Open AccessA method of accelerating molecular dynamics simulation for atomic diffusion in metallic interface
Author(s) -
XianMing Zhang,
Lihong Yang,
Wu Yong-Quan,
Tong Shen,
Zheng Shaobo,
Guilin Jiang
Publication year - 2008
Publication title -
wuli xuebao
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.199
H-Index - 47
ISSN - 1000-3290
DOI - 10.7498/aps.57.2392
Subject(s) - molecular dynamics , diffusion , embedded atom model , potential energy , nanosecond , saddle point , interface (matter) , statistical physics , lennard jones potential , computer science , materials science , chemical physics , physics , mechanics , chemistry , thermodynamics , computational chemistry , classical mechanics , optics , laser , geometry , mathematics , bubble , maximum bubble pressure method
Many interesting long-time dynamic properties of solid interface, such as deep diffusion, pervasion and phase forming, cannot be simulated directly using traditional molecular dynamics (MD) because of nanosecond timescale limitations. Thus, a simpler bias potential form has been proposed within the Voter's hyper dynamics scheme. In this method, the potential energy wells are raised by adding a coefficient, which was defined as the accelerating factor, to the original potential. So, the escape rate from potential wells was enhanced, which extends the timescale by several orders of magnitude comparing to the traditional MD simulations. What's more important, the features of potential surface are reserved even without any in_advance knowledge of the location of either the potential energy wells or saddle points. We demonstrate this method by applying it to the mutual diffusion of atoms in Mg/Zn interface with different accelerating factors using a simple Lennard-Jones potential. The results showed that long-time MD simulation can be realized very easily by our approach.