Efficient design method for terahertz broadband metasurface patterns via deep learning

• Published in: Optics and laser technology Vol. 160; p. 109058
• Main Authors: Teng, Yan, Li, Chun, Li, Shaochen, Xiao, Yuhua, Jiang, Ling
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01.05.2023
• Subjects: Deep learning; Inverse design; Metasurface; Terahertz; Deep learning; Metasurface; Terahertz; Inverse design
• ISSN: 0030-3992; 1879-2545
• DOI: 10.1016/j.optlastec.2022.109058

•The fast design method proposed in this paper is divided into the forward prediction of electromagnetic response and the inverse design of metasurface patterns.•The forward prediction model can obtain the corresponding amplitude and phase response of the metasurface within 3 ms.•The inverse design model can obtain the metasurface structure based on the target response within 10 min.•Multi-objective optimization combined with genetic algorithms is used to improve the design efficiency even more.•The method proposed in this paper allows easy design of high-Q value, multi-resonant, and ultra wideband THz-enabled devices. The traditional design method of metasurfaces is the trial-and-error process with full-wave electromagnetic simulation. Recently, as an effective method, deep learning has been widely used in a variety of fields to solve complex problems. Here, forward prediction and inverse design methods for terahertz (THz) random metasurfaces are proposed based on deep Convolutional Neural Networks (CNN) and Genetic Algorithms (GA). According to the metasurface pattern, the forward prediction model accurately obtains the reflection amplitude and phase response. Compared with the Full-wave solver, the calculation speed is increased by 40,000 times. Furthermore, the THz random structure can be accurately and quickly derived from the target response, operating in a broadband range of 0.2–2 THz. Then, we discuss the advantages of single-objective and multi-objective optimization in the inverse design of metasurface patterns. By combining with the GA, the design efficiency is greatly improved. This can serve as an efficient method for global optimization in complex designs. Finally, we obtain two meta-atoms used to encode metasurface in only 10 min and built a three-beam splitter. The model proposed in this paper provides a new approach to metasurface design at THz frequencies.