CCL:G: Double hybrid functionals for excited states with Gaussian09

 Sent to CCL by: Lars Goerigk [lars.goerigk _ uni-muenster.de]
 TD-B2PLYP works as follows:
 First, you perform a ground state B2LYP calculation. The perturbative
 correction  is not needed if you are not interested in any ground state
 properties, because you are just interested in the resulting KS-orbitals.
 Then a standard TD-DFT treatment (i.e. TD-B2LYP) is carried out, which gives
 you excitation energies and transition moments of TD-BHLYP quality.
 The B2LYP orbitals and TD-B2LYP excitation energies and amplitudes are then
 used for a CIS(D) type treatment to correct the TD-B2LYP energies.
 I do not know if such a treatment is possible in Gaussian by adjusting some
 IOPs. I assume that the CIS(D) routine will crash if TD-DFT results are used
 as an input.
 However, the routine may not fail with an CIS like input. The Tamm-Dancoff-
 Approximation is basically CIS. I do not know if Gaussian does TDA-DFT.
 Basically you can then carry out the CIS(D) correction based on your TDA-B2LYP
 result. This gives you TDA-B2PLYP energies. This was also discussed in the
 2007 Grimme/Neese paper. TDA-DFT usually gives you very similar energies to
 TD-DFT. But again, I do not know whether Gaussian can handle this with IOPs,
 but it's worth testing it.
 Maybe the Gaussian people can help you with the IOPs if you contact them;
 maybe it's even possible to do full TD-B2PLYP.
 Another option would be ORCA, which is freely available. It does TDA-B2PLYP,
 but not full TD-B2PLYP.
 I hope these information are somehow useful for you.
 Cheers,
 Lars
 On Wednesday 27 October 2010 05:08:27 pm you wrote:
 > Sent to CCL by: "Lara Martinez Fernandez"
 [lara.martinez*o*uam.es]
 > I am trying to run some TD-DFT calculation (for excited states) using
 >  double hybrid functionals with Gaussian09. I have some doubts about how
 >  TD-B2PLYP,TD-B2GP-PLYP work. If I dont misunderstood, first I have to run
 >  a B2LYP calculation for the ground state and then use this orbitals and
 >  amplitudes to run the CIS(D) correction for the rest of excited states. I
 >  dont know if this procedure is possible with Gaussian09. Do anyone know
 >  how to do that? I guess there must be any keyword to read the B2LYP (of
 >  the ground state) orbitals and not optimise them in the CIS(D)
 >  calculation.
 >
 >
 > Thanks in advance, Lara>