Hi Matt,

Thank you very much for answering my mail J

FEFF predicts the first shell to be around 1.9Ang, but processing the data in athena (with various attempts in background subtraction, k weighting and so on) always gives me a strong peak in |chi(R)| around 1A. The data is quite noisy, do you think it is possible that this peak is caused by errors or noise in |chi(k)| for high k-values? I am not sure if this is a valid argument, but high k values correspond to short wavelengths which then are responsible for peaks close to the central atom (this is more of a physical idea)? Can it be that a peak which is too close (like on 1 Ang instead of almost 2Ang) is caused by outliers in high k-ranges?

If so, that would NOT mean then that if I only need information on the first 3,4Ang I only need |chi(k)| for high k values right? Do I understand correctly that the small features in |chi(k)| determine the |chi(R)| for high R values and the slow oscillations in chi(k) determine the R-spectrum for low R-values (more of a mathematical thought)?

Best,Julian

Von: Ifeffit [mailto:ifeffit-bounces@millenia.cars.aps.anl.gov] Im Auftrag von Matt Newville
Gesendet: Mittwoch, 11. April 2018 05:04
An: XAFS Analysis using Ifeffit <ifeffit@millenia.cars.aps.anl.gov>
Betreff: Re: [Ifeffit] Phase shift

Hi Julian,

On Tue, Apr 10, 2018 at 1:30 PM, Julian Ehwald <jehwald@gmail.com> wrote:

Dear all,

I have a very general question about the phase shift. I tried to fit a rather noisy sample of Li2IrO3, where I get a first main peak around 1 angstroem, according to FEFF calculation there should be something at 1.9 and not any closer. This phase shift seems a bit too big to be true, is there something like an upper bound for a reasonable phase shift?

Best, Julian

It's a little hard for me to tell what you mean.  Are you saying that Feff predicts the peak of |chi(R)| to be at 1.9 Ang, or that the first shell distance should be 1.9 Ang?

Phase-shifts for single-scattering peaks typically cause the peak of |chi(R)| to be ~0.5 Ang below the near-neighbor distance. That can vary some, but I would be surprised for it to be as big as 1 Ang. 

It might be useful to post the data.

--Matt