Geometry optimization of the lowest triplet state

[lasermichel]

Hi,

I'm trying to optimize the lowest triplet state of 3-cyanoindole using CC2 (using spin component scaling) with the cc-pVTZ basis set. Control file reads:

$ricc2
  cc2
  geoopt model=cc2       state=(a{3} 1)
  scs   cos= 1.20000   css= 0.33333
$response
operators=diplen,qudlen
$excitations
irrep=a multiplicity=3 nexc=2
preopt=4
thrdiis=3

There is an error in the gradient step: cc_parse_states> inconsistency in state input!

Also the energies of the lowest two triplets, which are given in the first run of ricc2 are both negative! What is going wrong here?

Connected to that problem: What I'm really interested in, ist the optimization of the second excited singlet state. While the first excited singlet was no problem, I always run into a conical intersection with the first state, when trying to optimize the second one. For other indoles I have a good guess about the structure of the second state, so that I can avoid the crossing by choosing a good starting geometry. This does not work here. A colleague gave me the advice that the structure of this second singlet state should be similar to that of the lowest triplet. Therefore I try to optimize the tríplet for a better idea about the second singlet structure. Any other ideas, of how to avoid running into the CI?

Thanks, Michael

[christof.haettig]

You start from a closed-shell SCF?

[lasermichel]

Yes, the start orbitals were from closed shell.

[christof.haettig]

Is it a calculation without symmetry? If you use Cs symmetry the symmetry label should be a' or a".

The optimization of higher excited states is a difficult problem. I don't have a good idea to improve this. Geometry optimizations will always follow the gradient downhill.
The from the FC point the optimization ends up an in CI, one needs to find a better starting point. I'm not sure why T_1 should have a similar structure as S_2, but if this is the case, that would be an option. Else a possibility could be to scan the PES along a few (internal) coordinates and use such a scan to find a starting point that is close enough so that an optimization will converge to the S_2 minimum.

[Pingping]

hello,

I would like to ask you how to set the ground state (S0 structure) to the spin triplet in Turbomole for structural optimization, and I want to emphasize that I want to optimize the spin triplet of the ground state and not the spin triplet of the excited state, thank you so much
pingping

[christof.haettig]

The MP2 codes in TURBOMOLE can't do geometry optimizations for restricted open-shell Hartree-Fock reference functions. That is only possible of unrestricted Hartree-Fock references. You can set that by requesting parallel spins for the ``unpaired`` electrons through $alpha shells and $beta shells and then do the geometry optimization as usual for UHF-based MP2.
But you need to check the Hartree-Fock output for the amount of spin-contamination...