[Ifeffit] Fwd: Questions about Athena & FEFF

Bruce Ravel bravel at bnl.gov
Wed Apr 15 16:36:14 CDT 2009


I am answering your question on the Ifeffit mailing list, which is the
most appropriate venue for questions of this sort.  Please read this


If I understand your question, you are asking about whether it is
appropriate to run Feff with a small cluster.  The canonical answer to
that question is that you want to use a large cluster whenever
possible so that the muffin tin potential is well constrained.
Sometimes, with a very small cluster, the algorithm Feff uses to
compute the muffin tin radii can result in radii that are unphysically
large, resulting in an unphysical calculation.

Of course, that is a somewhat hand-waving answer.  It is hard to know
without actually running a calculation whether the problem with muffin
tin radii will actually happen.  I have comfortably published results
which use very small clusters indeed.

In your case, as in so many cases, the best advice I can give is to
try it both ways and see what happens.  That is, you can try doing a
XANES calculation (Are you using Feff8?  You didn't say...) with the
small cluster and comparing it to the Feff8 calculation which uses a
large cluster to compute the potentials and small cluster to compute
the full multiple scattering XANES.  See the Feff document at


for details about how to set parameters in the feff.inp file to make
that happen.


----------  Forwarded Message  ----------

Subject: Questions about Athena  & FEFF
Date: Wednesday 15 April 2009
From: "Pascal G. YOT" <pascal.yot at univ-montp2.fr>
To: bravel at bnl.gov

Dear Prof. Ravel,
I'm working on data coming from XAFS measurements carried out at Pb-L3 
edge on lead containing glassy samples.
Three reducing agent are used to reduce lead(+II) inside the original 
glass at high temperature.
Our samples contain then a fraction is in fcc lead(0) form and one other 
fraction  of lead(+II) still remaining inside the glassy framework.
EXAFS analysis show a decrease of the Pb-O coordination number from 3 
(for original glass) to lower CN (1 in extreme case).
My major concern is the analysis of XANES, some modifications are 
observed on the finger prints.
I've try to fit the spectra using linear combination fitting module from 
Athena using lead and original glass spectrum with poor result as you 
have understand.
I need then to generated model XANES spectra for the CN. As I'm not an 
expert of FEFF, I need some advices.
Can I use a cluster PbOx ((d=2.23 A) with coordinate for Pb and O coming 
from lead oxide) containing few atoms? Is the obtained spectra reliable 
to give a "good" fit?
I hope my questions are clear for you.
Thanks in advance for your remarks.
Best regards
Dr. Pascal Yot



 Bruce Ravel  ------------------------------------ bravel at bnl.gov

 National Institute of Standards and Technology
 Synchrotron Methods Group at NSLS --- Beamlines U7A, X24A, X23A2
 Building 535A
 Upton NY, 11973

 My homepage:    http://xafs.org/BruceRavel
 EXAFS software: http://cars9.uchicago.edu/~ravel/software/exafs/

More information about the Ifeffit mailing list