3.4 Detector

  In a departure from the norm for far uv instruments, IMAPS uses an image sensor that does not employ a microchannel plate as the primary detection element (Lampton 1991; Joseph 1995). Instead, the photosensitive medium is a thin layer of KBr on a smooth, nickel plated, solid surface about 2.5 cm in diameter. The quantum efficiency of such a photocathode, reaching about 77% in the middle of the IMAPS wavelength band (Carruthers 1981, 1987), is superior to that of a microchannel plate coated with the same compound (Siegmund et al. 1986). Photoelectrons emitted off the photocathode are accelerated by an electrostatic potential difference of 18 kV toward an RCA (SID-502) CCD that has been specially thinned on its back side. At the same time, the electrons are focused by an axial magnetic field created by the cylindrical permanent magnet that surrounds the image section. Fig. 2 shows that the direction of the electric field is inclined with respect to the direction of travel of the electrons and the magnetic field, because of the need to have the incoming optical beam and its image plane at an angle that is offset from the line to the CCD. This offset in the field directions causes very little degradation in the quality of the focus (Lowrance 1985).

The active area of the CCD has 320 pixels parallel to the direction of the echelle dispersion, spanning a linear dimension of 9.6 mm (equivalent to an angle of 18'20'' in the sky), and 256 pixels in the cross dispersion direction (spanning 7.68 mm 14' 40''). Each pixel is $30\mu{\rm m}$ square (3''.45). When the 18 keV electrons collide with the CCD, they each generate about 2500 secondary electrons within the silicon layer, and these secondaries are collected by the charge wells, in much the same manner as are the electrons that are created by direct photoillumination. The CCD has two frame registers: one is the active area that is being bombarded and the other is a storage area of the same size (and covered by a metal mask) that holds the previous frame's pattern to be read out.

The large signal gain created by the avalanche of electrons in the CCD allows the detector to be used as a photon counting device, since each event stands out as a bright spot with a charge equivalent to 15 times the rms noise in each pixel. This in turn means that with no great effort one can achieve a signal-to-noise ratio that is as good as the limit determined by the statistics of detected photons, rather than by the readout noise of the CCD or fluctuations in the dark current. However, this statement does not apply if one is coadding many independent readouts with very few events per frame, with no explicit detection of the events as they arrive. See Jenkins, et al. (1988) for the mathematical details.