Do fractionally incremented nuclear charges improve time-dependent density functional theory excitation energies as reliably as fractional orbital populations?

Gaiduk et al. (Phys Rev Lett 108:253005, 2012) showed that one can improve local, semilocal, and hybrid approximations to the Kohn-Sham effective potentials of atoms and molecules by removing a system-independent fraction of electron charge from the highest occupied molecular orbital (HOMO); if the corrected Kohn-Sham potential is used for adiabatic linear-response time-dependent density functional theory (TDDFT) calculations, accurate Rydberg excitation energies are obtained. One may ask whether the same effect could also be achieved by fractionally increasing the positive charges of the nuclei. We investigate this question and find that a small increase in nuclear charges can indeed substantially reduce errors in TDDFT Rydberg excitation energies. However, the optimal magnitude of the charge increase is system-dependent. In addition, the procedure is ambiguous for molecules, where one has to decide how to distribute the additional charge among individual nuclei. These two drawbacks of the fractional nuclear charge method make it disadvantageous compared to the HOMO depopulation technique.