[Pw_forum] On Lowdin charges (again)

[glevita at units.it]

Dear Users,
in my investigations on the electronic structure of bulk MoS2 I  
encountered a surprising outcome from my Lowdin charge analysis.
Using for Mo an ultrasoft PBE pseudo with semi-core state in valence  
(I think I have correctly converged all the calculation parameters) I  
get the odd result that the metal atom is slightly negatively charged  
(about 0.46 e-) w.r.t. the sulfur centre (which is about 0.29 +).
The p orbitals of the Mo atom are filled by 7 electrons, which is odd as well.
A similar reversed behaviour is shown by considering a single S atom  
on a 3-layered iron slab.

While I do not expect the absolute numbers to be too accurate, I am  
nevertheless confused by the reversed, counterintuitive polarity. I  
also noted that if I just take into account the outer valence orbitals  
(that is, 4d for Mo and 3p for S) the situation gets slightly better,  
as in this case the S atom shows a 0.12 e- accumulation on the 3p.

Do you think that for covalent compounds such as MoS2 a reliable  
Lowdin analysis is rather unfeasible or - as I think - I am missing  
some point in evaluating its outcome? When I consider charge  
transfers, should I focus on the outer valence states and leave the  
total charge?

Thanks a lot for any comment. Have a nice WE!
  Giacomo Levita

(post-doc, CNR-Nano, S3 Modena, Italy)