Moving the Fe atom off-center as if it were the Ti atom in a perovskite isn't a good way to simulate JT distortion. In Fe3+ compounds such as goethite (I assume Fe3+, not the rare Fe4+), the distortion consists of the two axial distances being shorter than the 4 equatorial. You could simulate the 1st shell by moving two of the O atoms artificially. Alternatively, in Artemis, you could clone the Fe-O 1NN path and set a parameter for the JT distortion, but that wouldn't get the MS or higher neighbors correct. Does anyone know of a program which would be like ATOMS, except with parameters, thus generating parameterized distances? The beat you get from the JT distortion will make a big difference in the fit quality once you get to something like 14A^-1. mam On 8/8/2011 6:21 PM, Matt Newville wrote:

Hi Enyuan,

On Mon, Aug 8, 2011 at 7:47 PM, Enyuan Hu wrote:

...

Dear all, I'm trying to fit the Fe K-edge EXAFS data and I appreciate that Fe in my sample mainly exists as Fe4+, which is associated with Jahn-Teller distortion. The one shell model that considers Fe residing in the standard octahedral apparently didn't give satisfactory result. I'm sorry that even though I found people do fitting considering Jahn-Teller distortion, I am still not aware of how to carry that out in Artemis. Can someone helps me to look at my fitting and give me some guidance on how to implement fitting that considers Jahn-Teller distortion in Artemis? Thank you.

I noticed a few things that might lead you to thinking about the problem better:

You started with the crystal structure for LiCoO3, which has octahedral symmetry, and calculated for the Co edge, even though your data is for Fe edge (and one can see the Co K edge showing in the chi(k) data at ~12.5 Ang^-1).

What you want to do is *edit* the feff.inp that results from running Atoms on the crystallographic data. When doing, you will definitely want to make the central atom Fe, by changing

POTENTIALS * ipot Z element 0 27 Co 1 3 Li 2 27 Co 3 8 O to

POTENTIALS * ipot Z element 0 26 Fe 1 3 Li 2 27 Co 3 8 O

While you're in there messing with feff.inp, you can introduce your own hypothetical Jahn-Teller distortion, by moving the absorbing atom away from 0, 0, 0, from

ATOMS * this list contains 103 atoms * x y z ipot tag distance 0.00000 0.00000 0.00000 0 Co 0.00000

to perhaps

ATOMS * this list contains 103 atoms * x y z ipot tag distance 0.00 0.04 0.00 0 Fe 0.00000

(though I'm not at all sure I have the direction for the distortion correct in the R -3 m structure). That will lead to a splitting of the Fe-O distances, for the distortion you're expecting.

I also guessed at the size of that distortion. Since the resolving power of EXAFS (especially with data only out to 12.5Ang^-1) is limited, so you probably will have a difficult time distinguishing the split in distance from an increased sigma2. That is, you may hae trouble proving that there is a Jahn-Teller distortion from the EXAFS data alone.

My final pieces of advice would be: don't use dk=1 with the Kaiser-Bessel window (it makes an otherwise ideal window function horrible) -- use dk=4 or dk=5.

Hope that helps,

--Matt

_______________________________________________ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit