Space and time dynamical heterogeneity in glassy relaxation. The role of democratic clusters

Abstract

In this work we review recent computational advances in the understanding of the relaxation<br />dynamics of supercooled glass-forming liquids. In such a supercooled regime these systems<br />experience a striking dynamical slowing down which can be rationalized in terms of the picture<br />of dynamical heterogeneities, wherein the dynamics can vary by orders of magnitude from one<br />region of the sample to another and where the sizes and timescales of such slowly relaxing<br />regions are expected to increase considerably as the temperature is decreased. We shall focus on<br />the relaxation events at a microscopic level and describe the finding of the collective motions of<br />particles responsible for the dynamical heterogeneities. In so doing, we shall demonstrate that<br />the dynamics in different regions of the system is not only heterogeneous in space but also in<br />time. In particular, we shall be interested in the events relevant to the long-time structural<br />relaxation or α relaxation. In this regard, we shall focus on the discovery of cooperatively<br />relaxing units involving the collective motion of relatively compact clusters of particles, called<br />?democratic clusters? or d-clusters. These events have been shown to trigger transitions between<br />metabasins of the potential energy landscape (collections of similar configurations or structures)<br />and to consist of the main steps in the α relaxation. Such events emerge in systems quite<br />different in nature such as simple model glass formers and supercooled amorphous water.<br />Additionally, another relevant issue in this context consists in the determination of a link<br />between structure and dynamics. In this context, we describe the relationship between the<br />d-cluster events and the constraints that the local structure poses on the relaxation dynamics,<br />thus revealing their role in reformulating structural constraints.