Usage of pattern recognition scheme in kinetic Monte Carlo simulations: Application to cluster diffusion on Cu(1Â 1Â 1)

We have invoked a simple pattern recognition scheme in kinetic Monte Carlo simulations of post-deposition evolution of two dimensional islands on fcc(1 1 1) surfaces. On application of the technique to the diffusion of small Cu clusters (8-100 atoms) on Cu(1 1 1) we find that, at room temperature, clusters with certain magic numbers show stick-slip type of motion with striking patterns rather than the random paths followed by the others. At higher temperatures all clusters display random motion. The calculated diffusion coefficients show dependence on size and temperature with an effective barrier ranging between 0.62 eV and 0.84 eV. Small asymmetries in diffusion barriers lead to a large difference in the frequencies of adatom diffusion along the two types of micro-facetted steps on Cu(1 1 1) leading to consequences in their shape evolution. The pattern recognition scheme revealed 49 basic periphery single atom diffusion processes whose activation energy barriers were calculated using the nudged elastic band technique and interatomic potentials from the embedded atom method.