Density Functional Theory augmented Multiplet Ligand Field Theory Applied to X-ray Spectroscopy of 3d Transition Metals

Abstract

The usual flow of new information within science is such that experiments are used to refine our understanding of the world, with theory playing a supporting or interpretive role. The rarer occurrence comes from when theory is used to motivate new directions for scientific exploration. The difficulty of course comes from the need for a theory that can provide predictions for novel phenomena that are sufficiently accurate and precise enough to merit experimental efforts. Here I explore methods for improving the standard multiplet ligand field theory (MLFT) model of x-ray spectroscopy with density functional theory (DFT), with the goal of making it more ab-initio so that it can be used for predictive applications instead of just interpretive. The main improvements come from reducing the number of free parameters that are used when fitting an MLFT calculated spectra to experiment, namely the Slater-Condon scaling, crystal field, and ligand hopping parameters. First, I apply the DFT + MLFT framework to core-to-core X-ray emission spectroscopy (CtC-XES), exploring how the charge transfer dynamics are affected by the core-hole in the intermediate and final states. In this work, I surveyed 8 different 3d transition metal systems across the periodic table and analyze how calculated spectra reproduce trends across peak width, spin state, and integral intensity to conclusively demonstrate the accuracy of this approach. Next, I utilized this framework to explore a resonant shake effect that manifests in 3d0 materials when the bonding-antibonding splitting matches the 2p spin-orbit splitting. Through this resonance I was able to predict and later experimentally verify a new spectral feature within the Kα XES of PbCrO4, which too my knowledge is the first ever example of multiplet theory being used to inform new experiment. Finally, I switched focus to studying how the linear polarization x-ray emission of single crystal systems varies depending on local geometry and chemistry. I compared the information contained both a core-to-core level where the local anisotropy was transmitted through Coulomb coupling between the core and valence states, with valence-to-core x-ray emission spectroscopy (VtC-XES) where the occupied density of states directly reflects the local anisotropy. This project was also extended to a study of how the anisotropy in polarized VtC-XES could be reproduced through supervised machine learning applied to a large dataset of crystal structures. This model highlights correlations between key chemical and geometric indicators such as the normalized quadrupole moment, and provides a continuous, quantitative method for characterizing spectral anisotropy.