Optics of semiconductors and insulators: Role of local-field effects revised

TL;DR

This paper demonstrates that neglecting local-field effects in TDDFT significantly improves agreement with experimental optical spectra of semiconductors and insulators, highlighting overestimations in existing approximations.

Contribution

The authors introduce a simple method that improves TDDFT optical spectra by nullifying local-field effects, reducing computational cost and enhancing accuracy.

Findings

Accurate optical spectra including excitons achieved

Overestimation of wing elements in existing theories identified

Method requires only static dielectric constant and head element

Abstract

We show that by nullifying the short-wave response to the long-wave excitation (local-field-effects), the adiabatic time-dependent density-functional theory (TDDFT) of optics of semiconductors and insulators can be brought into excellent agreement with experiment. This indicates that the wing elements [($\Gv,\0v)$ and $(\0v,\Gv)$, $\Gv\ne\0v$] of both the Kohn-Sham (KS) density-response function $\chi^s$ and the exchange-correlation kernel $f^{xc}$ are greatly overestimated by the existing approximations to the static DFT and TDDFT, respectively, to the extent that zero is a better approximation for them than the corresponding values provided by current theories. The head element of $f^{xc}$ is thereby fixed by the static macroscopic dielectric constant $\epsilon_M$. Our method yields accurate optical spectra including both the weakly and strongly bound excitons, while its computational cost is extremely low, since only the head element of the KS response matrix and the static dielectric constant are needed.