This directory contains the crystallographic data for Iron Sulfide
(pyrite, FeS2) along with a transmission scan of FeS2 taken at room
temperature at beamline 13BM at the APS.  The data was taken from
Matt's data archive at
     http://cars9.uchicago.edu/~newville/ModelLib/search.html
     
This is sufficient information to begin a fitting project for FeS2.
In this readme file, I'll write steps to take to work through this
example.  You should not consider this to be a recipe -- at each step
you should play around with the setting in both Athena and Artemis to
understand fully these data.

1.  Fire up Athena.  Import the data file.

2.  Find a set of parameters that gives a good background removal and
    Fourier transform.

3.  Save the project and export the chi(k) file (if necessary)
    a. If you are using a version of Artemis that can read the Athena
       project file, then it is unnecessary to save the chi(k) file.
    b. If you are using a version of Artemis which cannot read Athena
       project files, then you will need to save the chi(k) by
       selecting the correct group in the Athena groups list and then
       selecting "Save as chi(k)" form the File menu.

4.  Fire up Artemis.

5.  Import the chi(k) data.
    a. If you are using a version of Artemis that can read the Athena
       project file, verify that you are still happy with the Fourier
       transform parameters.
    b. If you are using a version of Artemis which cannot read Athena
       project files, set the Fourier transform and fit range
       parameters to sensible values.

6.  Import the crystallographic data contained in the file FeS2.inp.
    Run Atoms.  Verify that the Feff input data is reasonable, then
    run Feff.

7.  When the path selection dialog comes up, any choice is ok.  You
    may find the "first 10" paths option to be the most convenient.

8.  Look at the various pages now available in Artemis.  Note that a
    parameter set has been defined on the Guess, Def, Set page.  Note
    that simple math expressions have been defined for each of the
    paths.

9.  Make plots of the data and the various paths using context menus
    on the Feff interpretation page or by control-clicking on the data
    and one or more paths.  Which paths contribute strongly in regions
    of the first peak in the data?  In the region of the second peak?
    Which multiple scattering paths seem to be strong contributers?
    Which seem like they can be neglected?

10. Set the fit range appropriate for a first shell fit.  Include only
    those path(s) which contribute strongly under t he first peak.
    Hit the big green button and examine your first shell fit results.

11. Extend the fit range to cover the split peak between 2.3 and 3.7
    Angstroms.  Include each of the first four SS paths in the fit.
    Press the fit button.  This fit should look somewhat like the
    data, but not quite.

12. The problem with the previous fit was that it used a parameter set
    appropriate to a first shell fit to fit data with four shells.
    While the amplitude and E0 parameters are probably ok to use for
    all those paths, it is physically unreasonable to use the same
    delta R and sigma^2 for each path.

13. FeS2 is a cubic material (it's space group is Pa3 -- a cubic
    group).  This means we can use the trick of modeling the delta R
    parameters by an isotropic expansion coefficient.  Discard the dr
    parameter on the GDS page and define a guess parameter called
    "alpha".  Define the delR parameter for each path to be
    "alpha*reff".  Remember that "reff" always evaluates to the
    nominal path length as used in Feff when a path parameter math
    expression is interpreted.

14. Define separate sigma^2 parameters for each shell.  Continue using
    the same amplitude and E0 parameters for each path.  You fit using
    the first four shells should now be using 6 parameters.  Try
    running the fit.  Are the values for the parameters reasonable?

15. What other paths contribute significant spectral weight in the
    region between 2.3 and 3.7 Angstroms?  How about the next SS path?
    How about path #6?  Try adding these paths to the fit.  You may
    need to adjust the fit range or define new guess parameters?

16. Can you think of a way to approximate the sigma^2 for the MS paths
    using math expressions, def parameters, and the sigma^2 parameters
    that are used to fit the SS paths?

17. The 2nd and 3rd shells are both S and are relatively close
    together in length.  Test to see if the data support measuring
    sigma^2 independently for those two paths by comparing a fit in
    which they are constrained to be to same to a fit in which the
    float freely.

18. If you look at the atoms information, you will see that this
    structure is actually defined by two parameters -- the lattice
    constant and the position of the S atom.  The variation in the
    lattice constant is accommodated in our fit using the isotropic
    expansion constant alpha.  How might you include the possibility
    of refining the S coordinate in the fit?

19. Explore the effect of different k-weights on the fit.  Try the
    fits using different single k-weights and compare the best fit
    values.  How do they change?  Is this variation outside their
    error bars?  Try doing multiple k-weight fits?  How do the best
    fit values compare?

20. Explore the effect of co-refining the background spline.

21. Try to fit the peak around 5 Angstroms using the model building
    techniques you have employed in constructing your fitting model
    out to 4 Angstroms.  That is, determine which paths are important
    (you may need to import more paths from the calculation), define
    appropriate fitting parameters and make appropriate constraints,
    and set the fitting range accordingly.

22. Are your values for the parameters all reasonable?  Is the
    amplitude a reasonable value for S02?  If it is rather low, can
    you think of any reasons to explain it?  (In the data file header
    it says the data was collected as powder on tape.)
