[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Unexpected direction of Charge Transfer
Dear FEFF users,
First, how the feff difines the Charge Transfer:
We take density say for Ni atom and S atom. We overlap atomic densities
and find Norman radius for Ni and S. (Radius of a neutral sphere
around Ni and S) correspondingly. After self-consistency loop we
look how much charge is inside each sphere. One expects at least direction
of the charge transfer be in agreement with electronegativity
of the element. For example, always when I run oxides I get -0.2 to -0.4
charge on oxygen.
I took ZnS feff.inp file, which you can get from Webatoms database.
There the charge transfer was the expected one.
When I put Ni instead of Zn, the charge transfer changed sign.
So the problem was with Ni and not with S.
Then I changed occupation numbers in subroutine getorb, which
defines the atomic configurations. (Initial density depends on them.
FEFF has 3d^8 4s^2. If I start with d^9 s^1 or d^8s^1p^1 configurations
the charge on S is negative, as one can expect.
Notice that if I ran self-consistency loop my finite charge density
should be the same, and x-ray absorption should be independent
of my initial atomic Ni configuration. however, my initial
atomic densities define Norman radii, and it seems that the choice
of atomic initial configuration is not good for Ni in compounds.
We already planned to fix future version of FEFF so it has more
appropriate (for solid state applications) initial atomic configuration.,
which are specified in subroutine getorb. So please, report
to us any other unexpected sign of Charge Transfer.
Best wishes
Alex Ankudinov
On Wed, 23 Jun 2004, Gilles Hug wrote:
>
> Le 17 juin 04, à 08:13, Siew Wei Goh a écrit :
>
> > Dear FEFF Users,
> >
> > I'm hoping someone can help me with a problem which continues to
> > baffle me. It involves charge transfer calculations on Fe, Ni and Co
> > atoms in sulfides, eg. NiS. For some reason, FEFF consistently
> > calculates the charge transfer as negative on the metal and positive
> > on the sulfur counterpart! I do not understand why FEFF does that.
> > Shouldn't S atoms be negative while the metal be positive?
> In principle you are right, electrons should be transfered to the S
> atoms to close their shells. However, this is a principle not a
> demonstrated rule. There is no reason why it should be like that all
> the time.
>
> >
> > I have tried modifying a number of parameters by changing the degree
> > of muffin-tin radii overlap using either AFOLP or FOLP cards, shifting
> > the Ef relative to DOS using EXCHANGE card as well as using the 5
> > different available exchange models. None corrected the problem I am
> > facing here. Interestingly, this problem is not observed when dealing
> > with copper and lead sulfides or even in Fe & Ni oxides. Can someone
> > shed light on this behaviour?
> Charge transfer is a rather weird parameter, which is not measurable
> btw. It's value and even sense can depend on its definition. Maybe the
> best thing to do is to compare FEFF's calculation with some other
> theory.
> Can you cross-check with some other code like LMTO, APW, pseudo-pot or
> whatever?
> If not we can certainly possible to run for you a calculation for
> reasonably small unit cell.
> Cheers
> Gilles
>
> >
> > Regards
> >
> > Siew Wei GOH
> > School of Chemistry
> > University of New South Wales
> > Sydney NSW 2052, Australia
> > Tel: +61 (2) 0385 5519
> > email: sw.goh@unsw.edu.au
> >
> >
> Dr. Gilles Hug
> LEM ONERA-CNRS
> BP 72
> 92322 Châtillon
> France
> +33 1 46 73 45 42
> gilles.hug@onera.fr
> http://www.onera.fr/visiononera/derniere/laboratoire.html
>