Regensburg 2025 – wissenschaftliches Programm

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

O: Fachverband Oberflächenphysik

O 81: Heterogeneous Catalysis III

O 81.8: Vortrag

Donnerstag, 20. März 2025, 12:15–12:30, H6

Machine Learning Assisted Realistic Description of Mo-V Mixed Oxide Surfaces — •Kyeonghyeon Nam, Y. Song, L. Masliuk, T. Lunkenbein, A. Trunschke, K. Reuter, and C. Scheurer — Fritz-Haber-Institut der MPG, Berlin

The activity and selectivity of realistic heterogeneous catalysts can be noticeably altered by subtle changes in factors such as bulk composition, dopants, defects, and reaction conditions. These effects are intricately interrelated. To systematically unravel them, we aim to understand their impact on the evolution of catalyst surfaces. Specifically, we focus on the M1 structural modification of (Mo,V)O_x and (Mo,V,Te,Nb)O_x as selective catalysts for the oxidative dehydrogenation of ethane to ethylene.

The large primitive cell of the M1 catalyst poses challenges for a detailed study of its surface terminations using conventional first-principles calculations. To overcome this, we trained machine learning interatomic potentials (MLIPs) using a staged training method from motifs to surfaces. By combining density-functional tight-binding (DFTB) calculations with simulations employing MLIPs, such as molecular dynamics and ab initio thermodynamics (AITD), we elucidated the influence of niobium and tellurium doping on enhanced surface structure stability and catalytic activity during the thermal activation. This was supported by experimental quasi-operando scanning transmission electron microscopy (STEM) images.

Keywords: M1 catalysts; mixed oxide; ODE; ODH