The different physical aspects of glass transitions are reviewed and models aiming at their explanation are described. The following three main aspects are distinguished:(i) the degree of stability of supercooled liquids with respect to crystallisation, (ii) the variation of physical properties of supercooled liquids in metastable equi-(iii) the arrest of structural relaxation at the glass transition.The physical significance of computer simulations of glass transitions in simple liquids and the question of a hidden phase transition underlying an observed glass transition are examined critically. In relation to the stability problem, the geometrical constraints operative in typical disordered structures, such as random sphere packings and random covalent networks, are also discussed. A general survey of glass transition phenomena and concepts is presented in an introductory section. librium above the glass transition, and

No abstract

The ultrasonic attenuation in glasses at low temperatures is calculated using the model proposed by Anderson, Halperin, and Varma. It is found that the resonant absorption of sound by the localized two-level systems saturates and can be observed only if the sound amplitude is extremely small. A second contribution to the sound absorption is derived which arises from the relaxation of the localized excitations and does not saturate. Qualitative agreement with recent measurements of the ultrasonic attenuation in fused silica is obtained.

AbsvaeL Self-diffusion in a lattice-gas model with two-vacancy assisted hopping on the eiangular lattice is investigated, by both Monte Carlo simulation and analythd calculation. A very rapid decrease of the tracerarrelation factor and marked size effece in finite laltica give evidence for Swng dynamic correlations in both space and time at high particle concentration. Although the decrease of the self-dihion coefficient over 3.5 decades for concentrations up to c = 0.77 is best fitted by a power law (0.835c)3,54, it is w e d that the model dces not have a sharp dynamical phase transition with a critical concentration lower than one. The argument is based on a p m f of absence of permanently blocked panicles in infinite lattices at all concentrations Mow one. The self-diffusion coefficient is calculated analytically within a pair approximation which gives good results for lower wncenwtims, but fails at Lhe highex concentrations. The approximation is in qualitative agreement with the Monte Carlo data far the hater-correlation factor at all concentrations for a variant of the model with one-vacancy assisted hopping, in which the dynamic conelations are less pronounced.