CAREER: Novel Green's function methods for predicting experimentally relevant quantities for solids and molecules

Regents Of The University Of Michigan - Ann Arbor, Ann Arbor MI

Investigators

Abstract

Dominika Zgid, of the University of Michigan, is supported by an award from the Chemical Theory, Models and Computational Methods program in the Chemistry Division to develop new computational tools to study large molecules and solids in which the correlated motion of electrons is very important. In modern science and technology, materials chemistry plays an big role in the production of advanced optoelectronic materials, semiconductors and superconductors, solar cell and battery materials. To enable the discovery of new materials and to answer experimental questions, theory has to predict experimentally relevant, measurable quantities. In the last fifty years, the majority of quantum chemistry research was focused on the development of methodological advances for molecular systems. Currently molecular problems can be described very accurately. However, for large strongly correlated molecules and solids, quantum chemistry still lacks computational tools that describe electronic correlation accurately in a systematically improvable manner and deliver experimentally useful predictions. The development of novel ab-initio theoretical methods that are at the interface of quantum chemistry and condensed matter physics and are capable of delivering useful experimental predictions for solids is the major aim of this research. This interdisciplinary project involves training and mentoring of graduate students and postdocs by allowing them to understand their research in the broadest possible sense. The research prepares them for a wide range of careers. Dr. Zgid is also actively engaged in public outreach for minorities in Science, Technology, Education and Mathematics (STEM) by organizing workshops for middle school girls. The Green's function language provides a natural link to experiment, since spectra can be readily calculated without the cumbersome excited state formalism present in wave function or density theories. Green's function methods are controlled, reliable, and systematically improvable and may easily be generalized by employing embedding methods to work for solids or large molecules. In order to calculate excitation spectra, this project implements the Bethe-Salpeter equation with a second order Green's function method and self-energy embedding approaches. The formalism is calibrated on small molecules and subsequently extended to solids by using embedding methods. Since the realism and predictive power of quantum mechanical simulations depend on the accuracy of modelling all electrons, significant attention is given to the investigation of effective Hamiltonian approaches that aim to make Green's function embedding methods quantitative for realistic molecular and crystalline systems. Finally, since the Green's function is a large object that can be calculated in parallel, the investigation focuses on efficient ways of expressing Green's functions in computer implementations. A major outcome of the project is software containing efficient, reliable and systematically improvable Green's function embedding methods for solids that is released to the public. Additionally, Dr. Zgid's research group is preparing a series of lecture notes for graduate students explaining Green's functions in order to reduce the language barrier frequently experienced by quantum chemists when working with the Green's function formalism. The proposed interdisciplinary research involves training and mentoring of graduate students and postdocs allowing them to understand their research in the broadest possible sense and prepares them for wide range of careers. Additionally, Dr. Zgid also takes part in the "Science for tomorrow" program for middle school students from underserved communities in Michigan.