Dear users and developers, I am writing here to address the performance of the state-sepcific perturbational corrections to the excited state energies, applied in the TD-DFT/PCM calculations. They are termed ptSS (and ptLR) and are specified by giving the keyword StateSpecific Perturb in the PCM section of the input file. This, optionally in combination with the NonEquilibrium True, suffices to produce the excitation energies of the 0-th order (with respect to PCM, namely excitations calculated using KS orbitals that are first calculated for the ground state in equilibrium with the solvent, or the PCM reaction field), corrected to the first order by approximately accounting for the solvent reacting to the change in the solute electronic density, but only due to the “fast” – optical - component of the dielectric constant. As a result, the vertical excitation energies from the ground state are modelled in solution.
My question is – what if I wanted to do similar thing for the emission, which of course requires the solven being in equilibrium with the solute in its excited state, but without going through the external interation scheme? Can it be done perturbatively, but with the static dielectric constant used instead of just its optical component? The similar, perturbational state-specific correction in the program-that-must-not-be-named (termed Corrected Linear Response) does seem to have this possibility. Also, one can save there the corrected reaction field (thus approximately equilibrated with the solute in its excited state) and apply it subsequently to the ground state calculations, in which only the fast component of the dielectric constant would be allowed to adjust to the ground state density of the solute.
Can Q-Chem do such a thing? The manuas says that one can “cheat” the program by substituting the static value as the optical dielectric constant in the input. It does produces a considerably stronger ptSS/ptLR stabilization energy due to the solvent polarization (with respect to one obtained with the proper optical dielectric constant), but I am not sure if this result is meaningful (there is nothing in the manual apart from the remark, that one can do it if one wants). I also wonder if the reaction field obtained in this way can be saved for subsequent calculation of the ground state?
I have also another question, this time concerning the performance of the external iterations scheme. The procedure is based on the work of Improta et al. The implementation of an analogous procedure in the program-that-must-not-be-named is based on the same reference. One might thus expect to obtain similar results with both programs. This is, however, not the case and the discrepancies are huge.
The implementation in Q-Chem seems to work excellently: when applied to the molecule of enormous excited state dipole moment od 24D, it was able to recover nearly 85% of the experimentally measured fluorochromism when shifting from hexane to acetonitrile. Furthermore, this result was only weakly dependent on the choice of functional (I used B3-LYP, CAM-B3LYP, and wB97M-V; the excitation energies were, of course, quite different for different functionals, but energy changes due to the solvent polarity were very much alike for all three of them).
The other program, however, yielded solvent-dependent shifts that were much greater than those from Q-Chem. Especially B3LYP was acting strangely, with the negative (!) excitation energy in the acetonitrile. Overpolarization of the solvent was huge. I asked the support of the other program directly, sending input and output of the calculations, and they replied that everything was OK, and that this overpolarization of the solvent by the strongly polar solute molecule is typical for the external iteration scheme and that it was only to be expected. They did not, however, comment on the surprisingly (in view of the words of prof. J. Herbert concerning the accuracy of PCM: " Keep your expectations low and then maybe you won’t be disappointed" 
good estimates of the fluorochromism yielded by Q-Chem. So my question is – are you aware (and if you can share the knowledge) of any advantage of the Q-Chem implementation over the one in the other program. Perhaps at some point the overpolarization problem was noticed and measures were taken to eliminate it?
I will be grateful for any advice or comment on the above issues.
Yours sincerely
Marcin Andrzejak