Parts | Days | Selection | Search | Updates | Downloads | Help

O: Fachverband Oberflächenphysik

O 94: Poster Session VII: Poster to Mini-Symposium: Electrified solid-liquid interfaces II

O 94.3: Poster

Thursday, March 4, 2021, 10:30–12:30, P

Resolving the structure of oxidized Cu surfaces with machine-learned Gaussian Approximation Potentials — •Nicolas Bergmann¹, Nicolas G. Hörmann^1,2, and Karsten Reuter² — ¹Technische Universität München, Garching, Germany — ²Fritz-Haber-Institut der Max-Planck-Gesellschaft, Berlin, Germany

Copper was recently shown to exhibit promising capabilities toward electrochemical CO oxidation [1], yet only after undergoing activating surface morphological changes. The detailed structure and composition of the formed surface oxidic layer is hitherto unknown, preventing further mechanistic analyses. Here we use the high computational efficiency of machine-learned Gaussian Approximation Potentials (GAPs [2]) to systematically investigate the Cu(111) surface structure with varying concentrations of adsorbed oxygen. The potentials are trained with density-functional theory data of bulk CuO_x and molecular dynamics generated slab structures.

While low oxygen coverages do not alter the Cu substrate significantly, we find dramatic morphological changes above a critical coverage of ∼25% monolayer: Surface copper atoms are extruded from the top layer, forming CuO_x islands, while at the same time the amount of subsurface oxygen increases. A detailed analysis of local atomic environments reveals predominant local structural motives resembling those in well known bulk copper oxides.

[1] A. Auer et al., Nature Catal. 3, 10, (2020). [2] A.P. Bartok and G. Csanyi, Int. J. of Quantum Chem. 115, 16 (2015).