[Ifeffit] ATOMS: Partial occupancies -how to edit .cif File

[Matt Newville]

Hi Kai,

On Tue, Apr 23, 2013 at 6:41 PM, Sanwald, Kai <kai.sanwald at tum.de> wrote:
> Hi,
>
> I would like to fit the NiFe2O4 spinel structure to EXAFS data of mixed
> Ni/Fe-(hydro)xide films.
> However, ATOMS cannot deal with structures exhibiting crystal sites with
> partial occupancies.
> I know that this problem has been adressed to some extent on the
> mailing-list already in May 2012
> (see
> http://millenia.cars.aps.anl.gov/pipermail/ifeffit/2012-May/010544.html)
> and Bruce Ravel shortly outlined how to edit the .cif file in order to run
> ATOMS and FEFF6.
> Attached you find the NiFe2O4.cif file that I like to employ. I tried the
> following:
>
> 1. changed occupancy from 0.5 to 1 in line
> Ni1 Ni2+ 16 d 0.625 0.625 0.625 . 0.5 0
>
> 2. subsequently deleted line
>
> Fe2 Fe3+ 16 d 0.625 0.625 0.625 . 0.5 0
>
>
> I got rid of the ATOMS warning concerning the partial occupancy but FEFF6
> crashes due to atoms being postioned too closely to each other. I am pretty
> sure that my editing resulted in a change of the structure.
>
> I would be very glad if someone could explain to me in some detail how I
> need to edit the .cif.
>
>
> Thank you very much for your help.
>
>
> Best,
>
> Kai
>
> _______________________________________________
> Ifeffit mailing list
> Ifeffit at millenia.cars.aps.anl.gov
> http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit


This question comes up quite often.  I think you even pointed to the
places that have good answers....

Partial occupancy has a reasonably straightforward meaning for
crystallography.   Well, a simple interpretation of partial occupancy
is reasonably straightforward for dealing with simple diffraction
data.   Partial occupancy is not so straightforward meaning for
moderately complex X-ray or neutron scattering, and has no meaning
whatsoever for XAFS.  Simply put, XAFS does not use concepts such as
"site" from crystallography.

With Feff and the analysis programs that rely on it, you calculate the
XAFS for exactly one selected absorbing atom in exactly one cluster of
atoms.  Each atom is either there or not.   There is no partial
occupancy, there is no randomness, and there is no averaging over
crystallographic sites, because crystallographic sites are not
relevant.   You can move atoms around in the cluster, introduce
variable occupancy, and average over multiple sites in a cluster by
running calculations for many clusters and averaging the XAFS
contributions.

So, remove the partially occupied site, as these don't make any sense.
  Generate a cluster that does not have overlapping atoms, these don't
make any sense either.   If you want, you can then change the
properties of the atoms (position and Z) in the cluster and run Feff
for a few different clusters and average the results.

Hope that helps,

--Matt