Multifunctional Metasurface Design with a Generative Adversarial Network

Sensong An,Bowen Zheng,Hong Tang,Mikhail Y. Shalaginov,Li Zhou,Hang Li,Myungkoo Kang,Kathleen A. Richardson,Tian Gu,Juejun Hu,Clayton Fowler,Hualiang Zhang

DOI: https://doi.org/10.1002/adom.202001433

IF: 9

2021-01-20

Advanced Optical Materials

Abstract:<p>Metasurfaces have enabled precise electromagnetic (EM) wave manipulation with strong potential to obtain unprecedented functionalities and multifunctional behavior in flat optical devices. These advantages in precision and functionality come at the cost of tremendous difficulty in finding individual meta‐atom structures based on specific requirements (commonly formulated in terms of EM responses), which makes the design of multifunctional metasurfaces a key challenge in this field. In this paper, a generative adversarial network that can tackle this problem and generate meta‐atom/metasurface designs to meet multifunctional design goals is presented. Unlike conventional trial‐and‐error or iterative optimization design methods, this new methodology produces on‐demand free‐form structures involving only a single design iteration. More importantly, the network structure and the robust training process are independent of the complexity of design objectives, making this approach ideal for multifunctional device design. Additionally, the ability of the network to generate distinct classes of structures with similar EM responses but different physical features can provide added latitude to accommodate other considerations such as fabrication constraints and tolerances. The network's ability to produce a variety of multifunctional metasurface designs is demonstrated by presenting a bifocal metalens, a polarization‐multiplexed beam deflector, a polarization‐multiplexed metalens, and a polarization‐independent metalens.</p>

materials science, multidisciplinary,optics

What problem does this paper attempt to address?

The paper aims to address the challenges in the design of multifunctional metasurfaces. Specifically, traditional design methods (such as empirical reasoning or trial-and-error) are inefficient and ineffective when faced with complex design requirements, as these methods require a large number of numerical full-wave simulations, which are not only time-consuming but also labor-intensive. Additionally, the design of multifunctional metasurfaces (e.g., multi-wavelength, multi-polarization, or reconfigurable metasurfaces based on phase-change materials) faces the challenge of exploring a vast parameter space with sufficient complexity to meet stringent design requirements.

To address these issues, the paper proposes a novel method based on Generative Adversarial Networks (GAN) for designing free-form all-dielectric metasurface devices. Unlike traditional iterative optimization design methods, this approach can generate the required structures in a single design iteration, and its network architecture and robust training process are not affected by the complexity of the design objectives. Furthermore, this method can generate multiple categories of structures with similar electromagnetic responses but different physical characteristics, providing more flexibility for manufacturing constraints and tolerances.

The proposed method's multifunctionality and scalability are validated by demonstrating examples such as bifocal metalenses, polarization-multiplexed beam deflectors, polarization-multiplexed metalenses, and polarization-independent metalenses. This indicates that the method has significant advantages in the design of multifunctional optical devices, capable of efficiently generating high-quality design solutions.