High-Throughput Calculations of Spectroscopic Properties of Solids using Optimally-Tuned Screened Range-Separated Hybrid Functionals

Wannier-localized optimally tuned screened range-separated hybrid (WOT-SRSH) functionals have been shown to predict band gaps and optical spectra for many semiconductors and insulators with accuracy rivaling many-body perturbation theory at lower computational cost, addressing a limitation of common exchange-correlation functionals in density functional theory (DFT) [1-3]. In WOT-SRSH, range-separation parameters are tuned for each system by enforcing a nonempirical ionization potential ansatz. Using a recently developed workflow for automated tuning, we compute WOT-SRSH band gaps and optical spectra for a large set of materials in the Materials Project database, providing the largest assessment to date of this functional. We also discuss machine learning of tuned parameters, bypassing hybrid DFT runs during tuning, as well as of spectroscopic observables. Finally, we discuss the prospect of generating and using high-quality datasets of DFT-level spectroscopic properties in light of our results.

[1] Sagredo et al., Phys. Rev. Mater. 8, 105401 (2024)

[2] Ohad et al., Phys. Rev. Mater. 7, 123803 (2023)

[3] Wing et al., Proc. Natl. Acad. Sci. 118 e2104556118 (2021)