[Ifeffit] Estimation of S02
Leandro Araujo
leandrolangie at gmail.com
Thu Oct 9 00:39:04 CDT 2008
Hi Hiroshi,
questions very similar to yours have already been asked in the mailing
list and I still think the best answers are those available at:
1- Sections 4,8, 11-13 at http://cars9.uchicago.edu/iffwiki/FAQ/FeffitModeling
2- This previous thread from the mailing list:
http://www.mail-archive.com/ifeffit@millenia.cars.aps.anl.gov/msg00182.html
3- This one as well:
http://www.mail-archive.com/ifeffit@millenia.cars.aps.anl.gov/msg00368.html
As Bruce, Matt, Scott and many others have said, there is no "one fits
all" solution, so I suggest you start from the considerations found in
the links above to select the best approach to your specific system.
And take a good look at other previous threads in the mailing list -
seek and thou shall find...
As far as correlations are concerned, "That is the nature of the exafs
fitting problem. A correlation of 0.8 for either of those pairs
(amp/MSRD and E0/dr) is quite common." (quoting
http://www.mail-archive.com/ifeffit@millenia.cars.aps.anl.gov/msg00523.html).
Excellent pieces of advice on how to handle this problem can be found
in the links mentioned above and the best I can do is direct you to
them. Sorry if this is not the straight answer you were hoping for.
Regarding your Athena project, I suggest tweaking the position of E0
and the k and R windows according to the advice provided in the links
above. The experimental data looks nice and clean up to k~14.5 A-1.
Best of luck,
Leandro
2008/10/9 Hiroshi Oji <oji-h at spring8.or.jp>:
> Hi Matthias and Leandro,
>
> Thank you very much for your reply.
>
> I checked the DW-factor (MSRD) and found it also varies significantly
> depending on Fit k-weight as follows.
>
> kw amp ss
> 1 0.8145660 +/- 0.0486470 0.0073230 +/- 0.0005240
> 2 0.9230700 +/- 0.0673820 0.0082070 +/- 0.0005070
> 3 1.0439100 +/- 0.0854950 0.0089560 +/- 0.0005100
> 1,2,3 0.9081570 +/- 0.0722180 0.0081660 +/- 0.0005920
>
> The correation between these variables reported by Artemis are very
> strong, especially for kw=3.
>
> kw corration between amp and ss
> 1 0.8469
> 2 0.8992
> 3 0.9543
> 1,2,3 0.8543
>
> So, it seems to me that the amps obtaied by kw=1 and 1,2,3 with less
> degree are more reliable.
>
> If the DW-factor is fixed, for example, to ss = 0.008, the variation of
> amp become small as follows,
>
> kw amp ss
> 1 0.8674610 +/- 0.0296350 0.008 (fixed)
> 2 0.8984650 +/- 0.0260200 0.008 (fixed)
> 3 0.8998030 +/- 0.0327720 0.008 (fixed)
> 1,2,3 0.8898850 +/- 0.0325530 0.008 (fixed)
>
> But I don't know how to determine the appropriate DW-factor. Could you
> give me advice or comment on this problem?
>
> By the way, I attached the Athena project file of Zr-foil which I
> analyzed. I am glad if you kindly check it.
>
> Thank you for your coopration in advance,
> Best regards,
>
> Hiroshi
>
> Matthias Bauer wrote:
>> hi hiroshi,
>> i did not have a look at the attached project, but what paramteres did
>> you vary? even if you use different k-weightings, there is still the
>> strong correlation of N,S02 and the Debye-Waller like factor. so as long
>> as you do not keep two of these three on a certain value, your S02 might
>> vary to some extend, which will be more significant for shells beyond
>> the first one. anyway, it would be a great thing to hear some comment on
>> a S02 value higher than one, which is by definition of S02 not possible.
>> in most analyses the electron mean free path is not iterated, which can
>> be an explanation of unphysical values of S02, but than it should be
>> considered as a "net" factor accounting for both intrinsic and extrinsic
>> losses. am i right?
>> best regards
>> matthias
>>
>> Leandro Araujo wrote:
>>> Hi Hiroshi,
>>>
>>> your variation in SO2 does sound a bit too big, so I'll try to help
>>> adding my 50c and hopefully more people will jump in.
>>> It seems to me that your window in k-space is going a bit lower than
>>> it should - you seem to be including signal from the xanes region in
>>> your fit and that is not a good thing.
>>> I did some really quick fits using your project and just changing the
>>> window from k ~ 2-14 to k ~ 4.8-14. Then I got the following values:
>>> kw amp
>>> 1 1.0065350+/-0.0788250
>>> 2 1.0359800+/-0.0888140
>>> 3 1.0749950+/-0.0946640
>>> 1-3 1.0322490+/-0.0866650
>>> That seems a more reasonable variation to me. But maybe I'm being
>>> tricked by the fact that you have a split first shell and the MSRDs
>>> are not supposed to be very similar...
>>> I would also suggest opening the r-window to the right side a little
>>> bit. (By the way, that "bump" in the FT at ~ 1 Angstrom seems
>>> unexpected for a Zr foil. Maybe check the background removal in
>>> Athena?)
>>>
>>> The above comments obviously do not address the fundamental questions
>>> "why does SO2 change with k-weighting?" and "what's the best way to
>>> determine SO2?". This was intentional, since much better qualified
>>> people have previously written about it.
>>> My guess is that you will find something if you search previous
>>> threads in this mailing list. I would also suggest you check these
>>> sites:
>>> http://xafs.org/Tutorials
>>> http://cars9.uchicago.edu/iffwiki/FAQ/FeffitModeling
>>>
>>> Hope this helps,
>>> Leandro
>>>
>>>
>>> 2008/10/7 Hiroshi Oji <oji-h at spring8.or.jp>:
>>>
>>>> Dear all,
>>>>
>>>> Could you give me some advice on estimating the intrinsic loss factor (S02)?
>>>>
>>>> I am trying to estimate the S02 value in the curve-fitting analysis for
>>>> Zr K-edge EXAFS.
>>>>
>>>> To do this, I did the curve-fitting analysis on the first shell of
>>>> Fourier transform of Zr-K EXAFS of Zr-foil by Artemis. However, there
>>>> was a marked difference between the amps obtained by the fitting,
>>>> depending on the "Fit k-weights" as follows.
>>>>
>>>> kw amp
>>>> 1 0.8145660 +/- 0.0486470
>>>> 2 0.9230700 +/- 0.0673820
>>>> 3 1.0439100 +/- 0.0854950
>>>> 1,2,3 0.9081570 +/- 0.0722180
>>>>
>>>> Is this a usual thing? I can't find the reasonable explanation why S02
>>>> depends on "Fit k-weights".
>>>>
>>>> I attached the Artemis project file.
>>>>
>>>> Thank you in advance for your cooperation.
>>>>
>>>> Hiroshi Oji
>>>>
>>>> =========================================================
>>>> Hiroshi Oji, Dr.
>>>> Industrial Application Division
>>>> Japan Synchrotron Radiation Research Institute (JASRI)
>>>> 1-1-1 Kouto, Sayo-cho
>>>> Sayo-gun, Hyogo 679-5198
>>>> JAPAN
>>>> E-mail oji-h at spring8.co.jp
>>>> =========================================================
>>>>
>>>> _______________________________________________
>>>> Ifeffit mailing list
>>>> Ifeffit at millenia.cars.aps.anl.gov
>>>> http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
>>>>
>>>>
>>>>
>>> _______________________________________________
>>> Ifeffit mailing list
>>> Ifeffit at millenia.cars.aps.anl.gov
>>> http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
>>>
>>
>> _______________________________________________
>> Ifeffit mailing list
>> Ifeffit at millenia.cars.aps.anl.gov
>> http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
>>
>>
>
>
> --
> =========================================================
> 陰地 宏
> （財）高輝度光科学研究センター（JASRI）
> 産業利用推進室
> 〒679-5198兵庫県佐用郡佐用町光都1-1-1
> TEL(PHS): 0791-58-0802-3713
> または 0791-58-0803-3713
> FAX: 0791-58-0830
> E-mail: oji-h at spring8.or.jp
>
> Hiroshi Oji, Dr.
> Industrial Application Division
> Japan Synchrotron Radiation Research Institute (JASRI)
> 1-1-1 Kouto, Sayo-cho
> Sayo-gun, Hyogo 679-5198
> JAPAN
> E-mail oji-h at spring8.co.jp
> =========================================================
>
> _______________________________________________
> Ifeffit mailing list
> Ifeffit at millenia.cars.aps.anl.gov
> http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
>
>
More information about the Ifeffit
mailing list