Efficient Density-Functional-Based Calculational Methods for Surfaces
During the last decades, computer simulations have become an important tool for the study of elementary surface processes, such as atomic and molecular adsorption, diffusion of ad-atoms and surface recombination. The knowledge gained by such studies can be used to understand and develop technologically important processes such as catalysis and crystal growth. Thanks to the exponentially increasing power of computers and the rapid development of efficient numerical algorithms, it is now possible to solve the equations of density functional theory (DFT) for realistic surfaces. This has opened up the possibility of calculating potential energy surfaces that governs elementary surface processes entirely from first principles. Moreover, DFT calculations allows us to predict chemical and physical properties of surfaces.
In the thesis, I discuss how a particular implementation of DFT, the plane-wave pseudopotential method, can be used for calculating properties of "real-world" surface processes. Several applications are discussed: adsorption and recombination of atomic hydrogen on copper surfaces, adsorption of molecular hydrogen on stepped copper surfaces, and the growth of the kappa phase of aluminium oxide on a titanium carbide substrate. A few noteworthy results, which are supported by both experimental and theretical studies, are the following: A hydrogen molecule that is adsorbed on a kink atom at a stepped copper surface is found to rotate in a two-dimensional manner. For instance, this unusual quantum rotor state provides a key ingredient for preferential adsorption of ortho hydrogen at this surface. In the study of growth of aluminium oxide using chemical vapor deposition, the previously unknown atomic structure of the kappa phase is determined, and a we present a simple argument for why growth of the meta-stable kappa phase is favoured on the titanium carbide substrate.
I also present details about the numerical algorithms and optimization techniques that I have implemented in the DFT code in order to make efficient use of modern computer architectures and reduce the execution time of large simulations.
potential energy surfaces
density functional theory