Article Dans Une Revue Scientific Reports Année : 2024

Back-propagation optimization and multi-valued artificial neural networks for highly vivid structural color filter metasurfaces

Résumé

We introduce a novel technique for designing color filter metasurfaces using a data-driven approach based on deep learning. Our innovative approach employs inverse design principles to identify highly efficient designs that outperform all the configurations in the dataset, which consists of 585 distinct geometries solely. By combining Multi-Valued Artificial Neural Networks and back-propagation optimization, we overcome the limitations of previous approaches, such as poor performance due to extrapolation and undesired local minima. Consequently, we successfully create reliable and highly efficient configurations for metasurface color filters capable of producing exceptionally vivid colors that go beyond the sRGB gamut. Furthermore, our deep learning technique can be extended to design various pixellated metasurface configurations with different functionalities.

Optical color filters are structures or materials designed to discriminate and manipulate distinct light wavelengths through the selective transmission or reflection of particular colors while simultaneously absorbing or attenuating undesired colors 1,2 . Conventional color filters rely on the manipulation of chemical composition to achieve the desired optical properties, which can lead to issues such as absorption losses, thermal effects, and alterations in chemical characteristics 3 . An alternative approach involves the utilisation of structural color filters, offering distinct advantages and applications in diverse fields such as photorealistic color printing, color holography, anti-counterfeiting devices, and much more 4-6 .

Metasurfaces have emerged as a promising platform for structural color filters 7,8 , owing to its peculiar capability of controlling all the light properties at the nanoscale, enabling a plethora of applications 9-12 . Dielectric metasurfaces play a crucial role in color filter applications, especially within the visible spectrum range where the plasmonic conterpart based on metals is less performing owing to intrinsic optical losses. The limited losses of dielectrics (e.g. Si 3 N 4 , GaN, TiO 2 , ZrO 2 , HfO 2 ) make them highly desirable for designing efficient devices with sharp resonance responses 13-16 . Resonant dielectric metasurfaces achieve precise control over the phase of reflected and transmitted light by leveraging various resonant phenomena (e.g. Mie resonances) 17,18 . Through meticulous engineering of the resonators, selective interaction with different wavelengths is enabled, leading to efficient and vivid color filters. Such kind of metasurfaces offer exceptional phase control, high-quality factors, and sharp resonances, resulting in enhanced color purity and spectral selectivity 19-21 . Yet, the design of an ideal color filter demands capability to selectively filter all colors across the optical spectrum. In other words, at each desired wavelength, it is crucial to eliminate any background resonances in order to achieve a pure color response characterized by sharp reflection or transmission amplitudes. Given the fabrication constraints, finding the appropriate resonator shape to achieve a desired response, is a challenging task that has garnered significant attention in the research community. Numerous studies explored this area, employing sophisticated optimization algorithms including advanced Deep Learning (DL) approaches to tackle the inherent complexity of the problem 20-29 . However, relying on classical optimization approaches requires several costly simulations when optimizing various color targets simultaneously 30-32 . A viable solution for the design of vivid metasurface color

Fichier principal
Vignette du fichier
s41598-023-48064-x.pdf (4.22 Mo) Télécharger le fichier
Origine Fichiers produits par l'(les) auteur(s)
Licence
HAL

Est une version de hal-04169093 Article Arthur Clini de Souza, Stéphane Lanteri, Hugo Enirique Hernández-Figueroa, Marco Abbarchi, David Grosso, et al.. Back-propagation optimization and multi-valued artificial neural networks for highly vivid structural color filter metasurfaces. Scientific Reports, 2023, 13 (1), pp.21352. ⟨10.1038/s41598-023-48064-x⟩. ⟨hal-04169093⟩

Correction to: Scientific Reports https://doi.org/10.1038/s41598-023-48064-x, published online 04 December 2023 The original version of this Article contained an error in the spelling of the author Hugo Enrique Hernández-Figueroa, which was incorrectly given as Hugo Enirique Hernández-Figueroa. The original Article has been corrected.

Dates et versions

hal-05005240 , version 1 (23-07-2023)
hal-05005240 , version 2 (25-03-2025)

Licence

Identifiants

Citer

Arthur Clini de Souza, Stéphane Lanteri, Hugo Enrique Hernández-Figueroa, Marco Abbarchi, David Grosso, et al.. Back-propagation optimization and multi-valued artificial neural networks for highly vivid structural color filter metasurfaces. Scientific Reports, 2024, 14 (1), pp.2018. ⟨10.1038/s41598-024-52554-x⟩. ⟨hal-05005240v2⟩
1113 Consultations
377 Téléchargements

Altmetric

Partager

  • More