The CCD is a solid-state imaging device based on Silicon. It was invented by Boyle and Smith at AT&T Bell Labs in 1969. Forty years later, CCD performance is now approaching that of an ideal detector: over 90% of the incident light can be recorded over most of the visible spectrum and electronic noise is so low that even single photons can be measured. Pixel count has steadily risen, the current record being 111 mega-pixels, limited only by the size of the wafer on which the device was fabricated. Willard S. Boyle was born in Canada in 1924. His studies at McGill University in Quebec were interrupted by his military service in the second world war during which he trained in the UK as an aviator with the Fleet Air Arm. After the war he returned to McGill where in 1950 he gained his PhD. He began work at Bell labs in 1953 where he remained until he retired in 1979 as the Executive Director of the Communication Science division. He died in 2011.
George S. Smith was born in the USA in 1930. After gaining his PhD in 1959 he moved immediately to Bell Labs, working under Willard Boyle studying the electronic properties of semi-metals , thermoelectric cooling and low temperature electronic devices. He later led research into novel lasers and semiconductor devices, retiring in 1986 as head of the VLSI Device department. The contribution of Boyle and Smith to science as well as to the economy was recognized in 2009 when they were jointly awarded the Nobel Prize for Physics.
Bell Labs in New Jersey was clearly a remarkable place to work, indeed both our protagonists were content to spend their entire careers there. To date it has produced seven Nobel prizes and is famous for the invention of the junction and field-effect transistors, the laser, the solar cell and the Unix operating system. Interestingly, for a commercially funded laboratory, it was also home to some fundamental discoveries in physics: information theory, the first demonstration of the wave nature of matter and the first observation of the Cosmic Microwave Background. Bell Labs´ success has been put down to the creative freedom given to its staff. The telecommunications monopoly enjoyed by AT&T in the post-war era allowed it to maintain a critical mass of thinkers and doers, free to indulge their curiosity without the constraints of commercial deadlines. This culture was largely the creation of Bell's research director, and later president, Mervin Kelly. At the time of the invention of the CCD, Jack Morton the head of advanced research at Bell was an important motivator. He was pushing to find a semiconductor analogue to the magnetic bubble memory (also developed at Bell labs!). On another front there was a need for a solid state imaging device for the "Picturephone" video-telephone project. Initially a 660 x 660 pixel photodiode array was proposed but this needed to be read out using a scanning electron beam that added to its complexity and cost. Thus both imaging and memory applications were in mind when the CCD was born. The CCD was first conceived by Boyle and Smith on October 19th 1969. It was rapidly implemented in Silicon and announced in their famous paper: "Charge Coupled Devices", "Bell Syst. Tech J." Vol49 1970. CCDs detect light using the photoelectric effect. If the wavelength of the incident photon is less than approximately 1um then it will be absorbed by the Silicon, in the process liberating an electron that is free to drift under the influence of whatever electric fields happen to be present. The challenge for CCD designers is to manipulate these fields to ensure that these so-called photoelectrons are safely collected and stored close to their point of generation and at the end of the exposure accurately measured with as little noise as possible. By accurately measuring the pattern of charge present across the area of the CCD at the end of exposure we can thus infer the pattern of light (i.e. image) that gave rise to it.
Physically, a CCD consists of a sandwich structure typically comprising a 15um thick base-layer of p-type Silicon overlaid by a 0.5um layer of n-type Silicon. An insulating layer of SiO2 is deposited on top of the n-type layer and further overlaid by a series of electrode structures, manufactured from polycrystalline Silicon. Potential wells are created in the underlying p-n sandwich, one well per pixel, by applying a positive voltage to these electrodes. Photoelectrons fall into these wells where they are stored until the exposure finishes and the read-out process begins. Here, the same electrode structure used to collect the charge is used to transfer the charge, pixel at a time, to the measurement amplifier at the edge of the device. This is done by modulating or "clocking" the voltages applied to the electrodes in a carefully controlled sequence, each pixel charge packet following the most positive potential in its vicinity. The amplifier outputs a voltage whose amplitude is proportional to the charge contained within the pixel under measurement. As the readout of the CCD progresses this sequence of voltages needs to be further amplified and digitized by external electronics and the data then stored as a digital image file.
An important catalyst at the time of the early CCD development was the need to find a suitable detector for the Hubble Space Telescope (HST) and various JPL solar system exploration missions. Photographic film was not an option due to the problems of retrieval. Image tubes were also problematic due to photocathode degradation over time. The first area-CCDs measured only 100x100 pixels and suffered from poor charge-transfer efficiency (CTE) and poor ultra-violet sensitivity. When illuminated with green and red light, however, they were superbly sensitive, exceeding image-tube performance by a factor of 5 and photographic plates by a factor of 100.
The CTE problems were addressed at JPL who worked together with RCA, Fairchild and Texas Instruments with the aim of developing a sensor for the Viking and Voyager missions. This work lead to the first commercial CCDs from Fairchild in 1974. It was now obvious that CCDs could find a large market in TV cameras which provided an additional commercial incentive.
Astronomical use of the new Fairchild CCD began almost immediately. It proved to be a revolution, with each observation almost guaranteeing a new discovery. By 1976 400 x 400 pixel devices were available and work had started on an 800 x 800 pixel device for the HST and the Galileo probe to Jupiter. The initial instrument suite of the HST was upgraded several times by visiting Shuttle missions. In 2002 a mosaic of two 2k x 4k pixel CCDs manufactured by SITe was installed as part of the Advanced Camera for Surveys. The final Shuttle servicing mission in 2009 saw the installation of Wide-Field Camera 3 which contains a mosaic of two 2k x 4k pixel CCDs in its ultraviolet and visible channel. As of 2013 this camera continues to provide spectacular images of the distant universe with a resolution and spectral coverage difficult or impossible to obtain from any ground-based observatory. CCDs have now almost entirely displaced the use of film and image tubes at professional observatories. The trend is to build mosaics in order to cover larger areas of sky in a single exposure thus overcoming the size limitations of the individual CCD devices. One very productive mosaic-based project has been the Sloan Digital Sky Survey (SDSS). It uses a camera containing thirty 2k x 2k pixel CCDs to catalogue the positions and colors of objects in large swathes, gradually building up a database of objects across the entire northern sky. The mosaic shares a dedicated 2.5m telescope, located in New Mexico, with a second follow-up instrument (also containing CCD sensors) that records the spectra of specific objects of interest. Since becoming operational in 2000 SDSS has catalogued over 230 million celestial objects. It has harnessed the joint technologies of large CCD mosaics and large-scale data processing/archiving to revolutionize our knowledge of astronomy across fields ranging from stellar evolution to the large-scale structure and dynamics of the universe.
In the consumer market CCDs have also become dominant. The first digital camera was the 0.4 Mega-pixel Fuji DS-1P in 1988. Progress has since been rapid, stimulated by the release of JPEG compression technology in 1991 and of compact-flash storage devices in 1994. The largest consumer CCD camera is now the 40 Mega-pixel Pentax 645D which has a sensor measuring 44 x 33mm.
CCDs are currently being displaced by the newer CMOS technology in many top-end consumer cameras. CMOS technology, which is based on photodiode arrays, is still not able to give the same sensitivity and low-noise performance as a CCD, at least for the moment. They are, however, more economical to produce and can simplify other aspects of the design, negating the need for a mechanical shutter for example. It is remarkable that in just 40 years CCD technology has matured from initial conception to a level which approaches the fundamental limits of performance imposed by the laws of physics.