Multispectral imaging enables the simultaneous capture of multiple spectral bands from real-world scenes. However, miniaturizing multispectral imagers while preserving high transmittance remains a significant challenge for conventional filter arrays. The emergence of nanophotonic spectral routers with full-band transmission offers a promising alternative, yet their design typically relies on optimization algorithms plagued by prohibitively long computation times and limited efficiency. Here, we introduce a topology optimization algorithm based on a hierarchical parametrization framework that efficiently explores the full design space, mitigates trapping in local optima, and achieves more effective solutions to complex design problems. Using this approach, we designed free-form nanophotonic spectral routers capable of both single-wavelength and broadband high-efficiency spectral routing. Experimental results demonstrate that the energy utilization efficiency was enhanced by 50 and 71.2%, respectively, compared with the 25% theoretical efficiency limit of conventional filter arrays. The proposed devices and design strategy offer an effective pathway for high-sensitivity, high-resolution near-infrared multispectral imaging.
Wei et al. (Wed,) studied this question.