An intriguing problem in condensed matter physics is understanding the glass transition, in particular the dynamics in the equilibrium liquid close to vitrification. Recent advances have been made by using hydrostatic pressure as an experimental variable. These results are reviewed, with an emphasis in the insight provided into the mechanisms underlying the relaxation properties of glass-forming liquids and polymers.

Upon decreasing temperature or increasing pressure, a noncrystallizing liquid will vitrify; that is, the structural relaxation time, τ R , becomes so long that the system cannot attain an equilibrium configuration in the available time. Theories, including the well-known free volume and configurational entropy models, explain the glass transition by invoking a single quantity that governs the structural relaxation time. The dispersion of the structural relaxation (i.e., the structural relaxation function) is either not addressed or is derived as a parallel consequence (or afterthought) and thus is independent of τ R . In these models the time dependence of the relaxation bears no fundamental relationship to the value of τ R or other dynamic properties. Such approaches appear to be incompatible with a general experimental fact recently discovered in glass-formers: for a given material at a fixed value of τ R , the dispersion is constant, independent of thermodynamic conditions (T and P); that is, the shape of the R-relaxation function depends only on the relaxation time. If derived independently of τ R , it is an unlikely result that the dispersion of the structural relaxation would be uniquely defined by τ R .

Viscosities eta and their temperature T and volume V dependences are reported for seven molecular liquids and polymers. In combination with literature viscosity data for five other liquids, we show that the superpositioning of relaxation times for various glass-forming materials when expressed as a function of TV(gamma), where the exponent gamma is a material constant, can be extended to the viscosity. The latter is usually measured to higher temperatures than the corresponding relaxation times, demonstrating the validity of the thermodynamic scaling throughout the supercooled and higher T regimes. The value of gamma for a given liquid principally reflects the magnitude of the intermolecular forces (e.g., steepness of the repulsive potential); thus, we find decreasing gamma in going from van der Waals fluids to ionic liquids. For some strongly H-bonded materials, such as low molecular weight polypropylene glycol and water, the superpositioning fails, due to the nontrivial change of chemical structure (degree of H bonding) with thermodynamic conditions.

Recent experimental results on the dynamics of glass-forming materials, particularly polymers, are surveyed. The focus is on aspects of the behavior that are connected to or correlated with structural relaxation. These results include the invariance to thermodynamic conditions (temperature, pressure, volume) of a number of propertiesbreadth of the relaxation dispersion, number of dynamically correlating molecules, Johari−Goldstein secondary relaxation time, onset of the dynamic crossover, and the product of temperature and specific volume with the latter raised to a material constantprovided the structural relaxation time is maintained constant. Additional salient experimental findings include the correlation of various high-frequency processes, usually measured in the glassy state, with properties of the equilibrium material above T g. These correlations indicate that the glass transition, although conventionally defined by the relaxation time becoming larger than experimental time scales (>100 s), has its beginning many orders of magnitude sooner. Also described herein are effects of spatial confinement on the glass transition; these can be dramatic, yet taken in toto are rather discombobulating. Such generally observed phenomena must be included in a comprehensive theory or model of the glass transition, since properties intimately connected to structural relaxation cannot be derived separately and be expected to exhibit such correlations by coincidence.

No abstract