Work

Inverse metasurface design enabling arbitrary spectral and polarization control

Year: 2025

Type: article

Abstract: Metasurface has unusual light-matter interactions and possesses a distinctive capacity to control the electromagnetic wave’s amplitude, phase, and polarization. It is challenging to come up with a well-designed metasurface that supports the desired optical response. The conventional design approach requires extensive parameter sweeps, trial-and-error in design space, and hours of computational efforts and resources. Deep learning has recently gained attention in optics and photonics and is increasingly being applied to solve design and optimization problems. In this work, a framework consisting of a conditional generative adversarial network and a residual network-based neural network simulator is presented that has been trained to predict the geometry and transmission spectra of the metasurfaces within a moment. The framework takes a set of transmitted complex electric fields. It can generate a potential candidate metasurface unit cell corresponding to the user-defined spectra. The simulator network can predict the transmission spectra of metasurfaces with complex shapes and diverse geometries. The framework has the potential to significantly accelerate the design and simulation of metasurfaces, with promising applications in spectral filtering, polarization conversion, and beam shaping in micro and nanoscale devices.

Source: Optical Materials Express

DOI: 10.1364/ome.576189

Authors: Md Imran Hossain, Linzhi Yu, Humeyra Caglayan

Institution: Tampere University

Language: English

Topic: Metamaterials and Metasurfaces Applications

Subfield: Electronic, Optical and Magnetic Materials

Field: Materials Science

Domain: Physical Sciences

Open Access status: gold

Estimated APC paid: $1,843

Funders: European Commission, Academy of Finland, Horizon 2020 Framework Programme