Computer Simulations in Materials Physics: Time-scales and Accuracy

Doctoral thesis, 1996

Computer simulations are expected to play an increasingly important role within materials physics in the future. Owing to a combination of accurate physical approximations, improved implementations of these approximations, and the exponentially increasing power of computers, problems can now be solved on length and time-scales which were unimaginable only a decade ago. I discuss this development and suggest probable future trends. I support my arguments with several applications and calculations: an investigation of crystalline B2O3 highlights the importance of carefully considering the limitations of the underlying physical approximations. Work on the structure of .kappa.Al2O3 demonstrates the possibility of performing structural analysis of complicated structures which goes beyond the mere eproduction of known experimental data. Calculations using pseudo-atomic-orbitals illustrate the limitations of incomplete basis sets. Simulations of H diffusion in Pd using transition state theory show the feasibility of treating phenomena occurring on time-scales which are otherwise not immediately accessible to accurate atomic-scale simulations. Quantum Monte Carlo calculations on H diffusion on Ni(100) show that also low-temperature tunneling processes can be handled. Combining these results with current and recent work of others I conclude that the future of computational materials physics looks very bright indeed.