For analyze extended XAFS data, we use Feff calculations to simulate the
EXAFS scattering factors and then refine structural parameters such as
near-neighbor distance and coordination number for those calculations to
match experimental spectra.   As a theoretical physics code, Feff is a
pretty straightforward program to run to simulate EXAFS, but using it well
can still be one of the more daunting tasks for novices starting out with
EXAFS Analysis.

Hi I'm M N and this is part of the video series for using Larch for XAFS
Analysis.

In this video I'll show you how you can use XAS Viewer to set up and run
Feff calculations for extended XAFS.   I should be clear that while the
Feff codes can be used to calcluate XANES, RIXS, and other X-ray core-level
spectroscopies, the verions of Feff --  are only able to do Extended XAFS
calculations.

For modeling extended EXAFS, we think mainly about the photo-electron
ejected from the absorbing atom in the X-ray absorption process.  This
photo-electron will scatter from the neighboring atoms and it is that
photo-electron scattering and returning to the absorbing atom that explains
the EXAFS oscillations.  It is most convenient to think about EXAFS as a
SUM of PATHS through the molecule or crystal that the photo-electron can
take from the absorbing atom  as it scatters.  That is, there will be a path
to the first neighbor and back, a path to the second atom and back, and
maybe multiple scattering paths where the photo-electron goes first to one
neigher, then scatters to a second and then scatters back home.    All of
the possible paths can contribute to the EXAFS, though typically only 
a dozen or so will have a significant impact.

With the sum-of-paths formalism, you could say that each individual atom
contributes separately.  Or you could group paths of nearly the same path
length into a "shell" of atoms -- say all sixe near neighbors in a
rock-salt structure.  The EXAFS from each path will depend on the
near-neighor distance, coordination number, and some measure of the
distrubution of distances for that shell.  But it will also depend on
non-trivial electron scattering factors that wil depend on the species of
the absorbing atom and the scattering atom.    That gives EXAFS some
sensitivity to Z of the scattering atom, but also means that we need to get
those scattering factors somehow.  Using Feff is the simplest way to get
those.

Feff can be run with an arbitrary cluster of atoms -- you give it the
positions of the atoms, tell it which one is the absorber, and it
calculates the contributions for each path.   This  is what we'll want for
modeling EXAFS, but the question becomes: how do we get those clusters of
aotms?

One convenient way is to start with a crystal structure.  You don't have to
start with a crystal structure - if you have a calculation for where atoms
should be or molecular structure , that fine too -- and for sure EXAFS does
not need to be done on a crystal.  But, in fact, a crystal structure we
start with does not to be that accurate -- we're simulating the local
structure, so the first shells ought be close, but as we'll see, we may
wanto modify this structure anyway.

So, one way to get clusters for Feff calculations is to start with crystal
structures from Crystallographic Information Framework or Format Files  -
CIF files.  Larch comes with a collection of these that can be a good
starting point for generating and running Feff calculations.