Local hybrid functionals
are evaluated in linear-response TDDFT
computations for a broad range of excited-state properties including
excited-state structures, fluorescence, and phosphorescence energies
and the vibronic shape of absorption and phosphorescence spectra.
Computation of such properties requires the optimization of excited
states, which is facilitated by the recent implementation of excited-state
gradients for local hybrid functionals in the TURBOMOLE program (J. Chem. Theory Comput.2019155508). Comparison with coupled-cluster
reference values reveals competitive performance of local hybrids
for excited-state bond lengths with particular advantages for carbon–halogen,
carbon–carbon, and carbon–nitrogen bonds. As with most
global and range-separated hybrid functionals, carbonyl and thionyl
bonds in n → π* excited states are found
to be too compact. For the emission energies, results depend on the
multiplicity of the excited state. While the local hybrid functionals
tested perform moderately well, comparable to global hybrids, for
singlet states (fluorescence energies), they provide outstanding accuracy
for triplet states (phosphorescence energies), only matched by those
from the highly empirical M06-2X hybrid functional. The assessment
of the shape of vibronic spectra reveals rather small differences
between local hybrid functionals and conventional hybrid functionals
with comparable exact-exchange admixture. The advantages for phosphorescence
energies and the robust performance for the shape of vibronic spectra
are combined to showcase the potential of local hybrid functionals
for the prediction of phosphorescence spectra.