Any tips for getting rid of small imag. frequencies?


Clicked A Few Times
Dear NWChem community,
When optimizing "floppy" molecules I often end up with a small negative frequency of about -20 cm-1 after projection of translations and rotations. Here is an example calculated at the B3LYP-D3/6-31+G* level of theory:
 ----------------------------------------------------------------------------
 Normal Eigenvalue ||           Projected Infra Red Intensities
  Mode   [cm**-1]  || [atomic units] [(debye/angs)**2] [(KM/mol)] [arbitrary]
 ------ ---------- || -------------- ----------------- ---------- -----------
    1      -22.059 ||    0.004045           0.093         3.943       1.105
    2       -0.000 ||    0.001142           0.026         1.113       0.312
    3       -0.000 ||    0.000296           0.007         0.288       0.081
    4       -0.000 ||    0.000091           0.002         0.089       0.025
    5        0.000 ||    0.000651           0.015         0.635       0.178
    6        0.000 ||    0.000741           0.017         0.722       0.202
    7        0.000 ||    0.000650           0.015         0.634       0.178
    8       55.136 ||    0.000230           0.005         0.224       0.063
    9       77.074 ||    0.000380           0.009         0.370       0.104
   10       98.087 ||    0.001099           0.025         1.072       0.300
   11      110.271 ||    0.002207           0.051         2.152       0.603
   12      141.968 ||    0.000884           0.020         0.862       0.241
...

Before projection the frequencies look like this:
 ----------------------------------------------------------------------------
 Normal Eigenvalue ||                 Infra Red Intensities
  Mode   [cm**-1]  || [atomic units] [(debye/angs)**2] [(KM/mol)] [arbitrary]
 ------ ---------- || -------------- ----------------- ---------- -----------
    1      -31.003 ||    0.003451           0.080         3.364       0.943
    2      -16.381 ||    0.000803           0.019         0.783       0.219
    3       -8.186 ||    0.000064           0.001         0.062       0.017
    4        8.536 ||    0.000643           0.015         0.627       0.176
    5       13.740 ||    0.001031           0.024         1.005       0.282
    6       14.686 ||    0.000315           0.007         0.307       0.086
    7       26.899 ||    0.001201           0.028         1.171       0.328
    8       54.709 ||    0.000193           0.004         0.189       0.053
    9       76.931 ||    0.000378           0.009         0.369       0.103
   10       98.600 ||    0.001223           0.028         1.192       0.334
   11      110.978 ||    0.002085           0.048         2.033       0.570
   12      142.041 ||    0.000882           0.020         0.860       0.241
...


I have tried to start from the geometries printed after stepping along mode 1 without success. I have also tried to tighten the geometry convergence criteria - still no success.

Any suggestions on what to do is greatly appreciated!

Regards,
stigrs

Forum Vet
You may have to tighten the convergence of the optimization, but also of the SCF/DFT.

driver
 tight
end

scf
thresh 1.0d-8
end

dft
tolerances tight
grid fine
convergence energy 1.0d-8
end

Bert

Quote:Stigrs Mar 27th 6:18 am
Dear NWChem community,
When optimizing "floppy" molecules I often end up with a small negative frequency of about -20 cm-1 after projection of translations and rotations. Here is an example calculated at the B3LYP-D3/6-31+G* level of theory:
 ----------------------------------------------------------------------------
 Normal Eigenvalue ||           Projected Infra Red Intensities
  Mode   [cm**-1]  || [atomic units] [(debye/angs)**2] [(KM/mol)] [arbitrary]
 ------ ---------- || -------------- ----------------- ---------- -----------
    1      -22.059 ||    0.004045           0.093         3.943       1.105
    2       -0.000 ||    0.001142           0.026         1.113       0.312
    3       -0.000 ||    0.000296           0.007         0.288       0.081
    4       -0.000 ||    0.000091           0.002         0.089       0.025
    5        0.000 ||    0.000651           0.015         0.635       0.178
    6        0.000 ||    0.000741           0.017         0.722       0.202
    7        0.000 ||    0.000650           0.015         0.634       0.178
    8       55.136 ||    0.000230           0.005         0.224       0.063
    9       77.074 ||    0.000380           0.009         0.370       0.104
   10       98.087 ||    0.001099           0.025         1.072       0.300
   11      110.271 ||    0.002207           0.051         2.152       0.603
   12      141.968 ||    0.000884           0.020         0.862       0.241
...

Before projection the frequencies look like this:
 ----------------------------------------------------------------------------
 Normal Eigenvalue ||                 Infra Red Intensities
  Mode   [cm**-1]  || [atomic units] [(debye/angs)**2] [(KM/mol)] [arbitrary]
 ------ ---------- || -------------- ----------------- ---------- -----------
    1      -31.003 ||    0.003451           0.080         3.364       0.943
    2      -16.381 ||    0.000803           0.019         0.783       0.219
    3       -8.186 ||    0.000064           0.001         0.062       0.017
    4        8.536 ||    0.000643           0.015         0.627       0.176
    5       13.740 ||    0.001031           0.024         1.005       0.282
    6       14.686 ||    0.000315           0.007         0.307       0.086
    7       26.899 ||    0.001201           0.028         1.171       0.328
    8       54.709 ||    0.000193           0.004         0.189       0.053
    9       76.931 ||    0.000378           0.009         0.369       0.103
   10       98.600 ||    0.001223           0.028         1.192       0.334
   11      110.978 ||    0.002085           0.048         2.033       0.570
   12      142.041 ||    0.000882           0.020         0.860       0.241
...


I have tried to start from the geometries printed after stepping along mode 1 without success. I have also tried to tighten the geometry convergence criteria - still no success.

Any suggestions on what to do is greatly appreciated!

Regards,
stigrs


Forum >> NWChem's corner >> Running NWChem