Excited-State Geometry Optimization of Small Molecules with Many-Body Green's Functions Theory.

Area of Science:

• Computational Chemistry
• Quantum Chemistry
• Theoretical Chemistry

Background:

• Accurate prediction of molecular geometries in excited states is crucial for understanding photochemistry and photophysics.
• Many-body Green's function methods, particularly GW-BSE, offer a promising route to describe excited-state properties.
• Benchmarking these methods against established techniques is essential for validating their applicability.

Purpose of the Study:

• To benchmark gas-phase geometry optimizations for excited states of small molecules using GW-BSE theory.
• To evaluate the impact of various approximations within the GW-BSE framework on optimized geometries.
• To compare GW-BSE results with established high-level computational methods.

Main Methods:

• Geometry optimizations were performed using Green's function approximation (GW) and Bethe-Salpeter equation (BSE) with numerical gradients.
• Investigated approximations included one-shot G0W0 and eigenvalue self-consistent evGW, analytic vs. plasmon-pole models, and Tamm-Dancoff approximation for BSE.
• Results were compared against CASPT2, VMC, CC2, and TDDFT calculations.

Main Results:

• GW-BSE optimized geometries show good agreement with CASPT2 reference data, with average relative errors around 1-1.5%.
• Approximations within the GW-BSE framework had a negligible impact on the obtained geometries.
• Relative errors for GW-BSE were smaller than for CC2 and TDDFT, indicating its reliability.

Conclusions:

• The GW-BSE method provides accurate excited-state geometries comparable to high-order wave function methods.
• It serves as a reliable tool not only for excitation energies but also for structural properties in excited states.
• This benchmark study validates GW-BSE for excited-state geometry optimizations in computational chemistry.