Dresden 2026 – scientific programme
Parts | Days | Selection | Search | Updates | Downloads | Help
MA: Fachverband Magnetismus
MA 13: Magnetic Heuslers and Semiconductors
MA 13.5: Talk
Monday, March 9, 2026, 16:00–16:15, POT/0361
Modelling the martensitic transformation in Ni-Mn-based Heusler compounds with machine-learning force-fields — •Markus E. Gruner, Mike J. Bruckhoff, and Olga Miroshkina — Faculty of Physics and CENIDE, University of Duisburg-Essen, D-47048 Duisburg
Functional properties of Ni-Mn-based Heusler alloys, such as Ni2MnGa, depend on the presence of hierarchically twinned, modulated structures in the martensite phase. These can be interpreted as an adaptive, self-organized arrangement of [101]-aligned nanotwins consisting of non-modulated tetragonal building blocks. Density functional theory (DFT) suggests that these martensites are connected to cubic austenite via a downhill transformation path. This is owed to a Fermi surface reconstruction, softening the corresponding acoustic phonons of austenite. Modeling free energy surfaces at finite temperatures or the dynamics of the martensitic transition as probed in recent ultrafast laser heating experiments [1] is too expensive for conventional DFT. Machine-learning force fields (ML-FF) trained on DFT data offer a possibility to account for electronic instabilities and allow to explore the impact of adaptive nanotwining on the martensitic transition using ML-FF in classical molecular dynamics simulations. Funding by the DFG via TRR270 (B06), SFB1242 (C02) and MI3273/1 is gratefully acknowledged.
[1] Y. Ge, F. Ganss, D. Schmidt, D. Hensel, M. J. Bruckhoff, S. Sadashivaiah, B. Neumann, M. Brede, M. E. Gruner, P. Gaal, K. Lünser, S. Fähler, arXiv:2509.06513.
Keywords: magnetic shape memory effect; magnetocaloric effect; martensitic transformation; density functional theory; machine-learning force-fields