Title Exciton self-trapping in low-dimensional organic metal halide hybrid materials from GW/Bethe-Salpeter calculations and machine-learning-based force fields

Dr. David A. Strubbe¹, Associate Professor

Co-PI(s): Dr. Liang Z. Tan²

1: University of California, Merced, Merced, CA 95343

2: Lawrence Berkeley National Laboratory, Berkeley, CA 94720

Organic metal halide hybrid perovskites (OMHHs) are of great interest for solar cells and light-emitting diodes (LEDs). These tunable materials show excellent light absorption in thin films, high carrier mobility, easy charge separation for current extraction, and good defect tolerance. A serious problem, however, is the typical rapid degradation under light, heat, and moisture. While initial research interest was on bulk 3D perovskites, recent research has moved toward low-dimensional materials in which the inorganic component forms sheets, wires, or isolated nanocrystals, because they have reduced degradation. They also show an intriguing phenomenon of strong exciton self-trapping, in which the electron and hole produced after absorption of light deform the surrounding atomic structure and become localized. This phenomenon, which does not occur in traditional hard inorganic semiconductors like silicon, leads to emission of white light which can be used for LED devices. A detailed understanding of exciton self-trapping is key for the emerging area of soft semiconductor science, but it remains so far elusive. There have been few theoretical tools to investigate the early-time behavior after absorption of light. However, recent work on excited-state forces in the GW/Bethe-Salpeter equation (GW/BSE) approach allows a single efficient calculation to give the direction that each atom will move after absorption of light and opens the prospect of accurate calculations of these phenomena. This project comprises three research thrusts:

1. GW/BSE calculations of self-trapped excitons,
2. Parametrization of machine-learning-based force fields for excited states, and
3. Study of exciton structure and transport on longer length scales via force fields.

Results will be compared to experimental data from collaborators. Recent work at UC Merced has provided a practical software implementation for excited-state forces and demonstrated the accuracy of the method. The GW/BSE method will provide accurate electron-hole interactions, and the calculations will be used to parametrize novel force fields that describe the excitation energy. Molecular dynamics with force fields will be used to study longer length and time scales. This work will be enabled by expertise at LBNL’s Molecular Foundry on OMHHs and machine-learning force fields and will build on previous collaborations. The project will contribute to the fundamental understanding of excitations in soft semiconductor materials, advance the methodology of excited-state structural dynamics and machine-learning force fields, and inform the design of materials for LEDs. Implementation of new methods will be provided in the widely used open-source code BerkeleyGW (which is DOE-supported) and presented in tutorials for broad availability to the community. The project will build research capacity for further work in this area and enable proposals for future BES funding. Graduate student collaboration with LBNL and summer research at the Molecular Foundry will build the research culture of UC Merced, an institution only 19 years old and still developing.

This research was selected for funding by the Office of Science – Basic Energy Sciences

_____________________________________________________________________________________