Enhancing near-infrared sensitivity of CMOS image sensors using a hemispherical photon-trapping structure

Abstract

CMOS image sensors are extensively utilized in applications ranging from consumer electronics to biomedical imaging and autonomous systems. Despite their high efficiency in the visible spectrum, their sensitivity in the near-infrared (NIR) region remains significantly low due to the limited absorption of silicon beyond 700 nm. To address this challenge, we propose a novel light-trapping strategy incorporating a hemispherical structure at the silicon interface. This design facilitates the direct transmission of normally incident light into the silicon layer while enhancing light scattering and redistribution. Additionally, a pyramidal structure positioned below the silicon layer refracts transmitted light, further improving absorption. To minimize optical crosstalk between adjacent pixels, a deep trench isolation (DTI) structure is implemented. The optical performance of the proposed structure is evaluated through finite-difference time-domain (FDTD) simulations, demonstrating up to a 36% enhancement in optical efficiency at a wavelength of 1100 nm compared to conventional BSI CMOS image sensor designs. These findings highlight the potential of hemispherical photon-trapping strategies for enhancing CMOS image sensor performance in NIR applications such as machine vision and biomedical imaging.