Molecular dynamics (MD) simulations can offer entirely new insights into microstructure evolution, however, their accuracy critically depends on the quality of the interatomic potential. Traditionally, empirical potentials such as EAM and MEAM have been used to model atomic interactions. While these approaches capture basic properties of simple systems, they often fall short when applied to more complex materials. Recent advances in computer science have enabled the use of machine learning algorithms to develop interatomic potentials that aim to achieve density functional theory (DFT)-level accuracy in MD simulations.

In this study, we present a high-dimensional neural network potential (HDNNP) methodology, which utilizes atom-centered symmetry function descriptors [1]. We apply this approach to the extensively studied shape-memory alloy NiTi, which poses significant challenges at both experimental and theoretical levels. Our focus is on developing a neural network potential for the martensitic B19’ phase, capable of simulating complex microstructure evolution involving both twinning and plastic slip [2]. We demonstrate that the developed potential accurately captures the strong plastic slip anisotropy and that the HDNNP approach is well-suited for modeling the evolution of martensitic microstructures. In particular, we address the well-known discrepancy between the DFT-predicted ground state BCO structure and the experimentally observed B19’ phase by investigating the temperature-dependent stability of the BCO phase through molecular dynamics simulations using the newly developed NN potential.


Ref.
[1] Behler, J., 2011. Neural network potential-energy surfaces in chemistry: A tool for large-scale simulations. Physical Chemistry Chemical Physics 13, 17930–17955. https://doi.org/10.1039/c1cp21668f


[2] Seiner, H., Sedlák, P., Frost, M., Šittner, P., 2023. Kwinking as the plastic forming mechanism of B19′ NiTi martensite. International Journal of Plasticity 168, 103697. https://doi.org/10.1016/j.ijplas.2023.103697

This work was financially supported by the Ministry of Education, Youth and Sports of the Czech Republic in the frame of the project 'Ferroic multifunctionalities' (project No. CZ.02.01.01/00/22_008/0004591).