Dresden 2026 – scientific programme
Parts | Days | Selection | Search | Updates | Downloads | Help
MM: Fachverband Metall- und Materialphysik
MM 9: Topical Session: Physics-driven Artificial Intelligence for Materials II
MM 9.1: Topical Talk
Monday, March 9, 2026, 15:45–16:15, SCH/A251
Atomistic simulations in the ternary Fe-O-H system: interatomic potential development and applications — •Baptiste Bienvenu1, Mira Todorova1, Matous Mrovec2, Ralf Drautz2, Dierk Raabe1, and Jörg Neugebauer1 — 1Max Planck Institute for Sustainable Materials, Düsseldorf, Germany — 2Interdisciplinary Centre for Advanced Materials Simulation, Ruhr Universität Bochum, Germany
Atomistic modeling of iron oxides is challenging, requiring accurate electronic structure calculations and extensive length and time scales to simulate elementary mechanisms such as the extraction of metallic iron from its oxides through hydrogen-based reduction. To enable atomic scale modeling of these and other technologically relevant processes within the ternary Fe-O-H system (e.g., hydrogen embrittlement, water splitting), an accurate yet efficient interatomic potential is needed, something that is currently lacking in the literature. First, we focus on the binary Fe-O system, for which we previously developed a robust and transferable Atomic Cluster Expansion (ACE) machine-learning potential with an explicit account of magnetism, to study bulk diffusion and the structure and stability of various surfaces of iron oxides. We then extend the model to include hydrogen and show that the resulting ACE potential can faithfully reproduce key mechanisms of the Fe-O-H system: (i) surface reactions and microstructure evolution during hydrogen reduction of iron oxides, (ii) surface reactions of iron and its oxides with water and (iii) hydrogen trapping, interaction with extended defects and permeation in metallic iron.
Keywords: Iron oxides; Iron; Machine learning interatomic potential; Hydrogen reduction; Atomistic simulations