Optimization of large format CMOS image sensors using contrast detection figure of merit

Abstract

This paper discusses specific SRI CMOS imagers that use for optimization a relatively new imaging performance parameter called the Contrast Detection Figure of Merit (CDFM). The large backside illuminated (BSI) stitched imagers come in two primary formats: a NMOS 10 um pixel x Mk x Nk imager where M x N are whole numbers and a PMOS 4um quad pixel x 4k x 4k and 8k x 8k. We briefly describe their pixel design and performance in addition to comparing NMOS and PMOS technologies discussing their advantages and disadvantages. Next we describe the CDFM technique where we join TCAD device and Monte Carlo simulations which together produce the modulation transfer function (MTF) and Quantum Efficiency (QE) as a function of photon wavelength, silicon epitaxial (epi) thickness and substrate bias given the pixel size. The two parameters are then multiplied together producing CDFM. The results show that quantum efficiency (QE) and the modulation transfer function (MTF) are competing variables which lead to an optimum epi thickness for maximum CDFM and contrast signal to noise (CSN) performance. The CDFM curves can also be expanded to MTFxQExBB where BB is the working Black Body radiation spectrum (BB) leading us to absolute performance given the strength of incoming radiation as a function of wavelength. Lastly, the paper compares CDFM modeling and testing results using standard measurement techniques (MTF, QE and Photon Transfer).