CCL: correction of my comment on MO's in Solvent

[Andreas Klamt <klamt(!)cosmologic.de>]

 Sent to CCL by: Andreas Klamt [klamt[*]cosmologic.de]
 Dear colleagues,
after some critical questions by colleagues, I need to correct myself
 with respect to my previous statement on the meaning of the MOs in
 continuum solvents:
The MOs do make sense and they will be the same in both of the two
 implementations, i.e  by either considering the total polarization
 charges as prividing an external field and aditing the penalty for the
 polarization energy costs at the end, or by considering the solvent
 response as an indirect charge-charge iteractions (in addition to the
 direct Coulomb interaction), i.e. representing it as additional
 nuclei-nuclei,  nuclei-electron, and electron-electron interaction. As
 long as we are on the SCF level (i.e. Hartree-Fock or DFT) in both
 cases
 the final FOCK matrix is identical and hence orbitals and orbital
 energies will be identical. Just for subsequent perturbation treatment
 it is useful to realize that the continuum interaction has one-electron
 and two-electron contributions.
 Sorry for any confusion which I may have caused
 Andreas
 Am 31.03.2013 16:17, schrieb Andreas Klamt klamt,,cosmologic.de:
>  Sent to CCL by: Andreas Klamt [klamt{=}cosmologic.de]
>  Since my first two attempts were not posted yet, here again my comment:
>  Dear Ramesh,
> this is a good question. In the way as most continuum models are
>  implemented nowadays, the orbitals do not make much sense, because the
>  polarization charges are completely treated as external field, and the
>  energy for generatig the field is added to the total energy of the
>  molecule after the SCF has achieved. In my original COSMO
>  implementation in MOPAC I followed a different concept, considering
>  the continuum contribution as an indirect charge-charge interaction.
>  In that way you have additional nuclei-nuclei-interactions, additional
>  nuclei-electron-interactions (added to the one electron Hamiltonian,
>  and additional electron-electron interactions, added to the
>  two-electron-interactions. Thus the continuum interactions enter the
>  orbitals in a perfectly analogous way to the Coulomb interactions. A
>  post-correction for the self energy of the polarization field is not
>  required, since all energy contributions already take it into account
>  by the factor 1/2. While the total energy, and all expectation values
>  will be exactly the same in both implementations, the orbital energies
>  produced in that way are considerably different from thse in the
>  standard way, but I am convinced they make more sense.  (Without
>  understand it in detail, Zerner and coworker had earlier called the
>  models as model A and B, see  Szafran, et al. J. Comp. Chem 1993 ....)
> Please note the difference of the models for the orbitals: In the
>  standard implementation, the electrons in the LUMO "see" the
>  polarization charge polarization charges produced by the ground state
>  electron density, i.e. the change of the polarization charges due to
>  the excitation to the LUMO is missing.
> But I am afraid, currently no implementation of PCM or COSMO following
>  the Coulomb energy analogy is available. While writing this I realize
>  that I may be wrong: The GAMESS COSMO implementation may still have
>  it. Kim Baldridge should know about this best.
>  Best regards
> Andreashttp://www.ccl.net/chemistry/sub_unsub.shtmlConferences:
>  http://server.ccl.net/chemistry/announcements/conferences/>;
>