Inverse design of visible-infrared multispectral camouflage with radiative cooling and microwave transmission

Multispectral detection poses significant challenges to conventional camouflage systems, which require compatibility with diverse spectral regions. However, reconciling multispectral camouflage, radiative cooling, and microwave transparency within a single platform remains constrained by conflicting electromagnetic requirements. Herein, we report an inverse design framework combining the Non-dominated Sorting Genetic Algorithm II (NSGA-II) with Transfer Matrix Method (TMM) to optimize the germanium/zinc multilayer. This multi-objective optimization can simultaneously generate multiple optimal solutions without using empirical weighting coefficients in single-objective optimization, which reduces the iterations handling multiple independent targets. In only 400 iterations, the multilayer selective emitter can realize a compatible camouflage for the visible and mid-infrared (3–5 and 8–14 μm) bands, radiative cooling for the non-atmospheric windows (5–8 μm), and microwave transparency for radar bands (2–18 GHz). Specifically, the fabricated selective emitter significantly reduces the infrared radiant signature of the object compared to the silicon substrate at a temperature of 100 °C, achieving temperature reductions of 38.5 % and 33 % in mid and long wavelengths, respectively. Benefiting from the radiative cooling design, an extra temperature reduction of 5 °C compared to the polished silicon substrate is demonstrated. The inverse design can readily extend to other multi-objective optimization problems for multispectral applications, potentially creating opportunities to tackle challenges in multispectral manipulation.