Their large dynamic range (> 104), high quantum efficiency (QE ≈, 50%), and large format 103x103 pixels) make charge coupled devices (CCD) ideally suited to the high signal to noise demands of differential imaging (e.g. polarimetry, narrow-band photometry, fast-chopping ultra-deep imaging). However, if the CCD is read once for every phase in the chopping cycle, the readout noise is comparable to the signal in most astronomical applications (σ ≈ 30 RMS electrons/pixel), F ≈ 10 electrons/pixel -s, chopping rate ≈ 0.1 Hz). We describe herein a technique which allows long integration periods between readouts (t ≈ 103s) together with moderate chopping rates (0.1 - 1.0 Hz). During each phase of a differential chopping cycle, one set of charge images is shifted to the center of the chip, which is light sensitive. At the same time the other set(s) of charge images is shifted under one of two masked areas which serve as charge-storage sites. Since the vertical shifting between phases can be accomplished in much less time than the typical exposure per phase, there is negligible image smearing and, therefore, no shuttering is required except during readout. In order to achieve efficient charge transfer efficiency in both the normal and reversed vertical directions, three phase or four phase vertical electrode structures are required. For a given light level and chopping rate the performance of this technique is limited primarily by the readout noise and the vertical charge-transfer inefficiency (η ≈10-4 per shift). If, in order to achieve sufficient signal to be photon-noise-limited (S > σ2/η, many transfer cycles are required between readouts, the charge transfer inefficiency will diffuse the accumulated charge images along the,vertical columns. The diffusion time scale in terms of vertical shifts is ≈ 0.5 λo2'in where λo is the intrinsic image scale in picture elements (pixels). The figure of merit for a given CCD is, therefore, proportional to QE/σ2η. A more detailed discussion of the theoretical performance of the technique, including the effects of poor charge-transfer efficiency at low light levels, is presented. We also present laboratory and astronomical data obtained with this technique using the Steward Observatory CCD Camera and the 2.3m telescope on Kitt Peak.
|