The poor quality of solution 7 produced by the two X11 data sets is the result of poor contrast
in 
between the data sets. This can be seen in Table 
which shows the relative
isomorphous and anomalous occupancies for Iridium site 3 output by MLPHARE from solution 9 using all seven
data sets.
Table: List of the refined isomorphous and anomalous occupancies for Iridium site 3 produced by MAD phasing
solution 9. The corresponding and 
values are shown along with 
,
the difference in with respect to the reference data set (4). (
indicate theoretical values obtained
from the program CROSSEC.) The relative magnitudes of the isomorphous and anomalous occupancies agree well
with those suggested by the respective anomalous scattering factors and .
Since no temperature factor refinement was performed for the heavy atoms the occupancy factors produced
by MLPHARE include contributions from the actual occupancy of the atoms, the anomalous scattering factors
and the atomic temperature factors. If however we restrict ourselves and inspect occupancies of only one
heavy atom site, in this case site 3, we can assume that the actual heavy atom occupancy and the temperature
factor contribution are the same for each of the seven cases shown. The isomorphous and anomalous occupancies (
and 
) should
therefore reflect only the magnitudes of the respective anomalous scattering factors at the seven X-ray energies
scaled by a constant multiplicative factor accounting for the overall scale of the data, the actual atomic occupancy
and the atomic temperature factor.
To allow direct comparison of the experimentally measured anomalous scattering factors and the refined
occupancy factors the latter were scaled as follows:-
and
where the average scale factors 
and 
are calculated using data sets 1, 2, 3, 5, and 6 only since
there are no estimates of the anomalous scattering factors for data set 7.
Figs.
Figure: Values of and plotted together with the isomorphous and anomalous
occupancies calculated by MLPHARE for all seven data sets. The occupancy factors have been scaled to allow
direct comparison with the experimentally measured anomalous scattering factors.
and
Figure: Experimentally measured curves of and plotted with scaled occupancy factor
calculated from MLPHARE for data sets 2, 3, 4, 5 and 7 in the neighbourhood of the iridium 
absorption
edge. The r.m.s deviations of the refined and scaled occupancies about the experimentally determined values of the
anomalous scattering factors are 1.36e for and 0.87e for . The poor agreement of 
for data set 7 with the experimental curve suggests an
error in the estimation of the X-ray energy for this data set.
show the anomalous scattering factors plotted along with
and 
as a function of the data set number and as a function of X-ray energy near the absorption
edge. The r.m.s. deviations of and about the experimental curve are 1.36e and 0.87e
respectively. The values of and for data set 7 correspond to anomalous scattering
factors of 
and 
.
It is immediately apparent that very little contrast in has been achieved in the X11 experiment. The values
of the refined isomorphous occupancies imply a difference in between data sets 6 and 7 of 
.
This compares
poorly with a difference of 
observed between data sets 1 and 4 collected on X31.
The effect of the broader bandpass of X11 on the anomalous scattering factors of Iridium is demonstrated in
Fig. . The experimental data measured on X31 is shown after convolution with a Gaussian
functions with FWHM's of 
the resolution of the X31 beam line using a Si(111) monochromator, 
the calculated
energy resolution of the X11 beam line and 
an arbitrary intermediate resolution.
Given the magnitudes of the isomorphous and anomalous occupancies for data set 7 and the expected bandpass of the X11
beam line we can conclude that data set 7 was in fact collected at an X-ray energy which was 
below the
iridium absorption edge. It is however also clear that given an energy resolution of the attainable
contrast in the scattering factor is severely limited. The minimum value of becomes 
.
In addition the white line feature in the curve is almost eliminated by the effect of the
broad band pass.
Figure: Experimental anomalous scattering data for the Iridium edge convoluted with three Gaussian functions
of varying widths chosen to approximate X-ray beam line instrument functions with different bandpasses. Results are
shown after convolution with Gaussians of FWHM 4, 15 and 33
.