Collaborative Research: SI2-SSI: Software Framework for Electronic Structure of Molecules and Solids

California Institute Of Technology, Pasadena CA

Investigators

Abstract

Many traditionally experimental disciplines such as chemistry and materials science are rapidly changing due to our increasing ability to predict properties of molecules and materials purely by simulation. This is particularly true when molecules meet solid surfaces - due to the particular challenges of experiments in such a setting. Yet the molecule-on-surface frontier encompasses a vast class of problems of tremendous practical importance: industrial applications facilitated by surface processes are estimated to produce globally more than than 15 trillion USD worth of goods and products. This research will improve our ability to simulate the physics and chemistry of molecules on surfaces by extending the advanced simulation methodologies that were originally developed by for modeling electrons in molecules. This project will not only advance our fundamental understanding of the surface science but also open a road to technological applications relevant to producing and storing clean energy and in designing improved catalysts. The research may result in a new computer software framework for simulating electrons in molecules and materials. This software will be a unique contribution to the U.S. cyberinfrastructure and spur further innovation by other researchers in the US and worldwide, who will be able to access its source code for free. The software framework will also serve as an education platform for training computational chemists and materials scientists. A frontier simulation challenge lies at the intersection of the two domains of chemistry and materials science - namely to determine, with predictive accuracy, the properties and chemistry of molecules on solid surfaces. The molecule-on-surface frontier encompasses a vast class of problems of tremendous practical importance: heterogeneous catalysis, photovoltaics, and emerging electronic materials. Yet, from a simulation perspective, it is not currently possible to efficiently combine the recent advances in highly accurate many-body molecular and periodic condensed phase methodologies in these problems, due to a significant gap between how the electronic structure theories of molecules and materials are formulated, as reflected in distinct algorithms and disjoint codebases. The goal of this project is to reduce and/or completely eliminate the gap between molecular and solid-state electronic structure methodologies, in theory, algorithms, and in usable community software implementations. This will be achieved by building an ambitious Electronic structure for Molecules and Solids (EMOS) software framework that will permit accurate computation of the first-principles electronic structure of both molecules and solids on an equivalent footing - and with the high efficiency necessary for high-throughput screening or ab initio molecular dynamics. These efforts build on the leading track-record of the principal investigators in developing open-source quantum chemistry software as well as automated computer implementation and high-performance parallel libraries. The project will allow the advances from molecular electronic structure - embedding, reduced-scaling many-body methodology, accurate excited-state electronic structure, and others - to be applied routinely to molecules, materials, and combinations of the two as relevant to surface chemistry. This has great potential to advance the state-of-the-art in treatment of electronic structure and open new lines of theoretical inquiry. The resulting open-source production-quality toolkit will be validated against experimental data for a host of surface phenomena, from exciton dynamics to surface spectroscopy and catalysis. An open-source US-based advanced materials code is a long-standing omission in U.S. cyberinfrastructure. As a high-performance framework for simulation of electronic structure of molecules, solids, and their interfaces with unprecedented accuracy, EMOS will be a significant contribution to this effort. Further, the modular component based structure will be able to be integrated with other major electronic structure packages through the reuse of the modules. This project will provide invaluable training opportunities to the students and postdocs who will develop the software framework under the direct supervision of principal investigators. In addition, each project site will contribute to the development of a stakeholder network for EMOS by hosting, each summer, visiting students and faculty representing the broader theoretical community, to train them on the use of EMOS in research and education. The project team will also use EMOS in teaching classes and summer schools, building on already established efforts in this area; these efforts will also be extended to an online setting.