Abstract The increasing global temperatures have escalated the demand for indoor cooling, thus requiring energy-saving solutions. Traditional approaches often integrate metal layers in cooling windows to block near-infrared (NIR) sunlight, which, albeit effective, lack the broad modulation of visible transmission and lead to heat accumulation due to sunlight absorption. Here, we address these limitations by developing cooling windows using ZnS/MgF2 multilayers, optimized through a binary optimization-based active learning process. We demonstrated that these multilayers, with a total thickness below 1 µm, effectively reduced indoor temperatures by blocking NIR sunlight while achieving desired visible transmittance. The designed multilayers exhibited visible transmittance ranging from 0.41 to 0.89 while retaining decent NIR reflectance between 0.37 and 0.52. These spectral characteristics remained consistent up to incident angles of >60°, ensuring their practical applicability for vertically oriented windows. Outdoor experiments showed substantial temperature reductions of up to 8.8 °C on floors compared to uncoated glass windows. The active learning-based multilayers exhibited superior performance compared to analytical ZnS/MgF2 distributed Bragg reflectors with equivalent thicknesses by improving NIR reflectance and modulating visible transmittance. In addition, multilayers with a greater number of bits extensively tuned transmission color, enabling customization for aesthetic purposes. These findings suggest that all-dielectric multilayers can provide a scalable, cost-effective alternative for reducing energy consumption in buildings and vehicles with large glass surfaces, supporting efforts to mitigate climate change through enhanced energy efficiency.
Read full abstract