Dear RG Community,
i am new to Gaussian and Computational Physics (/Chemistry).
My goal is to calculate the 'IR spectrum' of liqiud Acetonitrile CH3CN (C3V point group) and its isotopologues (respectively the anharmonic vibrational spectra).
The best (scaled) results obtained for CH3CN have percent errors
Thank you for your quick reply.
So this is possibly just a problem of bad initial geometry?
I just tried and extended all bond lengths (somewhat exaggerated) and calculated the anharmonic frequencies with the same scheme as before. The optimized structure and spectrum are nearly the same as before and the same warnings occur.
Are there some general tips for variations of the initial geometry?
Lars,
The error that you are getting are related to the potential energy surface and you point out that they are due to some issue involving the symmetry. If I was in your position, I would use SYMMETRY=LOOSE to relax the structure a bit.
Hope this helps!
Regards,
Atiya
Thank you very much Atiya.
I give it a try.
Another approach i am thinking of is using NoSymm and deactivating one of each degenerate mode in the readanharm section (wich have nearly the same frequencies in the harmonic data).
NoSymm (without deactivation) is leading to the same warnings as before (except 3)), and
7) in 'Anharmonic Data: Displacement Vectors Definition'
WARNING: Deactivating the symmetry may cause some errors in the identification and treatment of degenerate normal modes.
Analysis will be done taking into account the symmetry.
only a few warnings in 2) (always a degenerate mode is part of i,j or k) and no warning in 3) but therefor
8) in 'Vibro-Rotational Analysis Based on Symmetry'
WARNING: Symmetry has been deactivated. Normal mode degeneracy will be ignored as per request. For this reason, the results below may be incorrect.
And unsurprisingly the degenerate modes are shifted in respect to each other and so the concerned normal modes as well as the combination bands are unpleasant and pointless.
So is there any meaning in the approach or would it just lead to some pointless data?
Regards,
Lars
Hi Lars,
Once you insert the anharmonic terms into the solution of the harmonic oscillator problem, it seems reasonable to me that some degeneracies (if not all) would be lifted. I definitely would use nosymm. Do you have any cubic terms for acetonitrile from the literature? If the cubic force constants are small you will only get small splitting of the degenerate states. I would recommend running the calculation with a small molecule which has a well documented set of experimental cubic force constants. To me the warnings are appropriate disclaimers indicating that there are limitations that may create a problem, but do not necessarily mean that you are wasting your time. This is what makes research a learning experience. It sounds like a very interesting investigation. I wish you well in this endeavor and I hope you find what you are looking for.
Good day,
try
# opt=calcall freq=anharmonic b2plyp/dgdzvp2 6d 10f scrf=(solvent=acetonitrile) geom=connectivity int=ultrafine
6d 10f will improve the upper frequencies, and int=ultrafine will improve the low ones.
But calculate anharmonisms in the solution is not good
Hello,
i got some information from the gaussian support.
The covergence problem of the geometry optimization is related to the grid used for PCM. The used Lebedev grids have octahedral symmetry and because of that there is a mismatch of the C3V structure and the octahedral symmetry of the Lebedev grid. So some integration errors arise, which are small, but enough to prevent the convergence of the geometry optimization with opt=tight or opt=verytight and enough to possibly lead to problems in noise senstive calculations - like anharmonic frequency calculations (cubic and quartic force constants).
However, i got an advice how to circumvent the problem. When using RADII=UA0 in the scrf-read section (along with opt=verytight and integral=(grid=superfine)), there are no off-axis-spheres of the PCM for Acetonitrile and so the integration errors are reduced.
And it worked for the optimization with opt=verytight. At the moment the anharmonic frequency calculation is running.
Thank you Shannon for your advice, until now i only found cubic force constants concerning the gas phase (or from dft calculations of the molecule), but i just searched a few moments now. If the running calculation is leading to the same problems as before, i am going to search for values or give another molecule a try - if need be.
And thank you Vladimir, once the running calculation is completed, i am going to add the 6d, 10f.
Regards,
Lars
Thank you Lars and Vladimir,
I find that the questions and responses to be very helpful to my own learning. I guess that is the real purpose of Research Gate.
---Shannon
- Badges
- Science topic