ICAMS / Interdisciplinary Centre for Advanced Materials Simulation


Maximum likelihood analysis of reaction coordinates during crystal nucleation in Ni

Date: 07.06.2019
Place: Mainz Materials Simulation Days, Mainz, Germany

Grisell Díaz Leines
Jutta Rogal

Fundamental knowledge of the mechanism of crystal nucleation and polymorph selection is essential for the synthesis and control of materials with targeted properties. However, understanding the atomistic mechanism of crystallization poses a major challenge as many materials exhibit complex transitions of multiple steps, forming polymorphic structures. In this work we employ transition path sampling simulations together with a maximum likelihood estimation (MLE) and perform a quantitative and multi-dimensional analysis of the reaction coordinate (RC) and the free energy landscape of the nucleation process in Ni. In particular, the MLE approach allows us to directly compare the quality of different order parameters as RCs based on their ability to model the committor function with respect to the given data of the path ensemble. The analysis of the reweighted path ensemble and the free energy surface on a set of candidate structural order parameters shows that the nucleation pathway in Ni is governed by the initial formation of mesocrystal regions and a subsequent emergence of fcc-hcp crystallites embedded within the core of these prestructured clusters. Our findings indicate that the preordered liquid region is an order parameter that en- hances the RC and therefore has an essential role in the structural description of the nucleus and its interfacial free energy. Moreover, we show that these regions of higher order than the liquid predetermine the coordination of the fcc-hcp polymorphs selected, acting as precursors of the crystallization. These results shed light on the prominent role of the prestructured liquid regions during crystallization and polymorph selection in Ni.

« back