Benchmark of GW Methods for Core-Level Binding Energies.

Area of Science:

• Computational chemistry
• Quantum mechanics
• Spectroscopy

Background:

• The GW approximation is a powerful tool for electronic structure calculations.
• Accurate computation of core-level binding energies is crucial for interpreting X-ray photoelectron spectroscopy (XPS) data.
• Existing GW methods require careful benchmarking for molecular systems.

Purpose of the Study:

• To comprehensively benchmark various GW methodologies for molecular inner-shell excitations.
• To identify the most accurate GW schemes for calculating absolute and relative core-level binding energies.
• To evaluate the performance of GW methods on the CORE65 benchmark set and ethyl trifluoroacetate.

Main Methods:

• Implementation and comparison of different GW approximation schemes: starting point optimized, partial and full eigenvalue-self-consistent, Hedin shift, and renormalized singles.
• Application of these methods to molecular inner-shell excitations.
• Validation against experimental XPS data and the CORE65 benchmark set.

Main Results:

• All tested GW methods consistently yield a unique solution for core-level energies.
• Partial eigenvalue self-consistency, single-shot GW with optimized hybrid functionals, and the Hedin shift demonstrate superior accuracy for absolute core-level energies (MAE of 0.3 eV).
• Eigenvalue self-consistent schemes and the Hedin shift achieve the highest accuracy for relative binding energies (MAE < 0.2 eV).

Conclusions:

• Partial eigenvalue self-consistency, optimized hybrid functional starting points, and the Hedin shift are recommended GW schemes for accurate core-level binding energy calculations.
• These methods provide reliable predictions for XPS interpretation in molecular systems.
• The study establishes a benchmark for evaluating GW methodologies in this domain.