An avalanche photodiode (APD) is a highly sensitive semiconductor photodiode detector that exploits the photoelectric effect to convert light into electricity. From a functional standpoint, they can be regarded as the semiconductor analog of photomultiplier tubes. The avalanche photodiode was invented by Japanese engineer Jun-ichi Nishizawa in 1952. However, study of avalanche breakdown, microplasma defects in silicon and germanium and the investigation of optical detection using p-n junctions predate this patent. Typical applications for APDs are laser rangefinders, long-range fiber-optic telecommunication, and quantum sensing for control algorithms. New applications include positron emission tomography and particle physics.
Principle of operation
By applying a high reverse bias voltage (typically 100–200 V in silicon), APDs show an internal current gain effect (around 100) due to impact ionization (avalanche effect). However, some silicon APDs employ alternative doping and beveling techniques compared to traditional APDs that allow greater voltage to be applied (> 1500 V) before breakdown is reached and hence a greater operating gain (> 1000). In general, the higher the reverse voltage, the higher the gain. Among the various expressions for the APD multiplication factor (M), an instructive expression is given by the formula
where L is the space-charge boundary for electrons, and is the multiplication coefficient for electrons (and holes). This coefficient has a strong dependence on the applied electric field strength, temperature, and doping profile. Since APD gain varies strongly with the applied reverse bias and temperature, it is necessary to control the reverse voltage to keep a stable gain. Avalanche photodiodes therefore are more sensitive compared to other semiconductor photodiodes.
If very high gain is needed (105 to 106), detectors related to APDs called SPADs (single-photon avalanche diodes) can be used and operated with a reverse voltage above a typical APD's breakdown voltage. In this case, the photodetector needs to have its signal current limited and quickly diminished. Active and passive current-quenching techniques have been used for this purpose. SPADs that operate in this high-gain regime are sometimes referred to being in Geiger mode. This mode is particularly useful for single-photon detection, provided that the dark count event rate and afterpulsing probability are sufficiently low.
Materials
In principle, any semiconductor material can be used as a multiplication region: Silicon will detect in the visible and near infrared, with low multiplication noise (excess noise). Germanium (Ge) will detect infrared out to a wavelength of 1.7 µm, but has high multiplication noise. InGaAs will detect out to longer than 1.6 µm and has less multiplication noise than Ge. It is normally used as the absorption region of a heterostructure diode, most typically involving InP as a substrate and as a multiplication layer. This material system is compatible with an absorption window of roughly 0.9–1.7 µm. InGaAs exhibits a high absorption coefficient at the wavelengths appropriate to high-speed telecommunications using optical fibers, so only a few micrometres of InGaAs are required for nearly 100% light absorption. The excess noise factor is low enough to permit a gain-bandwidth product in excess of 100 GHz for a simple InP/InGaAs system, and up to 400 GHz for InGaAs on silicon. Therefore, high-speed operation is possible: commercial devices are available to speeds of at least 10 Gbit/s. Gallium-nitride–based diodes have been used for operation with ultraviolet light. HgCdTe-based diodes operate in the infrared, typically at wavelengths up to about 14 µm, but require cooling to reduce dark currents. Very low excess noise can be achieved in this material system.
Performance limits
APD applicability and usefulness depends on many parameters. Two of the larger factors are: quantum efficiency, which indicates how well incident optical photons are absorbed and then used to generate primary charge carriers; and total leakage current, which is the sum of the dark current, photocurrent and noise. Electronic dark-noise components are series and parallel noise. Series noise, which is the effect of shot noise, is basically proportional to the APD capacitance, while the parallel noise is associated with the fluctuations of the APD bulk and surface dark currents.