cc2 triplet excitation energies

[feuerswolf]

Dear all,

I'm trying to calculate (Turbomole 7.0) singlet and triplet excitation energies of some organic molecules with the ricc2 module.
Unfortunately it doesn't really work due to an abnormal end of ricc2. At the end of the out-Datei it always states

===========================================



  === F matrix transformation for    6 state(s) ===


                    ========    F MATRIX MODULE    ========

               calculating     3 F * RE0 transformations
 
 


                    ========    F MATRIX MODULE    ========

               calculating     3 F * RE0 transformations
 
 ========================
  internal module stack:
 ------------------------
    ricc2
    cc_rspdrv
    cc_rspvec
    cc_fmat
 ========================

 cc_fmat not implemented for triplet
 ricc2 ended abnormally

What does this cc_fmat stand for?
In the control file I have the following input for the cc2-excitation-part:

$freeze
 implicit core=   20 virt=    0
$cbas    file=auxbasis
$ricc2
  cc2
  conv= 7.00
  oconv= 6.00
$excitations
  irrep=a  multiplicity=  1  nexc=  3
  irrep=a  multiplicity=  3  nexc=  3
  spectrum  states=all  operators=diplen,dipvel
$last step     define

Thanks in advance for any help!

feuerswolf

[Arnim]

Dear feuerswolf,

for the excitation energies you would have to remove the 'spectrum' key from $excitations.

'spectrum' triggers the calculation of oscillator strengths and they are zero for the triplet excitations.

Cheers,

Arnim

[feuerswolf]

Dear Arnim,

thanks a lot for your reply.
But why am I able to calculate oscillator strengths for transitions between ground and excited triplet states with other methods, e.g. TDDFT (module rpat)?

Kind regards

feuerswolf

Ps: After removing the spectrum keyord the calculations worked 

[Arnim]

I don't really know, what these numbers from TDDFT mean. But it is not phosphorescence strength. For that, you would have to run relativistic 2-c calculations.

Best wishes,

Arnim

[feuerswolf]

Actually I'm trying to set up some relativistic calculations to get transition moments, but TurboMole seems not to be the best choice for this purpose. Maybe I should switch to DIRAC...

Anyway, if I understand the TurboMole manual (7.0) correctly transition densities are calculated without taking spin into account (chapter 8.2). So the oscillator strengths for triplet excitations "make sense". However, in what why I am able to interpret them is not quite clear to me.

Best wishes,

feuerswolf

[Arnim]

Hi,

can calculate relativistic 2-component oscillator strength on CC2 or ADC(2) level.
Just switch on ri or rijk and soghf, then run ridft and ricc2 with such settings:
$ricc2
 adc(2)
$excitations
 irrep=a  multiplicity=  1  nexc= 3
 spectrum states=all operators=diplen,dipvel

And you have to use 2-c basis sets or have to set $rx2c to account for spin-orbit coupling.

Cheers,
Arnim

[evgeniy]

Dear Arnim,

I am interested in adc(2) excited state calculations with account for spin-orbit coupling, using ecp with a 2-c basis.
I wonder what auxiliary basis set is to choose in the case of 2c (dhf-SVP) "main" basis?

Best regards,
Evgeniy

[Arnim]

Hi Evgeniy,

actually for now (version 7.1) there are no optimized cbasis sets for the dhf-series.
For some elements the dhf are equal to def2. In these cases the def2-cbas can be chosen, but in the other cases not.
Then you could try the def2-TZVPP or def2-QZVPP cbasis. But do some convergence tests on smaller model systems to ensure the performance.

Best,

Arnim

[evgeniy]

Hi Arnim,

Many thanks for your reply. In relation to my question, I wonder if there is
a  well-defined and reliable recipe  what cbas is to use in general if there
is no optimized cbas for a particular basis set. For example, for Pople's
basis sets, like 6-311++G**.

Best regards,
Evgeniy

[Arnim]

Hi Evgeniy,

I don't think there is a general recommendation, but the recipe is quite simple. Take for a start really huge cbasis and then remove functions (or optimise them with the cbasopt script) while monitoring the errors.

In your case, I would say you could set up a small model system and start with the aug-cc-pV5Z cbasis, then go down to aug-cc-pVQZ and aug-cc-pVTZ. If the error with the TZ with respect to the 5Z is acceptable, then you are done.

Hope that helps,

Arnim