If I used 194 atoms (Al-Si-O) for simulation of the large molecules
the Gaussian did not converge. I want to know whole atoms are needed in simulation?
That is a rather unspecific question so there are many reasons of which the most important ones are:
1) If you use this model as part of a solid, you either have to use periodic boundary conditions (not sure how good Gaussian does that) or you have to saturate your crystal surface (e.g. with hydroxyl groups). Otherwise you have unpaired electrons which tend to cause trouble with SCF calculations.
2) If your starting geometry is too far from a rasonable one, SCF calculations also tend to have convergengy difficulties. Have you preoptimized your slab with a semiempiric method?
3) Your initial guess wavefunction is not appropriate. You can look into the Gaussian manual what other guesses you might use.
Dear Weippert
Thanks for your reply.
1) This large molecule is a part of solid. In some literature
the Si atoms in border were saturated with hydrogen, I did it. But didn't converge. But I didn t hydrogenate oxygen atoms.
I don't know how unpaired electrons.
2) Yes I used semi-empiric method but didn't work. I used ONIOM method and divided zeolite to two layers but ...
3) I try it.
I attached input file.
Hallo, Morad!
Your cluster has a huge positive charge (+62 if I correctly summarized its composition) - its SCF cannot converge if you force this cluster to be electroneutral. You must prepare a model cluster with a much smaller total charge (preferably electroneutral) and to use the correct charge value in your calculations.
Best wishes
mb
The charge is probably the 0 that comes before your geometry input. Your SCF probably still won't converge if you correct it to +62, that's why I recommended hydrogenation (that's the easiest way to get electroneutral). Apart from charge you still have to mind the multiplicity (probably the 1 after the 0) and if your cluster isn't neutral you can't guarantee the most stable configuration is a singulet.
- Badges
- Science topic