Crystallization of hard-sphere glasses

TL;DR

This study uses molecular dynamics to explore how hard-sphere glasses transition to crystals, revealing that crystallization can occur within non-ergodic glass states, challenging previous assumptions.

Contribution

It demonstrates that spontaneous crystallization can happen in glassy states, showing that crystallization does not necessarily indicate an ergodic fluid.

Findings

Crystallization occurs at volume fractions above the glass transition point.

Crystallization can happen within aging, non-diffusive glass states.

Spontaneous crystallization does not imply the system was an ergodic fluid.

Abstract

We study by molecular dynamics the interplay between arrest and crystallization in hard spheres. For state points in the plane of volume fraction ($0.54 \leq phi \leq 0.63$) and polydispersity ($0 \leq s \leq 0.085$), we delineate states that spontaneously crystallize from those that do not. For noncrystallizing (or precrystallization) samples we find isodiffusivity lines consistent with an ideal glass transition at $\phi_g \approx 0.585$, independent of $s$. Despite this, for $s<0.05$, crystallization occurs at $\phi > \phi_g$. This happens on time scales for which the system is aging, and a diffusive regime in the mean square displacement is not reached; by those criteria, the system is a glass. Hence, contrary to a widespread assumption in the colloid literature, the occurrence of spontaneous crystallization within a bulk amorphous state does not prove that this state was an ergodic fluid rather than a glass.