1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
|
This directory contains the crystallographic data for Iron Sulfide
(pyrite, FeS2) along with a transmission scan of FeS2 taken at room
temperature at beamline 13BM at the APS. The data was taken from
Matt's data archive at
http://cars9.uchicago.edu/~newville/ModelLib/search.html
This is sufficient information to begin a fitting project for FeS2.
In this readme file, I'll write steps to take to work through this
example. You should not consider this to be a recipe -- at each step
you should play around with the setting in both Athena and Artemis to
understand fully these data.
1. Fire up Athena. Import the data file.
2. Find a set of parameters that gives a good background removal and
Fourier transform.
3. Save the project and export the chi(k) file (if necessary)
a. If you are using a version of Artemis that can read the Athena
project file, then it is unnecessary to save the chi(k) file.
b. If you are using a version of Artemis which cannot read Athena
project files, then you will need to save the chi(k) by
selecting the correct group in the Athena groups list and then
selecting "Save as chi(k)" form the File menu.
4. Fire up Artemis.
5. Import the chi(k) data.
a. If you are using a version of Artemis that can read the Athena
project file, verify that you are still happy with the Fourier
transform parameters.
b. If you are using a version of Artemis which cannot read Athena
project files, set the Fourier transform and fit range
parameters to sensible values.
6. Import the crystallographic data contained in the file FeS2.inp.
Run Atoms. Verify that the Feff input data is reasonable, then
run Feff.
7. When the path selection dialog comes up, any choice is ok. You
may find the "first 10" paths option to be the most convenient.
8. Look at the various pages now available in Artemis. Note that a
parameter set has been defined on the Guess, Def, Set page. Note
that simple math expressions have been defined for each of the
paths.
9. Make plots of the data and the various paths using context menus
on the Feff interpretation page or by control-clicking on the data
and one or more paths. Which paths contribute strongly in regions
of the first peak in the data? In the region of the second peak?
Which multiple scattering paths seem to be strong contributers?
Which seem like they can be neglected?
10. Set the fit range appropriate for a first shell fit. Include only
those path(s) which contribute strongly under t he first peak.
Hit the big green button and examine your first shell fit results.
11. Extend the fit range to cover the split peak between 2.3 and 3.7
Angstroms. Include each of the first four SS paths in the fit.
Press the fit button. This fit should look somewhat like the
data, but not quite.
12. The problem with the previous fit was that it used a parameter set
appropriate to a first shell fit to fit data with four shells.
While the amplitude and E0 parameters are probably ok to use for
all those paths, it is physically unreasonable to use the same
delta R and sigma^2 for each path.
13. FeS2 is a cubic material (it's space group is Pa3 -- a cubic
group). This means we can use the trick of modeling the delta R
parameters by an isotropic expansion coefficient. Discard the dr
parameter on the GDS page and define a guess parameter called
"alpha". Define the delR parameter for each path to be
"alpha*reff". Remember that "reff" always evaluates to the
nominal path length as used in Feff when a path parameter math
expression is interpreted.
14. Define separate sigma^2 parameters for each shell. Continue using
the same amplitude and E0 parameters for each path. You fit using
the first four shells should now be using 6 parameters. Try
running the fit. Are the values for the parameters reasonable?
15. What other paths contribute significant spectral weight in the
region between 2.3 and 3.7 Angstroms? How about the next SS path?
How about path #6? Try adding these paths to the fit. You may
need to adjust the fit range or define new guess parameters?
16. Can you think of a way to approximate the sigma^2 for the MS paths
using math expressions, def parameters, and the sigma^2 parameters
that are used to fit the SS paths?
17. The 2nd and 3rd shells are both S and are relatively close
together in length. Test to see if the data support measuring
sigma^2 independently for those two paths by comparing a fit in
which they are constrained to be to same to a fit in which the
float freely.
18. If you look at the atoms information, you will see that this
structure is actually defined by two parameters -- the lattice
constant and the position of the S atom. The variation in the
lattice constant is accommodated in our fit using the isotropic
expansion constant alpha. How might you include the possibility
of refining the S coordinate in the fit?
19. Explore the effect of different k-weights on the fit. Try the
fits using different single k-weights and compare the best fit
values. How do they change? Is this variation outside their
error bars? Try doing multiple k-weight fits? How do the best
fit values compare?
20. Explore the effect of co-refining the background spline.
21. Try to fit the peak around 5 Angstroms using the model building
techniques you have employed in constructing your fitting model
out to 4 Angstroms. That is, determine which paths are important
(you may need to import more paths from the calculation), define
appropriate fitting parameters and make appropriate constraints,
and set the fitting range accordingly.
22. Are your values for the parameters all reasonable? Is the
amplitude a reasonable value for S02? If it is rather low, can
you think of any reasons to explain it? (In the data file header
it says the data was collected as powder on tape.)
|
