Research Objectives
We are studying, at a microscopic level, the damage produced in metal irradiated by an energetic atom, and how this damage is distributed into vacancies and interstitials. In particular we are interested in the clustering of the different defects after the irradiation and their effects in the later evolution of these defects under annealing of the material. Our final goal is to understand how microstructural changes occur under irradiation conditions.
Computational Approach
We use a classical molecular dynamics code to study radiation damage in metals. Our molecular dynamics code has been implemented to be used on the Cray T3E. It uses standard PVM for message passing, and the programming language is Fortran 90. This code was previously running on a Cray T3D and was easily implemented to run on the T3E.
Accomplishments
From the simulations performed on the T3E we are extracting a data base of the number and distribution of the vacancies and interstitials produced by different energetic ions. This data base will be used as the input for a later Monte Carlo simulation of defect diffusion at elevated temperatures, facilitating the study of defect clustering and, therefore, microstructure evolution and mechanical property changes. Other parameters needed for this Monte Carlo simulation have been extracted using the parallel molecular dynamics code, such as vacancy and interstitial diffusivities and binding energies of vacancies and interstitial clusters.
In particular, we have studied the damage produced by self-irradiation of lead with energies between 10 keV and 30 keV, both in the bulk of the material and at the surface. The low melting point of this metal, together with its low thermal conductivity and softness, results in very large cascades produced by the energetic atoms. In order to model this damage in a realistic manner with molecular dynamics, it is necessary to use simulation boxes that include between 250,000 and 106 atoms. A typical simulation of 30 keV Pb irradiation on Pb requires a simulation box with 60 lattice units in each direction (864,000 atoms). The forces on each of these atoms are calculated at every time step and the location of the atoms updated according to those forces. Each time step is on the order of femtoseconds, and the total simulation time is typically of 50 picoseconds, due to the large relaxation times for this material. The CPU time for a time step using 64 processors and 864,000 atoms is approximately 3.5 seconds. Therefore, a total of 48 CPU hours are needed to complete each one of these large cascades.
Significance
We are now performing Monte Carlo simulations of defect diffusion using the results from the parallel molecular dynamic code. These results provide a physically based model for defect production and migration. These simulations will increase understanding of the controversial problem of void swelling in metals.
 |
Vacancies (light color) and interstitials (dark color) produced by a 30 keV Pb in Pb. Observe the large interstitial clusters produced. The interstitial clusters are forming dislocation loops as can be observed by looking at a single plane on the lattice. |
| |
|