Eugenio,


On Sun, May 24, 2015 at 5:26 AM, Eugenio Paris <eugenio.paris@gmail.com> wrote:
Dear all,

i have a question about polarization vector in performing the FEFF calculation in Artemis.
I am studying the system CeOBiS2 using Bi L3-edge EXAFS. Starting from *.cif file i'm creating the *.inp file which atoms list come out as follows:

 ATOMS                  * this list contains 35 atoms
 *   x          y          z     ipot tag           distance
    0.00000    0.00000    0.00000  0  bi            0.00000
    0.00000    0.00000    2.50509  4  s2.1          2.50509
    2.81711    0.00000   -0.12006  4  s1.1          2.81967
   -2.81711    0.00000   -0.12006  4  s1.1          2.81967
    0.00000    2.81711   -0.12006  4  s1.1          2.81967
    0.00000   -2.81711   -0.12006  4  s1.1          2.81967
    0.00000    0.00000   -3.31180  4  s1.2          3.31180
    ...

Bismuth has a S atom sitting on top of him (s2) and it is square coordinated with other four S atoms (s1).
I made measurements on single crystals using fluorescence yield. The experimental geometry consist of the x-ray beam coming at (almost) normal incidence which corresponds to electric field vector lying on the crystal's ab plane.

If I perform the FEFF calculation without taking account of polarization I obtain the first scattering paths which are

1) @S2.1@, Degen = 1, Reff = 2.5051, Rank = 100
2) @S1.1@, Degen = 4, Reff = 2.8197, Rank = 100
3) @S1.2@, Degen = 1, Reff = 3.3118, Rank = 14.78
4) ......

When I insert the line "POLARIZATION 1 0 0" in the imput file i get

1) @S2.1@, Degen = 1, Reff = 2.5051, Rank = 100
2) @S1.1@, Degen = 4, Reff = 2.8197, Rank = 100
3) @S1.2@, Degen = 1, Reff = 3.3118, Rank = 3.67
4) ......

which is exactly the same result for the first 2 scatterings, even if the rank of higher shells is changed...

Now, my question is:

If the components of the polarization vector has to be written in the same reference frame of the atom list i'm expecting to see a strong suppression of the rank factor for the @S2.1@ scattering whose bond is along the z-direction, is it correct?
 


There are two different things going on that can help explain this.    But first:  it wasn't clear what you used to specify the polarization in feff.inp.  You probably used either
     POLARIZATION 1. 1. 0.

or   
    POLARIZATION  1. 0. 0.

If that assumption isn't correct, please clarify.

First, Feff's reporting of "Rank" is really "Intensity of this path compared to the most intense path seen so far".   The first path is always 100.0, and if any later path is 100.0 it indicates that it had more intensity than the previous highest-intensity path, but you have no idea of how much more intensity.   It's a really lame report on intensity and should be fixed.  Because of this, the only meaningful number in your list is the Rank of path 3, which does actually go down.

Second, you're asking for a calculation of an L3 edge.  The L3 edge includes both p->s and p->d transitions, which means the polarization dependence is a mixture of a dipole-like term a constant term.  So, polarized measurements at the L3 edge does not completely suppress contributions normal to the polarization vector, as it would for a K or L1 edge.    For your system, it looks like the out-of-plane contribution is suppressed by about a factor of 4 relative to the contributions from the S1.1 atoms in the X-Y plane.    I think that's not unusual for L3 edges, but the literature on this is scant.

Hope that helps,

--Matt