[molpro-user] ScO bond length and basis sets

[Kirk Peterson]

Dear Emine,

just selecting a relativistically contracted basis set like cc-pVTZ-DK does not change the underlying Hamiltonian, so yes you need to add another keyword.  I think this is probably most of the problem since the DK contracted sets will perform very poorly in non-relativistic calculations.  Somewhere before the rhf directive, just put a line that has   dkho=2
That will turn on the 2nd-order Douglas-Kroll-Hess Hamiltonian.  

I think the rest of your input looks ok, however I have not double-checked your occ, closed, and core cards.

best regards,

-Kirk


On May 15, 2014, at 7:36 AM, Kucukbenli Emine <emine.kucukbenli at epfl.ch> wrote:

> Dear All,
> 
> I am not sure if this is the appropriate place for this question but I would be very happy if you could find the time to answer:
> 
> I am trying to determine the bond length of Scandium oxide dimer at its ground state. The ground state is known to be a sigma state with 1 unpaired electron. The experimental bond length is 
> 1.668 Angstrom. 
> 
> There is a very nice paper on the whole 3d transition metal oxide dimers by Bauschlicher and Maitre (http://link.springer.com/article/10.1007%2FBF01113847) that reports the bond length with UCCSD(T) as 
> 1.680 Angstrom. 
> 
> I perform UCCSD(T) calculations as well, and as they did, I also correlate 3s and 3p electrons using the cc-pwcvXz kind of basis sets and using the "core" directive.
> 
> That paper also suggest that at the MCPF level of theory, relativistic effects are very small. To account for them anyways, I use basis sets that are like cc-pwcvXz-DK (but I do not add any other directive to the input file - is it necessary?)
> 
> Considering it is an ionic environment for oxygen, for it I use the same basis set but augmented with diffuse functions, such as (aug-cc-pwcvXz-DK). In Bauschlicher work the diffuse g function is deleted but I don't know why that is done or how, so I simply used the aug- basis set as is.
> 
> My resulting input files look like the following:
> 
> ***,ScO
> geometry={sc;o,sc,r}
> r=1.66 ang
> basis,sc=cc-pwcvqz-DK,o=aug-cc-pwcvqz-DK
> {rhf; wf,29,1,1}
> {uccsd(t); occ, 9,3,3,0; closed, 8,3,3,0; core, 4,1,1,0;}
> {optg, GRADIENT=1d-5, ENERGY=1d-6}
> 
> The geometry optimization yields the following results:
> 
> basis set  vs  bond length in angstrom
> TZ : 1.657 A   #cc-pwcvTz-DK/aug-cc-pwcvTz-DK 
> QZ : 1.659 A    #cc-pwcvQz-DK/aug-cc-pwcvQz-DK
> 
> Before going and doing the time consuming 5Zeta basis set calculation I wanted to understand whether I am doing something wrong here, as my results are far from the previous calculations of Bauschlicher and Maitre.
> So I tried to following to see the convergence behaviour of the basis set:
> {uccsd(t); occ, 9,3,3,0; closed, 8,3,3,0; core, 6,2,2,0;}
> i.e. using the same basis set but not correlating the 3s and 3p; for which the calculations are cheaper. 
> This is what I get in this case:
> 
> basis set vs Bond length in Ang.
> Tz:  1.680   #cc-pwcvTz-DK/aug-cc-pwcvTz-DK 
> Qz:  1.689   #cc-pwcvQz-DK/aug-cc-pwcvQz-DK
> 5z:  1.678   #cc-pwcv5z-DK/aug-cc-pwcv5z-DK
> 
> Can this even be concluded to be a convergent behaviour?
> 
> Being quite inexperienced in both the theory and the code I cannot disentangle whether it is a problem on the theoretical level or my usage of the code. 
> If you could point to what I am doing wrong here I would be very happy!
> 
> Many thanks in advance,
> 
> Emine Kucukbenli,
> Institute of Materials, EPFL, Switzerland
> _______________________________________________
> Molpro-user mailing list
> Molpro-user at molpro.net
> http://www.molpro.net/mailman/listinfo/molpro-user

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.molpro.net/pipermail/molpro-user/attachments/20140515/6c159d81/attachment.html>