DPG Phi
Verhandlungen
Verhandlungen
DPG

Dresden 2017 – scientific programme

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

MM: Fachverband Metall- und Materialphysik

MM 60: Topical session: Data driven materials design - machine learning

MM 60.4: Talk

Thursday, March 23, 2017, 12:45–13:00, BAR 205

Optimizing Materials Properties with Machine Learning Techniques: A Case Study on Hard-Magnetic Phases — •Johannes J. Möller, Georg Krugel, Wolfgang Körner, Daniel F. Urban, and Christian Elsässer — Fraunhofer Institute for Mechanics of Materials IWM, Freiburg, Germany

Machine Learning (ML) is an emerging field in materials science, in which a numerical model is built in order to predict a certain feature, for instance a materials property. The model building is typically based on a (large) data set of e.g. crystal structures or chemical compositions, for which the property is already known. The beauty of this approach is that it not only allows us to predict the properties for unknown compositions, but also to determine the composition that optimizes the desired property. Furthermore, ML models are inherently independent of how the original input data sets were determined, i.e. by experiments or by simulations.

In this presentation, we use ML techniques to predict optimal chemical compositions for new hard-magnetic materials. The underlying data set was determined in a combinatorial high-throughput-screening approach based on density-functional theory calculations [Drebov et al., New J. Phys. 15 (2013); Körner et al., Sci. Rep. 6 (2016)]. The developed ML models allow us to predict promising structure-composition combinations for substitutes of state-of-the-art materials like Nd2Fe14B with similar intrinsic ferromagnetic properties but no or less amounts of critical rare-earth elements. Finally, we discuss possible perspectives for further applications of ML in materials science.

100% | Mobile Layout | Deutsche Version | Contact/Imprint/Privacy
DPG-Physik > DPG-Verhandlungen > 2017 > Dresden