Logo image
All‐Dielectric SERS Metasurface with Strong Coupling Quasi‐BIC Energized by Transformer‐Based Deep Learning
Journal article   Peer reviewed

All‐Dielectric SERS Metasurface with Strong Coupling Quasi‐BIC Energized by Transformer‐Based Deep Learning

Wei Chen, Yuyang Li, Yineng Liu, Yuan Gao, Yiming Yan, Zhaogang Dong and Jinfeng Zhu
Advanced optical materials, Vol.12(4), p.n/a
01/02/2024

Abstract

bound states in the continuum deep learning field enhancement metasurfaces strong coupling
Researchers have paid great attention to pursuing metal‐free nanostructures as alternatives to metallic counterparts in the field of surface‐enhanced Raman scattering (SERS) for label‐free sensing. However, these metal‐free investigations are hindered by the tiny enhancement factor of the local near field. Here, the design of all‐dielectric SERS metasurfaces is proposed with quasi‐bound states in the continuum (Q‐BIC), inspired by transformer‐based deep learning. By manipulating the incident angle, the mechanism of strong coupling Q‐BIC is introduced with a large Rabi splitting of ≈105 meV, which opens a bandgap and forms an anti‐crossing behavior. Compared to conventional approaches, the strong coupling Q‐BIC scheme not only boosts an extraordinary SERS enhancement factor of ≈107 but also extends the field‐enhancing scale up to ten‐fold. The theoretical optimization implies overwhelming dominance versus the conventional metallic nanostructure design for SERS. The study denotes an approach to utilize the strong coupling effects of Q‐BIC in all‐dielectric SERS metasurfaces and will provide essential design guides for more powerful sensing applications based on SERS. By manipulating the incident angle, the mechanism of strong coupling quasi‐bound states in the continuum (Q‐BIC) is introduced with a large Rabi splitting of ≈105 meV, and the strong coupling Q‐BIC scheme boosts an extraordinary surface‐enhanced Raman scattering enhancement factor up to 1018.
url
https://doi.org/10.1002/adom.202301697View
Published (Version of record) Open

Metrics

1 Record Views

Details

Logo image