High instrumental resolution X-ray scattering experiments were performed
at the Cornell High Energy Synchrotron Source (CHESS) on the F3 station.
A double bounce Si monochromator was calibrated to 1.2147 Å X rays, the
scattering angles were selected by Bragg diffraction from the (111) face
of a Si analyzer crystal, and the intensity was measured by a NaI scintillation
detector. A diagram of the experimental set-up is shown in Fig. 4.
The beam size, as defined by the S1 slit, was about 1.0 mm high and 1.5
mm wide. The beam flux on the sample was about 
photons/sec/
.
The detector IC1 was used to monitor the incoming X-ray flux. The automatic shutter SH was open only during counting in order to minimize unnecessary exposure of samples to the X-ray beam. The ion chamber IC2 was used to detect the main beam. The NaI scintillator detector was used to detect the much weaker diffracted beam.
The longitudinal resolution (see Als-Nielsen et al, 1980 for
definitions used in this paragraph) is determined by the combination of
monochromator and analyzer crystals. Half of our data was taken in the
non-dispersive configuration (Fig. 4a)
which had a longitudinal resolution with half width at half maximum (HWHM)
of 1.0
Å
.
The other half of our data was obtained in the dispersive configuration
which had longitudinal resolution of 3.3
Å
(HWHM). The
out-of-scattering plane resolution is mainly determined by the horizontal
angular acceptance of the slits S1 and S3 and was 6
Å
. The relatively
poorer out of plane resolution leads to a small amount of slit smearing,
but this is easily accommodated in the data analysis. (Incidentally, it
may be noted that the term resolution in this and most physics x-ray scattering
studies refers to instrumental resolution in q-space and this should not
be confused with the direct space resolution of a structure determination.)
It may be useful to emphasize that the above experimental configuration, which uses an analyzer crystal, is intrinsically different from position sensitive detection, including film. These latter methods incur geometric artifacts involving the sample size and the length of the path for the scattered radiation that broaden peaks with increasing h as was carefully analyzed by Franks and Lieb (1979), but such artifacts are absent in our configuration. Also, to achieve the resolution in our configuration with even the best area or position sensitive detector, geometric artifacts would require scattering path lengths of 30 meters for a beam that is 1 mm in height.
With this experimental configuration, radiation damage, indicated by
systematic changes in the shape and position of the first order peak, did
not become apparent until after half an hour of exposure time in the non-dispersive
configuration. As a precaution, the maximum exposure time for typical data
was 15 minutes after which the capillaries were translated to expose a
fresh section of the sample. Also, thin layer chromatography on exposed
samples indicated only 0 to 2% lysolecithin formed per capillary, and the
amount of lipid degradation was not simply proportional to the exposure
time or to the time the sample remained near 50 
C.
Normalization of the scattering from two different exposures of the same
capillary was obtained by comparison of brief scans of the intense first
order peaks. Experimental protocol for the dispersive mode employed single
scans through the peaks with equally spaced angular increments. In the
non-dispersive mode several scans were made through each peak with different
angular increments for each scan.
The weak background scattering from capillaries containing air, pure water, 25% PVP solution and 50% PVP solution was measured using long counting times. Estimated uncertainties in background scattering are roughly 10% and error bars for the background subtracted data include the uncertainties from background determination. The angular range of the reported data for the tails of the peaks was restricted so that the average intensity was at least twice as great as the background. Also, the ratio of the maximum peak signal to the background was roughly 200 for the first order peak and roughly 50 for the second order peak.