On Simulating Ultra-fast Chemical Processes and their Spectroscopic Signatures

This Ph.D. thesis explores different theoretical methods to describe chemical processes via X-ray spectroscopy with an emphasis on X-ray absorption spectroscopy (XAS). This includes method development, comparisons of methods, and analysis of experimental data. The thesis reviews the theory for describing electronic structure and nuclear dynamics, and the properties of interest are presented. These rely on a single molecular geometry, which is determined by using a molecular gradient. An efficient implementation of such a gradient has been carried out as part of the presented work at the coupled cluster singles and doubles (CCSD) level of theory for both ground and excited states. Thus, the corresponding equilibrium structures can be determined. Also, the implementation lays the ground work for describing nuclear dynamics at the CCSD level of theory. Nuclear dynamics is here simulated with time-dependent density functional theory (TDDFT) using, e.g., trajectory surface hopping (TSH), where nuclei are treated with classical mechanics. This gives a qualitatively good description of the nuclear dynamics, even for a limited number of trajectories. A TSH calculation is also found to be a good basis for a quantum dynamics simulation. The CCSD method was employed to calculate different static X-ray spectra. It showed an overall good agreement with experiment for, e.g., XAS, X-ray photoelectron spectroscopy and resonant inelastic X-ray scattering. This indicates that the method yields a good description of the studied processes. The CCSD method was, for instance, used to analyse experimental ground state spectra of the heterocycle oxazole. This is a first step towards analyzing future time-resolved XAS (tr-XAS) experiments of the system. In the present study, TDDFT was also employed to calculate XAS spectra. Here, this less computationally expensive method turned out to yield results of similar accuracy to CCSD. This was also noticed in the analysis of experimental tr-XAS data for a larger molecule. Moreover, simulations of tr-XAS based on the above mentioned methods, showed that for rigid molecules, tr-XAS can be simulated based on optimized geometries alone. Indeed, when the molecule remains in the same electronic state for a long time, one might even avoid a nuclear dynamics simulation. Several states of interest will, however, require such a simulation to predict the evolution of state populations. These conclusions must be further investigated for less rigid molecules.