ri-dft (bp86/def-SVP) calculations for AF coupled spin states

[wmc_sameera]

Hello,

I am doing some ri-dft (bp86/def-SVP) calculations for Mn(V) –oxo porphyrin dimer system(s) with TURBOMOLE 5.9.0

I managed to optimize the ferromagnatically (F) coupled nonet spin state (F copled quintets of the individual Mn sites) for the above system. Then I tried to optimize the anti-ferromagnatically coupled singlet (AF copled quintets of the individual Mn sites) using the orbitals, geometry of the F coupled nonet and assigning the exact electron occupancy. But SCF convergence for the AF singlet was not succeeded. After that, I followed the same procedure with dft (b3-lyp/def-SVP), but it did not work.

According to my previous calculations with GAUSSIAN 03 for the same system(s), AF coupled electronic configurations are quite important than F coupled spin states. But G03 calculations are time consuming. If I would manage to optimize the AF coupled spin states using ri-dft, I can save my time.

Any comments would be appreciated.


W. M. C. Sameera (PhD student)
WestCHEM
Department of Chemistry
University of Glasgow
Joseph Black Building
University Avenue
Glasgow, G12 8QQ
United Kingdom.

+44 (0) 141 330 8121

[antti_karttunen]

Hello,

have you already tried to change keywords such as $scforbitalshift and $scfdamp? They can be used to control the SCF procedure and increasing both values might help with more difficult systems. Increasing the number of iterations ($scfiterlimit) can also help, but often it is more economical to first adjust $scforbitalshift and $scfdamp.

[uwe]

Hello,

the SCF convergence in Turbomole can be controlled by:

• the maximum number of SCF iterations ($scfiterlimit)
• using a higher DIIS damping  ($scfdamp), I would recommend to start with a damping factor between 4 to 8.
  If that does not help, try:
  
  • set the damping factor to a high value like 15, and limit the number of SCF iterations to, say, 20.
  • after that, reduce the damping factor to something like 4 and continue with $scfiterlimit of 300 (or so).
    You can check the convergence graphically (if gnuplot is installed) by calling
         cgnce <output-filename>
    If the energy starts to oscillate, the damping factor is too high...
  
• using another orbital shift ($scforbitalshift)

All other procedures might alter your occupation, and that is something you do not want to have in your case.

There is a short chapter (6.9) about SCF convergence also in the Tutorial of TURBOMOLE, which can be found in the DOC directory of your Turbomole installation. A detailed description of the keywords can be found in the documentation.

Yours,

Uwe