Deep Learning Inverse Design of Broadband Dual-Frequency Metasurfaces Using  Additive Manufacturing

Abstract

Multi-material additive manufacturing (AM) is a promising technique for creating functional electronic devices with complex geometries and properties. In this paper, we propose a novel inverse design method based on deep neural networks for conductive and dielectric multi-material printing to achieve broadband dual frequency metasurfaces that can produce two different orbital angular momentum (OAM) states, simultaneously, at the V-band and W-band. These OAM states can be used for high-speed wireless communications. We demonstrate the effectiveness of our inverse design method by designing, printing and testing multilayer metasurface using the lights-out digital manufacturing process, which can seamlessly sinter silver nanoparticles and acrylates inks. Our results show that the proposed method can achieve high efficiency in design optimisation and precision printing. This opens new avenues for the design and fabrication of multilayer metasurfaces using multi-material AM for future wireless electronic devices.

Keywords: Multi-material 3D printing, multi-metal-layer metasurfaces, ultra-thin profile, inverse design, deep learning, elements distributions, sintering process, piezoelectric printing