Pixel-integrated thin-film filter simulation and scaling trade-offs

TL;DR

This paper investigates the limitations of miniaturizing pixel-integrated thin-film optical filters, revealing a fundamental trade-off between pixel size, transmittance, and filter bandwidth due to diffraction effects, supported by analytical models and validation.

Contribution

It introduces analytical wave and ray optics models to predict transmittance in miniaturized filters, highlighting an intrinsic limitation and aiding future hardware and computational design.

Findings

Miniaturization reduces transmittance due to diffraction.

A trade-off exists between pixel size and filter bandwidth.

Models are validated with real sensor data and simulations.

Abstract

Thin-film optical filters can nowadays be integrated onto pixels of commercial image sensors used for spectral imaging. A drawback of having more filters on an image sensor is a loss in spatial resolution which could be regained by using smaller pixels. However, this work shows that miniaturization causes a reduction in transmittance and requires a trade-off between pixel size and filter bandwidth. This is caused by diffraction and the reduced number of reflections in the filter. Analytical wave and ray optics models are developed and used for fast transmittance prediction and this is validated using a commercial spectral sensor and numerical simulations. The identified trade-off is an intrinsic limitation of pixel-integrated thin-film filter technology and the developed models will be useful tools for exploring new hardware and computational solutions.