Molecular dynamics simulations were employed to study atomic motion within stationary and migrating asymmetric tilt grain boundaries. We employ several measures of the “complexity” of the atomic trajectories, including the van Hove correlation function, the non-Gaussian parameter, and dynamic entropy. There are two key types of dynamical events within the grain boundaries (i) a stringlike cooperative motions parallel to the tilt axis and occurring on a characteristic time scale of $\approx 25\mathrm{ps}$ and (ii) atomic motion across the grain boundary plane occurring on a characteristic time scale of $\approx 150\mathrm{ps}$. The characteristic times associated with each type of event decreases with increasing driving force for boundary migration. We present evidence as to how the driving force biases these types of events, leading to boundary migration. While the stringlike atomic motion is an intrinsic feature of grain boundary dynamics and is important for grain boundary migration, it is the second type of event that controls grain boundary migration rates.

• Received 4 May 2006

DOI:https://doi.org/10.1103/PhysRevB.74.115404