Developing Neural Network Machine Learning Interatomic Potentials for Molecular Dynamics Simulations of Complex Ternary Materials

While molecular dynamics (MD) is a very useful computational method for atomistic simulations, modeling the interatomic interactions for reliable MD simulations of real materials has been a long-standing challenge. We show that an artificial neural network machine learning (ANN-ML) scheme opens a new paradigm for developing accurate and efficient interatomic potentials for reliable MD simulation studies of the thermodynamics and kinetics of materials. Using ternary La-Si-P system as a case of study, we demonstrated that accurate and transferable interatomic potentials can be developed for reliable MD simulations of complex materials. The developed ANN-ML potential accurately describes not only the energy versus volume curves for all the known elemental, binary, and ternary crystalline structures, but also the structures of La-Si-P liquids with various compositions. Using the developed ANN-ML potential, the melting temperatures of several crystalline phases in La-Si-P system are predicted with good agreements with experiments. The developed ANN-ML potential also enables large-scale atomistic simulations to study the complex nucleation and growth kinetics of new materials.