CC2 with COSMO for excited states

[user20212707]

Dear Turbomole forum,
I am currently trying to calculate the excited state energy for small superoxides at the approximate CC2 level of theory including solvation effects via COSMO in the ricc2-module. These systems are open shell systems, containing symmetry elements of the C_2v-group. I am simply considering an array of prepared geometries, so the calculations are mostly single-point calculations without a previous optimization of the molecule to the electronic ground state (or any excited state).

Without invocing COSMO (via cosmoprep) these calculations converge readily and I obtain all required excitation energies and properties.
When I include COSMO however the computation always fails in the optimization of the ground state cluster amplitudes.

I oriented myself on the examples in the manual for the calculation of the ricc2-energies:

Code: [Select]

$ricc2
  cc2
$excitations
  irrep=a1 nexc=3 npre=3 nstart=3
  exprop states=all relaxed operators=overlap,diplen,angmom,soc
$response
  fop relaxed

and for the COSMO-part:

Code: [Select]

$cosmo
  epsilon=4.33
  rsolv=1.30
  refind=1.3524
  cosmorel state=(a1 2)
$cosmo_atoms
  ...
$cosmo_correlated

Notably, whichever cosmorel state I use, the error is always the same.
A dscf calculation of the molecule runs smoothly, but the ricc2 calculation after that one gives the following error (given below is the end of the file):

Code: [Select]

   **************************************************************************
   *                                                                        *
   *          OPTIMIZATION OF THE GROUND STATE CLUSTER AMPLITUDES           *
   *                                                                        *
   **************************************************************************

   Reaction field calculation using the COSMO solvation model
        Reaction field scheme  : post-SCF                                                                       
        Reaction field density : full unrelaxed                                                                 
        epsilon                :     4.330
        f(epsilon)             :   0.689441
     start CC2 from scratch because restart not enabled


   Iter. MP2 energy       Norm(Omega) Norm(Ombar)   Norm(t1) Norm(t2)  cpu    wall
   ................................................................................

========================
 internal module stack:
------------------------
    ricc2
    cc2_solve_t0
    cc_lhtr
========================

 unfinished (do_t1res and open-shell)
 ricc2 ended abnormally


I am not sure how to fix,  or even interpret this error as something other than CC2 is having internal problems. At first I thought it might have something to do with the OMP_stacksize, but the latter is unlimited on the server I am using. I think with generous $maxcor settings (> 5100 MiB per_core) there should also not be a lack of memory. In addition, decreasing the basis from cc-pVQZ to cc-pVTZ or even cc-pVDZ did not avoid the error.

I am very grateful for your suggestions and help. If additional input/output is required, I will hurry to deliver it.

Thanks,
Dennis

[christof.haettig]

COSMO-CC2 is not implemented with symmetry, also the post-SCF reaction field coupling scheme doesn't work together with the cosmorel option in the COSMO input. Also, excited-state properties are not yet implemented for COSMO-CC2.

[shafikoff]

Dear Christof,

Are there any plans to implement CC2-COSMO for excited states in Turbomole in near future?

Thanks,
Marsel

[christof.haettig]

COSMO-CC2 is available for excitation energies (not gradients) in ricc2 since some years within the post-SCF coupling scheme. The problem of your calculation was that your input asks for the PTED coupling scheme which is only implemented for ADC(2).