Let's take the frequently cited example of 'magnetite': the lattice spacing
and relative intensity of the (allowed!) 331 reflection are not listed in PDF
card #19-629. This card tells us that the pattern was taken in 1967 and was
produced using Ni-filtered Cu Ka radiation. The pattern can be verified in
the NBS Monograph 25(5) 1967 (there, on page 31 one finds the original NBS
data, copied number by number, and later sold to you as PDF card #19-629).
The information on the card tells me immediately that the pattern was not
produced under favorable conditions and 1967 was
pre-graphite-monochromator-time. There is no way that the 331 peak with a
relative intensity of <1 was to be detected under these conditions (no matter
if a scint. or prop. counter were used). Therefore, what the NBS researchers
saw they reported, no more no less - the information is correct and
exhaustive under the given experimental conditions.
This leads me to another question: how well does the Rietveld code behave if
the dispersion correction (which is not so well defined for X-rays in the
first place) comes into play by being too close to the absorption edge
wavelength? Has anybody any experience, does it matter at all? It appears
that with the advent and convenience of the diffracted beam graphite
monochromator, anything which diffracts, fluorescent or not, is probed with
CuKa. At least, one has to keep in mind that the use of CuKa radiation, with
or without graphite monochromator, reduces the penetration depth by a factor
~8 in (pure) Fe-samples when compared to MoKa radiation and the diffracted
intensity is diminished accordingly. Thus, precious sampling volume is lost
and surface effects due to absorption are substantially enhanced. With all
the enthusiasm for Rietveld analysis (which I greatly share), the basics of
the diffraction experiment must not be ignored.
Ludwig Keller
CAMET Research, Inc.
Goleta, CA