>I wonder if Grimme's dispersion correction (D3) can be used (can be recommended to use) in TDDFT geometry optimizations? I would appreciate some references to the works where it was used if there are some. Many thanks!
It appears to me that the answer is "it is of no use": I've accidentally found it in a reply of prof. Granovsky in one old thread (link below) while browsing the forum.
"As DFT-D correction is the same for all electronic states provided the nuclear geometry is the same, it does not affect vertical excitation energies and in particular TDDFT spectra."