Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe
SYMD: Symposium AI-driven Materials Design: Recent Developments, Challenges and Perspectives
SYMD 1: AI-driven Materials Design: Recent Developments, Challenges and Perspectives
SYMD 1.3: Hauptvortrag
Montag, 17. März 2025, 16:00–16:30, H1
Multiscale Modelling & Machine Learning Algorithms for Catalyst Materials: Insights from the Oxygen Evolution Reaction — •Nong Artrith — Debye Institute for Nanomaterials Science, Utrecht University, NL
Machine learning (ML) has emerged as a powerful tool to accelerate the discovery of catalytic materials by integrating information from computation and experiment. While ML excels at pattern detection in large, uniform datasets, many catalyst studies rely on small, experimentally measured datasets. Our approach combines ML and first-principles calculations to extract insights from such small experimental datasets by training a complex ML model on a large computational library of transition-state energies and combining it with simple linear regression models fitted to experimental data. We use this approach to explore the catalytic activity of monolayer bimetallic catalysts for ethanol reforming, identifying key reactions and predicting promising catalyst compositions. For the explicit modeling of catalytic reactions, we performed ML-driven molecular dynamics and metadynamics simulations of the oxygen evolution reaction (OER) over oxide materials. Using a neural network potential, trained using transfer learning, we captured the dynamic mechanistic details of OER, elucidating the impact of nickel doping on the catalytic activity of BaTiO3, a perovskite oxide synthesized from earth-abundant precursors. The combined insights from these case studies illustrate the versatility of ML in guiding the design of efficient and sustainable catalysts, ranging from ethanol reforming to water-splitting reactions.
Keywords: Machine Learning Interatomic Potential; Oxygen Evolution Reaction; Metadynamics Simulation; DFT; Atomistic Simulation