Abstract

When studying glassy materials and supercooled liquids, the disordered molecular structure is important. Unlike crystals, each molecule in a liquid has a slightly different local environment, and these subtle differences can have important implications for material properties. I will present a method [1] for classifying particles into groups (``structural communities’‘) based on their local environments, as characterised by the distances to their neighbours (radial distribution function) or their bond angles. The method is applied to three different binary mixtures, made of spherical particles. This is an unsupervised approach using only structural information; nevertheless the resulting communities are found to be correlated with particles’ dynamical behaviour.

[1] J. Paret, R. L. Jack, and D. Coslovich, J. Chem. Phys. 152, 144502 (2020).