Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe
O: Fachverband Oberflächenphysik
O 34: Catalysis and surface reactions I
O 34.3: Vortrag
Dienstag, 10. März 2026, 11:00–11:15, HSZ/0204
Modeling Solvothermal Reactions at Surfaces with Machine Learning Interatomic Potentials — •Maciej Baradyn1, Nils Gönnheimer1,2, and Johannes T. Margraf1 — 1University of Bayreuth, Bayreuth, Germany — 2Fritz-Haber Institute, Berlin, Germany
Dynamical processes such as adsorption, desorption and diffusion are vital elementary reactions involved in heterogeneous catalytic reactions. In solvothermal reactions, the presence of solvent molecules has significant effects on the dynamics of the adsorbates, which must displace hydration layers to diffuse or adsorb on the surface and undergo resolvation during desorption. The presence of solvent molecules also affects the interactions between surfaces and adsorbates, and can contribute to the (de-)stabilization of the adsorbed species.
In this contribution, we explore the robustness of machine learning potentials based on the MACE-MP-0 foundation model for describing dynamical processes under these conditions. This is demonstrated on a representative system of a glycerol molecule interacting with the Cu(111) surface in explicit water. Enhanced sampling methods are used to compute free energy profiles of the studied processes, and to reflect the experimental reaction conditions under which they occur in industrial applications.
Keywords: heterogeneous catalysis; machine learning; explicit solvation; enhanced sampling; dynamics