EFP-PCM hugely overestimating excitation energies

Hello everyone,

I was performing a TDDFT/EFP/PCM calculation for a system with the first solvation shell. The results seem that the solvation effect is being counted twice, that is, EFP+PCM instead of EFP/PCM. I have read a paper on the implementation and it says in Bandyopadhyay et al 2002:

" Recall that in the EFP/PCM approach the molecular cavity is obtained in terms of a generalized solute represented by the solute itself plus the solvent molecules introduced by the EFP model"

which seems like what a user might expect but the results do not seem so. Although, after lowering the SAS radius to the solvation shell radius, the results get considerably better.
So, is it a bug, or the method is supposed to have a bad accuracy?


Can you explain what you mean by “lowering the SAS radius to the solvation shell radius”, and maybe provide an example?

Also, is EFP+PCM described in the manual? My group implemented PCM and QM/MM/PCM but I don’t think we extended it to EFP, could have been done by someone else but I’m not certain that capability exists. One (crude) way that you could do it is to use a single spherical cavity that is large enough to encompass both QM and EFP atoms – that capability is described in the manual, just need to set some options in $pcm section. You could also contact Q-Chem support and make this a feature request, I don’t think it would be that hard to implement.


I see. I thought it was implemented already since it works great (except for the glitch I mentioned earlier). Then if it is not implemented, it is just an artifact of the existing code. I will check and contact Qchem support in that case then.

Thanks for the discussion!

The EFP+PCM functionality is not implemented as of Q-Chem 5.4. If their combination is requested, the PCM cavity will ignore EFP fragments. Even if one manually sets up a cavity (spherical for example) to include the EFP fragments, polarization due to EFP-PCM will not be evaluated and correctly accounted for.