Luca,
I understood Gerard's problem to have a measurement of a powder (of
unknown particle shape) in Bragg-Brentano geometry, for structure
refinement. As you said the graininess problem can be minimized
primarily by grinding and to a certain extent by rotation and enhancing
the divergence, axial as well as equatorial. It is obvious that overmuch
graininess would prohibit any reasonable structure refinement or
quantitative phase analysis, with or without the application of a
harmonic model for intensity correction.
Regarding the PO of powder mounts produced from minerals (big crystals)
by grinding (McCrone mill, particle size 1-5 µm) and front- back- or
sideloading the powder we must state that it is frequently present, of
course strongly dependent from the particle shape (by cleavage) of the
crystals. Not only platy or needle-like crystals tend to PO, but also
cubes (e.g. halite), octahedra (e.g. fluorite), rhombohedra (e.g.
calcite), pinakoids (e.g. feldspars) show PO because such crystals tend
to stacking and orientation "face by face", like bricks, in a real
powder mount. The variation of the preparation technique (say
sideloading instead of frontloading) does induce changes in PO what can
be used to identify PO, as I understood Frank's recommendation. This is
indeed the "trick" we apply every second week if a sample is suspicious
to be not adequately prepared. I remember also that an intentionally
enhancement of PO by preparation may also be applied for indexing and to
add information in structure refinement (of multiple patterns, McCusker..).
Of course the degree of orientation in such prepared powder samples is
probably not as high as in your "real textured" samples, but m.r.d.'s in
the magnitude of 3 (or down to 0.5) may appear. I believe this is weak
texture in the context of texture analysis? For such (in my opinion also
real and textured) powder samples, a model similar to the harmonics
works fine for PO correction, provided that enough measurable peaks are
in the pattern. E.g. the application of a model having 3 parameters to a
cubic material showing only 2 different lattice planes in the
one-dimensional scan must fail of course. Other problems may arise when
two directions (e.g. h00 and 0k0) show strongly overlapping peaks by
pseudo-symmetry, or when the intensity of a lot of hkl is very week for
structural reasons. In all these cases the one-dimensional powder
pattern contains not enough information to apply such a harmonic model,
the model must be deactivated, a better (more random) sample must be
prepared, the number of peaks must be enhanced... Thus, any PO
correction in Rietveld refinement is second-best solution only.
But in the daily routine work in Rietveld phase analysis, we can see
that in the most cases the calculated PO correction factors are
"reasonable" (reflecting the shape of the crystals, biggest faces get
highest, directions inbetween lowest...), and the quantitative results
are acceptable for such "realisticly oriented" powder samples what can
produced by standard techniques. I believe this is an argument to apply
these models, even though they are perhaps not able to decribe stronger
PO or maybe unable to do a texture analysis (what is anyway impossible
from a single Bragg-Brentano scan).
Greetings
Reinhard
Luca Lutterotti schrieb:
Reinhard,
I stick with what Gerard said:
"But i have no other information that supports the existence of
preferred orientation"
so what information give you the confirmation it is the powder mount
responsible of preferred orientation. I work almost exclusively with
image plate detectors and I can assure you that the graininess
problem is appearing more often than the preferred orientation case.
I am working on texture mostly so I am happy when you find them, but
this case is not s frequent as people think and for sure not s
frequent as graininess.
I wait also confirmation from Gerard that his sample is a powder and
it has plate like or fiber like particles. Otherwise I will
investigate the graininess case that with a proper grinding or a
spinner is easily resolvable.
Also, for who think that because the harmonic model can fit it is for
sure preferred orientation. I can just suggest to work for a while
with real textured sample and the harmonic and see if there is really
this relationship, you may be surprise by the result.
cheers,
Luca
On May 8, 2008, at 3:08 PM, Reinhard Kleeberg wrote:
Luca,
speaking about powder samples, Frank is right. The PO of powder
mounts is seldom reproducible and the filling technique is
responsible for particle orientation, depending on particle shape,
filling direction, pressure... In practice it is a nice trick to
repeat the filling of the powder holder with different filling
techniques to look for PO. Of course, sample graininess may be also
a reason for not reproducible intensity, but these effects ("rocks
in the dust") ar mostly hard to correct successfully by spherical
harmonics as Gerard stated for his problem. In any case, the problem
sounds to be related to sample preparation.
Reinhard
Luca Lutterotti schrieb:
On May 8, 2008, at 12:30 AM, May, Frank wrote:
You can check for texture effects (preferred orientation) by
obtaining multiple patterns of the material. It's realistic to
expect some differences, but preferred orientation is manifest by
not being able to replicate the pattern.
Not true,
preferred orientation or texture are perfectly reproducible,
provided you use the same sample orientation. What is not
reproducible and probably what Frank May is referring to is not
preferred orientation but graininess or few big grains that do not
guarantee the correct statistic. So if you need to check for
graininess, you just move a little your sample, so the beam covers
a different area on the sample. If you think you have texturte, to
check for it you have to change the sample orientation to see a
change. Beware that in a Bragg-Brentano instrument turning around
the axis normal to the sample surface is not a valid change in
orientation as nothing will change for texture; you have to change
the sample inclination instead (omega or chi).
Best Regards,
Luca Lutterotti
That's the simple test. Let us know what you find.
Another issue for "improper intensities" is when the specimen is
not sufficiently wide enough at low angles (typically below 20-
degrees 2- Theta with copper radiation) and the x-ray beam does
not fully impinge on the specimen. The observed reflections in
the low angle region will be less than calculated by a modelling
program.
Frank May
Research Investigator
Department of Chemistry and Biochemistry
University of Missouri - St. Louis
One University Boulevard
St. Louis, Missouri 63121-4499
314-516-5098
________________________________
From: Gerard, Garcia S [mailto:[EMAIL PROTECTED]
Sent: Wed 5/7/2008 8:57 AM
To: rietveld_l@ill.fr
Subject: Preferred orientation?
Dear all,
I have a laboratory Bragg-Brentano X-ray (Cu) pattern that shows
intensity mismatches only at low angles, ie 20-50 2theta or 1.8 to
4 Angstroms.
There are overestimated peaks and also underestimated peaks.I
have tried to discard factors that might cause this problem:
The thermal parameters look sensible. Moreover, the data at high
angle looks ok, so intensity transfer from low angle to high
angle or vice versa does not seem to be the cause.
Atomic positions also look sensible. And again, data at high
angle looks ok. Is the scattering angle dependence of the atomic
positions the same as for the thermal parameters? (I cannot
remember that, but i am pretty sure it is not).
Following the advice published in J. Appl. Cryst. 32, 36 (1999),
the other factor that might cause this problem is preferred
orientation:
I have tried to find a hkl dependence in the overestimated and
underestimated peaks but i could not find any. If i try to model
preferred orientation with spherical harmonics the problems
disappears nicely. The problem is how to justify the existence of
preferred orientation. The crystal system is orthorhombic. But i
have no other information that supports the existence of
preferred orientation.
Is there any other problem that I cannot think of?Is the
preferred orientation correction masking any of these other
problems I cannot think of?
Regards
Gerard
________________________________
Heriot-Watt University is a Scottish charity registered under
charity number SC000278.
