[Ifeffit] consultation

[Welter, Edmund]

Dear Mingliang,

> Thanks for your explain. But I am still puzzled. Following your 
> instruction, I performed *Self-Absorption *for the samples, choosing 
> *XANES(Fluo) *from *algorithm*, but I did not know the real meaning of  
> *Formula*. Does it mean the species formula which I wish to detect? Or 
> the whole composition of the sample? 
The composition of the whole sample. You do not need to know the 
species, you "only" need to know thew elemental composition of your 
sample. A hairy task if you don't digest your sample completely and put 
it into an ICP-MS or anything like that. I don't know which precision is 
necessary to yield usable results. Maybe someone else tested or 
calculated it.
In principle the matrix produces just a smooth background absorption and 
knowing the composition it is easy to calculate the amount of this 
absorption using tabulated scattering factors. But it should also be 
possible to determine this background experimentally, for instance 
extrapolating the background before the edge. But for this you would 
need a blank spectrum to "calibrate" the mued value.
If you follow the link to the "document section" in Athena a very 
instructive webpage will open and among other things you will find 
citations of the original papers underlying the four different 
algorithms used for the self absorption correction.


> I used natural mineral to interact 
> with selenite, then I wish to check the speciation of the selenium 
> reaction product. For my case, should I fill the blank with Se or FeSe 
> or both of them (because currently I do not know the exactly containing 
> speciation)? 
As said before, the "chemical Formula" of the entire sample. "Chemical 
formula" here does not mean you have any kind of chemical bond, it just 
means molar ratio.

> For the angles, normally are they 45? Becuase I measured 
> the samples in fluorescence mode, and didnot measure the angles between 
> sample, incoming beam and detector. I tried to input *Se *in the 
> *Formula* entry, and select 45 for the angles, then I got a corrected 
> spectra which was higher than the original one. 

45 degrees is the usually used geometry. However, the program asks for 
the geometry which you used not for the usually used geometry! If it was 
45 degrees than its fine. Obviously the algorithm does what it is 
supposed to do, it amplifies the oscillations. But this does not mean 
that the result is correct! It will only be "correct" if the input 
parameters were correct.

> But when I using this 
> spectra for LCF analysis, I still could not got overlapping line for the 
> fitted spectra with the sample spectra...
Using pure Se in the self absorption correction is definitely wrong, it 
overestimates the effect dramatically. Actually the effect more or less 
disappears for diluted samples. But even if the spectra were properly 
corrected I can not see any reason why the used set of reference spectra 
  MUST fit the spectrum of your sample. There are many possible reasons 
for this, the most likely is that you are missing one or more important 
reference spectrum (Or you are not using enough references/free 
parameters in your fit...;-)

Best regards,
Edmund Welter