High‐Performance Huygens' Metasurface for Multifunctional Near‐Field Modulation via Flexible and Efficient Pixelated Pattern Design

Abstract The wavefront manipulation capabilities of metasurfaces critically depend on the electromagnetic (EM) performance of meta‐atoms. Previous design optimization methods typically rely on large‐scale parameter sweeping within confined design spaces or data‐driven approaches, leading to limited EM performance and substantial training overhead. However, the integration of data‐driven design and parameter sweeping, combining expansive design spaces with embedded structural knowledge, remains largely unexplored. To bridge this gap, a new paradigm termed Pattern‐based Iterative Path Generation (PIPG) is proposed. This method iteratively introduces pixel‐level modifications to classical unit structures that have already exhibited high performance, thereby enriching the design space with additional, yet physically meaningful, degrees of freedom. Compared to traditional approaches, the proposed algorithm achieves reducing the optimization time by over 90% and extending the phase shift range of the arc‐like Huygens' metasurface by 13 and that of the I‐like Huygens' metasurface by 36. In addition, the effectiveness of the PIPG algorithm is further demonstrated through the multifunctional metasurface composed of the designed meta‐atom, for orbital angular momentum vortex beam generation and beam focusing, enabling dynamic control of focal vortex beams with three different topological charges. Benefiting from its strong generalizability and flexibility, PIPG can be extended for rapid meta‐atom design with arbitrary initial geometries and across a wide range of frequency bands, including terahertz regimes. These advantages pave the way for broader deployment of metasurfaces in advanced imaging, intelligent sensing, wireless communication, and beyond.