The sensitivity of the calibrator to shifts in the incident X-ray energy (neglecting the method of
feedback) is dependent on the rate of change of the signal in the two calibrator detector channels
with changing X-ray energy. This in turn is ultimately dependent on the width of the reflections
recorded in the calibrator as a function of energy. The narrowest reflections observed are those
which are diffracted with the largest Bragg angles since their widths are less affected by the
vertical and horizontal divergences of the X-ray beam. The width of reflections is however also
dependent on the energy bandpass of the monochromator. Although this effect reduces the calibrators
energy sensitivity, it simultaneously reduces the necessity for energy stability by damping out the
sharper features in the 
and 
spectra. This is not the case however
where the energy resolution is less than or comparable to the typical width of feature in the
spectra (
say). At this level reductions in the energy resolution effect the experiment
directly.
The shifts in X-ray energy which have been observed due to beam instability and/or monochromator
heating were typically 
or greater at the Ir 
edge. These shifts are small compared
with the widths of the features present in typical and spectra but
are meaningful when changes in and as a function of energy on the
rising and falling edges of a white line are considered. An energy change of may typically
result in a change of more than 2 electrons in the anomalous scattering factors in the vicinity of a
white line. These changes are of the order of 
of the and
. In turn observed anomalous differences in measured structure factors are
typically between 
. This implies that fluctuations in the X-ray energy where the
anomalous scattering factors are most sensitive can introduce errors into the measured structure
factors of the order of a few percent. The magnitude of these errors are not negligible compared to
the level of statistical error one expects in a MAD experiment, i.e. of the order a 
for
stronger reflections. The application of the stabilisation techniques reduced the level of these
fluctuation by at least a factor of three and this could be further improved, with the use of an
automatic feedback system, to a level which is insignificant with respect to other experimental
errors.
For the case of the Ir edge the calibrators 
and 
reflections were
used. Their Bragg angles at that energy were 
which resulted in negligible line
broadening, i.e. widths are dictated by the monochromator only. This allowed stability to within
to be achieved. However only the specific case of the Iridium absorption edge
has been studied in this thesis and the question remains whether all absorption edges could be
treated in a similar manner using the same apparatus to obtain energy stability of a similar
standard. Preliminary experiments at absorption edges of the other 3rd transition series
metals have suggested that it may not always be as straightforward to find a calibrator line with
such a narrow diffraction profile at the required X-ray energy. The scarcity of calibrator
reflections at lower X-ray energies (below 
) makes it especially difficult at the
absorption edges of the transition series metal having lower atomic number and the Lanthanide group
metals. The low number of lines
available at low energy could be increased by using a different calibrator crystal. The use of a
perfect quartz crystal, for example, has been suggested in [93]. Also a different cut
of the silicon crystal could be used. Cutting the crystal so that the (100) axis, for example,
points back to the source would systematically offset 
by 
making available a new
region of the 
vs. X-ray energy map for application to other absorption edges. Allowing that
axis always points vertically maintains the possibility of using 
and 
pairs
for the stabilisation procedure. A cut of crystal could in principle be found to produce a
diffraction line map which best matched the positions of as many absorption edges as possible. An
additional possibility would be the use of two reflections of differing indices, and 
, on condition that they appear in separate detector channels.
The calibrator could in theory be used at more extreme X-ray energies, for example at the 
and edges of Uranium which are found at 
and 
respectively. The
density of calibrator lines recorded by the calibrator per unit of X-ray energy space is 
times greater around the edge of Uranium compared with that of Iridium thus increasing the
chance of finding a suitably narrow calibrator reflection at the desired energies. The high energy
X-rays may create the need for additional shielding around the body of the calibrator detector which
may need to be incorporated into a new design. At the other extreme Uranium also has characteristic
absorption edges which are accessible using synchrotron radiation [49]. These lie at
energies below 
which make them unfavourable from the point of view of sample absorption
and radiation damage but advantageous because of the large anomalous scattering factors
(
between 
and 
and 
between 
and
[25]). Working at these energies may necessitate the use of a quartz
calibration crystal with a higher line density.
Still further improvement in the X-ray energy stability could be achieved by using an automated
feedback system of some description. Energy shifts observed as discrepancies in the signals recorded
from the two calibration channels could be directly countered by adjusting the monochromator crystal
angle with respect to the incident beam. The control system could either feed single pulses to the
stepper motor controlling the motion of the monochromator's arc or provide a change in voltage which
could be applied to a piezo stack mounted to provide fine tuning of the monochromator angle. The
feedback would operate to keep both calibrator signals equal. The photon counts rate typically
observed in the calibrator detectors during the stabilisation procedure was between 
and
. Given this count rate it would be possible to update the monochromator orientation
every 1/100th of a second while still retaining a reasonable signal to noise ratio from the
calibration channels.
In its present state the calibrator does not allow the X-ray energy to be repeatedly changed and
monitored so that each segment of reciprocal space can be collected at a number of energies within a
short space of time before proceeding to the next oscillation range. To perform such a task the
orientation of the calibrator crystal would need to be repeatedly and reproducibly changed so as to
shift the positions of the reference reflections in energy. The X-ray energy would then be changed
to the reference energy established by the calibrator reflections and the diffraction image would be
measured. Some preliminary tests have shown the orientational reproducibility of the calibrator
crystal's goniometer to be well within that required if the X-ray energy is to be repeatedly
readjusted to within 
. If the goniometer motor positions were calculated at each X-ray energy
before the experiment began, changing the X-ray energy could in principle be performed automatically
by a computer. The image plate read-out would provide sufficient time for such a change to be
performed.
The calibration apparatus lends itself well to installation on other beam lines. The requirements are that the beam line has spectrometer type geometry, i.e. that it allows scanning of the X-ray energy over a significant range without the need for realignment of the experiment and that there is sufficient space to allow installation of the apparatus. The X-ray beam dimensions at the calibrator crystal must allow for a small portion of the X-rays to be used by the calibrator. A beam cross section smaller than that of X31 would require a smaller hole in the silicon crystal or a tapered hole such that rotation of the crystal does not obstruct the through beam.