Abstract
The migration of point defects, e.g. crystal lattice vacancies and self-interstitial atoms (SIAs), typically occurs through three-dimensional (3-D) random walk. However, when vacancies and SIAs agglomerate with like defects forming clusters, the migration mode may change. Recently, atomic-scale computer simulations using molecular dynamics (MD) codes have reported that nanometer-sized two-dimensional (2-D) clusters of SIAs exhibit one-dimensional (1-D) fast migration1-7. The 1-D migration mode transports the entire cluster containing several tens of SIAs with a mobility comparable to single SIAs3. This anisotropic migration of SIA clusters can have a significant impact on the evolution of a material�fs neutron-irradiation damage microstructure, which dominates the material�fs lifetime in nuclear reactor environments8-9. This is also proposed to be a key physical mechanism for the self-organization of nanometer-sized sessile vacancy cluster arrays10-13. Given these findings for SIA clusters, a fundamental question is whether the 1-D migration mode is also possible for 2-D clusters of vacancies. Preceding MD results predicted that 1-D migration of vacancy clusters is possible in body-centered cubic (bcc) iron, but not in face-centered cubic (fcc) copper2. Previous experimental studies have reported 1-D migration of SIA clusters14, but there have been no observations of 1-D vacancy cluster migration. Here we present the first experimental transmission electron microscopy (TEM) dynamic observation demonstrating the 1-D migration of vacancy clusters in fcc gold. It was found that the mobility of the vacancy clusters via the 1-D migration is much higher than single vacancies via 3-D random walk and comparable to single SIAs via 3-D random walk. Hence, the mobility of the glissile clusters is not associated with the character of their constituent point defects. Dynamic conversion of a planar vacancy loop into a 3-D stacking fault tetrahedron geometry was also observed.
