(323b) Liquid-to-Solid Transition: Understanding Nucleation Using Computer Simulations | AIChE

(323b) Liquid-to-Solid Transition: Understanding Nucleation Using Computer Simulations

Authors 

Sarupria, S. - Presenter, Princeton University
Shi, Z. - Presenter, Princeton University


How the stable solid phase is nucleated in the metastable liquid phase is an important question for liquid-to-solid phase transitions. In principle, conventional molecular dynamics simulations, which sample nanometer lengthscales and nano-to-microsecond timescales are ideally suited to probe this question. However, since nucleation is a rare-event, large part of the conventional molecular dynamics time is spent on simply waiting for the nucleation event to occur. This makes it computationally challenging to generate statistically relevant number of trajectories of the nucleation event. To this end, recently several advanced sampling techniques have been developed that enable computationally viable access to nucleation events. Using one such technique, called forward flux sampling, we study the nucleation of two systems: (1) We explore the nucleation in a family of Lennard-Jones particles with tunable softness. We elucidate how features of intermolecular interactions affect the kinetics of nucleation, namely the properties of critical nuclei and rate of nucleation. (2) We study homogeneous nucleation of ice and carbon dioxide hydrates. There are only a few simulation studies that successfully nucleated ice and methane hydrates using brute-force molecular dynamics. Although, these studies provide important insights into nucleation, they are limited to small number of trajectories. Techniques like forward flux sampling enable us to general statistically significant set of trajectories. We will comment on the rate of ice and hydrate nucleation and characteristics of the critical nuclei. Together, our studies provide a quantitative and predictive framework for nucleation of solid phase from the metastable liquid phase.