(305a) Optimal Reaction Coordinates for Heterogeneous Nucleation of Densely Packed Atomic and Colloidal Crystals

Crystallization is a ubiquitous phenomenon, and a wide variety of atomic, molecular, polymeric, and colloidal building blocks can assemble into crystals over different time and length scales. Most crystals form through a mechanism known as heterogeneous nucleation in which external surfaces (mostly provided by insoluble impurities) facilitate freezing by decreasing nucleation barriers. While molecular simulations have emerged as attractive tools for understanding the physics of crystal nucleation, there is still a considerable gap in our understanding of how surface chemistry and topography impacts the kinetics and mechanism of heterogeneous nucleation. Part of the challenge to achieve such an understanding is to identify reaction coordinates that accurately describe the nucleation process.

In this work, we use molecular dynamics (MD) simulation and jumpy forward flux sampling (jFFS) [1] to probe heterogeneous nucleation of the face-centered cubic (FCC) crystal in the Lennard-Jones system (LJ). By high-throughput screening of a large number of reaction coordinates and using data science approaches, we demonstrate that traditional reaction coordinates based on Steinhardt bond order parameters are not good descriptors of the heterogeneous nucleation process. This is a stark contrast to the heterogeneous nucleation of tetrahedral crystals such as ice that can be decently described by such traditional RCs [2]. We also provide a recipe for constructing optimal RCs for heterogeneous nucleation in the LJ system. Such OPs can be widely utilized for studying heterogeneous nucleation of densely packed crystals in atomic, metallic, and colloidal systems.

[1] Haji-Akbari, J. Chem. Phys., 149: 072303 (2018).

[2] Lupi et al., J. Phys. Chem. Lett., 8: 4201 (2017).