CC2 Convergence Excited States

Dear all,
i am trying to calculate some spectra (and later excited state surfaces) with the DF-LR-CC2 method for a series of complexes. These complexes contain 2 iron atoms and 2 thiophenes with 57 atoms overall and i am aware, that a transition metal complex is not the best prerequisite for CC2. Nevertheless, there is a paper ( pubs.acs.org/doi/10.1021/jz5022087 ), demonstrating that TDDFT fails badly for thiophene compounds, so my choices are limited.

So now i am trying to calculate 20 roots with CC2 and reduced cost settings, but cannot manage to get any root converged. The norm of the residual vector never drops below 0.05. Here is my input file

Is there possibly some tweak i could apply to get the excited states converged?

Best wishes and thank you in advance
Phillip

Dear Phillip,

could you, please, attach the output file? Which version of the MRCC program do you use?

Best wishes,

Dávid

Dear Dávid,
this was done with the MRCC version from February 2019. The synthetic chemists allowed me to share the structure, so here are my MINP and output files.

Best wishes
Phillip

Dear Phillip,

the RHF solution is unstable and the system has a strong multireference character. The canonical CC2 calculation did also not converge to a stable solution. Unfortunately, a UHF-CC2 implementation is currently not available in the MRCC program.

Best wishes,
Dávid

Thank you Dávid. I tried a stability analysis with ORCA and indeed the RHF solution is instable. This is quite interesting for me, as a restricted DFT calculation is stable in the singlet and the complex is a singlet experimentally. I will try some DMRG calculations ...

Thank you and best wishes
Phillip