EXTI x [0]
An extinction parameter x is refined by least-squares, where Fc is multiplied
by:
-1/4
k [ 1 + 0.001 * x * Fc^2 * lambda^3 / sin(2theta) ]
where k is the overall scale factor. Note that it has been necessary to
change this expression from SHELX_76 (which used an even cruder approximation)
and SHELXTL (which used 0.002 instead of 0.001*lambda^3). The wavelength
dependence is needed for HKLF 2 (Laue) data. The program will print a warning
if extinction (or SWAT) may be worth refining, but it is not
normally advisable to introduce it until all the non-hydrogen atoms have been
found. For twinned and powder data, the Fc^2 value used in the above
expression is based on the total calculated intensity summed over all
components rather than the individual contributions, which would be easier to
justify theoretically (but makes little difference in practice). For the
analysis of variance and '.fcf' output file, the Fo^2 values are brought onto
the absolute scale of Fc^2 by dividing them by the scale factor(s) and the
extinction factor. The above expression for the extinction is empirical and
represents a compromise to cover both primary and secondary extinction; it
has been shown to work well in practice but does not appear to correspond
exactly to any of the expressions discussed in the literature. The article by
A.C. Larson in Crystallographic Computing (1970), Ed. F.R. Ahmed, Munksgaard,
Copenhagen, pp. 291-294 comes closest and should be consulted for further
information.