|
Results: Rietveld Refinement Results |
Course Material Index
Section Index
Previous Page
Next Page
Results: Rietveld Refinement Results
It is essential that evidence of the quality of a Rietveld refinement be
presented along with the final parameter values.
The article should therefore include the following:-
- Final Plot
A plot of observed and final calculated intensities
versus 2q (or d spacing
for time-of-flight or energy dispersive studies) is required. A plot of the
difference in the two profiles (observed intensity minus calculated intensity)
should also appear on the same figure, as well as markers showing the positions
of all reflections allowed by the space group for the structure. Examples of
such plots were shown earlier (e.g. the best fit to
the PbSO4 data in which the observed intensities are represented
by the black dots, the calculated intensities by the red line, the difference
plot by the pink line and the reflection markers are shown in green).
- R Factors
The values and definitions of all R factors should be provided, these include
Rwp, Rexp, RI, and
c2(=(Rwp/Rexp)2.
- Final Values
The final values of all refined parameters, with their estimated standard
deviations should be provided. These are normally provided in tabular form.
Examples of such tables are:
Crystallographic data, instrumental parameters and final residuals for
Pd1.53(63)[(Al32Si64)O192]
Formula (Z=2) |
Pd1.53(63)[(Al32Si64)O192
a (Å) |
24.6980(2) |
V (Å3) |
15065.6(3) |
Number of parameters refined |
33 |
Space group |
Fd-3m |
Scale factor |
0.1874(6) |
Zero point |
0.0123(3) |
Gaussian peak width parameters (°2) |
U 0.40(2) |
V -0.20(1) |
W 0.021(2) |
Rwp |
6.10 % |
Rp |
5.50 % |
RI |
6.01 % |
c2 |
1.10 |
|
Final atomic coordinates, isotropic temperature factors and
fractional occupancies for
Pd1.53(63)[(Al32Si64)O192]
Atom |
x |
y |
z |
Uiso (Å2) |
Occupancy |
Si(1) |
-0.05327(8) |
0.12613(9) |
0.03633(10) |
0.0104(5) |
0.66 |
Al(1) |
-0.05327(8) |
0.12613(9) |
0.03633(10) |
0.0104(5) |
0.33 |
O(1) |
0.0000 |
-0.1024(2) |
-0.1024(2) |
0.011(2) |
1.0 |
O(2) |
0.0001(2) |
0.0001(2) |
0.1442(2) |
0.026(2) |
1.0 |
O(3) |
0.0771(2) |
0.0771(2) |
-0.0281(2) |
0.024(2) |
1.0 |
O(4) |
0.0702(2) |
0.0702(2) |
0.3221(2) |
0.016(2) |
1.0 |
Pd(1) |
0.0436(3) |
0.0436(3) |
0.0436(3) |
0.025 |
0.096(2) |
Interpretation
In any article reporting the results of a Rietveld refinement, the final plot,
the R factors and the refined parameters table all contain important
information relating to the quality of the fit of calculated to observed data,
and so they should be studied carefully.
The plot of the difference between observed and calculated intensities should
be examined with particular note. There are several features to look for:
- the quality of the fit of the background (the difference plot should be
zero in
regions of background only).
- look for any reflections whose observed and calculated intensities don't agree,
the larger the difference and number of such reflections, the worse the
quality of the final structure. Differences between the observed and calculated
intensities indicate that the refined structure is not entirely consistent with
the observed diffraction intensities. This means that there might be atoms
incorrectly assigned or missing, or that additional corrections need to be
applied to give a better fit.
- if any particular classes of reflection don't match well this might
indicate that preferred orientation or peak broadening effects have not been
considered fully.
- if there is a systematic shape to every reflection in the difference fit then
the zero point, lattice parameters or the peak shape might not be correctly
fitted.
The R factors reported for the refinement should be as low as possible and
c2
should be close to one. If the value of c2 < 1, then the refinement has been
over-parameterised or the peak-to-background ratio of the data is low. If the
values of Rwp and Rp are high then you should not immediately class the fit
as poor, as these values are partly dependent on how the factors are defined by
the Rietveld program, and by the intrinsic quality of the diffraction data. If
any R factors are not quoted in an article, then look more closely at the
values of the refined parameters and the final profile plot. In summary,
consider all the R factors together when assessing the quality of fit.
The tables of final refined parameter values provide much information on the
quality of the Rietveld refinement presented and how the refinement was done.
You should look closely at the values of the parameters and the errors
(esd's). The smaller these are, the more precisely the corresponding parameters
have been determined, and the more precise any derived values will be. For
example, the lower the error on atomic positions the more precise will be the
values of the calculated bond distances and angles. When the value of the
refined parameter is equal to or less than approximately 3 times the esd then
the parameter value is not more significant than its error and must be treated
with caution, especially if the value is used to justify an argument later in
the discussion section of the article, for instance, on the presence of an atom
in the structure with a low site occupancy.
Certain points concerning the information presented in the table of atomic
parameters should be scrutinised carefully. These include the following:
- Atomic coordinates - look to see if any general position sites are not
refined
and why (maybe, for instance, the space group is polar), that atomic
coordinates on any symmetry elements are not refined, and whether any atoms are
constrained to have the same coordinates or to have the same shift in their
direction vector.
- Temperature factors - check that atoms of the same element have similar
temperature factors if they are in similar coordination environments and that
heavier atoms have the lower temperature factor values (for X-ray data). If one
atom has a much higher temperature factor than others of the same type check to
see if there is an explanation provided in the discussion section; for example,
is that atom not tightly bound in the structure or is there disorder?
If anisotropic temperature factors have been refined, check that the values for
the temperature factor along the three axes of the thermal ellipsoid are all
positive (may need a program to do this easily) and again, if the magnitude along
one direction is larger than any other, see if any explanation is proposed.
Also check whether the temperature factors are fixed, constrained or refined
freely.
- Site Occupancies - see if they are fixed or refined, and whether the
values,
when multiplied by the site multiplicity, are as predicted by the molecular
formula.
Examples of several deductions that can be made about the refinement process
from the above table of atomic coordinates, isotropic temperature factors and
occupancies are:-
- Si(1) and Al(1) have been constrained to have the same position, isotropic
temperature factor (all of which have been refined) and a total occupancy of
100 % (for which the values have been fixed at 66.7 % of Si and 33.3 % of Al).
- The isotropic temperature factors and and atomic coordinates of each of the
oxygen atoms have been refined independently, but their occupancies have been
fixed at 100 %.
- The Pd atom's position and site occupancy have been refined but its
temperature factor has been fixed at a pre-defined value.
Finally, as mentioned earlier, all the relevant information given should be
considered as a whole, in order to assess the quality of the Rietveld
refinement presented.
Course Material Index
Section Index
Previous Page
Next Page