As discussed above, charge stored within a pixel is moved from pixel to pixel during readout via changes in electrical potential between the biased gates. For surface channel CCDs, this charge movement occurs "on the surface" of the CCD, being transferred between overlapping gates. For example, each pixel in a three-phase CCD has three overlapping gates for each imaging cell or pixel. Each cell has an associated transfer efficiency and maximum possible transfer rate. The efficiency is near to, but not quite, 100% for each cell (we will discuss this charge-transfer efficiency (CTE) further in Chapter 3). Typical rates of transfer for CCD television cameras (or video cameras) are several megahertz, while low-light level applications with cooled devices (i.e., those occurring in astronomy) use rates nearer to the kilohertz range (Eccles, Sim, & Tritton, 1983).
The major drawback to the surface channel CCD is the presence of trapping states that occur at the boundaries of the gates. These traps, which are caused by imperfections within the silicon lattice or the gate structures, cause a loss of transfer efficiency by trapping some of the electrons from other pixels as they pass by during readout. Traps within CCDs are therefore undesirable for faint light level applications such as astronomical imaging. One method that can be used to eliminate traps (although not in general use anymore because of the advent of buried channel devices; see below) is to raise the entire surface charge level of the CCD above the level needed to fill in any traps. This is accomplished by illuminating the CCD at a low level prior to exposure of the astronomical source. This technique, called a pre-flash or a fat zero, allows any nonlinearities resulting from traps at low light levels to be avoided while only slightly increasing the overall noise level of the resulting image (Djorgovski, 1984; Tyson & Seitzer, 1988).
Deferred charge is another source of nonlinearity sometimes present at low light levels (Baum, Thomsen, & Kreidl, 1981; Gilliland, 1992). Often referred to as charge skimming, charge depletion, or low light level nonlinearity, this particular worry has all but disappeared in modern, low-readout-noise devices. We will address nonlinearity in CCDs in a more general manner in Section 3.8.
A better solution than those just discussed is to move the charge from pixel to pixel during readout via a channel of semiconductor material that is away from the gate structures. This "buried channel" architecture results from the application of an additional layer of semiconductor material placed under the CCD surface. The buried channel greatly enhances the charge movement through the device by reducing the number of possible trap sites and by decreasing the transfer time between adjacent cells. Higher transfer rates (up to 100 MHz) and high CTE values (>99.995%) are easily accomplished with buried channel devices.
The price paid for the addition of the buried channel is that the total charge storage capacity for each pixel is reduced by 3 or 4 times that of a surface channel detector. However, since the operation at very low signal levels is much improved, the overall dynamic range and sensitivity of a buried channel device become much higher.
Was this article helpful?