Previous Thread
Next Thread
Print Thread
molvib problems
#14118 04/20/07 10:33 AM
Joined: May 2005
Posts: 18
M
myriad Offline OP
Forum Member
OP Offline
Forum Member
M
Joined: May 2005
Posts: 18
I try to parameterize the triple C-=C bond. I used the model CH3-CH2-C-=C-H. I believe I have made the UMAT correct, as the vibrational frequencies from guessed+exist Charmm force field are tidy (only two negative eigenvalues; 1-3 modes contribute to one frequency; no negative contribution; reasonable frequencies for known vibrational modes). However, when I tried to use the same UMAT to calculate the frequencies from Gaussian results, it is a mess (too many negative eigenvalues; many negative contributions to some frequencies; the frequencies for the known modes (exist bonded parameters) are not correct). I tried using different levels of theory and/or different initail conformations in Gaussian calculations, but no luck.

In g03, I used
#MP2/6-31G* OPT POP=(MK,DIPOLE) DENSITY=CURRENT
for optimization and
#MP2/6-31G* GEOM=ALLCHECK FREQ IOP(7/33=1)
or
#MP2/6-31G* GEOM=ALLCHECK Opt=CalcAll IOP(7/33=1)
for freqency calculation.

Any suggests are highly appreciated.

Shunzhou

btw, months ago, Tina asked about C-=C bond parameters, any luck for him?

Re: molvib problems
myriad #14119 04/21/07 07:38 PM
Joined: Sep 2003
Posts: 250
Forum Member
Offline
Forum Member
Joined: Sep 2003
Posts: 250
I'd start optimizing the geometry at a lower level of theory (including the frequency routine) and then use the resulting forces in an iterative refinement procedure - i.e. a checkpoint-file, and then redo the optimization until all negative frequencies (at least those that are significant in magnitude) are washed out.

This would look something like:

#p hf/6-31g* geom=check opt(rcfc) scf=tight "etc etc and so on blablabla"

Also, you may want to check the symmetry in your molecule...

Good luck.

Last edited by jb007; 04/21/07 08:19 PM.
Re: molvib problems
jb007 #14120 04/22/07 08:26 PM
Joined: Sep 2003
Posts: 250
Forum Member
Offline
Forum Member
Joined: Sep 2003
Posts: 250
0ne more thing:

Since you have one s and two p orbitals that overlap in the triple-bond, you should probably expand the basis-set to account for diffuse and polarization functions, e.g. hf/6-31++g(2d, 2p) - or whatever level of theory you want to use - and then push the button that says "enter" on the keyboard.

Re: molvib problems
jb007 #14121 04/23/07 04:17 PM
Joined: May 2005
Posts: 18
M
myriad Offline OP
Forum Member
OP Offline
Forum Member
M
Joined: May 2005
Posts: 18
Thank you very much for your suggests.

I don't know how to do the iterative refinement procedure as I am not familiar with Gaussian. As this model is small, I think I can try different input options from the same structure. I tried HF and MP2, with 6-31G* or 6-31++g(2d,2p). The molvib calculations gave similiar frequency results from these Gaussian outputs. I couldn't find any significant improvement for a high level of theory/large basis set. Is there anything needs correcting in my Gaussian/Charmm calculations?

Thanks.

Shunzhou

Re: molvib problems
myriad #14122 04/23/07 06:24 PM
Joined: Sep 2003
Posts: 250
Forum Member
Offline
Forum Member
Joined: Sep 2003
Posts: 250
Provided you have done at least one geometry-optimization and a frequency calculation, you'll simply just use the checkpoint file created from that calculation; again:

#p hf/6-31g* geom=check opt(rcfc) [other options]

It might be a good idea to start on a lower level of theory and then optimize at that level until the structure is stable.

Re: molvib problems
myriad #14123 04/26/07 04:00 PM
Joined: May 2005
Posts: 18
M
myriad Offline OP
Forum Member
OP Offline
Forum Member
M
Joined: May 2005
Posts: 18
To make sure, in the *_molvib_g03.inp, the coordinates under CART are from Gaussian frequency calculation's output "Standard orientation"; while the F matrix is the data under "Force Constants in Cartesian Coordinates" in the gaussian output (delete the first three blanks). Is it correct?

Thanks.

Re: molvib problems
myriad #14124 04/28/07 10:08 PM
Joined: Oct 2003
Posts: 780
A
Forum Member
Offline
Forum Member
A
Joined: Oct 2003
Posts: 780

no, use the coordinates from the Input or Zmatrix orientation for molvib. deleting the first 3 characters in the Force constants is correct.

alex


School of Pharmacy
University of Maryland
20 Penn Street
Baltimore, MD, 21201

Moderated by  alex, chmgr, John Legato 

Link Copied to Clipboard
Powered by UBB.threads™ PHP Forum Software 7.7.4
(Release build 20200307)
Responsive Width:

PHP: 5.6.33-0+deb8u1 Page Time: 0.014s Queries: 28 (0.008s) Memory: 0.9419 MB (Peak: 1.0486 MB) Data Comp: Off Server Time: 2020-05-28 19:11:58 UTC
Valid HTML 5 and Valid CSS