Elucidating the origin of decoupling in glass-forming liquids through high-temperature activation free energies

Abstract It is well established that many glass-forming liquids exhibit a deviation from the Stokes–Einstein relation that is commonly quantified by a power-law relation (so-called fractional Stokes–Einstein relation) between the mass diffusion coefficient D and the momentum diffusion coefficient, corresponding to the shear viscosity η. This ubiquitous phenomenon is often attributed to dynamic heterogeneity upon cooling, characterized by the progressive growth in the average size and lifetime of dynamic clusters of particles having excessively high or low mobility. However, a predictive theoretical understanding of the non-universal power-law decoupling exponent ζ has been elusive. Here, for a wide variety of simulated glass-forming systems, we determine the power-law relation between the structural relaxation time τα and the peak time t* of the non-Gaussian parameter, i.e., $t^* \sim \tau _{\alpha }^{1 - \zeta }$, analogous to the power-law relation between D and η in the widely studied Kob–Andersen glass-forming liquid. We confirm the prediction from the string model of glass formation that ζ is determined by the ratio of the high-temperature activation free energies for t* and τα in the temperature regime where dynamic heterogeneity is minimal. This finding suggests that dynamic heterogeneity is not the cause of decoupling, but rather a symptom. Our study emphasizes the role of the high-temperature liquid dynamics in understanding the fundamental mechanisms of glass formation. Teaser Text The ‘breakdown’ of the Stokes-Einstein relation is shown to originate from the difference in the activation energies for momentum and mass transport in the high-temperature regime where dynamic heterogeneity is minimal.