Journal of Non-Crystalline Solids 172 174 (1994} 575 579
Physical aging phenomena in an amorphous polymer at temperatures far below the glass transition E. Muzeau 1, G. Vigier*, R. Vassoille Groupe d'Etudes de MOtallurgie Physique et Physique des MatOriaux, URA CNRS no. 341, INSA, 69621 Villeurbanne, codex, France
Isothermal agings have been carried out on poly(methyl methacrylate) at temperature far below the glass transition temperature region. Aging effects have been investigated by calorimetric and dynamic mechanical techniques; influence of quenching rate, time and temperature of aging have been studied. It is seen that the 13-relaxation is not affected by physical aging and that the a-relaxation spreads out in the fl-relaxation temperature domain. Comparison of the results obtained from the two techniques shows the occurrence of a specific mechanism for aging temperature far below the glass transition temperature.
Physical aging in polymer glasses is a phenomenon which has been extensively studied [1-4] through several techniques amongst which the calorimetric and mechanical ones. Physical aging consists of a spontaneous evolution of a non:equilibrium system towards a metastable equilibrium (supercooled liquid), through a diffusion process. The physical aging phenomenon is known to be time- and temperature-dependent. When aging is performed close to the glass transition temperature, Tg the system evolves towards its equilib-
] Present address: Laboratoire d'Etudes des Mat6riaux Plastiques et des Biomat6riaux - URA CNRS no. 507, Universit6 'Claude Bernard' Lyon I, 43, Boulevard du 11 Novembre 1918, 69622 Villeurbanne c6dex, France. *Corresponding author. Tel: +33 72 43 81 50.
rium within a duration of the order of minutes. However, as the aging temperature is significantly lowered, the time for reaching equilibrium becomes exponentially greater so that it is often out of scale of the experimental time. Most of the studies on physical aging have therefore been performed at temperatures close to Ts. In this work, we have focused on isothermal aging effects occurring at temperatures much lower than Tg, i.e. Tg - 100 K. It happens that this aging temperature is close to room temperature for poly(methyl methacrylate) (PMMA). It is therefore of significant interest to determine and interpret the experimental features of physical aging on P M M A at very low temperature compared to Tg. A previous work presented dynamic mechanical results of low temperature aging on P M M A which emphasised an unusual process . We report here complementary results obtained from both calorimetric and dynamic mechanical
0022-3093/94/$07.00 © 1994 Elsevier Science B.V. All rights reserved SSDI 0 0 2 2 - 3 0 9 3 ( 9 4 ) 0 0 1 2 9 - B
E. Muzeau et al. /Journal of Non-Crystalline Solids 172-174 (1994) 575 579
measurements. Qualitative examination and comparison of these results lead to physical implications which are discussed here.
ta Z 9 l O s i , r I
2. Experimental A R6hm and Haas (Germany) and a Norsolor (France) poly(methyl methacrylate) or PMMA, in the form of 2 mm thick sheet, have been studied. Their calorimetric glass transition temperatures (considered as the temperature at which the metastable equilibrium is reached), measured for a heating rate of 10 K/min, are 396 and 4 0 4 K , respectively. Their molecular weights Mw are 400000 and 67 000, respectively. Thermal treatments were performed as follows: (i) the sample was maintained above the glass transition temperature domain for more than 15 min and thereafter cooled to 125 K at a determinate rate of Vq (ii) isothermal aging was performed at a temperature T~ for a certain time ta (iii) then the dynamic or calorimetric measurements were carried out on increasing temperature at a rate, Vm. Dynamic mechanical measurements were obtained by means of an inverted forced torsional oscillation pendulum at frequencies of 0.1 and 1 Hz . To avoid contamination of the samples by moisture, all measurements and agings were made in hermetically sealed conditions. Vq and Vmcorresponding to dynamic mechanical measurements were therefore low and equal to 6 and 0.6 K/min, respectively. Calorimetric measurements were performed on a Mettler TA 3000 equipment. The rate Vq was either 1000 or 6 K/min to allow for comparison with the mechanical results and the rate vm was 10 K/min.
3. Results The results of specific heat, Cp, versus temperature were obtained for samples of different thermal histories. Fig. 1 shows the comparison of the data collected for an unaged sample and aged samples of
K Fig. 1. The DSC measurements on R6hm and Haas P M M A for different aging temperatures T~ and different aging times t~, with vq = 1000 K/min.
different aging times and different aging temperatures. Aging at 300 K produces a peak which occurs at a temperature much below Tg; it is therefore rather referred as pre-peak. Its amplitude is rather small but clearly detectable. As the aging time increases, two simultaneous phenomena are observed: the pre-peak shifts towards a higher temperature and its height increases, as observed by Hodge and Huvard . The curves corresponding to aging at higher temperatures also follow this general trend. Moreover, these results clearly show that the pre-peak in Cp shifts towards a higher temperature and its amplitude becomes higher the higher the temperature of aging. These general features are a coherent extension of the Cp peaks obtained by Gomez-Ribelles et al. 1-7] on P M M A for aging temperatures somewhat higher. The same aging temperatures have been used for the dynamic mechanical spectrometry (DMS) characterisation. The results are shown in Fig. 2(a) where the logarithmic scale enables us to see both the c~- and 13-mechanical relaxations. Again effects of aging at 300 K are clearly detectable and appear as a decrease of tan~b in a temperature domain much lower than Tr As the aging temperature is increased, similar effects are seen to occur at
E. Muzeau et al. /Journal o[Non-Crystalline Solids 172-174 (1994) 575-579
 T 150
300 T K
Fig. 2. D M S m e a s u r e m e n t s on R 6 h m a n d H a a s P M M A at 0.1 H z for different t h e r m a l histories, with Vq = 6 K / m i n : [13 unaged,  77, = 300 K a n d ta = 2.6 x 106 s,  Ta = 322 K and ta = 2.8 x 105 s,  Ta = 363 K and t, = 3.3 × 103 s. A tan ~b is the difference between u n a g e d and aged samples. The d a s h e d A tan ~ curve is explained in the text.
a higher temperature domain and with a greater amplitude. Fig. 2(b) shows the difference A tan q~ of unaged and aged samples. A tan q5 curves all exhibit a peak that shifts towards a higher temperature and has a greater amplitude when aging temperature is increased. Plot of the Log(tan q~) curve for aging at 300 K exhibits a particular phenomenon between the [3-and ~-relaxation peaks which is enlarged in Fig. 3: the appearance of a maximum in tan ~b at about 330 K. The temperature position of this maximum is independent of the frequency as shown by our measurements at 0.1 and 1 Hz.
Differential scanning calorimetry (DSC) and dynamic mechanical spectrometry (DMS) measurements, as presented previously, have shown changes on physical aging at aging temperatures as low as Tg-100 K. First, a qualitative analysis of the dynamic mechanical results is given. Then the mechanical and calorimetric results are compared in order to further develop our concept of physical aging phenomena occurring far below Tg.
Typical dynamic mechanical results of amorphous P M M A exhibit two relaxation processes which are referred on decreasing temperatures as the ~- and 13-relaxations. The ~ or main relaxation is related to the glass transition and involves long range molecular movements. It is therefore affected by physical aging. The 13-relaxation is a secondary or sub-T g-relaxation which involves thermally activated motion on a very localised scale. Our DMS results on aging at 300 K are particularly valuable because they emphasise aging effects below the 13-relaxation maximum. It seems therefore important to determine whether this mechanical relaxation is affected by physical aging, as the subject is controversial in the literature [8-9]. The [3-relaxation of P M M A has been found to have a very broad distribution of relaxation times . At 0.1 Hz, it extends down to temperatures much below 200 K, and its maximum in tan q5 occurs at 285 K (Fig. 2) Since aging at 300 K does not affect the mechanical results below 250 K, the 13molecular movements having the shortest relaxation times are not involved in the physical aging process. Moreover, it seems very much unlikely that only the slowest movements involved in the [3-relaxation would be affected by physical aging. Plots of A tan ~b show that aging effects are maximum at a temperature depending on the aging temperature. Since these temperatures do not
E. Muzeau et al. / Journal o f Non-Crystalline Solids 172- 174 (1994) 575-579
correspond to T~, Johari's criteria  enables us to believe that physical aging does not affect the fl-relaxation process. We therefore conclude from the two previous developments that the 13-relaxation process is not involved in the physical aging phenomenon. Therefore, only modifications on aging of the at-relaxation features account for the apparent aging effects on DMS results. Then it becomes quite clear that the Qt-relaxation contributes to the global tan ~b in a very large temperature range, i.e. down to at least 250 K and 0.1 Hz. Only this low temperature tail of the Qt-relaxation is affected by physical aging, in a temperature range that is lower, the lower the aging temperature is. Let us now focus on the significance of the maximum in tan ~b at 330 K, maximum which is not frequency dependent (Fig. 3). A separate experiment, aimed to determine the nature of this phenomenon, was carried out on a Norsolor P M M A (Fig. 4). We have measured tan q~ on increasing temperature for a 300 K-aged sample up to the point where the maximum occurs (see curve (b)). Sample was immediately after cooled and measured again on a second scan (see curve (c). This
t,,. ¢ 0.09
Fig. 3. The D M S measurements as 0.1 and 1 Hz on R6hm and Haas P M M A aged 2.6 × 106s at 300 K.
Fig. 4. The D M S measurements at 0.1 Hz of Norsolor P M M A , with vq = 6K/rain: (a) unaged, (b) T~ = 3 0 0 K and ta = 4.4 × 106s. When the sample had reached 3 3 6 K , temperature was lowered and measurement was carried out up to 346 K (c).
experiment constitutes a thermal treatment which enables a partial recovery of the mobility, as the second scan features tend towards these of an unaged sample (curve (a)). These results underline the kinetic effect of the phenomenon and show the link between the occurrence of the maximum and the thermal history of the sample. The position in temperature of this maximum, just as Tg, is a function of the heating rate during t.he measurement. The difference A tan ~b between curves (a) and (c) is plotted in dashed line in Fig. 2(b). The thermal treatment constituted by the two consecutive DMS scans are comparable to an isothermal aging at a temperature of about 325 K. !n order to compare DSC and DMS aging results, we performed a quenching on a DSC sample with a same quenching rate as for DMS measurements, i.e. vq = 6 K/min. The DSC sample was aged f o r 1 0 6 s at 300 K and measured on heating at a rate of 10 K/min. The Cp curve exhibits no detectable change from that of the unaged sample. It is rather puzzling that, under the same aging conditions, the DMS measurements are affected whereas the DSC measurements are not. One has to think of a possible explanation of this apparent
E. Muzeau et al. / Journal of Non-Crystalline Solids 172 174 (1994) 575-579
inconsistency and we suggest the following interpretation. When aged at a temperature much below Tg, a system can hardly evolve towards equilibrium through diffusion but very local rearrangements might still be possible. This localised motion would lead to a decrease of molecular mobility, i.e. a diminution of tan ~b, without any detectable change of specific volume or enthalpy, i.e. no Cp peak. This phenomenon is likely to occur during any aging experiment. However, under the usual aging conditions, it is negligible compared to movements occurring on a greater scale; it remains the only possible mechanism when quenching is slow and aging temperature is much below Tr Although purely qualitative, this analysis of the results is a first step towards a better understanding of a very complex phenomenon. It will be completed by a theoretical analysis of the experimental results from both calorimetric and mechanical techniques . Indeed, a physical model developed in our laboratory [12-14] is currently applied to account for the observed phenomena and is aimed to provide a physical interpretation of physical aging based on elimination of quasi-point defects. A quasi-point defect corresponds to a localised fluctuation of density, as it can be observed through light scattering technics . The DSC measurements are directly related to the concentration of these defects, which concentration does not evolve for a slow quenching rate at a low aging temperature. The DMS measurements are sensitive to both the concentration of defects and local rearrangements affecting the global mobility. Dynamic mechanical spectrometry appears therefore to be more sensitive than DSC for physical aging studies.
The use of several independent techniques is particularly valuable to obtain a complete knowledge of the physical aging phenomenon. In this work, changes in thermodynamic and anelastic properties have been explored on isothermal aging at particularly low temperatures compared to Tg. The resuits have been compared with those of agings at temperatures just below Tr It has been concluded
from a qualitative analysis of the mechanical results that only the main or Qt-relaxation is affected by physical aging. As a consequence, the 0~-relaxation is found to exhibit a low temperature tail which extends down to the 13-relaxation domain. Aging effects on the s-relaxation, and also on the specific heat features, occur in a temperature range that is lower the aging temperature is. At a fixed aging temperature, an increase in the duration of the isothermal aging leads to both a shift towards higher temperatures and a greater amplitude of the observed effects. A comparison of calorimetric and mechanical results for the same slow quenching rate and low temperature aging has led us to some apparent inconsistencies which have been interpreted in terms of local molecular rearrangements.
References  A.J. Kovacs, J.M. Hutchinson and J.J. Aklonis, in: The Structure of non-crystalline Materials, ed. P.H. Gaskell (Taylor and Francis, London, 1977) p. 167.  L.C.E. Struik, in: Physical Aging in Amorphous Glassy Polymers and Other Materials (Elsevier, Amsterdam, 1978).  A.R. Berens and I.M. Hodge, Macromolecules 15 (1982) 756.  I.M. Hodge and G.S. Huvard, Macromolecules 16 (1983) 371.  E. Muzeau, J.Y. Cavaill~, R. Vassoille, J. Perez and G.P. Johari, Macromolecules 25 (1992) 5108.  S. Etienne, J.Y. Cavaill6, J. Perez, R. Point and M. Salvia, Rev. Sci. Instrum. 53 (1982) 1261.  J.L. Gomez-Ribelles, A. Ribes-Greus and R. Diaz-Calleja, Polymer 31 (1990) 223.  G.P. Johari, J. Chem. Phys. 77 (1982) 4619.  L.C.E. Struik, Polymer 28 (1987) 57.  E. Muzeau, J. Perez and G.P. Johari, Macromolecules 24 (199l) 4713.  E. Muzeau, G. Vigier, R. Vassoille and J. Perez, submitted to Polymer.  J. Perez, J.Y. Cavaill6, S. Etienne and C. Jourdan, Rev. Phys. Appl. 23 (1988) 125.  R. Diaz-Calleja, J. Perez, J.L. Gomez-Ribelles and A. Ribes-Greus, Makromol. Chem. Macromol. Symp. 27 (1989) 289.  G. Vigier and J. Tatibouet, Polymer 34 (1993) 4257.  C.T. Moynihan and J. Schroeder, presented at the 2nd Int. Discussion Meeting on Relaxations in Complex Systems, June-8 July, 1993, Alicante, Spain.