Heterogeneous and Homogeneous Crystallization of Soft Spheres in Suspension

D Roehm and K Kratzer and A Arnold, HIGH PERFORMANCE COMPUTING IN SCIENCE AND ENGINEERING'13: TRANSACTIONS OF THE HIGH PERFORMANCE COMPUTING CENTER, STUTTGART (HLRS) 2013, 33-52 (2013).

DOI: 10.1007/978-3-319-02165-2_3

Nucleation, i.e., the onset of a phase transition like crystal growth, is a rare event with waiting times in the order of days. Yet, it is an event on the molecular scale, and therefore difficult to study, both experimentally and by computer simulations. Our interest is in the role of long range interactions in nucleation, in particular electrostatic and hydrodynamic interactions mediated by solvent molecules. In order to model the solvent, we use a lattice fluid that is propagated by the fluctuating Lattice Boltzmann (LB) method. Our implementation uses a graphics card (GPU) to propagate the solvent and is coupled to the Molecular Dynamics (MD) simulation package ESPResSo. Using this code, we study the heterogeneous crystallization in Yukawa-like colloidal systems. Our simulations allow to observe the growth of a crystal in a channel with and without hydrodynamic interactions, and indicate that hydrodynamic interactions slow down the crystallization. Additionally, we present results on the homogeneous crystallization of Yukawa particles. While heterogeneous nucleation can be observed directly in simulations, homogeneous nucleation requires special sampling techniques. We use our own Forward Flux Sampling implementation, the Flexible Rare Event Sampling Harness Systems (FRESHS). FRESHS can control popular MD simulation packages as back-end, making it a versatile tool to study rare events. Our simulations confirm previous results at higher supersaturations, which show that the nucleation mechanism involves two steps, namely the formation of a metastable bcc phase and the transformation to a stable fcc phase.

Return to Publications page