Answers are grounded in this paper and cite their sources.
CMOS image sensors require compact color filters, but conventional dye-based filters face fabrication and long-term reliability constraints, while plasmonic designs incur Ohmic loss that limits transmission efficiency. This work numerically investigates an all-dielectric RGB filter set, with no metal in the optical path, combining TiO₂ (blue) and Poly-Si (green, red) gratings on Si₃N₄/SiO₂ spacer stacks. TiO₂ avoids short-wavelength silicon absorption in the blue channel. Fano-lineshape and phase-matching analysis shows that the spacers tune the leaky guided-mode resonance while partially participating in it. Coupled multi-parameter optimization by response-surface modeling defines the blue and green designs. With wavelength-dependent material dispersion, the final designs reach peak transmittances of 83.3% (blue), 83.7% (green), and 79.01% (red), with FWHM of 70.9, 96.1, and 91.6 nm (Q = 6.3, 5.7, and 6.8). The peak wavelength stays within 5 nm of its normal-incidence value for both TE and TM polarization up to 15° (blue) and 20° (green and red). With silicon-photodiode weighting, the filter set covers 21.6% of the sRGB gamut, against 14.7% for the Horie et al. baseline under the same procedure, and gives a mean CIEDE2000 of 2.48 over 24 ColorChecker patches. A potentially CMOS-compatible fabrication route using localized excimer laser annealing and unannealed sputtered TiO₂ is proposed. All results are simulated for infinite periodic arrays.
No takes yet. Share an insight, caveat, or question.
El-Metwally et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: