Hello,
do not mind statistical precision (your error at the oxygen coordinate)
and look at the errors (e.s.d.) of the interatomic distances, especially
of those between bonded atoms. Precision of CW neutron machines is
frequently not so high as X-ray diffractometers, but this problem is
masked in Rietveld by extremely (artficialy) low e.s.d.'s of lattice
parameters. On the other hand, what you are looking for is not PRECISION,
but ACCURACY. Heating a sample on a standard BB X-ray machine is a good
source of systematic errors in the data which "transform" to the lattice
parameters. This is not the case in neutron experiment.
My advice is to look also at O-O distances.
Best wishes,
Lubo
On Sun, 9 Mar 2008, Davide Levy wrote:
Dear All,
I have a philosophic question for you: the structure obtained by NPD are
better than the ones found by XRPD?
I explain a little bit more my question, I studied a MgAlFeO4 spinel both by
NDP and XRPD from RT to 1000°C. The O coordinate it is different for two
data set (0.2581 vs.0.25953). The found error on the oxygen coordinate is
about 10-5 and 10-4 for NPD and XRPD data respectively. The coordinate
calculated from cation partitioning and cation radii is similar to the XRPD
result.
At the moment I did not found a justification of this and I do not know
which result is the right one.
Do you have any suggestion?
Thank you
Davide