I'd add to Bruce's explanation that the numbers usually end up being qualitatively similar to what you'd expect from oxidation states, but much smaller than the oxidation number. Saying iron is in a +2 oxidation state, for instance, doesn't really mean two electrons have been completely transferred from the iron atom to the ligands. You'll generally see FEFF come up with a much smaller positive number than +2 for iron in that case, but it will still recognize that the iron is positive. And although I haven't paid close attention to this, I'd expect in a system with mixed +2 and +3, that FEFF would probably assign a bigger charge transfer to the +3 in most cases. --Scott Calvin Sarah Lawrence College On Jan 2, 2009, at 6:30 AM, Bruce Ravel wrote:

On Friday 02 January 2009 09:15:05 am Bruce Ravel wrote:

...

what does charge transfer mean in FEFF? I cant understand what the number of charge transfer like 0.058, -0.073 means. The number indicate the number of electron in atom?

If anybody knows it, please answer

Charge transfer: iph charge(iph) 0 0.058 1 -0.073 2 0.072

Jeong,

Feff8 does a trick called "self-consistent potentials". It works like this: Feff8 sarts by computing the muffin tin potentials in a many rather similar to Feff6. It then computes the electronic densities of states for a reasonable set of low-value angular momentum states (usually up to l=3). With those DOS calculation, it integrates in energy to find the energy at which the integral of the DOS equals the number of valence electrons in the system. It is likely that each atom has a different number of electrons than the free atom.

Now each atom is differently charged than the starting configuration. The muffin tin potentials are recomputed, the DOS is recomputed, and the integral is done again. This process is repeated until the charge on each atom stops changing. That is the sense in which it is self-consistent.

At the end of this self-consistency loop, Feff8 reports the net change in charge (in units of number of electrons) on each atom type. In your case, to attain self-consisteny, a small bit of charge is added to the absorber and a small bit of charge is taken away from the other two atoms.

B

--

Hi Gleb, As a workaround, can you check the "fix step" box and adjust the edge step by hand that way? --Scott Calvin Sarah Lawrence College On Jan 4, 2009, at 2:00 AM, Gleb Pokrovski wrote:

Dear Bruce,

Since the 2 or 3 latest athena versions, I have noticed a somewhat strange behavior of the 'plot multiplier' command in athena. I'm using the latest horae and iffefit releases on linux (Madriva 2007) and windows (XP, SP3).

The plot multiplier works for chi and fourier-transformed spectra, but fails for xanes and xmu spectra: changing the plot multiplier to a value different from 1 makes no effect of the last two types of spectra. In contrast, 'y-axis offset' command works for all types of spectra.

Personally, I found the possibility to multiply xanes spectra extremely convenient (as far as I remember, it worked in older than 2006 versions of athena....), for example for quick comparison and superposition short-energy range XANES spectra for which the athena energy normalisation procedures do not work perfectly in some cases.