Nonequilibrium Interface Dynamics:
Fundamental Physical Issues in Nonequilibrium Interface Dynamics

CSIC Building (#406), Seminar Room 4122.
Directions: home.cscamm.umd.edu/directions

Towards a Realistic Approach in Atomistic Studies of Diffusion and Fluctuations on Metal Surfaces

Dr. Talat S. Rahman

Department of Physics at Kansas State University

Abstract:   In atomistic studies of nanoscale phenomena the kinetic Monte Carlo (KMC) technique appears to have an edge over the standard molecular dynamics (MD) method because of the limitations of length and time scales in the latter. The KMC method, however, requires a prior knowledge of diffusion mechanisms, their activation energies, and their pre-exponential factors. To make the KMC method more self sufficient, we have recently introduced two complementary approaches which rely on specific pattern recognition techniques. In one a minimal recognition scheme together with a large number of predetermined peripheral diffusion processes is used, while in the other a more extensive recognition scheme is supplemented with the provision to calculate activation energy barriers and diffusion paths on the fly. In the latter, the system thus builds its own data base of the energetics of possible single and multiple particle diffusion processes, as it evolves over time. The application of these two techniques to several diffusion and fluctuation related processes on Cu(111) will be presented. Usage is made of interaction potentials from the embedded atom method for the calculation of activation energy barriers. Particular emphasis in this talk will be on the diffusion of small two-dimensional Cu clusters on Cu(111). Results from the first method involving only single atom peripheral motion very interesting behavior. Clusters containing 6n+3 atoms, ([(n*(n+1)/2)*6]+1) atoms, which form a compact hexagon, as well as, a compact hexagon+1 are found to have much less diffusivity as compared to the others. The peculiarities in the motion of these "magic" sized clusters, and the scaling of the diffusion coefficient of the others with size will be discussed. Comparisons will be made with the results obtained from the time evolution of the same system using the open data-base, "self teaching" method. The relative frequencies of the events appearing in the two KMC simulations will provide the basis for judging the importance of many-particle processes and the ability of the system to sample a multitude of diffusion mechanisms.

*Work supported in part by NSF under grant EEC-0085604.