DFT+U

[apple]

Hi all,

Is the GGA(LDA)+U method applicable only to systems with partially filled d or f electrons? What about the systems with fully filled d or f orbitals? Could someone please explain. I asked this question before. Someone replied saying it should not matter, but I noticed that most of the published papers with DFT+U consider only partially filled d or f shells. Can someone comment on why it is the case (or especially the case) for partially filled shells?
I would greatly appreciate your help.

[boris]

Hi

I see no reason why DFT+U should work better with partially filled orbitals than fully filled orbitals. As soon as electrons are localized, DFT+U should work.

Maybe it has something to do with magnetism? I know that the double counting correction term in DFT+U has an influence on the magnetic moments.

Regards

[rgc]

I think the reason why GGA+U is usually applied in systems with partially filled d or f orbitals is that its effect is more dramatic in this case. The method works by penalising the mixing with the ligand orbitals, in such a way that the d or f density ends up being more localised (opposing the artificial trend of DFT to delocalise them). Normally the mixing of filled shells with the ligands' orbitals is less important, or less interesting in chemistry. Does this make sense?

[apple]

Thanks for the reply. It became clearer to me now.