| This dissertation involves development, testing, and application of an embedded correlated wavefunction (ECW) theory, which has been in development in the Carter group for over a decade. Electronic structure methods are valuable tools for predicting properties of molecules and materials. In this work, we aim to accurately model local features in crystals for which a mean field description such as density functional theory (DFT) does not capture the physics of either the feature and/or the phenomenon. Although DFT has proven to be rather accurate for describing many materials, situations exist where it fails to provide even a qualitatively correct picture, e.g., strong electron-electron correlations, excited states, and open-shell systems. In the ECW theory, the total system is partitioned into a cluster of atoms containing the feature of interest, treated with CW theory, and a periodically infinite background region, treated with DFT. The effect of the background on the cluster is represented by an embedding potential derived from orbital-free DFT. With a pseudopotential-based version of the ECW theory, we study the adsorption of carbon monoxide (CO) on Cu(111) and Pt(111), well-known systems where standard DFT fails to predict the correct binding site preference and overbinds CO to the surface. The ECW predictions for the binding site preference are consistent with experimental finding for CO on both surfaces. Predicted binding energies for CO on Cu(111) are in excellent agreement with experiment, in contrast to DFT and non-embedded CW theory. We then develop a new all-electron implementation of the ECW method and apply it to the study of cobalt (Co) adatoms on coinage metal surfaces, systems that display the surface Kondo effect. Here, the spin of a magnetic impurity within a metal antiferromagnetically couples to the conduction electrons, creating a resonant state and producing anomalous low temperature behavior. This type of interaction arises from strong correlation effects that cannot be properly described by a mean field theory. The ECW predicted Co d-electronic structure, combined with earlier predictions for Co on Cu surfaces, provides an explanation of trends in tunneling behavior observed with scanning tunneling spectroscopy. Finally, we examine excited state energetics within bulk magnesium oxide as predicted with the all-electron ECW method and embedded CW methods that employ a classical description of the background crystal, in order to determine the best approach for the description of excited states within metal-oxide crystals. Preliminary results indicate that both CW treatment and embedding are necessary to describe the ionic crystal accurately, and that a classical point charge description of the background is superior to a DFT-based embedding potential. We also discuss some extensions of the ECW method that were considered but not pursued. |
