To attain high sensitivity, particularly at the highest energies, designs of the X-ray optics for the major observatories mentioned in section 1 require focal lengths of 3 - 10 m. A typical focal plane coverage of 30 arcmin diameter therefore requires a detector size of 25mm to 80mm. A standard TV format CCD is only 15mm across its diagonal, and hence substantial effort has had to be expended to increase the size of individual chips, and to develop techniques for mosaicking many chips, efficiently to cover the focal plane areas.
These concerns are intimately related to those of packaging the devices. It is vital to ensure that the packages can accommodate the thin silicon die with the correct thermal contacts, while providing immunity against the vibration environment of launch, and avoiding the inclusion of radioactive components sometimes found in traditional semiconductor packages. The efficient packaging of CCD mosaics is therefore a non-trivial problem. Lincoln Labs devices baselined for ACIS are made 3-side abbuttable, which allows very efficient packing of a 4-chip assembly, whilst the mosaic is facilitated by reducing the pixel sizes in the store sections to allow more space to run the aluminum clock tracks along the chip sides [Burke et al]. Figure 8 compares the current mosaic designs for the AXAF ACIS and XMM EPIC cameras.
The focal planes of the X-ray telescopes are curved and at field angles of several arc minutes, the optimum focus is several millimetres nearer the mirror than the plane of the focus at the field center. Planar detectors are slightly defocussed at large field angles, which degrades the resolution and sensitivity. Utilizing relatively small chips or units provides the capability to better match the focal plane curvature, but with decreased efficiency of area coverage, and greater requirements on multiple electronic readout chains or multiplexing. The greater efficiency of coverage offered by larger units, with the contrasting advantages or disadvantages, also takes longer to read out, with the penalty of degraded image smearing or count rate limits. This aspect alone hints at the complex trade-offs involved in the design of a CCD focal plane. For example, to maximize detection sensitivity for survey-type science requires an efficient of field of view coverage, but the angular resolution versus off-axis angle must be optimized, while image smearing of faint sources and background degradation must be considered. The trade-offs require detailed study, yet also impact on other areas of scientific performance.