Dr. Alessandro Fortunelli: Machine-Learning-Accelerated Conformal First-Principles Sampling of Dynamical Processes in Catalysis and Materials Science

Nanocatalysis Department Seminar
Thursday, November 14, 2024, 13:00
J. Heyrovský Institute, Seminar Room 108

Dr. Alessandro Fortunelli, PhD, Italian National Research Council (CNR), Institute of Chemistry of OrganoMetallic Compounds, CNR-ICCOM, Pisa, Italy

Abstract
I will present our recent results on computational catalysis and modeling of dynamical processes in materials science, focusing on how to accelerate first-principles computational predictions of catalysis at heterogeneous interfaces by exploiting Machine-Learning (ML) techniques. We have combined stochastic sampling of activated mechanisms with ML force-fields, in the form of Neural Network Potential (NNP), within an integrated set of codes enabling static and dynamic sampling of configurations and transformation paths, and we have coupled this with conformal techniques to exploit information from existing data. We first use methanol decomposition (of interest as hydrogen production process) as a test of our strategy. Starting from worked-out Pt-based systems, we construct conformal databases that are exported to other systems to quickly derive with only two steps of Active Learning chemically-accurate NNPs, including bimetallic systems (CuPd) and cases exhibiting a change in adsorption sites as tests of catalytic path modifications. Dealloying in the PtMnO system (a paradigmatic example of catalyst restructuring) is then used to compare different NNPs and sampling approaches (m-Dyn, Pathsample, disconnectivity diagrams, and RGO).

Dr. Alessandro Fortunelli is Research Director at CNR-ICCOM in Pisa, Italy. He received his MSc and PhD in Chemistry at Scuola Normale Superiore in Pisa, and also holds a guest scientist position at the California Institute of Technology in USA. His research interests span various fields in chemistry and materials science, with the major goal of achieving a predictive computational science of the structural, catalytic, magneto-optical, mechanical, and transport properties of confined and low-dimensional materials (nanoalloys, 2D oxides, electrocatalysis, amorphous carbon, transport and growth  henomena), with applications in the field of energy and environment.