[Ifeffit] EXAFS fitting method

[Matt Newville]

Abhijeet,

On Sun, Jan 30, 2011 at 2:23 AM, Abhijeet Gaur <abhijeetgaur9 at gmail.com> wrote:
> Hi all,
>           As I understand, the EXAFS fitting procedure using Artemis, for
> generating a theoretical model requires the crystal structure parameter like
> space group, cell parameters and specially atom coordinates etc.

No, generating a theoretical model for EXAFS does not require
information about a crystal structure.  It does require a roughly
realistic model for the atomic positions of a cluster of atoms.
Here, "roughly realistic" means that it have inter-atomic distances
for the near-neighbors to the element of interest that are within 0.1
Angstroms or so.   Of course, the species of the neighboring atoms
should be approximately correct too.

It is often easy to start with a crystal structure to generate the
appropriate cluster of atoms.   It is not uncommon to have to modify
the cluster (say, to model Mn in ZnO, you might start with the crystal
structure of pure  ZnO, then change the absorbing atom to Mn).

For many organic and biological molecules, it is easy to find an entry
from the Protein Data Bank that has a similar molecular structure.

For some cases, non-crystalline structures predicted from DFT (or MD)
calculations are most appropriate, and can certainly be used.

> For this we have to perform XRD measurements of single crystals of the samples. I
> have some questions
> (1) In case we dont have the XRD data of a sample then how can the EXAFS
>       analysis be done  of that sample?

Make a guess of the local structure, model that local structure with
x, y, z coordinates of the atoms in that structure, and run Feff.  If
you expect the sample is crystalline, starting with the expected
crystal structure is an easy approach.  If the sample is not
crystalline, you could still start with a crystal structure close to
the expected structure.

> (2) If we have already performed the XRD on the sample then by doing
>      EXAFS what additional information are we getting ?

That depends of the sample.   Perhaps you're looking at an impurity in
that crystal.  Perhaps you're trying to refine the structure beyond
what the bulk crystal structure gives.    It is always a good idea to
know when XRD or EXAFS or some other technique is most likely to
answer the question you're asking.

> (3) If the answer to the first question is that - use the crystal structure
> of a similar sample, then what are the conditions under which we can use
> the known crystal structure of one sample for other samples ?

To the extent that the refinement of the structure makes sense.

--Matt