Re: Cluster Calculations

["John J. Rehr" <jjr@leonardo.phys.washington.edu>]

Hi Norbert,

  Thanks for your question.

> I just got into the calculation of XANES and LDOS for different Iron 
> containing clusters. One question sprang to my mind, which I'd like to get 
> your opinion on:
> 
> If I'm doing something like an Fe(OH)3-O-Fe(OH)2-O-Fe(OH)3 chain, then the Fe 
> atoms in the centre and the ones at the sides have different chemical 
> surroundings. So do I have to run FEFF twice, each time setting one of the 
> two as absorbing atom and average over these two calculations? Or does one 
> run with one of the two centers do the trick (I don't actually think so...)?

   The short answer is YES, as you have already guessed. 

   In general the total x-ray absorption is obtained by averaging over all
absorption sites. If all sites are identical then only one calculation is
needed. Otherwise one has to do a weighted average, based on on the fraction
of a given site.

   One technical point to keep in mind is that the results printed out
by FEFF are normalized by the absorption at 50 eV past the edge, with
a normalization constant printed out in the header of xmu.dat. If the
normalization constant is different for the different sites you also have
to weight the calculations by the normalization constants).

# xsedge+ 50, used to normalize mu           1.9126E-03
# -----------------------------------------------------------------------
# e, e(wrt edge), k, mu=(1+chi)*mu0, mu0, chi @#

  Finally, for EXAFS, one can do a configurational average with the
CFAVERAGE card and only one calculation.

  Cheers,
  John Rehr