| The ability to reproduce a high quality image depends strongly on the image sensor's light sensitivity. This sensitivity depends on the electrical and optical properties of the image sensor pixel.; In the first part of this dissertation, we calculate the optical efficiency of a CMOS image sensor pixel using a geometrical-optics phase-space approach. We compare theoretical predictions with measurements made using a 0.35-μm CMOS digital pixel sensor (DPS), and find them in agreement within 3%. We then use optical efficiency calculations to trade off pixel sensitivity and functionality as CMOS technology scales.; The materials used in CMOS technology are chosen to optimize electronic performance. It is therefore surprising to discover that these materials are also suitable for controlling the transmission of optical signals. In particular, as feature sizes approach the sub-100nm regime, it becomes possible to integrate wavelength selectivity into the CMOS image sensor circuit design.; In the second part of this dissertation, we explore an integrated color pixel (ICP) design. We prototype ICPs based on sub-visible wavelength patterned metal layers using a standard 0.18-μm CMOS technology. The integration of optical filters using patterned metal layers inside the pixel is motivated by the goal of increasing integration of capture and processing to reduce system power and cost. Transmittance of the patterned metal layers is measured under various illumination conditions. Several 1D and 2D patterns are found to exhibit selectivity in the visible wavelength range. Electromagnetic field simulations, using a 2D finite-difference time-domain (FDTD) method, yield good agreement with measured transmittances of 1D patterned metal layers. Based on these results, ICPs may become an alternative to pixel designs that employ a color filter array (CFA). |
