Problems with negative frequencies

Just Got Here

9:19:41 AM PST - Thu, Dec 20th 2018

Dear NWchem community,

I ran a DFT calculation on a small molecule in solvation model using nwchem 6.8. It was converged but it has a negative frequency of about -162 cm-1. I tried to re-optimize the geometry and using a tighter energy convergence of 5d-9 but still no success.

Here is the input file:

geometry units angstroms autosym 0.2

   C                  -0.30975950     1.32138254     0.00000000

   C                   0.04718087    -0.20210007     0.00000000

   H                   0.13238927     1.79961154     0.88134209

   H                  -1.39114031     1.48379999     0.00000000

   H                   0.13238927     1.79961154    -0.88134209

   O                   1.27722401    -0.45562499     0.00000000

   O                  -0.92245943    -0.99906777     0.00000000

end

basis

 * library 6-311++G**

end

dft

 xc b3lyp

 mult 1

 direct

 grid fine

 smear

 convergence energy 5d-9 damp 45 ncydp 0 dampon 1d99 dampoff 1d-4

end
set tolguess 1d-3

driver

 tight

 clear

 maxiter 200

end

cosmo

 dielec 8.51

end

task dft optimize
task dft freq numerical

Any suggestions will greatly appreciated.

Regards,
Sophia

Forum Vet

4:12:36 PM PST - Wed, Jan 16th 2019
Sophia If you look at the frequencies output, you will find the section " Negative Nuclear Hessian Mode 1 " If you use as new geometry the one as listed under " Geometry after -100.0% step for mode 1;" C -0.09310508 1.34795637 0.00258095 C 0.00214260 -0.18798365 0.00404920 H 0.67531313 1.77100773 0.72045709 H -1.12474372 1.70215805 0.27599438 H 0.16807654 1.71637592 -1.03984743 O 1.16957720 -0.66899511 0.02412262 O -1.06729404 -0.85799128 -0.02636247 Then, if you re-run the calculation with this new geometry, all frequencies will be positive

Clicked A Few Times

9:48:32 AM PDT - Wed, Aug 14th 2019
Quote:Edoapra Jan 16th 11:12 pm Sophia If you look at the frequencies output, you will find the section " Negative Nuclear Hessian Mode 1 " If you use as new geometry the one as listed under " Geometry after -100.0% step for mode 1;" ... Then, if you re-run the calculation with this new geometry, all frequencies will be positive I have a similar problem with a negative-frequency mode (even the frequency value is similar -167.765 [cm**-1]) despite the tighter energy convergence and optimisation used. I have tried to follow this advice to put new geometry from " Geometry after -100.0% step for mode 1;" in the input file and re-run the calculation but it did not help. It converged to a bit lower energy but the mode is still at the negative frequency. My input file: start PCO echo scratch_dir . permanent_dir . memory 6000 mb charge -5 geometry units angstroms O 2.63694460 3.94597618 -0.00000016 C 2.90591488 2.64073309 -0.00000013 O 1.89749032 1.74606799 -0.00000006 O 4.16999971 2.21992605 -0.00000016 O -0.01986707 0.00426229 1.90601275 Pu -0.02688600 -0.00141010 0.00000009 O -0.01986716 0.00426243 -1.90601265 O -4.00241237 2.53107489 -0.00000018 O 0.53932580 -2.50059322 -0.00000008 C -3.73858336 1.22669983 -0.00000014 C 0.81223868 -3.82105873 -0.00000012 O -4.73281162 0.34093790 -0.00000017 O -2.45996385 0.79869299 -0.00000004 O -0.18408865 -4.70381661 -0.00000013 O 2.07680464 -4.23684930 -0.00000013 end BASIS Pu library "Stuttgart_RLC_ECP" C library "Stuttgart_RLC_ECP" O library "Stuttgart_RLC_ECP" END ECP Pu library "Stuttgart_RLC_ECP" C library "Stuttgart_RLC_ECP" O library "Stuttgart_RLC_ECP" END dft xc b3lyp noio direct vectors input PCO_01.movecs output PCO.movecs iterations 5000 convergence energy 5d-9 damp 40 ncydp 0 dampon 1d99 dampoff 1d-4 rabuck 20 levlon 0.2 smear grid fine odft mult 4 end driver tight MAXITER 600 end set int:txs:limxmem 25621940 set driver:linopt 0 task dft optimize task dft freq

Clicked A Few Times

4:29:56 AM PDT - Fri, Aug 16th 2019
Following the advices in these threads http://nwchemgit.github.io/Special_AWCforum/st/id2535/Trouble_with_negative_freque... http://nwchemgit.github.io/Special_AWCforum/st/id2763/Frequency_Analysis_Grid_Dens... I also used "grid huge" but it did not help either. Furthermore, the use of this option increased the calculation time to the extent that the numerical frequency calculation is faster with the "grid fine" option. The latter calculation gives only positive frequencies. There seems to be a problem with analytical frequency calculations in NWChem.

Forum Vet

3:07:26 PM PDT - Mon, Aug 19th 2019

Quote:Sergbuto Aug 16th 3:29 am

Following the advices in these threads
http://nwchemgit.github.io/Special_AWCforum/st/id2535/Trouble_with_negative_freque...
http://nwchemgit.github.io/Special_AWCforum/st/id2763/Frequency_Analysis_Grid_Dens...

I also used "grid huge" but it did not help either. Furthermore, the use of this option increased the calculation time to the extent that the numerical frequency calculation is faster with the "grid fine" option. The latter calculation gives only positive frequencies.

There seems to be a problem with analytical frequency calculations in NWChem.

Please try the following input. The major change is the line set dft:no_prune t. This is required for the correct integration of the contribution to XC potential from the Pu atom.
The associated output does not contain any negative frequency.
The input change increases the time to solution. We are working at a better fix that does not increase the time to solution.

start pco_new
echo

memory 6000 mb noverify
charge -5

geometry units angstroms
 O      4.72777782     0.00036022     0.00000000
 C      3.80305541    -0.95770261     0.00000000
 O      2.49547411    -0.62952538     0.00000000
 O      4.16561764    -2.23897376     0.00000000
 O      0.00000000     0.00000000     1.91098610
 Pu     0.00000000     0.00000000     0.00000000
 O      0.00000000     0.00000000    -1.91098610
 O     -0.14380067     4.72701758     0.00000000
 O     -1.79292203    -1.84638129     0.00000000
 C     -1.07213291     3.77239390     0.00000000
 C     -2.73092249    -2.81469129     0.00000000
 O     -2.36420087     4.09419559     0.00000000
 O     -0.70255209     2.47590666     0.00000000
 O     -4.02181697    -2.48804382     0.00000000
 O     -2.36357695    -4.09455580     0.00000000
end
BASIS spherical
Pu library "Stuttgart_RLC_ECP"
C library "Stuttgart_RLC_ECP"
O library "Stuttgart_RLC_ECP"
END

ECP
Pu library "Stuttgart_RLC_ECP"
C library "Stuttgart_RLC_ECP"
O library "Stuttgart_RLC_ECP"
END
dft
xc b3lyp
direct
vectors input atomic output PCO_new.movecs
iterations 5000
convergence energy 1d-10 damp 40 ncydp 0 dampon 1d99 dampoff 1d-4 
grid fine
smear
odft
mult 4
end

driver
xyz
tight
end

set int:txs:limxmem 25621940
set quickguess t
 
set dft:no_prune t
task dft optimize 

task dft freq

Forum >> NWChem's corner >> Running NWChem