -7.5-8 -8.5-9.157.16.163.166.169.172.175 1/T [K -1 ] Figure 11. Arrhenius plot of k II for the vinyl ester resin. DCPD Iso VE a k I [in -1 ] k II [in -1 ] k [in -1 ] k II [in -1 ] 3 C.13.32.22.28.21 32 C.2.78.37.92.648 34 C.25.193.144.31.22 36 C.28.45.96.97.69 Ea [kj/ol] 38 134 191 18 175 TI [ C] 196 17 193 Table 1. Data on the different resins tested. 6