Scalable Methods for Electronic Excitations and Optical Responses of Nanostructures: Mathematics to Algorithms to Observables

Abstract

Kohn-Sham density functional theory (DFT) is a powerful, well-established tool for the study of condensed phase electronic structure. However, there are still a number of situations where its applicability is limited. The basic theme of our research is the development of first principles electronic structure approaches for condensed matter that goes beyond what can currently be done with standard implementations ofKohn-Sham DFT. Our efforts to this end have focused on two classes or' methods. The first addresses the well-lmown inability of DFT to handle strong, many-body electron correlation effects. Our approach is a DFT -based embedding theory, to treat localized features (e.g. impurity, adsorbate, vacancy, etc.) embedded in a periodic, metallic crystal. A description for the embedded region is provided by explicitly correlated, ab initio wave function methods. DFT, as a fo1n1ally ground state theory, does not give a good description of excited states; an additional feature of our approach is the ability to obtain excitations localized in this region. We apply our method to a first-principles study of the adsorption of a single magnetic Co ada tom on non-magnetic Cu( 111 ), a known Kondo system whose behavior is governed by strong electron correlation. The second class of methods that we are developing is an orbital-free density functional theory (OFDFT), which addresses the speed limitations ofKohn-Sham DFT. OFDFT is a powerful, O(N) scaling method for electronic structure calculations. Unlike Kohn-Sham DFT, OFDFT goes back to the original Hohenberg-Kohn idea of directly optimizing an energy functional which is an explicit functional of the density, without invoking an orbital description. This eliminates the need to manipulate orbitals, which leads to O(N{sup 3}) scaling in the Kahn-Sham approach. The speed of OFDFT allows direct electronic structure calculations on large systems on the order of thousands to tens of thousands of atoms, an expensive feat within Kohn-Sham. Due to our incomplete knowledge of the exact, universal energy density functional, this speedup comes at the cost of some accuracy with respect to Kohn-Sham methods. However, OFDFT has been shown to be remarkably accurate with respect to Kohn-Sham when used in the study of nearly-free-electron-like metals, e.g., AI, for which good density functionals have been derived. Examples of past applications of OFDFT include the prediction of properties of bulk crystals, surfaces, vacancies, vacancy clusters, nanoclusters, and dislocations, as well as OFDFT -based multiscale simulations of nanoindentation in AI and Al-Mg alloys.