problem in optimization of molecule

[rose]

Hi everyone,
I want to find the optimized structure of attachment molecule and surprisingly after few step in optimization the system collapses and everything attract to each other?
why does such thing happen?
thanks
Rose

[Jess Wellendorff]

Dear Rose,
your structure looks basically like a finite chunk of CeO turned into a molecule. The initial forces on atoms are large, > 10 eV/Ang, so it's only natural that you see a significant atomic rearrangement. Two suggestions:
1) you are using SingleZeraPolarized basis sets. Are you certain these are accurate enough for your present purpose?
2) You use non-polarized GGA (GGA.PBE). Are you certain that the true ground state is not spin-polarized (SGGA.PBE)? Just a thought...

[rose]

Hi,
Thanks for the response.

 I thought it might be that so I used a stable reported structure of CeO2 and used DZP- SGGA- PW- with Hubbard 4 on Ce. But the system is reducing bonds so that all atoms are compact to each other.... I really can not understand what is going on that such odd behavior happens. The attachment is the input file.  Would it be possible that this related to charge or sth like that?
thanks
Rose

[Jess Wellendorff]

I do not know exactly what density functional method should be used for CeO2 nanocluster (LSDA? +U ? etc), but it seems pretty clear to me that your initial structure has the atoms too close to each other. Some Ce-O distances are only 2 Å. A unit conversion error?
The attached script shows the LSDA calculation I did for your system. I believe the ground state is spin-polarized. The initial forces are ca 20 eV/Angstrom on each Ce atom, which is quite a lot. The cluster expands towards equilibrium, does not collapse.

BTW: for rare-earth elements, I would try out the SG15 pseudopotentials also. May be better for this sort of studies than the FHI ones.

Hope this helps,
Jess