Author Topic: Gas phase frequencies from COSMO optimised structures  (Read 4239 times)


  • Jr. Member
  • **
  • Posts: 19
  • Karma: +0/-0
Gas phase frequencies from COSMO optimised structures
« on: July 13, 2023, 04:33:38 PM »
Dear Users,

I have optimized structures using RI-DFT (PBE0/def2-SVP/D3-BJ) while using COSMO. This means I have to calculate my frequencies numerically using NumForce -central -ri. Is it possible to recalculate these modes in the gas phase without reoptimizing my structure? I have tried the following:

1. Optimize in COSMO, then remove the COSMO information from the control file and run aoforce. aoforce failed with no error message, it got stuck (shown below) and increasing $maxcor did not help
Code: [Select]
                       Nonorthonormal Krylov Space Iteration

 total number of roots to be determined: 216

 maximum core memory set to   500 MB,
 corresponding to       20 vectors in CAO basis

 maximum number of simultaneously treated vectors (including degeneracy):      216

 Recommendation for optimum performance: Increase the maximum amount of
 allocatable core memory (e.g. $rpacor, $maxcor) to     5400 MB or to the
 maximum amount of core memory available on your machine.

 Iteration IRREP Converged      Max. Euclidean
                 roots          residual norm

2. The same as above but run NumForce -central -ri. In this case the calculation fails when checking the gradient norm
Code: [Select]
   gradient norm                  : 0.97432950E-01
No stationary point of energy, therefore no 2nd derivatives,   dempsey!

3. I then tried the same but skipping the gradient check using NumForce -central -ri -c. This progresses the NumForce calculation but unsurprisingly I get imaginary modes. NumForce tried to warn me!

Is it possible to do what I am trying, or is it necessary to reoptimize in the gas phase? The reason I am trying this is because some of my thermodynamic data does not follow a trend we expected, and so I am thinking it could be that it could be and issue with COSMO frequencies and/or numerical error, hence me wanting to compare to gas phase analytical frequencies. Unfortunately, my structures do not optimize well in the gas phase.