I think (in agreement with Matt's comment) that if the standards are not
known to be calibrated to the same energy scale then LCA may not be as
useful if done in energy space. But, depending on the problem of course,
there could be a good spectral contrast between the data not only in energy
but also in k-space. The choice of the energy origin in k-space does not
depend on the energy scale but on the position of the "E0" point within the
edge region. Hence, even if the standards and the experimental spectrum are
not known to be calibrated to the same energy scale, the LCA can be done
for the data in k-space.
Of course, the k-space analysis may not necessarily be possible (if the
data quality is poor or k-range is too short or the background removal
process affects the results too strongly etc etc) or meaningful (if the
experimental data are obtained at different temperature, for example, than
the standards - then XANES analysis will be more meaningful by LCA than
EXAFS...) but there are definitely many cases out there where XANES and
EXAFS analysis by LCA results in similar values of mixing fractions.
Anatoly
On Fri, May 12, 2017 at 11:42 AM, Matt Newville
Hi Pamela,
On Thu, May 11, 2017 at 12:19 PM, Carrillo Sanchez, Pamela < pcarrillo@bnl.gov> wrote:
Hi all,
I have gone through the mail archives as well as following the instructions of Athena user’s manual as well as Scott Calvin’s book but still I find myself with the doubt of how correctly “choose E0”. I have aligned the data and standards with the merge of reference foils measurements of the metal (Mn). We did not have a reference foil placed while measuring the data so I used the merge of the references scans when the standards were measured and they showed a consistent energy shift.
I have set the same E0 for all the data and standard to the same E0, I chose the first peak of the derivative in the data (6MnRh_RED). When I do the LCF analysis without ‘fitting E0’ , I get a worse fitting than when I choose the fit E0 on Athena as well as different types of components on the fitting. Which one is the best choice for it ? From what I have read I think that without fitting E0 would be the proper one regardless of the “worse fit”.
The recommendation to not fit E0 in linear combination fitting is generally preferred, but also assumes that you have all data calibrated to the same energy scale. The concern for fitting E0 in such a linear analysis is that energy shifts are often a sign of real chemical effects, not an indication of a change in energy calibration.
But, you can definitely use Athena's linear combination fitting to figure out what the energy shift is between two spectra on the same material measured with different calibrations.
Generally, it's best to measure several standards at the same time, so as to be able to assert that their energy scales (and resolution) are consistent. It's OK to measure some samples / standards at different beamlines or different sessions at the same beamline as long you're careful about calibrating energy consistently (say with the same sample such as metal foil or hard-to-alter metal oxide) and the resolutions are similar. These days most beamlines are pretty stable in energy calibration and resolution.
If the standards are well-aligned internally, applying very different E0s to the standards is probably not what you want. Your "fit E0" example shows E0 shifts varying between ~0 and -6 eV for the different standards. I think that's kind of large - it's possible the energy scales for the standards vary by that much, but you might want to double-check that too.
But, if you don't have calibrated standards, I might suggest asking the beamline scientist if they have an idea what kinds of energy shifts they expect.
Hope that helps,
--Matt
_______________________________________________ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit Unsubscribe: http://millenia.cars.aps.anl.gov/mailman/options/ifeffit