Rigorous Three-dimension Electromagnetic Simulations and Optimization for CMOS Image Sensor pixels

The size of pixels in complementary metal-oxide-semiconductor image sensor has steadily decreased during the last decade, which leads to a challenge on maintaining the pixels' optical performance. The diffraction effects result in increased spatial crosstalk and decreased optical efficiency, so rigorous simulation and optimization are important. FDTD method is used to simulate the optical performance and a diffuse-like source is used to reproduce real conditions. A commercial tool from Lumerical Solutions is used to do three-dimension FDTD simulation. The diffuse-like sources are simulated by the uniform sum of several focused beams characterized by a given f-number of the objective len. The typical 1.75 mu m, 1.45 mu m and 1.1 mu m pixels are simulated and analyzed. The results show that the optical efficiency of the pixels decreases dramatically when the pixels size scaling down. Several approaches to produce better device performance for sub-2um pixels are analyzed. Micro-lens optimization, dielectric stack height reduction can decrease the optical power loss and optical confinement realized by an air-gap can reduce the spatial crosstalk. The results show that the optical performances of the optimized 1.1 mu m pixel are comparable to those of conventional 1.75 mu m pixel.