I use Gaussian to obtain the partial charges from a slab of aluminum oxide (corundum) with a hydroxide surface.
the keywords that I use is
# b3lyp/6-31+g(d,p) nosymm pop=npa geom=connectivity scf=xqc
but the convergence is not met. I want to know if im using a proper basis for this crystal
It depends on what properties you're interested in. The non-convergence may not necessarily be directly related to the basis set, though this statement needs to be qualified.
I would first just run a regular HF calculation and see if you can get that to converge, if it does, save that solution as your initial guess and read it in to start your DFT calculation. If HF doesn't converge, back off to a smaller basis like 6-31G* and try that. For DFT you really should be running unrestricted and once you have a converged HF solution you can run a stability check as well to make sure you have the proper solution for your system to begin with.
As Francesc mentioned, you might also want to consider a custom basis set with different functions on different atoms. Again, this will depend on what property(ies) you are interested in.
Hi the only result that I'm interested is the partial charges from the NBO analysis.
I'll try a custom basis separating the Al from the O, and for the Hydrogen also without diffusion functions?
I would just use 6-31G* (others may disagree, though this is tricky since most charge analyses are not unique and can vary depending on the basis used). If the solution looks like it's converging, try to use the point where it fails as your restart and/or up the number of scf cycles (forgot to mention this before).
- Badges
- Science topic