Bayesian Optimized High-Figure-of-Merit Broadband Directional Thermal Emitters.

Broadband directional thermal emitters have attracted significant attention due to their potential applications in infrared camouflage and radiative cooling. However, existing broadband directional thermal emission (BDTE) multilayer structures rely heavily on the Berreman modes of epsilon-near-zero (ENZ) materials, usually requiring a substantial number of stacked ENZ thin films for broader spectral coverage. Moreover, the lack of optimized thicknesses fails to achieve the optimal figure of merit (FOM) of BDTE. Here, we have realized a high-FOM BDTE structure with a reduced number of ENZ layers based on Bayesian optimization. By coupling epsilon-near-pole (ENP) resonance with the Brewster effect of the dielectric spacer, we extend the BDTE bandwidth by 2 μm (from 7.9-12 to 7.9-14 μm). The optimized structure shows unprecedented performance, achieving an average directional emissivity of 0.94 and an FOM of 8.087, which are also validated by experimental measurements. Notably, by integrating our emitter with low-emissivity covers, we develop a series of patterned devices for infrared information encryption and deception applications, which exhibit angle-dependent distinct, even contradictory, infrared information. This work not only provides theoretical guidance for the design and optimization of BDTE structures but also paves the way for their applications in infrared information technologies.