Doped Crystals and Alloys
For samples which are doped crystals, there are a couple of methods people have used. For purposes of this article, I'll consider cases where the dopant is substitutional as opposed to interstitial (maybe someone could edit this article to include that case?).
As an example of two methods, let's consider FeS2 substitutionally doped with molybdenum. (I have no idea if such a material is possible...I'm using it because FeS2 is included as an example in the Ifeffit and friends distrribution.)
Method 1:
Run atoms for FeS2.
Now look at the feff.inp file that is generated. Under "Potentials," it says the following:
POTENTIALS
* ipot Z element
0 26 Fe
1 26 Fe
2 16 S Add another line for the Mo, which is atomic number 42 (the atomic number is required):
POTENTIALS
* ipot Z element
0 26 Fe
1 26 Fe
2 16 S
3 42 MoImportant: Do not skip numbers in the "ipot" column, and make sure "0" is the absorber!
Next, take the list following the word "ATOMS" in the feff.inp file, and arbitrarily change roughly the right number of iron atoms to moly atoms. Make sure to change the "ipot" column to match...it's the part feff will actually use:
ATOMS * this list contains 71 atoms
* x y z ipot tag distance
0.00000 0.00000 0.00000 0 Fe1 0.00000
2.07514 0.62686 0.62686 2 S1_1 2.25657
0.62686 -2.07514 0.62686 2 S1_1 2.25657
-0.62686 0.62686 2.07514 2 S1_1 2.25657
-0.62686 2.07514 -0.62686 2 S1_1 2.25657
-2.07514 -0.62686 -0.62686 2 S1_1 2.25657
0.62686 -0.62686 -2.07514 2 S1_1 2.25657
-3.32886 0.62686 0.62686 2 S1_2 3.44488
0.62686 3.32886 0.62686 2 S1_2 3.44488
0.62686 -0.62686 3.32886 2 S1_2 3.44488
3.32886 -0.62686 -0.62686 2 S1_2 3.44488
-0.62686 -3.32886 -0.62686 2 S1_2 3.44488
-0.62686 0.62686 -3.32886 2 S1_2 3.44488
-2.07514 -2.07514 2.07514 2 S1_3 3.59425
2.07514 2.07514 -2.07514 2 S1_3 3.59425
2.70200 2.70200 0.00000 1 Fe1_1 3.82121
-2.70200 2.70200 0.00000 3 Mo1_1 3.82121
2.70200 -2.70200 0.00000 1 Fe1_1 3.82121
-2.70200 -2.70200 0.00000 1 Fe1_1 3.82121
2.70200 0.00000 2.70200 1 Fe1_1 3.82121
-2.70200 0.00000 2.70200 3 Mo1_1 3.82121
0.00000 2.70200 2.70200 1 Fe1_1 3.82121
0.00000 -2.70200 2.70200 1 Fe1_1 3.82121
2.70200 0.00000 -2.70200 1 Fe1_1 3.82121
-2.70200 0.00000 -2.70200 3 Mo1_1 3.82121
0.00000 2.70200 -2.70200 1 Fe1_1 3.82121
0.00000 -2.70200 -2.70200 1 Fe1_1 3.82121
-2.07514 3.32886 2.07514 2 S1_4 4.43776In this case, I changed 3 of the 12 nearest iron neighbors into moly...reasonable if I have about 25% doping.
IF you are doing a feff calculation for the moly edge, then also change the very first iron to moly, and change potential 0 in the ipot list to moly with ipot 0.
POTENTIALS
* ipot Z element
0 42 Mo
1 26 Fe
2 16 S
3 43 Mo
ATOMS * this list contains 71 atoms
* x y z ipot tag distance
0.00000 0.00000 0.00000 0 Mo1 0.00000
2.07514 0.62686 0.62686 2 S1_1 2.25657
0.62686 -2.07514 0.62686 2 S1_1 2.25657
-0.62686 0.62686 2.07514 2 S1_1 2.25657
-0.62686 2.07514 -0.62686 2 S1_1 2.25657If you are doing the calculation for the iron edge, leave the first iron alone, since it is still the absorber.
Now run feff, and you'll get the iron scattering paths listed separately from the moly scattering paths.
One more step...correcting for the actual doping fraction. Suppose there is actual 20% moly and not 25%, as we implied. We couldn't have handled that just through feff, because we can't change exactly 20% of 12 atoms...we have to change 2, which is 17%, or 3, which is 25%.
The fix for this is to change the S02 in the moly and sulfur scattering paths to account for this. You could, for example, use the following gds parameters:
s: MolyPercent 0.20 d: IronPercent 1-MolyPercent
Then go to the individual path representing the scattering off of nearest neighbor moly, and assign it an S02 of
amp*MolyPercent/(3/12)
That way, if the MolyPercent is 20%, it will reduce the amplitude of those paths by 20/25%, as is proper.
Of course, the iron scatterer would get an S02 of
amp*IronPercent/(9/12)
That's more or less it!
You could, of course, guess the MolyPercent instead of setting it, if for some reason it was unknown in your sample.
Method 2:
Suppose we want to analyze the iron edge.
Run atoms for FeS2 and then run feff.
Then make a new atoms page, type or read in the FeS2 file, and just change the Fe to Mo. Run atoms again.
If you're doing the iron edge, then change the absorber to iron in the feff.inp file (this requires changing the potential list; see the description under "Method 1" for how to do this.) Run feff.
(If you want to analyze the moly edge, then of course you change the feff.inp file in the first calculation to moly and leave it as moly in the second.)
You now have TWO sets of feff files associated with one data set.
Make gds parameters; e.g. use:
s: MolyPercent 0.20 d: IronPercent 1-MolyPercent
Now make the S02 for all paths calculated with the original atoms file:
amp*IronPercent
and for all paths calculatged with the new atoms file:
amp*MolyPercent
Again, you can guess the MolyPercent if it's unknown.
Discussion of methods:
Which method you use is largely a matter of taste. The first method is easier to screw up, since there's a lot of counting involved. On the other hand, it generates many fewer paths, and thus makes for smaller files and may fit faster (you're not wasting time and effort counting sulfur paths twice, for example). The first method also gives you the potential of finding a few multiple scattering paths that involve both iron and moly (in this example) that you can't probe at all by the second method. This is most likely to be true when the dopant is in low concentrations but is high-Z...it's possible that there may be a moly-iron multiple-scattering path that is significant, and it's not going to be modeled so well by the weighted average of iron-iron and moly-moly paths used in method 2. But the price for this is that properly incorporating multiple-scattering paths via method 1 requires an annoying amount of counting and thinking.
This page is part of CategoryNotBulkCrystals. Pages in this category give tips on using Artemis to analyze materials that are not simple macroscopic crystals.