POINT RESPONSE WITH THE BORON FILTER
An apparent count rate deficiency has been evident in data for all observations carried out with the Boron filter (ref. Express no. 8 p.31) and a clear explanation of the effect has now been established viz: the point spread function (PSF) is much wider because of scattering of the X-rays within the boron filter. This scattering is energy dependent and strongest at low energies. Figure 1 shows a comparison of the integral PSF for a number of point sources observed during the mission. Clearly, the PSF for HZ43, a soft X-ray emitter with kT ~0.006 keV and low column density ~1018 cm-2, is so wide that an integration box size of ~16 arcmin is necessary to retrieve all source photons. Her X-1 with a column density of ~1020 cm-2, shows the on-set of the effect in comparison with 7 harder and/or more cut-off sources, which comprise all strong sources (0.3 cts/s) observed with the boron filter. Note that for these hard sources, the PSF is still wider than the PSF for a lexan filter measured in flight. All other filters have comparable PSF's to that of lexan.
Using a single integration box size for all filters, for example as for lexan, will thus give an underestimate of the source counts for observations with the boron filter. The onset of this effect at ~1020 cm-2 is further illustrated in figure 2, which shows the ratio of predicted to measured boron filter count rates for a sample of bright AGN's. A photon spectral index of 1.7 was assumed and the corresponding NH was found from the lexan/Al-Parylene count ratios. An increase of the count rate deviation at lower NH is clearly visible. Note that this figure merely illustrates the energy dependence and cannot, because of the underlying assumptions, be regarded as a calibration.
In view of the above, we give some guidelines to aid the analysis of the boron filter data. Strong sources with good signal-to-noise ratio should be integrated with a large box size, which can be estimated from figure 1. Increasing the box size will indicate when the full source flux is collected. Source fluxes to an accuracy of ~20% for a hard or highly cut-off (NH 1021 cm-2) source can be found with a correction for the almost constant shape of the PSF (see figure 1). The CCF PSF is fairly representative in this case. For soft and/or fairly nonabsorbed sources the CCF currently does not contain sufficient information to accurately process the data. The PSF as a function of energy has been calculated at the Space Research Lab. Leiden, using predictions of the scattering in the boron filter. This data is being implemented at ESOC. A comparison with in-flight calibration sources such as HZ43 will be carried out to validate the model data. Finally, the PSF for the boron filter will be updated in the CCF, probably in the form of the scattering data and an algorithm to construct the boron PSF for a given source spectrum.
J. Davelaar & P. Giommi