Dear All, I acquired Cd L3-edge spectra of some binary and ternary mixtures in varying proportions and for the individual components. The mixtures were created on Cd-mass basis. Then, I tried to fit the reference spectra to the spectra of the mixtures using linear combination fitting of Athena to get their abundance. However, the results were disappointing despite all spectra were carefully energy calibrated and normalized, so I decided to create simple mathematical binary and ternary mixtures by summing up the spectra of the individual reference spectra. After that I did an edge-step normalization in excel and imported the normalized calculated mixtures into Athena. Then, I tried the fitting again to exclude mixing-failures and check sensitivity of LCF with the idealized spectra. Even though the results of the LCF of the mathematical mixtures were better compared to the real mixtures, LCF was also not able to reliable deconvolute these spectra into the individual reference spectra. Does anybody have an explanation for that? It would be nice if somebody could give me information about the mathematical fitting algorithm implemented in Athena. Attached is a data file of three mixtures (two ternary and one binary mixture) including the mathematical mixture created in excel (named calculated at the end). Mixing ratios are named 1to1to1 (meaning 1:1:1 of the components in the same order). For the 1:1:1 ternary mathematical mixture the deconvolution was very good, but the others need improvement. I hope I made my problem clear this time. Thanks a lot! Wishes, Nina
Dear Nina, you used for this linear combinations very, very short scans. In my humble opinion, the scans are much to short to get a reasonable background correction and normalization. That is not a problem of athena or other solutions, there are just not enough data points for a reasonable background fit. Best regards Stefan Mangold Am 15.08.2011 um 11:35 schrieb Nina Siebers:
Dear All,
I acquired Cd L3-edge spectra of some binary and ternary mixtures in varying proportions and for the individual components. The mixtures were created on Cd-mass basis. Then, I tried to fit the reference spectra to the spectra of the mixtures using linear combination fitting of Athena to get their abundance. However, the results were disappointing despite all spectra were carefully energy calibrated and normalized, so I decided to create simple mathematical binary and ternary mixtures by summing up the spectra of the individual reference spectra. After that I did an edge-step normalization in excel and imported the normalized calculated mixtures into Athena. Then, I tried the fitting again to exclude mixing-failures and check sensitivity of LCF with the idealized spectra. Even though the results of the LCF of the mathematical mixtures were better compared to the real mixtures, LCF was also not able to reliable deconvolute these spectra into the individual reference spectra.
Does anybody have an explanation for that? It would be nice if somebody could give me information about the mathematical fitting algorithm implemented in Athena.
Attached is a data file of three mixtures (two ternary and one binary mixture) including the mathematical mixture created in excel (named calculated at the end). Mixing ratios are named 1to1to1 (meaning 1:1:1 of the components in the same order). For the 1:1:1 ternary mathematical mixture the deconvolution was very good, but the others need improvement.
I hope I made my problem clear this time.
Thanks a lot! Wishes, Nina
_______________________________________________ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Nina, I agree with Stefan. Your data are too short for adequate normalization and your criterion for choosing the post-edge line is ill-conceived. Normally, we choose a region well beyond the edge as the basis for the post-edge line. This is because the differences between the bare atom and the atom in a solid are small well above the edge. Thus regressing a line to that region well above the edge is likely to provide a consistent normalization which is insensitive to chemistry. Your criterion, on the other hand, is highly sensitive to chemistry. You chose a post-edge line that follows the slope of half of one particular oscillation in the region near the edge where the sample-to-sample variation due to chemistry are as large as they get. In my opinion, your method is guaranteed not to work. The only solution that I can recommend is that you abandon this data set and return to the synchrotron to remeasure your data. Take the time to measure at least 100 or 150 volts before the edge and many hundreds of volts after the edge. If you need to save time, you can measure sparsely in the pre- and post-edge regions -- but you have to measure something out there if you want to normalize your data in a way that is defensible. As for your question about the algorithm, I answered that the first time you asked. If you have a specific question, please ask and I will happily answer it. But please do not repeatedly ask the same vague, open-ended question. You won't get a better answer by simply re-asking the same question. B On Monday, August 15, 2011 05:56:18 am Stefan Mangold wrote:
Dear Nina,
you used for this linear combinations very, very short scans. In my humble opinion, the scans are much to short to get a reasonable background correction and normalization.
That is not a problem of athena or other solutions, there are just not enough data points for a reasonable background fit.
Best regards
Stefan Mangold
Am 15.08.2011 um 11:35 schrieb Nina Siebers:
Dear All,
I acquired Cd L3-edge spectra of some binary and ternary mixtures in varying proportions and for the individual components. The mixtures were created on Cd-mass basis. Then, I tried to fit the reference spectra to the spectra of the mixtures using linear combination fitting of Athena to get their abundance. However, the results were disappointing despite all spectra were carefully energy calibrated and normalized, so I decided to create simple mathematical binary and ternary mixtures by summing up the spectra of the individual reference spectra. After that I did an edge-step normalization in excel and imported the normalized calculated mixtures into Athena. Then, I tried the fitting again to exclude mixing-failures and check sensitivity of LCF with the idealized spectra. Even though the results of the LCF of the mathematical mixtures were better compared to the real mixtures, LCF was also not able to reliable deconvolute these spectra into the individual reference spectra.
Does anybody have an explanation for that? It would be nice if somebody could give me information about the mathematical fitting algorithm implemented in Athena.
Attached is a data file of three mixtures (two ternary and one binary mixture) including the mathematical mixture created in excel (named calculated at the end). Mixing ratios are named 1to1to1 (meaning 1:1:1 of the components in the same order). For the 1:1:1 ternary mathematical mixture the deconvolution was very good, but the others need improvement.
I hope I made my problem clear this time.
Thanks a lot! Wishes, Nina
_______________________________________________ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit _______________________________________________ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
-- Bruce Ravel ------------------------------------ bravel@bnl.gov National Institute of Standards and Technology Synchrotron Methods Group at NSLS --- Beamlines U7A, X24A, X23A2 Building 535A Upton NY, 11973 My homepage: http://xafs.org/BruceRavel EXAFS software: http://cars9.uchicago.edu/ifeffit/Demeter
Dear Nina, in addition to what Stefan Mangold has mentioned (energy range of your spectra too small for proper normalization, for both pre-edge and post-edge), there is another (maybe) critical issue in your data. Your "A" spectrum has an edge step value of \Delta µx=0.03, while your "C" spectrum has \Delta µx=2.92. In other words, your "A" and "B" standards are extremely dilute, and your "C" and "D" standards are almost too concentrated. There is almost 2 orders of magnitude difference in Cd concentration between A and C. In principle, normalization of the spectra should take care of the largely different concentration of Cd in your reference samples. This assumes that Beer-Lambert law holds, i.e. µx linearly depends on the concentration. In reality however, there is a number of effects that lead to a non-linear relation between concentration and µx ("thick sample effects", e.g. largely different harmonics content, pinholes for transmission experiments, self-absorption for fluorescence). Grant Bunker shows all the relevant effects in this set of slides: http://gbxafs.iit.edu/training/XAFS_sample_prep.pdf Maybe you can tell us a little more about your experiment, i.e. sample preparation, XAS experimental setup, and data treatment? Maybe we can then come up with some more advice. Best regards, Dominik On 15.08.2011 11:35, Nina Siebers wrote:
Dear All,
I acquired Cd L3-edge spectra of some binary and ternary mixtures in varying proportions and for the individual components. The mixtures were created on Cd-mass basis. Then, I tried to fit the reference spectra to the spectra of the mixtures using linear combination fitting of Athena to get their abundance. However, the results were disappointing despite all spectra were carefully energy calibrated and normalized, so I decided to create simple mathematical binary and ternary mixtures by summing up the spectra of the individual reference spectra. After that I did an edge-step normalization in excel and imported the normalized calculated mixtures into Athena. Then, I tried the fitting again to exclude mixing-failures and check sensitivity of LCF with the idealized spectra. Even though the results of the LCF of the mathematical mixtures were better compared to the real mixtures, LCF was also not able to reliable deconvolute these spectra into the individual reference spectra.
Does anybody have an explanation for that? It would be nice if somebody could give me information about the mathematical fitting algorithm implemented in Athena.
Attached is a data file of three mixtures (two ternary and one binary mixture) including the mathematical mixture created in excel (named calculated at the end). Mixing ratios are named 1to1to1 (meaning 1:1:1 of the components in the same order). For the 1:1:1 ternary mathematical mixture the deconvolution was very good, but the others need improvement.
I hope I made my problem clear this time.
Thanks a lot! Wishes, Nina
_______________________________________________ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
-- Dr. Dominik Samuelis d.samuelis@fkf.mpg.de Max-Planck-Institut für Festkörperforschung Max Planck Institute for Solid State Research Heisenbergstr. 1 70569 Stuttgart Germany Phone +49-711-689-1769 Fax +49-711-689-1722 Web http://www.fkf.mpg.de/maier/
Nina, Dominik, The issues Dominik raises are relevant for measuring and fitting experimental data, but they shouldn't affect your test of the LCF function using a simulated combination of spectra. Bruce's suggestion that the data normalization could be a problem also should not affect the LCF test - as long as you didn't change the normalization for the fitting standards compared to the spectra you used to make the simulated combination. Jeremy
-----Original Message----- From: ifeffit-bounces@millenia.cars.aps.anl.gov [mailto:ifeffit-bounces@millenia.cars.aps.anl.gov] On Behalf Of Dominik Samuelis Sent: Monday, August 15, 2011 8:53 AM To: XAFS Analysis using Ifeffit Subject: Re: [Ifeffit] Athena: problems with LCF
Dear Nina,
in addition to what Stefan Mangold has mentioned (energy range of your spectra too small for proper normalization, for both pre-edge and post-edge), there is another (maybe) critical issue in your data. Your "A" spectrum has an edge step value of \Delta µx=0.03, while your "C" spectrum has \Delta µx=2.92. In other words, your "A" and "B" standards are extremely dilute, and your "C" and "D" standards are almost too concentrated. There is almost 2 orders of magnitude difference in Cd concentration between A and C.
In principle, normalization of the spectra should take care of the largely different concentration of Cd in your reference samples. This assumes that Beer-Lambert law holds, i.e. µx linearly depends on the concentration.
In reality however, there is a number of effects that lead to a non-linear relation between concentration and µx ("thick sample effects", e.g. largely different harmonics content, pinholes for transmission experiments, self-absorption for fluorescence). Grant Bunker shows all the relevant effects in this set of slides: http://gbxafs.iit.edu/training/XAFS_sample_prep.pdf
Maybe you can tell us a little more about your experiment, i.e. sample preparation, XAS experimental setup, and data treatment? Maybe we can then come up with some more advice.
Best regards, Dominik
Dear All,
I acquired Cd L3-edge spectra of some binary and ternary mixtures in varying proportions and for the individual components. The mixtures were created on Cd-mass basis. Then, I tried to fit the reference spectra to the spectra of the mixtures using linear combination fitting of Athena to get their abundance. However, the results were disappointing despite all spectra were carefully energy calibrated and normalized, so I decided to create simple mathematical binary and ternary mixtures by summing up the spectra of the individual reference spectra. After that I did an edge-step normalization in excel and imported the normalized calculated mixtures into Athena. Then, I tried the fitting again to exclude mixing-failures and check sensitivity of LCF with the idealized spectra. Even though the results of
the mathematical mixtures were better compared to the real mixtures, LCF was also not able to reliable deconvolute these spectra into the individual reference spectra.
Does anybody have an explanation for that? It would be nice if somebody could give me information about the mathematical fitting algorithm implemented in Athena.
Attached is a data file of three mixtures (two ternary and one binary mixture) including the mathematical mixture created in excel (named calculated at the end). Mixing ratios are named 1to1to1 (meaning 1:1:1 of the components in the same order). For the 1:1:1 ternary mathematical mixture the deconvolution was very good, but
On 15.08.2011 11:35, Nina Siebers wrote: the LCF of the others need improvement.
I hope I made my problem clear this time.
Thanks a lot! Wishes, Nina
_______________________________________________ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
-- Dr. Dominik Samuelis d.samuelis@fkf.mpg.de Max-Planck-Institut für Festkörperforschung Max Planck Institute for Solid State Research Heisenbergstr. 1 70569 Stuttgart Germany Phone +49-711-689-1769 Fax +49-711-689-1722 Web http://www.fkf.mpg.de/maier/
_______________________________________________ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Hi Nina, As others have said, normalization is critical for getting reasonable answers from LCF. One of the drawbacks of a linear analysis is that it always gives an answer, even when the assumptions going into the question are poor.
Even though the results of the LCF of the mathematical mixtures were better compared to the real mixtures, LCF was also not able to reliable deconvolute these spectra into the individual reference spectra. Does anybody have an explanation for that?
I'm not sure what your tests entailed, but I would have expected that your 'A_D_1to1' or 'A_D_1to1_calculated' spectra would have been a 1-to-1 mixture of the A and D spectra. A 1-to-1 mixture would a) have several isobestic point (for all E where A(E)=D(E), all linear combinations of A and D have the same value), and b) be half-way between A(E) and D(E) where the difference is largest, say near E=3568 eV. From the attached plot (simply plotting the data from your project), it's pretty obvious that neither of these is true. This might be related to poor normalization, a simple calculation mistake, or maybe I misunderstand your intent.
It would be nice if somebody could give me information about the mathematical fitting algorithm implemented in Athena.
Non-linear least-squares, using the Levenberg-Marquardt algorithm. See http://en.wikipedia.org/wiki/Levenberg-Marquardt_algorithm and do feel free to read the docs: http://cars9.uchicago.edu/~ravel/software/doc/Athena/html/analysis/lcf.html http://cars9.uchicago.edu/~ifeffit/refman/node63.html Admittedly, Levenberg-Marquardt may not be the most obvious choice for linear analysis, but is usually quite robust and fast for finding optimal solutions for linear problems, and is needed for other parts of XAFS analysis. --Matt
participants (6)
-
Bruce Ravel -
Dominik Samuelis -
Kropf, Arthur Jeremy -
Matt Newville -
Nina Siebers -
Stefan Mangold