Materials processing: unleashing the power of computer simulations

Abstract

In the past ten years, the status of computer modeling has changed dramatically. Increases in computer performance have made it possible to solve difficult materials problems, including first-principles calculations of semiconductor surface properties, and finite element simulations of the diffusion of impurities in three-dimensional semiconductor structures. However, in addition to these applications of advanced supercomputers, other avenues have opened up that may have equally important consequences. It is now possible for groups and individuals on modest budgets to solve challenging problems and perform technologically useful research. Two main factors are responsible for this. First, PCs and workstations now cost only a few thousand dollars and provide significant computational power together with graphics. Second, advances in communications are making it possible for researchers in remote locations to have access to the latest ideas, software and experimental data. The accessibility afforded by Internet makes possible close collaborations between scientists using models that span the range from the macroscopic scale to electronic structure calculations. The free exchange of computer codes, model parameters, data and ideas can greatly increase the ability to understand phenomena, and to develop models with predictive capability. This type of computer modeling can be matched to industrial needs for understanding and predictions. We propose the implementation of an electronic bulletin board to provide a link to industry and between these scientists. The conditions are ripe for the growth of a cottage industry of scientists performing computations in support of industrial research and development. These ideas are presented strictly as an individual viewpoint.