Fig. 1: Machine learning-assisted meta-absorber for enhanced broadband absorption.

a Schematic of the structure of the proposed meta-absorber consisting of four concentric silver nanorings embedded in an a-Si active layer (thickness 20 nm) on the top, a dielectric spacer SiO₂ layer in the middle, and a silver mirror acts as a back reflector at the bottom. The structure is illuminated by incoming light covering the spectrally dense region of the solar spectrum S(λ), as shown in the lower panel. The inset on the top right depicts the front view of the silver nanorings pattern. b The proposed RPN can predict the absorption spectra from the parameters of the meta-absorber. In turn, the inverse DPN can predict the design parameters of the meta-absorber for a given absorption spectra. Blue and pink circles represent design and response spaces, respectively. c Broadband absorption spectrum of the 20 nm silicon layer of the meta-absorber