Two-body potential model based on cosine series expansion for ionic materials

[Display omitted] •Two-body potential composed by a cosine series for ionic materials is proposed.•The cosine coefficients are determined by fitting to energy/force/stress of DFT.•The constructed potential model for MgO works nicely in MS and MD simulations.•This method can systematically generate a...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Computational materials science 2016-01, Vol.111 (C), p.54-63
Hauptverfasser: Oda, Takuji, Weber, William J., Tanigawa, Hisashi
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:[Display omitted] •Two-body potential composed by a cosine series for ionic materials is proposed.•The cosine coefficients are determined by fitting to energy/force/stress of DFT.•The constructed potential model for MgO works nicely in MS and MD simulations.•This method can systematically generate a quality potential for ionic materials. A method to construct a two-body potential model for ionic materials with a Fourier series basis is examined. In this method, the coefficients of cosine basis functions are uniquely determined by solving simultaneous linear equations to minimize the sum of weighted mean square errors in energy, force and stress, where first-principles calculation results are used as the reference data. As a validation test of the method, potential models for magnesium oxide are constructed. The mean square errors appropriately converge with respect to the truncation of the cosine series. This result mathematically indicates that the constructed potential model is sufficiently close to the one that is achieved with the non-truncated Fourier series and demonstrates that this potential virtually provides minimum error from the reference data within the two-body representation. The constructed potential models work appropriately in both molecular statics and dynamics simulations, especially if a two-step correction to revise errors expected in the reference data is performed, and the models clearly outperform two existing Buckingham potential models that were tested. The good agreement over a broad range of energies and forces with first-principles calculations should enable the prediction of materials behavior away from equilibrium conditions, such as a system under irradiation.
ISSN:0927-0256
1879-0801
DOI:10.1016/j.commatsci.2015.08.055