July 2022
Spotlight Summary by Emmanuel Centeno and Antoine Moreau
Deep-learning-assisted designing chiral terahertz metamaterials with asymmetric transmission properties
Despite their sub-wavelength thickness, metasurfaces are powerful devices that allow manipulating electromagnetic waves with an almost complete level of control, explaining why such structures are currently envisioned for so many applications. This is particularly true for the development of terahertz technologies, a domain where efficient and compact components are much needed. However, the traditional approach for the numerical design of devices with a precise response can turn out to be quite cumbersome, since solving Maxwell's equations for such complex structures is likely to consume a lot of time and computational power. This drawback and the lack of design rules for such devices are driving a growing interest in artificial intelligence as a systematic method to conceive efficient photonic components. A lot of effort is currently devoted to the conception of deep learning architectures able to produce designs with optical properties as close as possible to the desired ones. Fen Gao and coauthors propose here an advanced artificial neural network conceived to accelerate the design of terahertz metasurfaces presenting asymmetric transmission properties. This work is a step towards the fast and reliable automated design of metasurfaces and can be considered as an inspiration for the design of even more complex photonic components.
You must log in to add comments.
Add Comment
You must log in to add comments.
Article Information
Deep-learning-assisted designing chiral terahertz metamaterials with asymmetric transmission properties
Feng Gao, Zhen Zhang, Yafei Xu, Liuyang Zhang, Ruqiang Yan, and Xuefeng Chen
J. Opt. Soc. Am. B 39(6) 1511-1519 (2022) View: Abstract | HTML | PDF