A CCD (charge-coupled device) stores charge produced by incoming photons in each pixel and transfers it pixel by pixel with a clocked voltage for readout. The first prototype worked in 1970, providing higher sensitivity and linearity than photographic film. It eliminated film in TV broadcasting, astronomy, and medical endoscopes, accelerating digitization. Although CMOS sensors now dominate consumer devices, CCDs remain preferred for scientific work requiring low noise and high quantum efficiency. More than 50 years after its invention, the CCD still underpins both science and industry.

A CCD uses an oxide layer and polysilicon gates on p-type silicon; photo-generated electron-hole pairs are separated and electrons are trapped in a potential well. Horizontal and vertical clocks shift the charge packets like a bucket brigade, align them into a serial register, and then convert them to voltage via an amplifier. Charge-transfer efficiency (CTE) must exceed 99.999 %; gate defects and radiation damage are major degradation factors. Back-illumination and deep-UV coatings raise quantum efficiency above 90 %. Cooled CCDs suppress readout noise to a few electrons and are employed in instruments such as the Hubble Wide-Field Camera and X-ray spectrographs.

CCDs are categorized into surface-channel (SCCD) and buried-channel (BCCD) devices, the latter employing a thin n-implant to deepen the potential well and improve CTE and dark noise. Three-phase/four-phase clocking modulates transfer barriers, while split-node architectures enable high-speed readout. The output stage uses a source follower and Johnson-correlated double sampling, achieving kTC noise near 1 e⁻. P-channel CCDs and Δ-doping are explored for radiation hardness and UV response, as adopted in projects such as LSST and Euclid. Boyle and Smith’s principle remains the gold standard for high-precision photon counting.