We are confident that we understand the accuracy of quantum efficiency models for the front-illuminated (FI) detectors, and we discuss this subject at some length here. On the other hand, the models we present for back-illuminated (BI) devices are not particularly accurate, and, perhaps worse, the errors in the back-illuminated models are not well characterized. Thus the BI models given here must be regarded as interim products. Fortunately, we have very recently identified the origin of at least a large fraction of the fairly large, (i.e. 10-20%) systematic errors in our current analysis of BI CCD data. In short, we now believe it is essential to use accurate models of both source spectrum and detector spectral redistribution to obtain reliable quantum efficiency measurements for these devices. This more accurate analysis has not been completed at this writing, but we are confident that it can be done. See section 4.7.3 for a brief quantitative discussion of these errors.
We have used the so-called ``slab and stop'' model of the ACIS CCD gate structure, described in section 4.1, to model detector quantum efficiencies. For clarity, the various slab-and-stop model parameters used in the model of FI detector quantum efficiency are represented in Table 4.62. Note that only four model parameters were varied in fits to measurements made with the flight detectors. The remaining three parameters were constrained by mesh measurements (see section 4.5) or by scanning electron micrograph (SEM) measurements of siblings of the flight devices.
| |||||||||||||||||||||||||||