Dresden 2026 – wissenschaftliches Programm

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

MM: Fachverband Metall- und Materialphysik

MM 13: Data-driven Materials Science: Big Data and Workflows I

MM 13.1: Vortrag

Dienstag, 10. März 2026, 10:15–10:30, SCH/A251

Surface reconstruction via automated LEED analysis based on Bayesian optimization — •Xiankang Tang and Hongbin Zhang — Institute of Materials Science, Technische Universitat Darmstadt, 64287 Darmstadt, Germany

Recent advances in machine learning have enabled the automation of many material characterization processes, which are essential for realizing autonomous experimentation for solid-state materials in the near future. Low-energy electron diffraction (LEED) is a fundamental technique in surface science, providing structural reconstruction information encoded in the energy-dependent intensity modulation of diffracted beams. However, the complexity of data analysis and the computational demands have limited the broader adoption of quantitative LEED in routine surface structure determination. In this work, we implement a Bayesian optimization-based approach to automatize the LEED I(V) analysis, where the best matching structures can be obtained for the experimental I(V) curves by minimize the R-factor between the experimental and simulated I(V) data. This approach can bee combined with density functional calculations or atomisitic simulations to further accelerate the recommendation of plausible structures by minimizing total energies.

Keywords: Low-Energy Electron Diffraction (LEED); Bayesian Optimization; Structural Optimization; Automated Materials Characterization