Soft Matter C Dynamic Article Links < Cite this: Soft Matter, 2012, 8, 5119 www.rsc.org/softmatter COMMUNICATION Tg depression and invariant segmental dynamics in polystyrene thin films† Virginie M. Boucher,ab Daniele Cangialosi,*b Huajie Yin,c Andreas Sch€onhals,c Angel Alegrıabd and Juan Colmeneroabd Received 23rd February 2012, Accepted 26th March 2012 DOI: 10.1039/c2sm25419k We investigate the segmental dynamics and glass transition temperature (Tg) of polystyrene (PS) thin films. The former is investigated by alternating current (AC) calorimetry and dielectric spectroscopy (BDS). The Tg, underlying the equilibrium to out-ofequilibrium crossover from the supercooled liquid to the glass, is obtained by differential scanning calorimetry (DSC) and capacitive dilatometry (CD). We show that the intrinsic molecular dynamics of PS are independent of the film thickness both for the freestanding and supported films, whereas Tg decreases with film thickness from several microns down to 15 nm. This result is found for complementary methods and in a simultaneous measurement in BDS and CD. This questions the widespread notion that segmental mobility and the equilibrium to out-of-equilibrium transition are, under any experimental conditions, fully interrelated. For thin films, it appears that the molecular mobility and Tg are affected differently by geometrical factors. If a viscous liquid cools down and crystallization can be avoided, a glass is formed. The temperature at which a dramatic increase of the mechanical modulus due to glass formation occurs can be taken as the signature of approaching the glass transition temperature (Tg). The associated dynamic phenomenon, the structural (a) relaxation, has been widely studied in bulk glass formers, including polymers. Nowadays a considerable knowledge has been achieved, though no generally accepted theory exists and the dynamics related to the glass transition remain a pending problem in soft matter research. The glass transition phenomenon in nanostructured systems, such as glass formers in nanopores,1 thin polymer films2 and nanocomposites,3 is a relatively new topic and object of intense scientific debate. The achievement of a consensus in the field may have considerable fundamental as well as technological impact. Concerning thin polymer films, much controversy exists due to apparently contrasting a Donostia International Physics Center, Paseo Manuel de Lardizabal 4, 20018 San Sebasti an, Spain b Centro de Fısica de Materiales (CSIC-UPV/EHU), Paseo Manuel de Lardizabal 5, 20018 San Sebasti an, Spain. E-mail: swxcacad@sw.ehu.es c BAM Federal Institute for Materials Research and Testing, Unter den Eichen 87, 12205 Berlin, Germany d Departamento de Fısica de Materiales, Universidad del Paıs Vasco (UPV/ EHU), Apartado 1072, 20080 San Sebasti an, Spain † Electronic supplementary information (ESI) available. See DOI: 10.1039/c2sm25419k This journal is ª The Royal Society of Chemistry 2012 results exhibiting both an increase, decrease or no change of Tg. Aspects such as the preparation conditions4 and the presence of a substrate,5 rather than clarifying the controversy, have generated further debate. Furthermore, on those numerous studies reporting Tg depression, the magnitude of such depression can significantly vary depending on the applied cooling rate6,7 and on whether the measurement is performed on heating or cooling. An issue that has been often neglected concerns the kind of technique employed to estimate Tg. Some works report deviating results, employing different techniques to the same sample.8–12 In particular, for methods—such as differential scanning calorimetry (DSC),13,14 ellipsometry,2,6 capacitive dilatometry (CD),8,15,16 fluorescence spectroscopy17 etc.— Tg is defined as the temperature where a first order thermodynamic property (or an associated quantity) undergoes a change in its temperature dependence. These methods deliver information on the way a glass former leaves thermodynamic equilibrium upon cooling becoming a glass or recovers it when heated. This equilibrium to outof-equilibrium crossover is usually called the thermal glass transition, which depends on the cooling rate.18 Contrariwise, techniques such as broadband dielectric (BDS)4,16,19,20 and specific heat (AC-calorimetry)4,16,21 spectroscopy are relaxation techniques working in the linear regime, i.e. the spontaneous fluctuations associated with the segmental relaxation are probed at temperatures usually above the thermal Tg discussed before. Thus, information is delivered on the intrinsic segmental dynamics of the glass former in the supercooled liquid-like state. From this a ‘‘dynamic’’ Tg is usually defined, for instance as the temperature at which the relaxation time reaches a given value, e.g. 100 s.18 From a fundamental point of view the two types of Tg(s) above defined deal with different—though likely related—phenomena.18 Thereby, at a first glance there is no reason why both quantities should coincide or behave in the same way. The aim of this work is to clarify this issue from an experimental point of view in the case of polymer thin films. To do so, we have carried out a detailed investigation and comparison of the thermal and dynamic glass transition of thin films of polystyrene (PS). Two high molecular weight PS samples (Mw ¼ 1400 kg mol 1 and 570 kg mol 1) have been investigated in thin films with thicknesses from several microns down to 15 nm by means of calorimetric and dielectric techniques. Both techniques are sensitive to the dynamic as well as thermal glass transition. In particular, AC-calorimetry and BDS characterize the segmental dynamics of PS. DSC and CD deliver thermal and volumetric information respectively on the equilibrium to out-of-equilibrium transition marked by Tg. We Soft Matter, 2012, 8, 5119–5122 | 5119 emphasize that all measurements have been carried out on identically prepared and treated samples and, in the case of the dielectric techniques, both the BDS and CD results are obtained simultaneously on the same sample by using the same experimental set-up during the same measurement. Our results show a Tg depression with decreasing film thickness by both DSC and CD, whereas the molecular dynamics, as probed by BDS and AC-calorimetry, remain unchanged compared to the bulk down to a thickness of 15 nm. Moreover, consistent results were obtained for different sample geometries like freestanding films, samples with one free surface and layers capped between electrodes. BDS measurements were performed for both aluminium (Al) capped and stacked films configurations. AC-calorimetry was performed on samples with one free surface, the other supported by silicon dioxide. The stacked film configuration was employed for all standard DSC experiments. High molecular weight PS films were employed for both DSC and BDS to prevent interpenetration of the stacked films.14,22 Each of the stacked films is therefore surrounded by a gas atmosphere and, therefore, can be considered ‘‘freestanding’’. Particular attention has been devoted to the efficiency of solvent and water removal (long annealing times at high temperatures in vacuum; see the ESI†). Fig. 1 presents results obtained simultaneously by BDS and CD, i.e. for the same specimen. Panels (a) and (b) report the loss part of the permittivity as a function of the temperature for freestanding (at 40 Hz) and Al-capped (at 105 Hz) films, respectively. The dielectric response of bulk PS is included for both displayed frequencies too.‡Fig. 1a and b prove that the segmental dynamics in the thin PS films is, within the experimental error, bulk-like for any film thickness, irrespective of the presence of a substrate. The temperature dependence of the real part of the permittivity (30 ) is depicted in panels (c) and (d) of Fig. 1, taken at a frequency (105 Hz) sufficiently high to guarantee that these values are not significantly affected by relaxational processes in the considered temperature range. This allows the monitoring of the temperature dependence of the film density.8,15,16 Tg is identified (as usual) as the intersection temperature of the linear fits of the 30 values in the glassy and melt state. Fig. 1 (c) and (d) show a systematic Tg decrease with decreasing film thickness for both film geometries. Interestingly the effect of the film thickness is more pronounced for freestanding than for Al-capped films. In particular, the freestanding ones display a Tg depression beginning at thicknesses in the order of microns, whereas for the Al-capped films noticeable effects are observed only for thicknesses smaller than several hundred nanometers. To summarize the BDS results, in the same experiment, an unchanged PS segmental dynamics and a decrease of Tg with decreasing film thickness have been simultaneously observed for the first time. Fig. 2 shows the results obtained by calorimetry. Panels (a) and (b) (right axes) of Fig. 2 respectively present the rescaled amplitude— proportional to the real part—and the phase angle—related to the loss part of the complex heat capacity21—of the measured differential voltage at 160 Hz as a function of the temperature measured by AC-calorimetry. These plots indicate that both the temperature corresponding to half of the amplitude step and that corresponding to the maximum of the phase angle of the complex differential voltage are independent of the film thickness, within the experimental uncertainty. Furthermore, the width of the phase angle peak does not depend on the film thickness. Since the AC-calorimetry experiments are carried out in the linear regime, these results indicate that the spontaneous thermal fluctuations associated to the segmental 5120 | Soft Matter, 2012, 8, 5119–5122 Fig. 1 Temperature dependence of the dielectric loss at: (a) 40 Hz for freestanding films; (b) 105 Hz for Al-capped films; and of the real part of the dielectric permittivity measured at 105 Hz for: (c) freestanding films; (d) Al-capped films. Symbols in panels (c) and (d) are the same as those in (a) and (b). The straight lines in panels (c) and (d) are the linear fits to the data of the glassy and liquid branches of the real part of the dielectric permittivity. dynamics are independent of the film thickness and identical to those of bulk PS, in agreement with the literature4,21 and the dielectric measurements discussed above. Panels (a) and (b) (left axes) of Fig. 2 display the temperature dependence of the specific heat as obtained by DSC and its temperature derivative for the freestanding films, respectively. The Tg values—estimated from the step in specific heat and the This journal is ª The Royal Society of Chemistry 2012 Fig. 2 (left axes) Temperature dependence of the specific heat (a) and its temperature derivative (b) obtained from standard DSC at a cooling rate of 20 K min 1. The inset of (b) is an enlargement in the micron thick region of the temperature derivative of the specific heat. (right axes) Temperature dependence of the amplitude (a) and the phase angle (b) of the complex differential voltage at a scanning rate of 2 K min 1 and a frequency of 160 Hz as measured by AC-calorimetry. corresponding peak in its derivative—decrease systematically with the film thickness. Deviations from the bulk Tg are visible already for several-micron thick films (see inset of Fig. 2b). To our knowledge we report significant and systematic deviations from the bulk Tg for films thicker than a few hundred nanometers for the first time. Furthermore, similar results as those reported in Fig. 2, shown later in the manuscript (see Fig. 4), were obtained for a lower molecular weight PS (570 kg mol 1). This implies that in our case the Tg depression is independent of the molecular weight. This result agrees with several studies where no molecular weight dependence was found,2,23 whereas it contrasts with the previously reported molecular weight dependence of the Tg depression.24 In summary, similar to BDS, calorimetry shows that unchanged segmental dynamics and a Tg depression take place simultaneously in PS thin films prepared employing exactly the same protocol. Fig. 3 Relaxation time obtained by BDS and AC-calorimetry versus the reciprocal temperature for all investigated systems. This journal is ª The Royal Society of Chemistry 2012 To emphasize the lack of thickness dependence of the dynamics, Fig. 3 displays the relaxation map obtained by both BDS and ACcalorimetry for the segmental dynamics of PS thin films. The typical relaxation time, determined as s ¼ 1/(2pf), is plotted versus the inverse of the temperature of the maximum of 300 or the phase angle for all systems. For all thicknesses the data collapse into one chart independent of both the applied perturbation (electrical or thermal) and the samples surfaces (freestanding films, one free surface and Alcapped). Considering the data reported in Fig. 1 (panels (a) and (b)) and Fig. 2 (right axes), this concerns also the width of the relaxation peak. The Tg data from DSC and CD as a function of the film thickness h are summarized in Fig. 4, as difference between the Tg of the film and the bulk value (376 K for DSC and 370 K for CD on cooling respectively at 20 and 0.2 K min 1). The main conclusion here is that the DSC and CD results on freestanding films show an identical Tg depression, whereas a weaker decrease of Tg is observed for the Al-capped films, probably due to the weak interaction of the segments with the Al substrate, in agreement with recent results.25 The present study shows that a Tg depression and unchanged segmental dynamics are simultaneously observed for PS films with thicknesses from several microns down to 15 nm. A Tg depression for films thicker than 100 nm has never been reported before, likely due to the fact that no systematic study in this thickness range has ever been performed. It is worth noting that the Tg depression found in the present work may substantially differ from that found in other studies (see e.g. ref. 14 and 26, where the Tg from DSC and CD respectively are reported). As previously discussed, this can be generally attributed to different measurement conditions, such as the heating/cooling rate6,7,14,27 and adsorption phenomena,25 often giving rise to a large scattering in Tg depression data.5 However, these aspects go beyond the goal of the current work: the comparison between thermal and dynamic Tg. In the framework of the present study, it is important to emphasize that our results have been found for samples prepared under identical conditions and, in the case of dielectric techniques, in a single measurement on the same sample. These findings are also independent of the type of the perturbation applied to the sample: thermal in the case of AC-calorimetry and electrical for BDS. Although DSC and CD data provide information on the thermal glass transition at temperatures somewhat lower than those relevant for BDS and AC-calorimetry, it is unlikely to expect dramatic changes of the relaxation time of thin films in comparison to the bulk Fig. 4 Deviation of the Tg of all investigated films from the bulk behavior as a function of the film thickness. Soft Matter, 2012, 8, 5119–5122 | 5121 at lower temperatures. The presence of a bulk-like relaxational component in freestanding thin PS films at temperatures comparable to the thermal Tg, recently found by Ediger and co-workers,28,29 corroborates the previous consideration. Our results can be understood in terms of the different information obtained by techniques probing the way a polymer melt leaves equilibrium when cooling down and those providing direct characterization of the spontaneous fluctuations occurring in the supercooled state.30 Thus the most obvious consequence is that the equilibrium to out-of-equilibrium transition occurring at Tg is not uniquely related to the intrinsic molecular mobility, with geometric factors,31 the nature of the interface2,32 and the heating/cooling rate6,7,27 also important in thin films and likely in nanostructured systems. V.M.B., D.C., A.A. and J.C. acknowledge the University of the Basque Country and Basque Country Government (Ref. No. IT-43607, Depto. Educaci on, Universidades e Investigaci on and Spanish Minister of Education (Grant No. MAT 2007-63681) for their support. V.M.B. acknowledges the JAE-Doc contract from CSIC, co-financed by the European Social Fund (ESF). H.Y. and A.S. thank Deutsche Forschungsgemeinschaft for financial support (SCHO470/20-1). Notes and references ‡ Dielectric spectra for freestanding films are displayed at a relatively low frequency since for these films no information can be obtained at relatively high frequencies, and therefore high temperatures, where interpenetration of the stacked films occurs, a process that should be prevented. 1 C. L. Jackson and G. B. McKenna, J. Non-Cryst. Solids, 1991, 131, 221–224. 2 J. L. Keddie, R. A. L. Jones and R. A. Cory, Europhys. Lett., 1994, 27, 59–64. 3 A. Bansal, H. C. Yang, C. Z. Li, K. W. Cho, B. C. Benicewicz, S. K. Kumar and L. S. Schadler, Nat. Mater., 2005, 4, 693–698. 4 M. Tress, M. Erber, E. U. Mapesa, H. Huth, J. Mueller, A. Serghei, C. Schick, K.-J. Eichhorn, B. Volt and F. Kremer, Macromolecules, 2010, 43, 9937–9944. 5 J. A. Forrest and K. Dalnoki-Veress, Adv. Colloid Interface Sci., 2001, 94, 167–196. 6 Z. Fakhraai and J. A. Forrest, Phys. Rev. Lett., 2005, 95, 025701. 7 M. Y. Efremov, E. A. Olson, M. Zhang, Z. Zhang and L. H. Allen, Phys. Rev. Lett., 2003, 91, 085703. 5122 | Soft Matter, 2012, 8, 5119–5122 8 V. Lupascu, S. J. Picken and M. W€ ubbenhorst, J. Non-Cryst. Solids, 2006, 352, 5594–5600. 9 J. S. Sharp and J. A. Forrest, Phys. Rev. E: Stat. Phys., Plasmas, Fluids, Relat. Interdiscip. Top., 2003, 67, 031805. 10 C. Soles, J. Douglas, W. Wu, H. Peng and D. Gidley, Macromolecules, 2004, 37, 2890–2900. 11 R. Inoue, T. Kanaya, K. Nishida, I. Tsukushi, M. T. F. Telling, B. J. Gabrys, M. Tyagi, C. Soles and W. I. Wu, Phys. Rev. E: Stat., Nonlinear, Soft Matter Phys., 2009, 80, 031802. 12 A. Clough, D. Peng, Z. Yang and O. K. C. Tsui, Macromolecules, 2011, 44, 1649–1653. 13 X. R. Wang and W. S. Zhou, Macromolecules, 2002, 35, 6747– 6750. 14 Y. P. Koh and S. L. Simon, J. Polym. Sci., Part B: Polym. Phys., 2008, 46, 2741–2753. 15 R. D. Priestley, L. J. Broadbelt, J. M. Torkelson and K. Fukao, Phys. Rev. E: Stat., Nonlinear, Soft Matter Phys., 2007, 75, 061806. 16 V. Lupascu, H. Huth, C. Schick and M. W€ ubbenhorst, Thermochim. Acta, 2005, 432, 222–228. 17 C. J. Ellison and J. M. Torkelson, Nat. Mater., 2003, 2, 695–700. 18 E. Donth, The Glass Transition, Springer-Verlag, Berlin Heidelberg New York, 2001. 19 D. Labahn, R. Mix and A. Sch€ onhals, Phys. Rev. E: Stat., Nonlinear, Soft Matter Phys., 2009, 79, 011801. 20 H. Yin, S. Napolitano and A. Sch€ onhals, Macromolecules, 2012, 45, 1652–1662. 21 H. Huth, A. A. Minakov and C. Schick, J. Polym. Sci., Part B: Polym. Phys., 2006, 44, 2996–3005. 22 S. J. Whitlow and R. P. Wool, Macromolecules, 1991, 24, 5926– 5938. 23 C. J. Ellison, M. K. Mundra and J. M. Torkelson, Macromolecules, 2005, 38, 1767–1778. 24 K. Dalnoki-Veress, J. A. Forrest, C. Murray, C. Gigault and J. R. Dutcher, Phys. Rev. E: Stat. Phys., Plasmas, Fluids, Relat. Interdiscip. Top., 2001, 63, 031801. 25 S. Napolitano and M. W€ ubbenhorst, Nat. Commun., 2011, 2, 260. 26 S. Napolitano, C. Rotella and M. Wuebbenhorst, Macromol. Rapid Commun., 2011, 32, 844–848. 27 D. Hudzinskyy, A. V. Lyulin, A. R. C. Baljon, N. K. Balabaev and M. A. J. Michels, Macromolecules, 2011, 44, 2299–2310. 28 K. Paeng, R. Richert and M. D. Ediger, Soft Matter, 2012, 8, 819– 826. 29 K. Paeng, S. F. Swallen and M. D. Ediger, J. Am. Chem. Soc., 2011, 133, 8444–8447. 30 V. M. Boucher, D. Cangialosi, A. Alegrıa, J. Colmenero, I. PastorizaSantos and L. M. Liz-Marzan, Soft Matter, 2011, 7, 3607– 3620. 31 V. M. Boucher, D. Cangialosi, A. Alegria and J. Colmenero, Macromolecules, 2010, 43, 7594–7603. 32 J. A. Forrest, K. Dalnoki-Veress, J. R. Stevens and J. R. Dutcher, Phys. Rev. Lett., 1996, 77, 2002–2005. This journal is ª The Royal Society of Chemistry 2012