Chemistry Reference
In-Depth Information
Now, if the density and thermal motion models result in strong net forces acting
onto the nuclei, these forces would shift the nuclei until equilibrium was obtained.
The model is unphysical in this case, as the crystal is in equilibrium.
Of course, it would be desirable to have vanishing electrostatic forces as the
result
of the refinement without forcing the model to obey such constraints in the
first place. It must be said, however, that the experimental refinement procedure is
not of that quality and virtually every experimentally obtained model is prone to
violate equilibrium conditions.
On the other hand, obeying the equilibrium conditions still does not
prove
the
model to be the best possible model. But, and that is important, the violation of any
equilibrium condition proves the model to be unphysical and therefore unaccept-
able for describing an equilibrium system. That is why physical measures of quality
should be available in the future.
4.2 Further Tools for Analysis, Advanced Theories,
and Adjustments to Charge-Density Needs
There has been a large progress in X-ray detection with the area detectors and in the
data processing capabilities by the improved IT technologies. These have been used
to make data collection and standard structure determination faster, and to enable
charge-density studies, however, for the special needs of charge-density applications,
there needs to be a similar progress in the data processing, error analysis, and model
building. Model building again comprises not only density and thermal motion
models but also the scattering model, which at present neglects TDS and Renninger
reflections. In future, it might be necessary to identify and quantify the Renninger
contributions to individual reflections and to handle the TDS more accurately if we
want to be in control of the sources of error. It might also be necessary to investigate
and recalibrate incident beam corrections and photon conversion rates for individual
experimental setups. Whether this is necessary or not depends on how accurate and
precise the experimental results are required to be. Also, scaling and merging
techniques should be revised to assure appropriateness for charge-density studies.
One example was given above where it has been shown that an empirical extinction
correction influences the data such that errors of 10% in this empirical correction
would amount to approximately 4 gross residual electrons, which may be the whole
difference between an IAM and a multipole model refinement.
Another example is given here: Suppose a weighting scheme is used for identi-
fying and subsequent downweighting of outliers. What does this imply for weak
and zero intensity observations? As the minimum error is given by the counting
statistics, this obviously means that there is no easy way of identifying outliers for
weak and zero intensity observations, as these do not have a sharp expectation
value. The concept of an outlier cannot be easily applied to weak and zero intensity
observations. This holds also in the case of zero background. If outlier treatment is
