Hi Lisa,
Without thinking too much about your particular structure (I wouldn't be surprised if someone on the list chimes in with that kind of info by the time I finish composing this!), I am very confident it is not a fault in the FEFF code.
Ad you probably know, chi(R) is NOT a radial distribution function, although they are related. For one thing, multiple-scattering can play a role, and it may be that in one of the sites multiple scattering may be enhancing the peak--perhaps there are more sulfur atoms collinear with the gold-gold paths, for instance. In addition, chi(R) is complex valued, meaning that different paths can reinforce or cancel in a non-linear way if you just look at the magnitude. In other words, the chi(R) magnitude of the combination of Au1 - Au2 and Au1 - Au3 might be larger than the sum of the individual magnitudes of the individual magnitudes or smaller.
About a decade ago, I analyzed the structure of elemental manganese using EXAFS. Elemental manganese has one of the nuttier crystal structures you'll find in a pure element, with four crystallographically distinct absorbing sites. Because of interference between scattering paths, the "importance" of each site to the fit varied by up to 400% from what would be expected due to abundance.
I never published that result, but I do have the pdf of my presentation to the APS on it, so I'm attaching that.
I'd guess your effect is either something like that or an effect of multiple-scattering.
--Scott Calvin
Sarah Lawrence College
On Feb 3, 2014, at 1:27 PM, Lisa Bovenkamp
Hello.
I am running some EXAFS simulations Au L3 on Au clusters with SR coating. In the outer shell Au sits between 2 S atoms. 4 Au atoms are in the neighborhood up to 3.5 Angström. Expecting two peaks in the chi(R) - one for Au-S and one for Au-Au (here it can be several little ones) I find that for 1 out of 5 sites the Au-Au peak is way too high. Is something in the FEFF code weighing an Au-Au path for this site too much? Does anyone have some ideas on that?
Thanks, Lisa