[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: With and without SCF and FMS on Cu

Dear Jerome,

  There were two problems with your feff.inp

  a)  you specified an L3 edge:
EDGE      L3

  whereas the 1992 PRL discussed the Cu K edge with the path expansion.

  b) your cluster size  of 5.61 Ang only has 55 atoms.
FMS        5.61585  0

   Typically one needs more for XANES, e.g., as many as the FEFF8's
default limit of 87 atoms -- or more.

  I was able to roughly reproduce the 1992 paper results with 87 atoms,
FMS  6.5  0


 Indeed we find that for L-shell spectra in transition metals one needs about
150 atoms, which requires a recompilation or a new executable. Several such larger
executables are available for WIN-PC and Macs from the FEFF WWW download site.

 Finally for L-shell and more shallow edges, one sometimes gets better results
without a fully screened core hole - i.e., the NOHOLE card.  We're currently
working on an improved treatment of the core-hole screening which with the goal
of handling this automatically.

  J. Rehr


On Wed, 11 Aug 2004, Jerome Kartham Hyun wrote:


I have been trying to output the lDOS for Cu metal using feff8.2. However
the results seem to be unreliable given that my chi and xmu data deviate
substantially from the published results (Rehr et al, 1992,  Phys. Rev.
Lett. 69, 3397). Without the SCF and FMS cards and using the MS path
expansion, I am able to reproduce the published data easily. Attached is
the feff input file I used to calculate the lDOS. Since my ultimate goal
is to output an accurate lDOS for Copper I would like to keep the SCF, FMS
cards. Any suggestions on what card parameters to tweak or whether I am
completely missing something? Thanks so much.