About this Abstract |
Meeting |
2022 TMS Annual Meeting & Exhibition
|
Symposium
|
Computational Thermodynamics and Kinetics
|
Presentation Title |
Data-driven Magnetic Materials Modeling; Advances in Classical Molecular Dynamics |
Author(s) |
Svetoslav Nikolov, Mitchell Wood, Aidan Thompson, Julien Tranchida |
On-Site Speaker (Planned) |
Svetoslav Nikolov |
Abstract Scope |
We outline a data-driven simulation method for constructing machine learned interatomic potentials that accurately capture both magnetic and phononic degrees of freedom in iron. This novel approach allows us to incorporate realistic spin effects, typically only accessible in costly first-principles models, into computationally efficient classical molecular dynamics. To test our framework, we examine the magnetoelastic, magnetostrictive, and thermal properties of alpha-iron, which serves as an analogue for any ferromagnetic material. We find that our MD simulations capture the experimental trends in the magnon/phonon thermal conductivities and elastic properties well, up to and above the Curie temperature. In our conductivity analysis, we probe the dominant heat carrying modes within the magnon/phonon subsystems and investigate how these are modified with changes in temperature and external magnetic fields. Using our novel simulation method we also examine how the magneto-crystalline anisotropy energy and magnetostriction coefficients vary across the alpha-phase. |
Proceedings Inclusion? |
Planned: |
Keywords |
Machine Learning, Magnetic Materials, Modeling and Simulation |