Polymer International 40 (1996) 239-249 Multiple Dielectric Relaxation Processes in an Unequilibrated Bisphenol-A Polyca r bonate Graeme J. Pratt* & Michael J. A. Smith Department of Physics, University of Warwick, Coventry CV4 7AL, UK (Received 15 December 1995; accepted 12 March 1996) Abstract: Complex relative permittivity data within the ranges 0-1Hz-3 MHz and -175°C to + 2 W C are presented for an unequilibrated compressionmoulded bisphenol-A polycarbonate of commercial origin (LEXAN 141). Comparisons are drawn with the nitrogen-equilibrated material in which eight distinct absorption regions have been identified. One low-frequency region common to both materials at low temperature is shown to be caused directly by a processing stabiliser. A second at high temperature is thought to originate in part from an interfacial polarisation associated with a second phase of polycarbonate. An absorption observed only in the unequilibrated polycarbonate is attributed to bound water. Two intermediate-temperature losses are differentiated and delineated for the first time. Their origins and reason for diminution by annealing are examined. The multiplicity of overlapping absorptions, including a low-temperature loss region which is almost continuous at higher frequencies, is consistent with the high impact strength of polycarbonate over an extended temperature range. K e y words: dielectric loss, permittivity, polycarbonate, DC-resistance, interfacial polarisation, low-temperature properties, intermediate-temperature processes, relaxation processes, impact strength. 1 INTRODUCTION ratively high, although some deterioration of properties occurs where exposure to ultra-violet radiation or other environmental influences is prolonged. The photodegradation of polycarbonate, as indicated by dielectric measurement, is the subject of a separate study.' Comprehensive measurements of dielectric response at sub-gigahertz frequencies can yield valuable information regarding modes of molecular and segmental motion in polymers. Characterization of the motion is often facilitated by the use of complementary techniques such as dynamic mechanical measurements, thermallystimulated depolarization (TSD), deuterium or carbon13 nuclear magnetic ~ ~ S O ~ a n(NMR), ce small-ande neutron scattering (SANS) and electron spin resonance (ESR). In particular, comparisons between absorption Polycarbonate is much used for its high impact strength, dimensional stability and low &electric loss over a wide temperature range. These properties, combined with its transmissibility and low crystallinity, have led to applications in recent technologies, such as the optical highway in car area networks (CAN) and the CD-ROM, in addition to established uses in, for example, safety-related products and film capacitors. Resistance to ageing and photo-degradation is cornpa- * To whom all correspondence should be addressed at: Department of Mechanical & Manufacturing Engineering, University of Melbourne, Parkville 3052, Australia. 239 Polymer International 0959-8103/96/$09.00 0 1996 SCI. Printed in Great Britain 240 peaks observed by dielectric and dynamic mechanical methods can elucidate details of the specific molecular, segmental and side-group motions responsible for transitions and secondary relaxations, as cited for polymethyl metha~rylate~.~ and for polystyrene and related polymer^.^ Dielectric response at the lower frequencies includes contributions reflecting the phase morphology of the polymer and its content of mobile charged species. Material differencesbetween phases cause charge accumulations at the phase boundaries which lead to dielectric spectra characterized by a Debye shape. Such interfacial polarisations, usually referred to as MaxwellWagner-Sillars (MWS) processes, can produce a high dielectric increment, particularly when conductivity differences between the phases are pr~nounced.~ The shape of a suspended phase may or may not be of particular significance. A further contribution to the imaginary permittivity at the lower frequencies is provided by any non-dispersive DC conductivity associated with translational charge motion, normally that of impurities in the continuous phase. The magnitude and inverse frequency dependence of this loss may cause the imaginary component of a low-frequency MWS process to be masked. In an earlier paper6 dielectric data extending over a wide and continuous range of frequency and temperature were presented for samples of the 141 (basic) and 143 (W-resistant) grades of LEXAN polycarbonate which had been equilibrated in dry nitrogen at 30-35°C for at least 24h prior to investigation. We now report a complementary dielectric investigation of unequilibrated LEXAN 141. Attention is directed towards the multiple secondary relaxations occurring in the glassy state, with particular emphasis on relaxations observed at intermediate temperatures. Although relaxations in the latter temperature region have been reported previously by Krum & MUllery7 Watts & Perry' and othersY9-' considerable disagreement persists regarding the nature, location and origin of these intermediate processes. One specific area of contention is the suggestion of Muller and other^^-'*'^ that the 'intermediate process' in polycarbonate is a nonequilibrium effect associated with residual stresses. 2 EXPERIMENTAL G. J . Pratt, M . J . A. Smith ginary permittivity caused by the quasi-DC conductivity was subtracted from the measured total permittivity. 3 RESULTS AND DISCUSSION 3.1 Presentation of experimental results The response of unequilibrated LEXAN 141 to an applied alternating field, less the small quasi-DC conductivity, is shown in Figs 1-3. Figure 1 is a contour map of loss tangent (tan 6)as a function of temperature and frequency. Figure 2 is the corresponding map for the real relative permittivity E'. The 3-D representation of the imaginary relative permittivity E" shown in Fig. 3 has axis scales and an aspect chosen to assist the visualisation of absorption regions. Nine distinct regions, identified as a peak or shoulder of significant extent, may be distinguished in this representation. The possible existence of a tenth relaxation process is suggested by the observation that one of the absorption regions shows a change of slope in its Arrhenius plot, implying that two processes may be operating over different temperature regimes. Figure 4 is an Arrhenius plot of the logarithm of the frequency ( v 3 of maximum loss against reciprocal temperature for the dielectric a-process. Corresponding plots for the 'intermediate' and b-processes are shown in Figs 5 and 6 respectively. Figure 5 also includes comparative data for intermediate-temperature absorptions which have been reported in the literature. The numerical values in Fig. 6 are derived activation energies for each linear section of the plot. Three distinct dispersion regions are evident in the contour map of E' (Fig. 2); a low-temperature region where the value of E' increases gradually with an increase in temperature or a decrease in frequency, an extensive plateau region at intermediate temperatures throughout which E' is essentially constant (3% total change), and a high-temperature region where e' increases rapidly with increasing temperature. The plateau region extends from about - 119°C to + 164°C at 0.1 Hz and from about -9°C to + 172°C at 3 MHz. More detail is apparent in the 3-D representation of E" (Fig. 3). The individual features are discussed separately in succeeding sections. Samples of LEXAN 141 clear bisphenol-A poly3 2 High -temperature processes carbonate manufactured by General Electric Plastics were prepared and tested according to the general experimental procedures described e l s e ~ h e r e . ~ * ' ~3.2.1 ~ ~ ~A Maxwell- Wagner-Sillars process. The very large losses reported in equilibrated LEXAN 141 and They were exposed to the laboratory atmosphere for at 143 at low frequencies and the highest temperatures least 1 month before investigation. Immediately prior to also occur in the unequilibrated material.6 The rise is a dielectric measurement, the quasi-DC conductivity particularly apparent in the 3-D representation of E" (v, was determined after allowing time, typically up to 1h, 7') shown in Fig. 3. for the decay of current transients. The equivalent imaPOLYMER INTERNATIONAL VOL. 40, NO. 4, 1996 24 1 Multiple dielectric relaxation processes Hr 101 10' 10 @.I -Z@@ -I@@ m 1@@ rro'e Fig. 1. Contour map showing the temperaturefrequency variation of tan 6 for unequilibrated LEXAN 141. The activation energy associated with the slope of the high tan S contours of Fig. 1 is the same as that of the quasi-DC conductivity (about 220 kJ mol- I), and the contour separation at a given temperature, where it can be discerned, is consistent with an approximate inverse frequency dependence of the loss. Moreover, the highest-temperature real-permittivity contour (E' = 3.8) in Fig. 2 also has a slope indicating a causal link with -200 0 200 OC Fig. 2. Contour map showing the temperaturefrequency variation of real relative permittivity E' for unequilibrated LEXAN 141. POLYMER INTERNATIONAL VOL. 40. NO. 4, 1996 242 G . J . Pratt, M . J . A. Smith 1 1 1 0 -100 I I 100 I I 200Oc Fig. 3. 3-D model representation of the temperature-frequency variation of imaginary relative permittivity ~''(v, 7') for a sample of unequilibrated LEXAN 141 polycarbonate. The annotations DC/MWS, a, i,, i, , pi, #I2,J,, LT and H,O refer to the individual dispersion regions discussed in the text. the DC conductivity. This contrasts with the E' contours between 3-1 and 3-6 at the higher frequencies which are associated with the more highly activated primary arelaxation. These observations suggest that, in addition 200 175 150 I I I 6 4 t E - 3. to a direct contribution from the DC conductivity, a polarisation process involving the DC conductivity indirectly is providing a significant contribution to the high-temperature permittivity at the lower frequencies. The extended form of the equations of Sillars given by van Beek5 for the case of a continuous phase containing a random suspension of a spheroidal second phase of lower conductivity and similar real permittivity can provide an explanation for the present observations. The continuous phase (matrix) is identified with amorphous polycarbonate and the second phase with a crystalline or a densified, variant. With the further assumption of a second phase in the shape of oblate spheroids (lamellae), the equations can be used to predict a high-frequency tail of an MWS polarisation of appropriate magnitude extending into the present lowfrequency region, even for an anticipated second-phase concentration of a few percent. 0) 0 - 2 0 1 1 21 22 10% - 1 23 24 K . ' Fig. 4. Arrhenius plot (logarithm of frequency v, of maximum loss plotted against reciprocal temperature) for the a-process in unequilibrated LEXAN 141. 3.2.2 Alpha process. The primary (a-) relaxation associated with the glass transition appears as an extremely narrow (about 25°C wide) and very prominent region of absorption which merges into the combined DC/MWS absorptions below a few hertz. The a-peak extends from 160°C at 3 Hz to 200°C at 1 MHz. The average activation energy derived from the Arrhenius plot for the aprocess in unequilibrated LEXAN 141 (Fig. 4) is 530 kJ mol-l over the reciprocal temperature interval (22-24) x 10-4K-'. This value compares with 490kJmol-I for equilibrated LEXAN 141 over the same interval given in the earlier paper6 where it was noted that values from 480 to 835 kJ mol-I have been reported in the literature. The curvature of the Arrhenius plot of Fig. 4 demonstrates -an inadequacy in the POLYMER INTERNATIONAL VOL. 40, NO. 4, 1996 243 Multiple dielectric relaxation processes 75 25*( I -zE 3 0 -0 -1 1 24 1 1 28 I 32 I(-’ 1 4T + Fig. 5. Arrhenius plot for the intermediate-temperature processes in bisphenol-A polycarbonate. The symbols 0 and refer to equilibrated and unequilibrated LEXAN 141 in the present work. The other symbols refer to previous work reported in the literature; the associated numbers correspond with those stated in the list of references. - UNEQUILIBRATED 5 - E z -3 P - 0 d c 1- - ; *3 Fig. 6. Arrhenius plot for the /Iand - related processes in equilibrated and unequilibrated LEXAN 141. POLYMER INTERNATIONAL VOL. 40. NO. 4, 1996 244 G . J . Pratt, M . J . A. Smith single activation energy, or single relaxation time, representation of the a-process and suggests a range of constraints and environments which affect the large scale motion of individual molecules and segments in the solid polymer. A functional decrease in activation energy with increasing temperature is a well-known phenomenon for polymeric a-relaxations and is often described by the semi-empirical Williams-Landel-Ferry (WLF) equation" or its generalised version." Characterisation by an average activation energy rather than by WLF or Rusch parameters is used here to facilitate comparison with other studies. 3.3 Lo w-temperatureprocesses 3.3.1 Beta and related processes. Differences in composition, thermal history and pre-treatment, and limitations imposed by the experimental method, have led to apparent conflict in the results of earlier investigations on polycarbonate. Many diele~tric'~~'~~'~~~~-~~ and mechanical' 2 ~ 1 4 * 2 8 ~ 2 9investigations report a single, broad and unresolved #?-peak. LeGrand & Erhardt" suggest that they were unable to resolve the 8-peak into its expected three constituent loss curves because similar activation energies caused identical temperature-frequency shifts of the individual curves. In the present work, as noted below, activation energies for the three processes are indeed essentially the same. Illers & B r e ~ e r , ~Nielsen3' " and Watts & Perry' report an asymmetrical 8-peak for polycarbonate, whilst others 1-33 three34.35 succeed in resolving the peak into or even four3" components. The applied inter-electrode field in a dielectric measurement induces, and responds to, motion in the chain segments containing electric dipoles, such as the carbony1 or carbonate groups in polycarbonate. However, local motions of non-polar segments may be detected indirectly provided they are coupled to the dipolar segments of the polymer chain. For instance, motions of methyl and phenylene groups in polycarbonate require cooperative carbonate motion^'^*^'^^^ and are, therefore, observable by dielectric methods. Sacher regards the structural changes induced in polycarbonate by annealing at temperatures well below the glass temperature as evidence that a high degree of segmental chain motion must occur in the glassy state." Open packing of the polymer chains provides considerable freedom of movement of the carbonate thus facilitating the molecular and segmental motions thought to be associated with the 8-process. Present measurements indicate a clear asymmetry and structure in the extremely broad 8-absorption which occupies about two-thirds of the total temperature-frequency regime for which permittivity measurements are reported. At each frequency the width of the absorption is at least 250°C. Its peak 'J extends from about - 113°C at 3 Hz to about - 15°C at 3 MHz in unequilibrated LEXAN 141 (Fig. l), and corresponds approximately with the low-temperature boundary of the plateau in E' (Fig. 2). Closer investigation indicates that at least five individual relaxation processes contribute to the breadth and shape of this broad /?-peak3' and that three of these components persist throughout most of the absorption region. Activation energies derived from Arrhenius plots are 21 f 1kJmol-I for each component. The three peaks may, therefore, be expected to experience identical shifts with change of temperature and/or frequency, as suggested by LeGrand & Erhardt." By comparing the dielectric behaviour of bisphenol-A polycarbonate and the related ester polyethylene terephthalate, Krum & Muller7 established the role of the carbonyl/carbonate group in the 8-absorption of polycarbonate. Subsequent debate has been concerned with the details of the molecular motions. Specific mechanisms considered include hindered local mode relaxation of carbonate groups restricted by adjacent phenyl g r o ~ p s ~ and ~ combined ~ ~ ~ motions ~ ~ ~ of carbonate groups and phenyl rings hindered by the glassy matrix.1.15.1 7.24.25,27-29.31-34.38 MGller & Huff7 and others8.9+15,17,23-26,32-34,36,39 have shown that the mechanical and thermal history of polycarbonate can affect the magnitude of the /?-peak. McCrum et a1.18 argue that the manner in which the 8-relaxation in polycarbonate is affected by crystallinity and other factors indicates that the 8-process does not arise from independent motion of carbonate groups, but that interactions with other groups must be involved. Overall, it would appear that these interactions lead to forms of restricted local motion on the low-temperature side of the peak whilst, at higher temperatures, combined motions occur which involve tight coupling of functional groups such as carbonyl and phenyl. Considering the 8-absorption as a whole, the E" isotherm response is considerably broader than is expected on a Debye representation. A closer first approximation might be to regard E" as independent of frequency within the absorption, thereby implying constant energy dissipation per cycle, in sharp contrast to the a-peak. The broad-banded nature of the 8-absorption almost certainly arises from a relatively strong interaction of the driven carbonyl dipole with the surrounding polymer matrix, and with adjacent phenyl groups in particular. On the lower-temperature side of the &peak the phenyl groups are immobile3' and segmental motion is hindered by adjacent phenyl rings. At higher temperatures the mobility of the phenyl groups facilitates combined motion of carbonate and phenyl groups. At moderate and low temperatures the loss component E" falls towards zero at low frequencies whereas the real component E' increases marginally, indicating that the dipoles readily follow the applied field at such frequencies. POLYMER INTERNATIONAL VOL. 40. NO. 4, 1996 ~ ~ , 245 Multiple dielectric relaxation processes 3.3.2 A water-related process. Above 30 kHz the path of the loss peak (Fig. 1) is affected by the admixture of an additional process found only in the unequilibrated material. The admixture appears in the 3-D representation (Fig. 3) as a subsidiary peak superimposed on the main /?-absorption and extending from about - 38°C at 100kHz to - 10°C at 3 MHz. The presence of the additional process is even more clearly indicated by the change in slope of the Arrhenius plot above 30kHz (Fig. 6). There is no corresponding change in slope for the equilibrated material. The activation energy derived from the Arrhenius plot is 62 kJmol-’ for the combined processes in unequilibrated LEXAN 141. All samples are cut from the same polycarbonate sheet after it has received prolonged exposure to ambient humidity. The unequilibrated sample is tested without any pre-conditioning whereas the equilibrated sample is placed in warm, dry nitrogen gas for at least 24 h before commencement of the dielectric measurement. It seems likely, therefore, that the additional absorption in unequilibrated LEXAN 141 involves a water-related relaxation. The influence of water absorption on the /?-peak in polycarbonate has been noted previously by Chung & SauerZ8and by Allen et al.” Both groups observed an increase in the magnitude of the /?-absorption and a shift in the peak to higher temperature on the addition of small amounts of water. The extent of the shift was independent of water content over the concentration range investigated. Allen et al.” report further that the activation energy increases from 36 kJ molto 52 kJmol-’. They conclude that the water molecules hydrogenbond to the polar groups in the polymer chain so as to participate in the motion causing the relaxation. Allen et al.” observed similar effects in other polar polymers containing para-phenylene linkages. Vanderschueren et al.34breport that TSD results in this temperature range are very sensitive to the presence of small amounts of water. They found a water-dependent peak on the higher temperature side of the /?-peak and attributed this feature to relaxation of the hydrogen-bonded water itself. In a detailed study of the influence of water absorption on the behaviour of injection-moulded LEXAN 141 polycarbonate, Bair et aLZ7 also report that the temperature and magnitude of the /?-peak are both increased by the addition of water. They suggest that the increase is due to correlated motion of water dipoles with the polymer molecules, and note further that the formation of water clusters in hydrolyticallydegraded polycarbonate causes additional dielectric ’ adjacent carbonyl and phenyl groups.” As the additional relaxation does not persist below about - 38”C, the interaction is deemed to be initiated at this temperature. Alternative models might be based on conformational changes around -38°C which lead to water structures of limited dimension and moderate conductivity in the persistence region. 3.3.3 A lower-temperature process. A distinctive dielectric absorption is apparent in the vicinity of - 120°C to -150°C below 3Hz (see, for example, Fig. 1). Aoki & B~ittain,~’Sikka” and other^'^,'^ suggest that local motions of methyl groups are the cause of loss peaks observed variously between - 140°C and -200°C. on the other hand, Linkens & Vander~chueren,~~ units are responsible suggest that motions of -COOfor the TSD peak they observed at - 140°C. For our particular absorption, the variation of 8’‘ with frequency appears narrow, possibly Debye-like, suggesting that highly-coupled motion is not involved. The low-frequency nature of the absorption indicates a mobile segment which is either large or highly constrained, or a form of interfacial effect. It is unlikely that motion of methyls is responsible, since any process involving so small a group would be expected to be seen as a high-frequency absorption, such as is reported by TOrma1a4’ at 40MHz. T0rmala40 attributes his (6) relaxation to methyl group orientation and points out that the effective activation energy (lOkJmol-l) is in agreement with that estimated for methyl group rotation in polycarbonate. In an earlier paper6 this low-frequency lowertemperature absorption was attributed tentatively to the presence of a processing stabiliser. More recently, permittivity measurements have been repeated for an equivalent additive-free grade of bisphenol-A polycarbonate (LEXAN 145) which contains no phosphite processing ~ t a b i l i s e r .In ~ ~ this material no separate absorption is observed in the vicinity of -120°C to - 150”C, confirming the hypothesis that the lowertemperature low-frequency absorption is associated directly with the processing stabiliser. At these low temperatures the stabiliser has condensed to form an additional phase and, hence, an interfacial polarisation with a peak position determined primarily by the conductivity of the phosphite condensate.5 The observed fall in the peak height of the absorption with increasing temperature is consistent with dispersal of the condensed phase as the temperature is raised. relaxation^.'^ In the present work, the higher activation energy of the water-related relaxation relative to the fundamental /?-process is consistent with the existence of hydrogenbonding or some other form of intermolecular interaction which hinders the cooperative movements of POLYMER INTERNATIONAL VOL. 40. NO. 4, 1996 3.4 Intermediate -temperature processes A number of workers7-17 have reported absorption regions at temperatures intermediate between those of the a- and /?-relaxations, as summarised in Fig. 5. 246 G.J. Pratt, M . J. A. Smith or tWo7 frequencies only, Observation has been by d i e l e c t r i ~ ~ - and ' ~ ~ ~ ~one8-14.16,17,24,25,32,35.39 ~~~~~ dynamic m e ~ h a n i c a l ' - ' ~ * ~ ~techniques, ~ ' ~ ~ ' ~ although and fail to distinguish two separate intermediateNeki & Geili4 report the absorption only for mechanitemperature relaxations in bisphenol-A polycarbonate. cal tests. The polycarbonate materials are obtained The earlier paper6 failed to recognise fully the range from a variety of sources and also differ in their preof the i,-absorption. Its extent is, however, readily treatment and thermal history. There are considerable apparent in the constant-frequency E"( T) plots and the variations in detail between data from the different 3-D model representation of ~ " ( v ,T) (Fig. 3). The i,sources. absorption appears as a separate loss peak at low and Dynamic mechanical measurements at approximately intermediate frequencies which gradually merges with 1Hz on Bayer polycarbonate show an absorption at the primary absorption region until, at the highest freintermediate temperatures near to 70°C for untreated' quencies, it becomes a faint shoulder on the lowor compression-moulded samples" and for injectiontemperature side of the a-peak. The i,-absorption most moulded film." At higher frequencies the absorption probably corresponds with that reported by Miiller & occurs in the vicinity of 100-llO"C, although Neki & c o - ~ o r k e r sSacherI5 ,~ and other^.^^'^^'^ GeilI4 report a somewhat lower temperature. Neki & At kHz frequencies around 100°C Krum & Muller7b Geil14 tested Mobay MERLON 60 polycarbonate observed an absorption of moderate intensity only in whereas LeGrand & Erhardt'' and Litt & Torp13 both polycarbonate samples which had been injectionused General Electric LEXAN materials in the respecmoulded or cold-drawn. Annealing diminished the magtive forms of compression-moulded and as-received nitude of the absorption, and they were unable to detect sheet. The temperature difference at the higher frethe absorption in fully-annealed samples. Watts & quencies might arise from the difference in materials. Perry,' Muller & co-workers7 and others'," conclude Alternatively, two different intermediate-temperature that the 'intermediate process' in polycarbonate is a processes may be involved, as is indicated by the evinon-equilibrium effect associated with stresses induced dence below. by the moulding process or subsequent mechanical treatment. The dielectric measurements reported in the literature are concentrated primarily in the vicinity of 1-3 kHz. At An interfacial origin for the i,-absorption is discounted at this time because a dispersed phase of suffisuch frequencies the quoted temperature for the 'intermediate' process(es) ranges from approximately 55°C to cient conductivity cannot be identified. Watts & Perry' 120°C. The lowest values are those of Muller & Huff7" and ~ t h e r ~ ~have ~ suggested ~ ' ~ *that ~ their ~ ~ interme* ~ ~ for quenched Bayer MAKROLON polycarbonate. The diate process is associated with a pre-T, phenomenon. At temperatures where there is insufficient thermal highest reported values are for LEXAN Variation in the source or manufacture of the polyenergy to excite micro-Brownian motion in complete the available energy polymer chains, that is below carbonate appears to produce greater differences in the temperature location of the absorption than are caused may be shared between restricted sequences of repeat units which could then undergo some form of motion by variation in thermal or mechanical pre-treatment. relative to each other. Such behaviour implies that the In the present work two distinct absorptions of lesser spatial variation of free volume is such that segmental magnitude than the high-temperature processes are motion is possible in some regions and not in others. As reported in the glassy state at temperatures above ambient. The first, referred to as the first intermediate the temperature is raised the energy and free volume fraction continue to increase until, at T,, continuous i,, is a region of moderate loss which appears in sequences of repeat units are excited into motion. Kilian constant-frequency plots of E"( T) (not presented here) as a shoulder from about 105°C at 0.1Hz through to & B ~ u e k ehave ~ ~ presented evidence from X-ray diffraction measurements for the existence in polystyrene about 140°C at 300Hz, but which cannot be discerned of organised coils which contain approximately 12 main above 1kHz. The loss corresponds closely with the 'intermediate' absorption reported previously by Watts chain carbon atoms and 50 atoms in total and which & Perry' for lower average molecular weight LEXAN move relative to each other at temperatures approach101. ing T,,where all traces of such organisation are lost.44 The second intermediate absorption region (i,) perThe i, relaxation process is now regarded as involving partially-correlated motion of a moderate number sists throughout the frequency range covered by the of molecular segments or repeat units. This process present work, and may be followed from about 60°C at could be viewed as a precursor to the cooperative but 0-1Hz to approximately 160°C at 3 MHz. Such a wide uncorrelated micro-Brownian motion which occurs frequency range for an intermediate-temperature above q . The primary a-relaxation itself is associated process has not been reported previously. Indeed, with with micro-Brownian motion of complete polymer the exception of the data of SacheP which covers chains which is sufficiently extensive that correlation three frequency decades (but which shows a somebetween the motions of the chain repeat units is no what unusual frequency-temperature dependence), longer discernible. most earlier studies report measurements at s, POLYMER INTERNATIONAL VOL. 40. NO. 4, 1996 247 Multiple dielectric relaxation processes The location of, and activation energy for, the i,peak/shoulder in the tan 6 contour map (Fig. 1) are intermediate between those of the ct- and /?-peaks. It has been noted above that the /?-process is a consequence of a range of restricted local motions involving coupled motions of carbonyl and phenyl groups in the repeat unit of bisphenol-A polycarbonate. The range of motions produces an exceedingly broad /?-absorption in polycarbonate. In contrast, the i,-absorption, as well as being of substantially lesser magnitude, is not as diffuse as the /?absorption and is shifted to higher temperature. These characteristics of the i,-peak are consistent with an extended local motion which resembles the restricted local motions associated with the /?-relaxation process but which involves the cooperative motion of a fairly specific sub-unit of the polycarbonate molecule consisting of a fixed number of repeat units or an almost constant number of such units. Sacher' attributes his intermediate (7) absorption in polycarbonate to a cooperative movement of several repeat units which resembles wagging and rocking vibrations of carbonate, methyl and phenylene groups in the polymer chain. On the other hand, on the basis of NMR and carbon- 13 chemical shift anisotropy, Jones & co-workers3' propose that phenyl groups undergo 180" flips and restricted oscillation. Although this relaxation process has usually been assumed to occur in the upper temperature region of the /?-absorption, the same temperature region also encompasses the i,-absorption of the present work. and other^^^-^' have also reported Schaefer et the occurrence of 180" ring flips of phenyl groups in bisphenol-A polycarbonate. The defect-diffusion ringflip mechanism proposed by Jones3' involves the interchange of cis-trans and trans-trans transformations (or equivalent transitions) between adjacent carbonate groups in the polycarbonate chain. These particular conformational transitions alter the relative positions of adjacent carbonate groups in a manner which can activate 180" ring flips of an intervening phenyl group. This motion produces a volume fluctuation which can diffuse along the chain as the cis-trans conformation diffuses along polycarbonate chains composed largely of transtrans units. In addition to such intrachain influences, the rotational freedom of the phenyl rings is limited by interchain steric interactions due to the proximity of rings in adjacent chains. Indeed, Schaefer, Perchak and other^^^,^^ suggest that it is the volume fluctuations associated with the dilation of the surrounding polycarbonate 'lattice' which relax interchain constraints and allow the rings to flip. Regardless of whether volume fluctuations are seen as the controlling or because constraint relaxation is a consequence of conformational e ~ c h a n g e , ~there is general agreement that such fluctuations must necessarily create ' POLYMER INTERNATIONAL VOL. 40. NO. 4, 1996 sufficient free space to allow a ring to flip. The magnitude of the i,-relaxation is substantially less than that of the /?-relaxation, being consistent with the statistical probability of occurrence and transformation of the necessary conformations, namely cis-trans trans-trans or equivalent, to facilitate 180" ring flips of intervening phenyl groups. Krum & Muller,7 Watts & Perry' and other^^^.^^,^^^*^^.^^ report that the magnitude of the intermediate relaxation in polycarbonate is diminished by annealing at temperatures below T,, and suggest that their intermediate relaxation arises from a nonequilibrium process associated with residual frozen-in stresses which are removed by annealing. The compression-moulded samples used in the present work would not be expected to retain frozen-in stresses from the moulding process, and are maintained at elevated temperature for a considerable period of time during measurements. No diminution of the intermediatetemperature absorptions was observed. The two separate intermediate relaxations which are observed and defined occur in all LEXAN grades which have been investigated, whether equilibrated or unequilibrated, UV-irradiated or unirradiated. It appears unlikely, therefore, that either of the intermediatetemperature absorptions observed in the present work is due to 'moulded-in' stresses. ~ ' ~ Computer simulations by Perchak et ~ 1 . suggest that phenyl ring motions are greatly affected by the surrounding environment. One consequence of the annealing process in polymers is a reduction in free-volume fra~tion.'l-~ Annealing and physical ageing lead to more efficient packing of chains with consequential inhibition of ring flip and phenyl rocking motions.47 Rather than the relief of frozen-in stresses, this reduction in free volume is considered to be the underlying cause of the diminution by annealing of the intermediate relaxa t i o n ~ . ~ - ~A *ring-flip ~ ~ * process ~ ~ ~ *such ~ ~as~ that ~ ~ described by Jones, Schaefer and others38,45-48would necessitate the existence of substantial packets of free volume each greater than a critical extent uCRIT to permit 180" rotation of a phenyl ring whether by conformational transformations or other process. Bisphenol-A polycarbonate normally contains free volume suflicient to permit substantial segmental motion even in the glassy ~ t a t e . ~ ~However, ~ . ~ ~ , ~ ~ - ~ annealing would reduce the overall free-volume fraction and, in particular, would reduce the proportion of free volume in packets of critical extent, thereby inhibiting the 180"ring-flip process. Furthermore, the &peak occurs at significantly higher temperatures (Fig. 5 ) in unequilibrated LEXAN 141 than in the equilibrated material. The &-relaxation is likely, therefore, to involve a process which is affected by the presence of absorbed moisture. The higher temperature in the unequilibrated material is consistent with the occurrence of hydrogen-bonding or similar 248 influences which hinder motion. Average activation energies are 172 kJmol-1 and 207 kJmol-1 for equilibrated and unequilibrated LEXAN 141 respectively. This difference in activation energies is similar to that associated with the water-related absorption in the /Iregion of unequilibrated LEXAN 141. Intermolecular and intramolecular hydrogen-bonding would interfere with the movement of the large sub-unit associated with the 180" phenyl ring flips which, on balance, are considered to be the origin of the dielectric i,-absorption. 4 CONCLUSIONS MWS interfacial polarisation arising from the conductivity mismatch between different phases or domains most probably contributes to the losses at high temperatures and low frequencies in polycarbonate. Acceptable agreement is obtained between experimental data and a model based on a two-phase morphology containing a small fraction of relatively non-conducting densified or crystalline lamellar domains in a predominantly amorphous matrix. Despite considerable conflict in detail between the various published studies, it is clear that secondary relaxation processes in bisphenol-A polycarbonate are affected by thermal and mechanical history. Evidence has been presented for at least nine dielectrically-active relaxations in unequilibrated LEXAN 141 polycarbonate. In many applications polycarbonate is chosen principally for its superior impact strength over an extended temperature range. The very broad low-temperature absorption arises from overlapping multiple loss processes involving a range of molecular and segmental motions and is well illustrated in the 3-D representation of ~ " ( v ,T) (Fig. 3). The one relaxation which is absent in the corresponding equilibrated material is thought to be water-related. Of the relaxations which occur at low temperatures, those associated with the B-processes provide significant energy-dissipating mechanisms, particularly at the higher frequencies. The semi-continuum of loss provided by the multiplicity of overlapping absorptions in polycarbonate at the higher frequencies offers a reason for the good impact behaviour at low temperature. The combined evidence from a range of experimental techniques indicates that suflicient free volume exists in the glassy state of unannealed bisphenol-A polycarbonate to permit a range of restricted local-mode motions activated by the strongly polar carbonyl groups. In the lower temperature range of the Babsorption the immobility of the phenyl groups severely restricts segmental motion. As temperature is increased mobility of the phenyl groups permits a range of carbonyl-driven cooperative localised motions of the polycarbonate chain backbone. G. J . Pratt, M . J . A. Smith In the earlier work6 it was tentatively proposed that the lower-temperature low-frequency process observed in commercial polycarbonate materials was due to the presence of a processing stabiliser. This hypothesis has been confirmed by measurements for the equivalent additive-free material.41 Disagreement has persisted regarding details of the intermediate-temperature relaxations in bisphenol-A polycarbonate. With the possible exception of the data of Sacher," previous conflicting reports in the literature failed to define adequately in the temperature-frequency regime the extent of either absorption, or even to recognise that the isolated data points arise from separate dielectric relaxations in bisphenol-A polycarbonate. In the present work two separate absorptions intermediate between the a- and /I-absorptions are positively identified and clearly delineated for the first time. The i,-absorption corresponds with that reported by Watts 8z Perry8 and others1'.17 below 1 kHz, and most probably involves a partially-correlated micro-Brownian motion. The i,-absorption corresponds with that reported by Muller and CO-workers,'S a c h e P and ~ t h e r ~ , ~ ~ . ~ ~ and persists throughout the extended frequency range of the measurements presented here. It is postulated that the origin of the i,-absorption is a relaxation process in which the phenyl groups undergo 180" flips as might arise when adjacent carbonyl groups interchange between cis-trans and trans-trans conformation^.^^ No evidence has been found to support the h y p o t h e ~ i s ~that . ~ the process underlying i, or i, involves frozen-in stresses of thermal or mechanical origin. The reduction in free volume by annealing is proposed as the underlying cause for the reported diminution by annealing of the intermediate absorption@)in bisphenol-A polycarbonate rather than the relief of frozen-in residual stresses as hitherto p r ~ p o s e d . ~ - ~ REFERENCES 1 Pratt, G. J. & Smith, M. J. A., Polym. Degr. Stub., 25 (1989) 267. 2 Heijboer, J., Kolloid-Z., 148 (1956) 36. 3 Heijboer, J., in Physics ofNon-Crystalline Solids. North Holland, Amsterdam, 1965, pp. 231-54. 4 Boyer, R. F., Rubber Chem. Tech., 36 (1963) 1303. 5 van Beek, L. K. H., Prog. Dielectr., 7 (1967) 69. 6 Pratt, G. J. & Smith, M. J. A., Brit. Polym. J., 18 (1986) 105. 7 (a) Miiller, F. H. & Huff, K., KoZZoid Z., 144 (1959) 34; (b) Krum, F. & Miiller, F. H., KoZZoid Z., 164 (1959) 81. 8 Watts, D. C. & Perry, E. P., Polymer, 19 (1978) 248. 9 (a) IUers, K. H. & Breuer, H., Kolloid Z., 176 (1961) 11; (b) Illers K. H. & Breuer H., J . Colloid. Sci., 18 (1963) 1. 10 (a) Heijboer, J., J . Polym. Sci., C, 16 (1968) 3755; (b) Bussink, J. & Heijboer, J., Proc. Int. Con$ IUPAC, Delft, 1964. 11 LeGrand, D. G. & Erhardt, P. E., J . Appl. Polym. Sci., 13 (1969) 1707. 12 (a) Allen, G., McAinsh, J. & Jeffs, G. M., Polymer, 12 (1971) 85; (b) Allen, G., Morley, D. C. W. & Williams, T., J. Muter. Sci., 8 (1973) 1449. 13 Litt, M. H. & Torp, S., J . Appl. Phys., 44 (1973) 4282. 14 Neb, F. & Geil, P.H., J . Mucromol. Sci.-Phys., B8 (1973) 295. POLYMER INTERNATIONAL VOL. 40, NO. 4, 1996 Multiple dielectric relaxation processes 15 (a) Sacher, E., J. Macromol. Sci.-Phys., B9 (1974) 163; (b) Sacher, E., J. Macromol. Sci.-Phys., B10 (1974) 319; (c) Sacher, E., J. Macromol. Sci.-Phys., Bll(l975) 403. 16 Kochi, M., Sasaki,T. & Kambe, H., Polym. J., 10 (1978) 169. 17 Sikka, S., Polymer Bull., 3 (1980) 61. 18 McCrum, N. G., Read, B. E. & Williams, G., Anelastic and Dielectric Effects in Polymeric Solids. Wiley, New York, 1967, pp. 52037. 19 Pratt, G. J. & Smith, M. J. A., J. Phys. E: Sci. Instrum., 15 (1982) 927. 20 Pratt, G. J. & Smith, M. J. A., Electrical, Optical and Acoustic Properties of Polymers, The Plastics & Rubber Institute, London, 1988, pp. 10/1-10. 21 Williams, M. L., Landel, R. F. & Ferry, J. D., J. Am. Chem. SOC., 77 (1955) 3701. 22 Rusch, K. C., J. Macromol. Sci.-Phys., B2 (1968) 179. 23 Mikhailov, G. P. & Eidelnant, M. P., Vysokomol. Soedin., 2 (1960) 287,295. 24 (a) Ishida, Y. & Matsuoka, S., Polym. Preprints, 6 (1965) 795; (b) Matsuoka, S. & Ishida, Y., J. Polym. Sci., C, 14 (1966) 247. 25 (a) Ito, E. & Hatakeyama, T., J. Polym. Sci.; Polym. Phys. Ed., 12 (1974) 1477; (b) Ito, E. & Hatakeyama, T., J. Polym. Sci.; Polym. Phys. Ed., 13 (1975) 2313. 26 Venkataswamy, K., Ard, K. & Beatty, C. L., Polym. Eng. Sci., 22 (1982) 955. 27 Bair, H. E., Johnson, G. E. & Memweather, R., J. Appl. Phys., 49 (1978)4976. 28 Chung, C. I. & Sauer, J. A., J. Polym. Sci., A-2,9 (1971) 1097. 29 Reding, F. P., Faucher, J. A. & Whitman, R. D., J. Polym. Sci., 54 (1961) S56. 30 Nielsen, L. E., Mechanical Properties of Polymers. Reinhold, New York, 1962. 31 Locati, G. & Tobolsky, A. V., Adv. Mol. Relax. Proc., 1 (1970) 375. 32 (a) Aoki, Y. & Brittain, J. O., J. Polym. Sci.; Polym Phys. Ed., 14 (1976) 1297; (b) Aoki, Y. & Brittain, J. O., J. Appl. Polym. Sci., 20 (1976) 2879; (c) Aoki, Y. & Brittain, J. O., J. Appl. Polym. Sci., 21 (1977) 199; (d) Hong, J. & Brittain, J. O., J. Appl. Polym Sci., 26 (1981) 2459. 33 (a) Pochan, J. M., Hinman, D. F. & Turner, S . R., J. Appl. Phys., 47 (1976) 4245; (b) Pochan, J. M., Gibson, H. W., Froix, M. E. & Hinman, D. F., Macromolecules, 11 (1978) 165; (c) Pochan, J. M., Gibson, H. W. & Pochan, D. L. F., Macromolecules, 15 (1982) 1368. 34 (a) Linkens, A. & Vanderschueren, J., Polym Letters, 15 (1977) 41; (b) Vanderschueren, J., Linkens, A., Haas, B. & Dellicour, J., J. Macromol. Sci.-Phys., B15 (1978) 449. 35 (a) Jain, K., Agarwal, J. P. & Mehendru, P. C., I1 Nuova Cimento, 55B (1980) 123; (b) Mehendru, P. C., Jain, K. & Agarwal, J. P., J. Phys. D : Appl. Phys., 13 (1980) 1497. 36 Garfield, L. J., J. Polym. Sci., C, 30 (1970) 551. POLYMER INTERNATIONAL VOL. 40. NO. 4, 1996 249 37 Pratt, G. J. & Smith, M. J. A., Polymers at Low Temperatures (Proceedings).PRI, London, 1987, pp. 9/1-10. 38 (a) Jones, A. A., Macromolecules, 18 (1985) 902; (b) O'Gara, J. F., Jones, A. A., Hung, C.-C. & Inglefield, P. T., Macromolecules, 18 (1985) 1117; (c) C O M O ~ ~ J.YJ.,, Inglefield, P. T. & Jones, A. A., J. Chem. Phys., 86 (1987) 6602. 39 Guerdon, L. & Marchal, E., Polymer, 22 (1981) 1199. 40 Tomalii, P., J. Macromol. Sci.-Reu. Macromol. Chem., C17 (1979) 297. 41 Pratt, G. J. & Smith, M. J. A., Plastics, Rubbers and Composites Processing and Applications, 16 (1991) 67. 42 Varadarajan, K. & Boyer, R. F., J. Polym. Sci.; Polym. Phys. Ed., 20 (1982) 141. 43 Kilian, H. G. & Boueke, K., J. Polym. Sci., 58 (1962) 311. 44 Brydson, J. A., in Polymer Science: A Materials Science Handbook, ed. A. D. Jenkins. North Holland, Amsterdam, 1972, Ch. 3. 45 (a) Schaefer, J., Stejskal, E. O., Perchak, D., Skolnick, J. & Yaris, R., Macromolecules, 18 (1985) 368; (b) Perchak, D., Skolnick, J. & Yaris, R., Macromolecules, 20 (1987) 121; (c) Poliks, M. D., Gullion, T. & Schaefer, J., Macromolecules, 23 (1990) 2678. 46 Wehrle, M., Hellmann, G. P. & Spiess, H. W., Colloid Polym. Sci., 265 (1987) 815. 47 Smith, P. B., Bubeck, R. A. & Bales, S. E., Macromolecules, 21 (1988) 2058. 48 Hutnik, M., Argon, A. S. & Suter, U. W., Polym. Preprints, 30 (1989) 36. 49 Havriliak, S. & Pogonowski, C. S., Macromolecules, 22 (1989) 2466. 50 Frank, W., Goddar, H. & Stuart, H. A., J. Polym. Sci., B ; Polymer Letters, 5 (1967) 711. 51 (a) W, A. J., Jones, P. L., Lind, J. H. & Pearsall, G. W., J. Polym. Sci. Pt A : Polym. Chem., 26 (1988) 1541; (b) Hill, A. J., Katz, I. M. & Jones, P. L., Polym. Eng. Sci., 30 (1990) 762; (c) Hill, A. J., Heater, K. J. & Agrawal, C. M., J. Polym. Sci. Pt B: Polym. Phys., 28 (1990) 387; (d) Hill, A. J. & Agrawal, C. M., J . Mater. Sci., 25 (1990) 5036. 52 (a) Smith, P. B., Bubeck, R. A. & Bales, S . E., Macromolecules, 21 (1988) 2058; (b) Bubeck, R. A., Yasar, H. Y. & Hammouda, B. H., Polymer Commun., 30 (1989) 25. 53 Haward, R. N., in Molecular Behaviour and the Deuelopment of Polymeric Materials, ed. A. Ledwith & A. M. North. Chapman & Hall, London, 1975, p. 441. 54 Golden, J. H., Hammant, B. L. & Hazell, E. A., J. Appl. Polym. Sci., 11 (1967) 1571. 55 Siegmann, A. & Geil, P. H., J. Macromol. Sci.-Phys., B4 (1970) 239,273. 56 LeGrand, D. G., J. Appl. Polym. Sci., 16 (1972) 1367. 57 (a) Yannas, I. V. & Lunn, A. C., Polymer Letters, 9 (1971) 611; (b) Lunn, A. C. & Yannas, I. V., J. Polym. Sci.: Polym. Phys. Ed., 10 (1972) 2189.