Grain-boundary diffusion in metals

The available models for grain-boundary diffusion are reviewed together with the relevant mathematical techniques for extracting the grain-boundary diffusivity from the various types of experimental data. The more useful experimental techniques are described, fo11owed by a presentation of results that shed light on the models and mechanisms of grain-boundary diffusion. A ‘pipe’ mechanism of diffusion based on the well established dislocation model seems most appropriate for small-angle boundaries. Open channels which have atomic configurations somewhat like dislocation cores, probably play a major role in high-angle grain-boundary diffusion. Dissociated dislocations and stacking faults are not efficient paths for grain-boundary diffusion. The diffusion and computer-modelling experiments are consistent with a vacancy mechanism of diffusion by a rather well localized vacancy. The atomic migration is a three-dimensional process with jumps away from, and back towards, the dislocations as well as jumps along the dislocation core. The effective width of a boundary for grain-boundary diffusion is about two atomic planes. The limited data on impurity effects on grain-boundary diffusion are discussed in term s of impurity segregation to grain boundaries. Diffusion along moving grain boundaries appears to be several orders of magnitude greater than along stationary boundaries. Additional studies of moving boundaries are required before detailed mechanisms for rapid diffusion can be identified.