`Strong' and `superstrong' liquids, and an approach to the perfect glass state via phase transition

Abstract

We show first that when the excitations described by the `bond lattice' model, and its `defect' model relatives, are allowed to interact, this simple class of model predicts the possibility of first-order transitions between viscous and fluid liquid states. Looking for support of this prediction, we examine the behavior of a series of `tetrahedral' liquids, all of which are famous for one or another aspect of their behavior. We show that when diffusivity data for these liquids, SiO 2, BeF 2, water, and liquid Si, are plotted on a reduced temperature scale with glass transition temperature as the scaling temperature, a systematic pattern is revealed in which a mild deviation, in the case of BeF 2, from the `strong' extreme of the normal strong/fragile glass-former pattern becomes a (contentious) strong deviation in the case of water and develops finally into the predicted first-order transition deviation in the case of liquid Si. Unfortunately, in the latter two cases, the systematic strengthening of the anomalous character is associated with a decrease in the temperatures of occurrence, such that in each case they fall below the melting point. The corresponding competition with crystallization makes their observation difficult. Our account of the phenomenology, therefore, depends heavily on computer simulation studies, but extensive links to experimental results are given. We relate our findings to the recent observations that the amorphous states of Si and water are unique among glassy systems in showing little or no trace of the very low temperature (glassy state as opposed to liquid state) anomalies formerly considered `ubiquitous' among glassy systems. We interpret this to mean that systems with strong cooperativity in their excitations are able to access the lower minima on their respective configuration space potential energy hypersurfaces and thereby to reach states, which are close to the ideal of the `perfect glass'. In this state the residual entropy is near zero, and the defect-related boson peak and two-level tunneling system excitations are weak or absent. Such systems require unusual routes to access their glassy states and their properties are more closely related to those of crystals than to those of ordinary glasses. A new designation may be required. The range of such systems is large, embracing all the 3:5 and many 2:6 semiconductors.