Dynamic Crosstalk Analysis for Branching Image Sensors

This paper reviews the concept of a backside-illuminated image sensor with a branching channel. Further, we present the analysis of the temporal crosstalk due to transient motions of signal electrons with varying voltages applied to the sensor. A Monte Carlo simulation shows that the average arrival time of signal electrons to a floating diffusion is 386 ps, which is equivalent to the frame rate of 2.59 Gfps. The frame count of the sensor is 12 frames. The sensor can operate in two modes: a burst imaging mode and a continuous imaging mode for a frame count of 100 to 1,000 frames. Burst imaging distributes signal electron packets to 12 in-pixel floating diffusions (FDs) in turn. Signal accumulation is also possible for weak incident light from reproducible events by repeating the image capturing process; theoretically, noiseless ultrafast imaging is possible, especially with system cooling. Continuous imaging further transfers the image signals from the FDs continuously to memory elements in a stacked memory chip. The signal transfer rate from each FD is 1/12 of that for image sensors with a single FD, and thus the signal transfer noise decreases by 1/√12. Keyword ultra-high-speed imaging, branching image sensor, super-temporal resolution, dynamic analysis, temporal crosstalk