Developed a method to create interatomic potential using neural network (NN) Perform ab initio molecular dynamics (AIMD) simulations to generate NN training data
Titanium Nitride (TiN) is used in a variety of applications due to its excellent wear and corrosion resistance. Ti N with a rock salt type crystal structure has been extensively studied, but has only recently been reported from first-principles calculations of the existence of non-rock salt type phases (eg Ti 2 N). Molecular dynamics (MD) simulations are powerful tools for predicting mechanical properties, but they usually require interatomic potential. The traditional many-body interatomic potential of rock salt TiN (eg modified embedded atom method (MEAM) potential) is not applicable to these other phases.
Therefore, we created a neural network potential (NNP) using a neural network (NN) -based method for creating an interatomic potential.
Regarding the mechanical properties of TiN, when the MD simulation (NNMD) using the created NNP and the MEAM potential were compared with the experiment, the result of NNMD was closer to the experimental value. NNMD also revealed that the strength and brittleness of other phases (Ti2N) are high.
”Development of neural network potential for MD simulation and its application to TiN”
T. Miyagawa, K. Mori, N. Kato, A. Yonezu, Comput. Mater. Sci , 206, ( 2022 ), 111303
Original Source from:https://ctc-mi-solution.com/ニューラルネットワークポテンシャルnnpを活用し/