De-coupling Detection Efficiency and Spectral Resolution by Choice of Event Selection Criteria

The charge diffusion and charge loss processes just described also affect the effective detection efficiency of the CCD. This coupling operates mainly by way of the ``event-shape-based'' event selection criteria which, as discussed above, are used to reject events for which charge collection is so poor that the event amplitude is not indicative of the incident photon energy. If each and every event rejected on grounds of shape resulted from interactions outside the depletion region, then the effect of the selection criteria could be modelled in a straightforward ``geometrical'' way. In fact, the selection criteria are not perfectly efficient; they accept a fraction of events occuring outside the depletion region, and reject a fraction of events occuring in the depletion region.

While the ultimate CCD model will predict this coupling by accurate simulation of the details of the charge diffusion process, for present purposes we have attempted to avoid the problem by choosing event selection criteria that minimize the influence of charge collection effects on detection efficiency. Specifically, we elect to compute the response of the CCD for ASCA event grades 0, 2, 3, 4 and 6, with a split event threshold of 15 electrons. (This criterion accepts all 1- and 2-pixel events, as well as ``L-shaped'' 3-pixel events and square or ``Quad-shaped'' 4-pixel events. More precisely, our standard grade selection criterion is select in accordance with the following ACIS grade map: 0x1d05 0x0047 0x0004 0x0047 0x1133 0x0003 0x1100 0x0000 0x1d04 0x0000 0x0004 0x0000 0x0000 0x0003 0x0000 0x0000. See the ACIS Software Requirements Specification for an interpretation.)

This criterion also turns out to be relatively insensitive to the precise value of the split-event threshold.

In principle, the event selection criteria we've adopted represent the high-detection efficiency extreme of detection efficiency-vs.-spectral resolution tradeoff. However, in practice, the readout noise of the ACIS detectors is so low, and the depletion depth so large, that, at least for front-illumianted detectors, there is little sacrifice in spectral resolution entailed by this choice. As is discussed elsewhere in this report, the situation is not so clear for the back-illuminated detectors.