Modeling amorphous silicon nitride: A comparative study of empirical potentials

A Dasmahapatra and P Kroll, COMPUTATIONAL MATERIALS SCIENCE, 148, 165-175 (2018).

DOI: 10.1016/j.commatsci.2017.12.008

We perform a comparative study of empirical potentials for atomistic simulations of amorphous silicon nitride (a-Si3N4). We choose 5 different parameterizations of the Tersoff potential, the Marian- Gastreich two-body (MG2) and three-body (MG3) potential, the Vashishta (V) potential, and the Garofalini (SG) potential. Amorphous models of Si3N4, comprising of 448 atoms, are generated by each empirical potential using a melt-and-quench procedure. Subsequently, models are optimized using Density Functional Theory calculations, and structures resulting from these DFT optimizations are compared. We emphasize local coordination of atoms and the enthalpies of formation (Delta H-f) relative to crystalline beta-Si3N4. The SG and MG2 potentials prove to be best options for modeling of a-Si3N4. Models generated with these potentials are close to their DFT local minimum, exhibit the smallest number of defects, and have realistic enthalpies of formation. (C) 2017 Elsevier B.V. All rights reserved.

Return to Publications page