Numerical method for computing the free energy of glasses

TL;DR

This paper introduces a numerical method to accurately compute the free energy of glasses by sampling their inherent structures, overcoming limitations of traditional extrapolation techniques, with potential applications in ultra-stable glassy materials.

Contribution

A novel numerical approach for calculating the free energy of glasses by sampling inherent structures, applicable when standard methods fail.

Findings

Free energy estimates are lower than slow cooling methods at low temperatures.

Method can be applied to compute chemical potentials of ultra-stable glasses.

Approach may enable solubility estimation of glassy materials.

Abstract

We propose a numerical technique to compute the equilibrium free energy of glasses that cannot be prepared quasi-reversibly. For such systems, standard techniques for estimating the free energy by extrapolation, cannot be used. Instead, we use a procedure that samples the equilibrium partition function of the basins of attraction of the different inherent structures (local potential energy minima) of the system. If all relevant inherent structures could be adequately sampled in the (supercooled) liquid phase, our approach would be rigorous. In any finite simulation, we will miss the lower-energy inherent structures that become dominant at very low temperatures. We find that our free energy estimates for a Kob-Andersen glass are lower than those obtained by very slow cooling, even at temperatures down to one third of the glass transition temperature. The current approach could be applied to compute the chemical potential of ultra-stable glassy materials, and should enable the estimation of their solubility.