Photoredox catalysts (PCs) have contributed
to the advancement
of organic chemistry by accelerating conventional reactions and enabling
new pathways through the use of reactive electrons in excited states.
With a number of successful applications, chemists continue to seek
new promising organic PCs to achieve their objectives. Instead of
labor-intensive manual experimentation, quantum chemical simulations
could explore the enormous chemical space more efficiently. The reliability
and accuracy of quantum chemical simulations have become important
factors for material screening. We designed a theoretical protocol
capable of predicting redox properties in excited states with high
accuracy for a selected model system of dihydroquinoxalino[2,3-b]quinoxaline derivatives. Herein, three factors were considered
as critical to achieving reliable predictions with accurate physics:
the solvent medium effect on excited-state geometries, an adequate
amount of Hartree–Fock exchange (HFX), and the consideration
of double-excitation character in excited states. We determined that
it is necessary to incorporate solvent medium during geometry optimizations
to obtain planar excited-state structures that are consistent with
the experimentally observed modest Stokes shift. While density functionals
belonging to the generalized gradient approximation family perform
well for the prediction of photoelectrochemical properties, an incorrect
description of exciton boundedness (spontaneous dissociation of excitons
or extremely weak boundedness) on small organic molecules was predicted.
The inclusion of an adequate amount of Hartree–Fock exchange
was suggested as one approach to obtain bound excitons, which is physically
reasonable. The last consideration is the double-excitation character
in S1 states. As revealed by the second-order algebraic
diagrammatic construction theory, non-negligible double excitations
exist in S1 states in our model systems. Time-dependent
density functional theory (TDDFT) is blind to doubly excited states,
and this motivated us to use spin-flip DFT (SF-DFT). We established
a theoretical protocol that could provide highly accurate estimations
of photophysical properties and ground-/excited-state redox properties,
focusing on the three factors mentioned above. Geometry optimization
with DFT and TDDFT employing the B3LYP functional (20% HFX) in solution
and energy refinement by SF-DFT reproduced the experimental redox
properties in the excited and ground states remarkably well with mean
signed deviations (MSDs) of 0.01 and −0.15 V, respectively.
This theoretical protocol is expected to contribute to the understanding
of exciton behavior in organic PCs and to the efficient design of
new promising PC candidates.