Hello Bruce,
This is to follow up the *test* experiment you suggested. Attached are
three Artemis files.
I chose the Fe (im-3m) structure to do the test. The normal crystal
structure has its first-shell distance at 2.48 angstrom. A large
structure was created so that the first-shell distance reaches 3.11
angstrom.
Both normal and large structure are calculated in JFEFF in the same
manner (i.e., same path number, same calculation procedure, same
sigma2=0.0045 for all paths) to generate the calculated chi data.
File #1 attached was exactly following the procedure you mentioned.
The quick first shell fits very well, four parameters except S02 give
normal results. The amplitude parameter S02 is very large (2.2+/-0.14).
I extended the test in #2 and #3 for comparison. In #2, calculated
data based on large structure is presented to fit to the same large
structure. As expected, we get reasonable fit looking with other
parameters normal. However, amplitude S02 fits to 1.80+/-0.03. In #3,
both data and structure are normal. In this case, it is not surprising
to get good fit and all parameters including S02 turn out to be close
to their true values.
I think it is clear that first-shell distance larger than 3 angstrom
does has effect in making amplitude artificialy large.
Best,
Yanyun
Quoting Bruce Ravel
On 03/20/2015 12:48 PM, huyanyun@physics.utoronto.ca wrote:
Thank you. Our group has one copy of your book, I'll read it again after my colleague return it to shelf. I still want to continue our discussion here:
If we treat S02 as an empirically observed parameter, can I just set S02=0.9 or 1.45 and let other parameters to explain the k- and R- dependence? Because S02 is not a simplistic parameter which may include both theory and experimental effects, I feel that S02 is not necessarily to be smaller than 1, although I admit S02 smaller than 1 is more defensible as it represents some limitations both in theory model and experiment, but I have a series of similar sample and all their S02 will be automatically be fitted to 1.45~1.55, not smaller than 1. Could this indicate something?
I actually found in my system, when I set S02=0.9 (instead of letting it fit to 1.45), other parameter will definitely change but the fitting is not terrible, it is still a close fit but important site occupancy percentage P% changed a lot. So how should I compare/select from the two fits, one with S02=0.9 and one with S02=1.45 with two scenarios showing different results?
Yanyun,
As I recall, you are looking at those bizarre skuttuderite materials which consist of a metal framework with an enormous gap. Sitting in the gap is your absorber atom. The center point of the gap is, as I recall, over 3 angstroms away from the nearest vertex of the framework. The point I am about to make hinges upon all that being more or less correct.
Feff drops neutral atoms into the specified lattice positions then does a rather simple-minded algorithm to overlap the charges and come up with the radii that are used to compute the muffin tin potentials. In the case of one of those atoms rattling about inside the cage, I am skeptical that Feff's model produces a highly reliable set of scattering potentials. Probably ain't bad -- as you said in your first email, your fits look good. But it probably ain't quite right either. As Scott hinted, mistakes in the theory can show up in surprising with surprising k- or R-dependence, and surprising amplitude and phase dependence.
I have absolutely no intuition for how Feff might introduce systematic error into a fit for the physical situation of a nearest neighbor at a distance of 3 or more angstrom, so I don't know how to "explain away" an oddly large S02.
That said, I can think of some experiments that /might/ give some insight. Pick something simple, like a metal oxide or a metal sulfide -- something with a cubic structure. You don't want this experiment to get to complicated.
1. Before generating the feff.inp file, make the lattice constant nonphysically large such that the near neighbor distance is about 3 angstroms.
2. Run Feff and add up all the paths to make a theoretical chi(k) spectrum for your nonphysically large crystal. For a later iteration of this, you might add some synthetic noise to the spectrum.
3. Treat the chi(k) you just made as your "data". Import it and the normal crystal data into Artemis. Run Feff on the normal crystal.
4. Use Artemis's single scattering path tool to make a path for the first shell scatterer at the distance you used to make your theoretical data.
5. Make a simple first shell, four-parameter fit using that SS path.
Can you make a reasonable looking fit? With sensible error bars? What happens with the amplitude? Is it very large or very small?
Perhaps try the experiment the other way around. Fit the "normal" theoretical data with the unphysical Feff calculation.
The point I am driving at that I wonder if you can figure out what happens to the amplitude in a decent fit when you contrive a situation with an unusually large first neighbor distance. If you see a trend in these "Feff experiments", perhaps that can help you understand the amplitude in your skuttuderite fits.
Again, I have no intuition about this. I have no idea if my suggestion will be fruitful or not. For that matter, I have no idea if my memory of your problem is correct.
But maybe this is a brilliant suggestion. Unlikely, but stranger things have happened :)
B
-- Bruce Ravel ------------------------------------ bravel@bnl.gov
National Institute of Standards and Technology Synchrotron Science Group at NSLS-II Building 535A Upton NY, 11973
Homepage: http://bruceravel.github.io/home/ Software: https://github.com/bruceravel Demeter: http://bruceravel.github.io/demeter/ _______________________________________________ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit