Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe

O: Fachverband Oberflächenphysik

O 81: Heterogeneous Catalysis III

O 81.5: Vortrag

Donnerstag, 20. März 2025, 11:30–11:45, H6

Exploration of Explicit Solvation Effects in Heterogeneous Catalysis Using Machine Learning Interatomic Potentials — •Maciej Baradyn and Johannes T. Margraf — University of Bayreuth

The presence of solvent affects many aspects of modeling elementary reactions involved in heterogeneous catalysis, such as stabilization of adsorbed species, the nature of interaction with the surface, as well as energetics and reaction rates, to name just a few. Unfortunately, treating solvent effects in atomistic simulations is a great challenge, since the solvent adds many additional degrees of freedom to the calculation. In first-principles calculations, implicit solvation models are often used to approximately include these effects with moderate computational effort. However, they do not take into account the explicit solvent-surface and solute-solvent interactions, and are therefore known to fail, e.g. when hydrogen bonding is important. More importantly, the solvent molecules can sometimes play a decisive role in the reaction's mechanism (e.g. as proton shuttles), which cannot be captured by implicit models, where solvent is treated as a continuous medium. In this contribution, we explore the efficiency of machine learning potentials based on the MACE-MP-0 foundation model for describing explicit solvation at catalytic interfaces. These potentials are used to capture the kinetic and thermodynamic parameters of small organic molecules interacting with metallic surface in an explicit water bath. Implications for our understanding of the underlying catalytic reaction network and improved design of catalysts will be discussed.

Keywords: catalysis; machine learning; explicit solvation