First-principles self-interaction free GWΓ simulations for first ionization potentials and electron affinities.

Area of Science:

• Computational chemistry
• Quantum chemistry
• Materials science

Background:

• Accurate prediction of molecular properties like ionization potentials (IPs) and electron affinities (EAs) is crucial in chemistry.
• Traditional GWΓ simulations suffer from self-interaction errors, limiting their accuracy.
• The local density approximation (LDA) reference in GWΓ calculations introduces significant self-interaction errors.

Purpose of the Study:

• To develop and demonstrate self-interaction-free GWΓ simulations.
• To improve the accuracy of calculating IPs and EAs for molecules.
• To investigate the impact of using a Hartree-Fock approximation (HFA) as a reference instead of LDA.

Main Methods:

• Implemented a one-shot GWΓ simulation framework with HFA as the reference.
• Incorporated self-interaction corrections to the GW and GWΓ terms.
• Applied the method to calculate IPs and EAs for 24 middle-sized molecules.

Main Results:

• Achieved good agreement between simulated and experimental IPs and EAs.
• Demonstrated significant improvements in computational accuracy compared to LDA: ~0.1 eV for IPs and 0.97 eV for EAs.
• Showed that HFA reference reduces GW contributions by 15%-38% and Γ contributions by 65%-81% compared to LDA.

Conclusions:

• The HFA-referenced GWΓ method effectively eliminates self-interaction errors, leading to accurate molecular property predictions.
• Self-interaction corrections to the Γ term are essential for maintaining accuracy, even with an HFA reference.
• This approach offers a more accurate and computationally efficient alternative for electronic structure calculations.