"If the potential applied to the gate G5 is 0.6 to 1.1 volts, the potential in the active layer 12 beneath the gate G5 will be positive, equal to around 0.2 volts for example. There then exists a strong local electric field beneath the gate at the surface of the silicon towards the edge of the photodiode which is maintained at 0 volts by the surface region 16. This electric field acts by lowering the forbidden band of the semiconductor and by therefore increasing the probability of electrons passing into the conduction band. This is a physical effect of band-to-band tunnelling, which creates a leakage current. Electrons are generated beneath the gate without the lighting being the cause; they will go to be stored in the photodiode with the highest potential. This current can be likened to a dark current since it exists independently of the lighting. This dark current, specifically due to the presence of a difference between the potential beneath the gate and the surface potential of the photodiode, is particularly bothersome when detection of weak lighting is desired. It can be several hundred times higher than if the potential beneath the gate was nil."
Whether this explanation is correct or not, the patent application proposes a pulsed anti-blooming bias to minimize the dark current:
