学术文献库

The photon counting imaging paradigm in the visible and the infrared comes from the very small energy carried by a single photon at these wavelengths. Usually to detect photons the photoelectric effect is used. It converts a photon to a single electron making it very difficult to detect because of the readout noise of the electronics. To overcome this there are two strategies, either to amplify the signal to make it larger than the readout noise (used in the so called gain or amplified detectors), or to lower the readout noise in a standard image sensor. For a long time, only amplified detectors were able to do some photon counting. Since the first photon counting systems in the visible, developed by Boksenberg and his collaborators in 1972, many groups around the world improved photon counting techniques. In the 2000's in the visible, EMCCDs (electron multiplying charge coupled devices) allowed to replace the classical image intensifier photon counting systems by solid state devices and improved a lot the QE. But EMCCDs suffer from several issues, and the most important of them is the excess noise factor which prevents to know the exact incoming number of photons in the case of multiple photons per pixel. In the infrared there was no equivalent to EMCCDs up to the development of e-APD sensors and cameras made with HgCdTe material (electron initiated avalanche photo diode). With an excess noise factor near 1 at low temperatures, photon counting is possible with these devices but only in the infrared. We will show that having excess noise factor prevents from being able to do multiple photon counting (quanta imaging) and the only solution is to lower the readout noise.