Since it’s invention at Bell Laboratories by Boyle and Smith [1970], the CCD gradually took over as the dominant form of imaging technology only recently being replaced by APS(CMOS) technology. In television applications, CCDs almost completely replaced scanned image tubes in both professional and domestic applications. Computer image scanners and facsimile machines also make extensive use of CCD technology. It is, however, in astronomical applications that we are most interested.

Unlike many day to day applications, detectors for astronomical use must operate with extremely low light levels and consequently often with exposure times measured in minutes, even hours. What then are the characteristics of an "ideal" detector and how does the CCD measure up to this ideal? How does photographic film compare?

Quantum Efficiency
The quantum efficiency is a measure of how many of the arriving photons are detected. For a CCD this corresponds to the creation of an electron-hole pair and for film a photochemical reaction. The CCD will typically have peak quantum efficiency of more than 60%, with values exceeding 90% being achieved. This is 10 to 100 times faster than film which has a quantum efficiency from about 3% to as low as 0.1%, depending on film type and how it is prepared prior to exposure.
Spectral Response
Although a basic CCD has little response at wavelengths much shorter than 400nm, special manufacturing and or processing techniques can extend the useful response to well beyond the atmospheric cutoff at 292nm. The band-gap energy of silicon (from which CCDs are made) sets an intrinsic long wavelength limit to the response of about 1100nm. In practice, the response has typically dropped to 10% of it’s peak by 950nm and to negligible amounts by 1000nm. In contrast with basic CCDs, most films have excellent response to shorter wavelengths but their response to red and IR is often minimal.
Dynamic Range
The dynamic range of a detector is the ratio of the largest to smallest signal that can be simultaneously detected. It is in this area that the CCD has the greatest advantage over photographic films. Photographic films typically have a dynamic range in the order of 30:1. The CCD on the other hand is, for many devices, capable of read noise limited dynamic ranges in excess of 10,000:1.
Linearity
This is the other main characteristic which makes CCDs significantly more useful than photographic films for quantitative measurements. The response of film is approximately logarithmic while the CCD is nearly ideal in that the detected output is, to a high degree of accuracy, proportional to the number of photons detected.
Uniformity and Repeatability
In the early days of CCD technology, CCDs often exhibited large variations in pixel to pixel sensitivity, although specially selected scientific grade devices were available for critical applications. Modern devices, however, are significantly better with standard devices having typical peak deviations in pixel sensitivity of less than 25%, with the RMS deviation significantly less than this. Selected devices may achieve better than 1% peak. Fortunately, these variations in sensitivity are relatively unimportant in terms of measurement accuracy. The reason for this is repeatability: For a given set of operating conditions, the sensitivity variations will be time invariant and can be determined by applying an even illumination to the detector then measuring the deviation from the average detected level. It is therefore possible to calibrate or flat-field the CCD and almost completely compensate for sensitivity, and other, variations in its characteristics.
 

Silicon, Electrons and Holes

Table 4.1: Some optical properties of silicon.
Wavelength
(um)
Refractive
Index
Absorption
Depth (um)
0.325.100.01
0.45.740.08
0.54.440.6
0.64.021.7
0.73.804.5
0.83.6710
0.93.5826

Like the majority of today’s semiconductor technology, CCDs are made from silicon. Silicon is an element with 4 valence electrons and a resulting diamond crystal structure. The semiconductor nature of silicon comes from the fact that relatively little energy is required to move an electron from the valence band, where it forms part of a covalent bond with neighbouring atoms, to the otherwise empty conduction band. For silicon this intrinsic band-gap energy is 1.12ev at 300K. As we will show later, the band-gap energy has important implications for the optical properties of a CCD detector. Some of the other optical properties of silicon are presented in Table 4.1.

When an electron is transferred from the valence band to the conduction band a vacancy, or hole, is created in the valence band structure. In the absence of any electric fields, the electron will jump back and recombine with the positively charged hole, effectively disappearing from the conduction band. On the other hand if an electric field is present then conduction will occur because of the “free” electron. Additionally a nearby valence electron may move into the positive hole leaving another positive hole in it’s original position. The process repeats, resulting in a net transport of positive charge. In its application to electronic devices, the electrical characteristics of pure (intrinsic) semiconductor silicon are controlled by that is the addition of trace amounts of impurities (doping).

ccd-f1.gif
Figure 4.1: Simplified view of the
addition of impurity atoms to
silicon.

As illustrated in a very simplified manner in Figure 4.1, N-type silicon is created by doping intrinsic silicon with a 5-valent element such as arsenic. Four of the five valence electrons participate in bonding to neighbouring silicon atoms. The fifth valence electron, being superfluous to the bonding structure, is only loosely bound to the positively charged arsenic ion. The thermal energy at room temperature is more than adequate to elevate the majority of these “extra” electrons into the conduction band. In a similar manner P-type silicon is created by doping intrinsic silicon with a 3-valent element such as boron, where the incomplete covalent bond (hole) is loosely bound to the boron atom.

As the temperature is reduced, the number of available majority charge carriers (electrons in N-type doped extrinsic semiconductor and holes in the P-type) is also reduced, until at some low temperature the properties of the material approach that for pure intrinsic silicon. For devices such as CCDs the resultant minimum functional temperature is typically about 150K. The importance of this will be discussed later.

Next: how CCDs operate