Development of Optimized Buckingham Potentials for Accurate Molecular Dynamic Simulations of Stishovite Silicon Dioxide Using the Niipfit Program

Silicon dioxide (SiO2) has emerged as a highly promising material that is currently under extensive investigation due to its remarkable capability to facilitate the generation of nanostructures, which are of significant interest in various scientific and technological applications [1]. In order to effectively conduct molecular dynamics simulations that involve the intricate properties and behaviors of SiO₂, it is imperative to accurately fit the potentials that characterize this particular material. However, it is regrettably the case that the existing literature provides limited information concerning the parameters A + R of SiO₂ when utilizing the DL_POLY simulation code [2]. Consequently, this limitation necessitated the development of our own set of potentials specifically tailored for these materials. To successfully achieve this objective, a total of nine pairs of parameters are generally required, yet it is noteworthy that the interatomic pair parameters for the oxygen–oxygen (O-O) interactions had already been fitted based on the established TiO2 potentials, which allowed us to focus on fitting only six pairs of parameters instead. More specifically, we concentrated our efforts on the short-range parameters, which include the values of {A_ij, ρ_ij, and C_ij} associated with each of the two distinct atomic pair interactions, namely Si-Si and Si-O. The reference crystal structure that served as the basis for our investigation of SiO₂ is the tetragonal stishovite structure, which is characterized by the space group notation p42/mnm. We sourced the necessary values for the elastic constants, lattice constant, bulk modulus, and static dielectric constant from a reputable study conducted by Cem Sevik and Ceyhun [3], while the cell parameters were obtained through the use of the MedeA software platform. It is worth mentioning that the process of manually adjusting the values of Aij, ρij, and Cij using the GULP program [3] poses significant challenges, making it a complex task to derive.