Berlin 2024 – scientific programme
Parts | Days | Selection | Search | Updates | Downloads | Help
MM: Fachverband Metall- und Materialphysik
MM 6: Interface Controlled Properties, Nanomaterials and Microstructure Design I
MM 6.1: Talk
Monday, March 18, 2024, 10:15–10:30, C 230
Learning the influence of chemistry on grain-boundary segregation — •Christoph Dösinger1, Oleg Peil2, Daniel Scheiber2, Vsevolod Razumovskiy2, and Lorenz Romaner1 — 1Montanuniversität Leoben, Department of Materials Science, Leoben, Austria — 2Materials Center Leoben Forschung GmbH, Leoben, Austria
The grain-boundary segregation energy is the central quantity for describing the process of grain-boundary segregation which influences interfacial properties. Usually, to obtain highly accurate values for segregation energies, density functional theory is employed, which incurs high computational costs. This makes it impractical to do a thorough study of segregation to multiple grain-boundaries for a range of solutes. To reduce the number of calculation needed for such a complete description, we apply machine learning methods to density functional theory data. In this talk I will show, how one can train machine learning models that cover the periodic table of elements. By combining element specific features and features of the local atomic structure, these models are able to generalize to different elements and grain-boundaries and accurately predict the segregation energies. The method is tested on a comprehensive data-set of segregation energies in W and then applied in an active learning loop for learning segregation in Cr.
Keywords: Grain-boundary segregation; data-driven