Comparative investigation of a newly optimized modified embedded atom method potential with other potentials for silicon

• Genetic algorithm is used to optimize the parameters of a potential for silicon.
• Better properties of surface, stacking faults and dislocation are reproduced.
• Suggestions for choosing suitable potential according to research targets are given.

Genetic algorithm (GA) is used to optimize the parameter set of the second nearest-neighbor modified embedded atom method (2NN MEAM) interatomic potential for silicon (Si). The optimization is carried out by tuning the parameters to match a set of physical properties including elastic constants, point defect formation energy, phase transformation, surface formation/relaxation and stacking faults. Besides the physical properties for optimization, other molecular dynamics (MD) predictions such as surface reconstruction, point defect diffusion, dislocations and thermal properties are also calculated to test the robustness of the new potential. Another purpose of this work is to compare various physical properties of available MEAM potentials for Si (and Tersoff and SW where necessary). It is shown that the new potential gives a better description in surface, stacking fault and dislocation. Finally, extensive discussion is given to specify the applicability of this potential and the validity of the potentials on fracture simulation.

Figure optionsDownload as PowerPoint slide