Hi Matt, Thank you for your answer and for taking time to explain me basic things again and again. I have completely different question as well, and hopefully the last one for a while: I just started using Artemis, and have a question regarding the amplitude reduction factor: I use different So^2 for different elements, correct? If I start fitting and have paths involving the same kind of atom but once at distance 2A, and once maybe 3A, is this then the same So^2 or am I supposed to model the behaviour of So^2 somehow? Similar for \Delta R: If I don’t have a nice simple cubic structure and can’t use isotropic expansion similar like alpha*R_eff, what would be a smart(or physically correct) way to connect \Delta R (as well as sigma^2) for the same element for different distances? The material I am looking at is Li2O2 and Li2IrO3, which don’t have a very simple structure. I am aware that maybe I am asking a bit much, but if you could help me in any way I would be very delighted. Best, Julian Von: Ifeffit [mailto:ifeffit-bounces@millenia.cars.aps.anl.gov] Im Auftrag von Matt Newville Gesendet: Mittwoch, 2. Mai 2018 17:51 An: XAFS Analysis using Ifeffit Betreff: Re: [Ifeffit] Truncating signal Hi Julian, On Wed, May 2, 2018 at 6:32 AM, Julian Ehwald mailto:jehwald@gmail.com > wrote: Hi all, I have basic question on background removal in Athena. My data has some strong changes very close to the absorption edge, and if I start the background removal at k=0, the spline is not able to follow the signal for higher k-values (even at Rbkg=1.7 or similar). The background then oscillates slowly (4 A^-1) around my signal, generating huge peaks in the low regions of chi(R). If I set the spline range to begin at k=2.7 only (around 28eV above absorption edge) then I get rid of it. Now my question in general is, do I loose essential information about the nearest neighbours this way, respectively is that a common thing to do? Best, Julian It should be fine to start the background as high as k=2.7 Ang^-1, though I think most of find that we don't really need to do that very often. You will lose that part of the chi(k), but that part of the spectrum is really XANES, and EXAFS analysis usually starts at around 2.5 to 3.0 Ang^-1 anyway. And, since we typically k-weight by k, k^2, or k^3, that de-emphasizes the low-k part of the spectrum anyway. So, you don't really lose much information about the near-neighbor coordination. It might point to a different or weirder problem (you didn't say which edge you were working with, or where on the spectrum E0 was set), but I'd say that it is probably fine. --Matt