Simulation Of Low Noise CMOS Image Sensor Based On Skipper Technology

In the fields of night vision imaging,astronomical observation,medical endoscope imaging,and other imaging applications,imaging faint objects is a common challenge.This poses high requirements for the signal-to-noise ratio of image sensors.In recent years,with the advancement of correlated multi sampling technology in readout circuits,the readout noise of CMOS image sensors has been effectively reduced.To further reduce the readout noise,in this work the Skipper technology is applied on a CMOS image sensor.Compared to multiple sampling technology in readout circuits,this technology is expected to further reduce readout noise by shortening the correlation time between reset and signal.This promotes the application of CMOS image sensors in the field of low light imaging.Based on the architecture and working principle of CMOS image sensors,the noise reduction principle of Skipper technology is firstly analyzed in this thesis.The pixel structure uses a pinned photodiode as a photosensitive element and realize charge transfer and sampling for multiple times by controlling the voltage of each gate in the pixel channel structure.The charge transfer timing within the pixel structure is also analyzed.A simulation model of the proposed pixel structure is then built in Silvaco TCAD and the charge transfer process within it is simulated.Firstly,the fabrication process of the pixel structure is planned following the standard CMOS process.Then the process simulation is conducted in the Athena module of the Silvaco TCAD,including silicon substrate preparation,isolation process,channel engineering,gate engineering,photosensitive region and sidewall formation,etc.The charge transfer process in the surface channel and the buried channel is simulated in the Atlas afterwards and the law of charge transfer within pixels is studied.From these simulations,the gate potential at each stage of the charge transfer and the charge distribution in the pixel structure are obtained.The characteristics of the noise,full well capacity,and charge transfer efficiency are also calculated and analyzed.Based on the aforementioned results,a CMOS pixel design is proposed.The simulation results show that the full well capacity of the pixel structure can reach34 ke-using a pinned photodiode as the photosensitive element.Compared to the surface channel,the buried channel can further improve charge transfer efficiency,with charge transfer efficiency higher than 99.9999% within 10 transfers.At the same time,charge transfer efficiency can be improved by reducing the gate distance.The analysis results of pixel noise characteristics show that increasing the number of multiple samples in Skipper improves the overall noise reduction effect,and the signal value after multiple averaging tends towards stability.And under different light intensity,the stable signal values have good linearity.Considering the effects of flicker noise,dark current and charge transfer efficiency,and the readout time,the number of charge transfers within a pixel is recommended to be set at 10.Based on this,the typical parameters based on this pixel structure are estimated.The number of pixels is 16384(64×256),the pixel size is 10 μm,the filling factor is above90%,the full well capacity is 34 ke-,and the frame rate is 48 fps.Finally,for the convenience of subsequent tape-out,a preliminary layout design is provided.This thesis focuses on the research on low noise CMOS image sensors based on Skipper technology.The results are beneficial for optimizing the design of low noise CMOS image sensors and promote its wider application in weak light imaging and other fields.