摘要

Action-derived molecular dynamics is applied to the simulation of self-diffusion processes on copper substrates. By minimizing a modified action with an energy conservation constraint, the method enables effective computations of minimum energy paths and activation energy barriers for the broad range of multiple timescale problems, including infrequent events and slow-mode systems. Single-adatom diffusions of hopping and exchange moves are first presented to demonstrate its performance. More complex diffusion mechanisms are simulated for hopping and exchange motions across a double-layer step on the Cu(111) surface, which are very difficult to explore by conventional molecular dynamics. Strain effects on diffusion energy barriers are also investigated for a Cu(001)flat surface. Finally, we propose an algorithm to incorporate a multiple length scale scheme into the current method, i.e., the combination of the action-derived molecular dynamics with the nonlocal quasicontinuum method. This hybrid scheme is expected to provide an efficient route to the simultaneous coupling of multiple length and timescales within a single algorithmic framework.