Hi Shan:
Sorry this took so long but I wanted to get my facts straight on the
structures.
From what I can see, Al2O3 and Y2O3 are both more or less octahedrally
(6-fold) coordinated. I would expect that you get very similar results
with both model (as you do). The only difference might be that since you
are starting with a different initial near neighbor distance, the DWF
might be different too. Since you do not see second shell coordination, I
think that the most you can tell from this model is that you have on the
order of 5 oxygen near neighbors to the Y atom in this system.
In looking at the structures of Al2O3 and Y2O3, I noticed a couple of
interesting issues. The first shell in the Al2O3 structure is
# Atom R
3 O 1.8519
3 O 1.9717
The Y2O3, however seems to have two diffente Y sites, and the first shells
are
Site 1 Site 2
# Atom R # Atom R
6 O 2.24752 O 2.2619
2 O 2.2796
2 O 2.3448
You said that you were using relaxed structural models from these two
crystal structures but because there are significantly different
environments (and you can't even tell which of the two Y2O3 environments
to use!), I think you may be getting caught up in bias from your initial
FEFF calculations.
Since you don't see any second shell coordination, why don't
you try this:
Take the final value of the distance in the two models (I hope that you
get the more or less the same thing from both but my guess is that the
final distances should be more like those in Y2O3!) and make a model using
the quick first shell fitting option using that specific value of near
neighbor distance and an octahedral environment. You will then have an
unbiased model to work from. You can then set a reasonable value of the
CN and see what you get. This way of doing things has two benefits. The
first is that your FEFF calculation will start with distances which are
close to the final values. This will actually make a small, but real,
difference. The second is that you simplify the first shell environment
and aren't biasing the fit by having more than one initial distance (you
did not say how those change in the Al2O3 model).
Once you have done this, you can judge how well the fit works and what
kind of results you get. It may indeed be that you need two slightly
differences in distances to fit the data but you have to figure out an
objective method of deciding this.
Hope this helps,
Carlo
On Wed, 20 Dec 2006, Shantanu Behera wrote:
Thanks Carlo and Scott for ur response.
Some numbers from my fits in this regard and some more queries.
When I make a cluster based on the Alumina structure (replacing the
central Al with Y with relaxed lattice parameters) I get DW 0.0025 +-
0.002, without any restraint. While I restrain the DW to 0.005, my CN
jumps from 4.8 to 5.2.
When I make a cluster based on Yttria structure, I get DW factor of
0.008 +- 0.002, unrestrained. While I restrain the DW to 0.01, it
again jumps from 4.8 to 5.2.
First query...if you can pin down on the CN change (Carlo..you wanted
some numbers)
Second...what are ur comments on the above 2 models and the change in
DW factors.
I might add that ther is no appreciable difference with k=2, 3 or 23
fittings. All the fits produce red-chi-sqr less than 100 (~50-70), all
R-factors are 0.01 and no unphysical reports after the fit. And I
have only 1 peak to worry about, there is no second shell correlation
seen.
I would appreciate your 2 cents on this.
Warm regards
Shan
--
Carlo U. Segre -- Professor of Physics
Associate Dean for Special Projects, Graduate College
Illinois Institute of Technology
Voice: 312.567.3498Fax: 312.567.3494
[EMAIL PROTECTED] http://www.iit.edu/~segre [EMAIL PROTECTED]
___
Ifeffit mailing list
Ifeffit@millenia.cars.aps.anl.gov
http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit