Diffusion at Grain Boundaries
J. Harding and D. Harris
Department of Physics and Astronomy
University College London
W. Smith
CSE Department, Daresbury Laboratory
The project will use the biased-potential hyperdynamics method of
Voter [1] and kinetic Monte Carlo methods to study diffusion processes
with rates too low for conventional molecular dynamics to be used. The
hyperdynamics methods will be incorporated in the DL_POLY program [2]
developed at Daresbury laboratory. The kinetic Monte Carlo methods
will be written as a stand-alone program. These programs will then be
used to study diffusion problems in grain boundaries in ceramics.
A number of generic problems will be addressed. First, we will
obtain values for the effective width of a grain boundary for
diffusion (considering how it varies for different boundaries and
crystal structures). Second, we will calculate the variation of the
diffusion coefficient with boundary tilt angle. In particular, we will
consider how the anisotropy of the diffusion coefficient in the
boundary varies with tilt angle and how useful it is to consider the
boundary as an array of dislocations. Third, we will consider
variations of the diffusion coefficient with structure and, for a
given structure, the variation with ion size. Finally, we shall
calculate the effect of impurities on the diffusion rate and
mechanism. In particular, we shall calculate the diffusion rate of
dopants in boundaries.
References
[1] A.F. Voter, J. Chem. Phys. 112 (2000) 9599.
[2] T.R. Forester and W. Smith, The DL_POLY Molecular
Simulation Package, J. Molec. Graphics 14 (1996) 136.
