Unconstrained Global Optimization of Molecules on Surfaces: From globally optimized structures to scanning-probe data

The adsorption of molecules on a surface plays a vital role in heterogeneous catalysis. For a proper unterstanding of the reaction mechanisms involved, the adsorption ge ometry of the molecules on the surface needs to be known. So far, experimental data from tunneling microscopes and spectroscopy, such as STM and IRAS are the main ways to obtain such knowledge. Due to the vast search space of adsorption geometries, especially for oligomers, optimizations using ab initio methods can be used to confirm the experimental data only if good initial guesses are available. Global optimization can serve two purposes in these situations. On the one hand it allows for a thorough investigation of the given search space, which can provide good initial guesses for subsequent high-level structural refinements. On the other hand, given a known reaction mechanism, it could also be used to find catalysts that influence e.g. the relevant bonds. With respect to this idea the topic of this thesis is to find a local optimization method cheap enough such that the total computational cost of global optimization does not exceed availability and yet good enough that the results are meaningful to the problem at hand. With this in mind multiple force field and semiempirical methods have been tested and evaluated mainly on benzene, acetophenone and ethyl pyruvate on Pt(111) surfaces. Some other adsorbates have also been tested shortly. In addition to these global optimization results, DFT geometry optimizations of ethyl pyruvate on Pt(111) have been performed and the structures of the best adsorption geometry from global optimization and from DFT are compared. Furthermore, from the DFT data STM images have been calculated that are compared to experimental results. The theoretical and experimental STM images agree well.