Environment-adaptive design of multilayer optical structures for surface polariton excitation via dynamic multi-objective optimization

The graphene–dielectric multilayer architecture constitutes a fundamental and widely utilized platform for sustaining surface polariton (SP) propagation. Owing to their extraordinary prospects in defence critical technologies, including radar-absorbing stealth coatings, high-power microwave shielding, and ultrafast optical switching, SPs have attracted intense and sustained interest. In this study, we develop an environment-adaptive design framework that models wavelength variation as a dynamic environmental change and automatically adjusts the design parameters in response. Our method employs a dynamic multi-objective optimization algorithm augmented with a predictive transfer strategy, optimizing SP coupling efficiency, structural compactness, and fabrication feasibility. Using a population history prediction mechanism, the framework not only adaptively generates multilayer designs across the full visible spectrum without full re-initialization, but also retains and exploits knowledge of how environmental variations influence the distribution of optimal solutions. This enables rapid adjustment of the optimization direction when parameters such as wavelength, angle, or doping change, thus avoiding the need to restart the search from scratch. Comprehensive comparisons demonstrate outstanding robustness under continuous wavelength shifts. The optimized graphene-coated distributed Bragg reflector (DBR) stacks achieve near-perfect absorption (> 98%) at each individual wavelength across the visible spectrum. This work not only provides theoretical guidance for SP excitation experiments, but also contributes to the optimization of polariton device design, which is crucial for enhancing the performance of defence-related optical systems.