Modelling Magnets: From Atoms to Bulk Properties



Presenter: Thomas Schrefl
Title: Modelling Magnets: From Atoms to Bulk Properties
Affiliation: Christian Doppler Laboratory for Magnet design through physics informed machine learning, University for Continuing Education Krems, Austria
Date: 16.11.2023
Time: 17:00 h
Place: F-SR-III


Contents of the Talk

Permanent magnets, essential for power generation and transformation, play a crucial role in transitioning to a carbon-free economy. The most powerful magnets today contain rare-earth elements such as Neodymium and Dysprosium. The scarcity of these critical elements, research activities focus on developing magnets with minimized reliance on critical raw materials, while simultaneously tailoring properties for specific applications. The material design process requires an in-depth understanding of how the atomic and granular structure of the magnet influences its hysteresis properties. The analysis of experimental data and simulation results helps to identify the relevant physical processes, thereby facilitating machine learning assisted design of magnetic materials. In my talk, I will cover simulation techniques covering atomistic simulations, finite element micromagnetics, and reduced order modelling of hysteresis properties. Reliable model prediction requires considering experimental data sets. I will discuss challenges and potential methodologies for merging experimental data with simulations and theoretical models.


Short CV

Thomas Schrefl is the head of the Center for Modelling and Simulation at the University for Continuing Education in Krems, Austria. He received his PhD from TU-Wien in 1993 and habilitated in “Computational Physics” in 1999. He has worked at IMB Research on parallel solvers for micromagnetic problems and served as a Professor of Functional Materials at the University of Sheffield. Currently, he leads the Christian Doppler Laboratory for Magnet Design through Physics-Informed Machine Learning. He has applied numerical micromagnetic simulations to address design questions in magnetic recording, magnetic sensors, and permanent magnets. His current research interests include the use of machine learning in materials science. He is an author or co-author of more than 340 publications.

Further details.