Kinetics and thermodynamics of Transformations Using Machine Learning
Accurately modeling transformations at the atomic scale is crucial for the determination of their thermodynamic and kinetic properties in materials science. These phenomena, including chemical reactions, nucleation, crystallization, and precipitation, play an important role in applications for energy storage, controlling for instance the performances and longevity of Li-ion batteries. A long-standing challenge is the identification of optimal collective variables able to describe and drive the transformation under scrutiny. Here I will present a new method for the generation, optimization, and comparison of collective variables that can be thought of as a data-driven generalization of the path collective variable concept. It consists of a kernel ridge regression of the committor probability, which encodes a transformation’s progress. The resulting collective variable is one-dimensional, interpretable, and differentiable, making it appropriate for enhanced sampling simulations requiring biasing. I will present applications to phenomena such as precipitation and ion pairing in electrolytes, which are crucial for the understanding of the solid electrolyte interphase formation in Li-ion batteries.
CNRS researcher at IMPMC, Sorbonne Université
Dr. Arthur France-Lanord is a CNRS researcher at the Institute of Mineralogy, Materials Physics and Cosmochemistry (IMPMC) in Paris. He holds a Ph.D. in physics from Université Paris-Sud and Materials Design, where he focused on nanoscale transport phenomena. He worked as a postdoctoral researcher with Jeffrey Grossman at MIT and with Mathieu Salanne at PHENIX (Sorbonne Université), studying materials for energy storage applications. His current focus is on combining machine learning techniques and statistical mechanics to investigate transformations at the atomic scale.
