SurfaceScience21 – scientific programme
Parts | Days | Selection | Search | Updates | Downloads | Help
O: Fachverband Oberflächenphysik
O 20: Poster Session II: Heterogeneous catalysis II
O 20.1: Poster
Monday, March 1, 2021, 13:30–15:30, P
Data-Driven Descriptor Engineering and Refined Scaling Relations for Predicting Transition Metal Oxide Reactivity — •Wenbin Xu1, Mie Andersen1, and Karsten Reuter1,2 — 1Chair for Theoretical Chemistry, Technical University of Munich, Garching, Germany — 2Fritz Haber Institute of the Max Planck Society, Berlin, Germany
Screening of transition metal (TM) and TM alloy catalysts is often carried out using the comparatively simple d-band model and related scaling relations to predict adsorption enthalpies. Unfortunately, these methodologies do not extend to TM oxides due to their more localized and intricate electronic structure. Multiple geometric and electronic properties could in principle govern the adsorption enthalpies at these more complex compound materials. Given the immense number of candidates, we here apply a data-driven compressed sensing method[1] to tackle the search for best-performing multidimensional descriptors expressed as nonlinear functions of intrinsic properties of the clean surface. Our descriptors largely outperform previously highlighted descriptors in terms of accuracy and computational cost. Furthermore, we identify properties related to the local charge transfer to be the missing ingredient in standard scaling relations and show that their inclusion is crucial for the correct identification of promising catalyst materials for the oxygen evolution reaction.[2]
[1]R. Ouyang et al., Phys. Rev. Mater. 2, 083802 (2018)
[2]W. Xu et al., ACS Catal. 11, 734 (2021)