Work

Reinforcement learning enabled high-efficiency DUV aluminum nitride metagrating

Year: 2025

Type: article

Abstract: Deep ultraviolet (DUV) light is essential for applications ranging from advanced spectroscopy to precision nanofabrication, driving demand for compact and efficient DUV optical elements. Here, we demonstrate high-efficiency aluminum nitride (AlN) metasurface gratings (metagratings) operating in the DUV regime, designed using reinforcement learning (RL) integrated with rigorous coupled-wave analysis (RCWA). This approach enables faster and more effective optimization compared to conventional forward design methods. The resulting metagrating achieves a deflection angle of 80° with a measured efficiency of 75 %, representing state-of-the-art performance in this spectral range. Structural analysis further provides insights into the key design parameters responsible for the observed performance, offering valuable guidance for accelerating future design iterations and improving device scalability. This work showcases the potential of AI-assisted design in advancing next-generation DUV photonic systems.

Source: Applied Surface Science Advances

DOI: 10.1016/j.apsadv.2025.100883

Authors: Po-Yen Yu, Yu Jie Wang, Sung-Lin Tsai, Hong-Han Shuai, Ming Lun Tseng

Institution: National Yang Ming Chiao Tung University

Language: English

Topic: Metamaterials and Metasurfaces Applications

Subfield: Electronic, Optical and Magnetic Materials

Field: Materials Science

Domain: Physical Sciences

Open Access status: gold

Estimated APC paid: $2,350

Funders: National Science and Technology Council, Ministry of Education, Industrial Technology Research Institute, National Yang Ming Chiao Tung University