Optimization of CMOS active pixels with high signal-to-noise ratio for high-resolution X-ray imaging

As the pixel size becomes smaller, the signal level generated in pixel will decrease. This small signal is critically affected by noise, and so the image quality also degrades. Therefore, it is very important to increase signal-to-noise ratio of the pixel especially in high-resolution imaging applications. We developed a time-dependent theoretical noise model considering integration noise due to the shot noise generated in photodiode and readout noise due to the thermal and flicker noise arising from the readout transistors. The developed model can take the input variables such as photocurrent, capacitance of the photodiode, integration time, each size of the transistors comprising active pixels with 3 T structure, load capacitance, etc. To validate our noise model, test structures having several sizes of the source follower and selection transistor in 20 μm pixel pitch have been realized using AMIS 0.5 μm (2P/3 M) CMOS process. The pixel type is the n+/p and the fillfactor is around 30%. The results of the noise measurement are agree well with model calculations, and the optimum values of in-pixel components can be extracted using this developed noise model.