The surprising utility of excess entropy in the study of supercooled liquids

In equilibrium liquids, excess entropy shows a surprising effectiveness in correlating with the dynamical properties of many computational liquids, yielding a simple scaling relation that collapses the trends observed across these systems known as Rosenfeld scaling. Thus, the dynamics seen in these equilibrated systems can be inferred directly from a thermodynamic quantity. In supercooled liquids however, this simple relationship breaks down, and the common two-body approximation for the excess entropy is insufficient to predict average particle mobilities. In this work we analyze a variety of binary Lennard-Jones mixtures to explore the utility of excess entropy in the supercooled regime. We find that under proper partitioning of systems by their rearrangement dynamics, we can collapse the dynamical properties of these ensembles across systems as a function of excess entropy. We show how the barriers for rearrangement change with the shape of the pair potential and interpret how these changes are related to the stiffness of the potential.