CMOS image sensors include control transistors in the pixel itself, which generally results in a square or rectangular shape in the design of the photo sensitive surface, which is thus much smaller than the total pixel surface. The CMOS process also results in the presence of a stack of layers (dielectric and metallic) with specific optical properties, deposited above the photo sensitive area. In order to focus the maximum number of photons on the optically sensitive area of the pixel, a micro lens is fabricated on the top of the stack.
The aim of this study is to optimise the micro lens to maximise photon collection in the photodiode. Especially we have evaluated the influence of micro lens cross sectional and base shapes on the light focusing and sensitivity.
For this study we used the optical modelling software ZEMAX (ray tracing) and modelled the CMOS image sensor by a simple optical system: micro lens, media ("stack") with particular optical properties and a photosensitive area.
Concerning the cross sectional shape, our results show that it is important to optimise the process to obtain a spherical one. Furthermore, even though the photosensitive area is rectangular, the best base shape for the micro lens is not necessarily a square: a significant portion of the photon gain due to the base shape (square or octagonal) is lost in the non-sensitive area of the pixel. Moreover, the gain in sensitivity on-axis due to the larger size of a square base is relatively small, and is offset by a significant loss in relative illumination versus incidence angle.