Author Topic: SCF convergence problems with COSMO  (Read 6766 times)

marand

  • Full Member
  • ***
  • Posts: 27
  • Karma: +0/-0
SCF convergence problems with COSMO
« on: September 23, 2022, 12:22:06 PM »
I am trying to assess the solvent effects on the equilibrium structure of a relatively large copper complex with over 100 atoms. The molecule is not really globular, so the cavity is bound to be
complicated. Turbomole did issue a warning of 28 disjunct cavities, so following its advise I have use the use_contcav option in the COSMO section of the control file.

I have use the optimised radii for all atoms except from Cu and P, which are, however, in the centre of the molecule so the non-optimum atomic radii in their case should be of secondary importance.

The SCF energy behaves quite chaotically (jumping up and down by even several Hartees) despite any efforts to stabilise the convergence (increased damping, virtual orbital shift, fermi smearing).
I have noticed that it is the Ediel value changes drastically (from 0.9 to 8 or 9) and rather unsystematically with every step of the SCF procedure. I gather that this may be the reason for the overall instability of the SCF procedure.

However, I do not know how to circumvent this problem. Would be grateful for any assistence in the matter.

My very best regards
Marcin


uwe

  • Global Moderator
  • Hero Member
  • *****
  • Posts: 575
  • Karma: +0/-0
Re: SCF convergence problems with COSMO
« Reply #1 on: September 23, 2022, 06:37:21 PM »
It could be that the use_contcav option is causing that behavior. This option is helpful for a very small number of cases only, as given in the manual:

Quote
in case of disjunct cavities only the largest contiguous cavity will be used
and the smaller one(s) neglected. This makes sense if an unwanted inner
cavity has been constructed e.g. in the case of fullerenes. Default is to use
all cavities.

You have 28 disjunct cavities, and adding this option will remove 27 of them, keeping just the largest one.

marand

  • Full Member
  • ***
  • Posts: 27
  • Karma: +0/-0
Re: SCF convergence problems with COSMO
« Reply #2 on: September 26, 2022, 11:29:10 AM »
Thank you for this suggestion. However, I initially tried to conduct the calculations without the 'use_contcav' option and the behavior was no different.

I have also tried to use the isodensity option with several threshold values but to no avail. This is most likely to do with the size (over 100 atoms) and complicated (bush-like) shape
of the molecule, but I have little experience with COSMO in turbomole.  I would greatly appreciate assistance or advice.

Thank you
Marcin