[Ifeffit] charge transfer by SCF in FEFF8 and simulation for charge disproportionation
Scott Calvin
SCalvin at slc.edu
Thu Aug 28 15:35:42 CDT 2008
Hi Ying,
I think maybe the problem isn't so much with FEFF as with a century of
conventions in chemistry.
Until fairly recently, no one really knew where electrons were sitting
in real materials. Chemistry "works" pretty well if you do things like
assign formal oxidation states and apply octet rules. But as a
physicist, you know that the actual wave functions aren't going to be
that simplistic.
As both theory and experiment advanced, it became very clear that even
the most stereotypically ionic substances don't actually consist of
isolated ions interacting with their classically assigned charges. I
guess you can say that implies a partially covalent character, but
electrons are indistinguishable, so there aren't really individual
"pairs" of electrons in covalent bonds either. The whole solid is
actually a great big messy quantum mechanical entity.
That does not mean that models like formal oxidation state, chemical
bonds, and the like aren't extremely useful. Personally, I consider
myself a chemical physicist, not a solid state physicist, and I much
prefer the bonding description to, for example, a model based on
"bands" for understanding in my own head what is happening.
But the bottom line, to me, is that you aren't going to find those 1.7
electrons that you think should be involved in the bonding somehow in
the FEFF SCF calculation, nor do they really belong. FEFF, as I
understand it, doesn't have interactions between electrons in
different atoms at all (bonding and antibonding orbitals, if you like
that kind of description), and thus only needs to know the net charge
transfer. "Covalent bonds" have no meaning in FEFF.
OK, FEFF experts, how'd I do? :)
--Scott Calvin
Sarah Lawrence College
On Aug 28, 2008, at 4:03 PM, Ying Zou wrote:
> Dear all,
>
> I am doing a SCF calculation by Feff8.40 for a model compound
> PbTiO3. However, the obtained charge transfer deviates far away from
> its well-known formal oxidation state. According to common electron
> negativity data, the oxidation state would be:
>
> Pb 2+
> Ti 4+
> O 2-
>
> The SCF run gave out charge transfer for those atoms as following:
>
> Pb 0.717
> Ti 0.181
> O -0.300
>
> After I read through the existing threads on charge transfer in this
> list, I know that transfered charge FEFF gives out is just a portion
> of general formal oxidation state. Can we undderstand that this is
> just ionic portion for entire bonding? Such as for oxygen, "-0.300"
> is ionic bonding, while the rest of valency (-1.700) exists as
> covalent bonding, which doesn't have apparent charge transfer by
> definition.
>
> Another related question is then: is there any way in FEFF to take
> into account this "covalence" part? (maybe deriving from overlapping
> area of mufftin-potential?) I think this may be crucial for a self-
> consistent analysis of charge disproportionation.
>
> Any comments on this would be greatly appreciated. Attached please
> see"atom.inp" and "feff.inp". Thanks so much!
>
More information about the Ifeffit
mailing list