[Ifeffit] Co fitting questions

Neil M Schweitzer neil.schweitzer at northwestern.edu
Fri Sep 2 16:18:11 CDT 2016

Thanks Carlo,
For this sample specifically, I am only trying to get a reasonable estimate of SO2 for use with my samples, which are supported, Co oxide clusters (catalysts).

You are right. If it take out the extra O scattering paths and reduce the R-range, the fit becomes significantly better (both R-factor and Reduced chi-square are reduced). However, now the SO2 is ~1.11. Does this make physical sense? What would be the cause of it being too big?


-----Original Message-----
From: Ifeffit [mailto:ifeffit-bounces at millenia.cars.aps.anl.gov] On Behalf Of Carlo Segre
Sent: Friday, September 02, 2016 3:42 PM
To: XAFS Analysis using Ifeffit <ifeffit at millenia.cars.aps.anl.gov>
Subject: Re: [Ifeffit] Co fitting questions

hi Neil:

The sigma squared values for your first Co-O path and the Co-Co path are not out of line with those I have seen for Co(OH)2.  Personally, I would not include the last Co-O path and cut the fitting range down to 3.5A or so.  It is possible that the second Co-O path is also not relevant.  The distance shift seems to be very large and simply compensating for the peak at 2.7A.  Again, I have seen this in my Co(OH)2 samples as well.

The question is what do you need to glean from these data.


On Fri, 2 Sep 2016, Neil M Schweitzer wrote:

> I have been trying to fit a Co(OH)2 reference sample and think I have a good fit, but I have a couple reservations about it and wanted to run it by some people for an expert opinion. If you are interested, I have included the Athena file, an image of the fit, and the fit log file. Please read on! If not, sorry I spammed you.
> As you can see, the fit contains four paths. They are all single scattering paths, and the paths with the four largest ranks according to feff. All paths share the same SO2 and deltaE. Each path has its own delR and ss. After some work, I got it to the point that the fit looks pretty (R-factor<0.02), but I have several reservations about it:
> 1)      Generally, the ss values are all higher than I would like. 0.007 even seems high to me for a metal-O bond. Am I right about this? Would this indicate that this may have been a poor choice for reference sample (I should mention it is a powder) and it is not very crystalline? Could it mean there was something we overlooked when we recorded the data?
> 2)      The biggest problem with the fit is the large error on ss_O2 and ss_O3. The largest feature at ~2.75 is obviously Co-Co scattering, but the Co-O paths help to fit the shoulder (or feature at ~3.1 in the Real part) in R-space. If I take them out, then the Co path has a large ss with an equally large error. So here is my question. If I add more paths, the error of ss_O2 and ss_O3 will go down, but then the errors on the new path will consequently be large, so when do I stop adding paths? For the purpose of my samples, I am only interested in the first Co-O path and the Co-Co path.
> Is this fit good enough you would publish it? Please feel free to critique any other aspect of the fit too.
> Thanks for your input!
> Neil

Carlo U. Segre -- Duchossois Leadership Professor of Physics Interim Chair, Department of Chemistry Director, Center for Synchrotron Radiation Research and Instrumentation Illinois Institute of Technology
Voice: 312.567.3498            Fax: 312.567.3494
segre at iit.edu   https://urldefense.proofpoint.com/v2/url?u=http-3A__phys.iit.edu_-7Esegre&d=CwICAg&c=yHlS04HhBraes5BQ9ueu5zKhE7rtNXt_d012z2PA6ws&r=oD5QfHEvBMM5YYrnfMF2YrJNX5aSKvlIRGd2B3iC9kQ&m=UHovCVboippuwjHS83MUbFyIwD3zbmNtLtJdSaRupGY&s=vR2qFm6nzIZ4Hvsa87KtvYjUzdKhqohA72GQpeTS-wg&e=    segre at debian.org
Ifeffit mailing list
Ifeffit at millenia.cars.aps.anl.gov
Unsubscribe: https://urldefense.proofpoint.com/v2/url?u=http-3A__millenia.cars.aps.anl.gov_mailman_options_ifeffit&d=CwICAg&c=yHlS04HhBraes5BQ9ueu5zKhE7rtNXt_d012z2PA6ws&r=oD5QfHEvBMM5YYrnfMF2YrJNX5aSKvlIRGd2B3iC9kQ&m=UHovCVboippuwjHS83MUbFyIwD3zbmNtLtJdSaRupGY&s=O7NfUTdRgH9IJMhcNbrkMTJBBXYNzXtNppPwcG2TkYc&e= 

More information about the Ifeffit mailing list