Computational design and AI optimization of a symmetry-engineered ultra-broadband metasurface solar absorber for renewable energy systems

Metamaterial solar absorbers have attracted significant attention for next-generation photothermal and energy-harvesting applications due to their ability to engineer light–matter interactions beyond conventional material limits, particularly in the near-infrared to mid-infrared (NIR–MIR) regime. In this study, a tunable ultra-broadband metamaterial absorber is proposed based on a symmetry-engineered metasurface architecture integrating refractory metals, a GaSb semiconductor substrate, and graphene-enabled plasmonic modulation. An AI-assisted optimization framework systematically explores the multidimensional design parameter space to enhance broadband absorption performance across the 200–3000 nm spectral range. The absorber’s electromagnetic response is modeled using the Finite Element Method (FEM) in COMSOL Multiphysics, with comprehensive validation under variations in incidence angle, geometry, and material properties. Ultra-broadband absorption is achieved through the synergistic interplay of multipolar resonance coupling, effective impedance matching, localized surface plasmon resonances (LSPR), propagating surface plasmon polaritons (SPPs), and Fabry–Pérot cavity interference. Simulation results demonstrate near-unity absorption (>99.9%) at incidence angles up to 30° for both transverse electric (TE) and transverse magnetic (TM) polarizations, while maintaining absorption above 95% up to 50° for TE polarization. Furthermore, machine learning models achieve high predictive accuracy (R² > 0.92) in estimating absorption performance under varying geometric and angular conditions. These findings highlight the proposed design as a scalable, tunable, and AI-optimized platform for high-efficiency solar energy harvesting and advanced photothermal applications.