Abstract
Abstract The possibility of arbitrary spatial control of incident wavefronts with the subwavelength resolution has driven research into dielectric optical metasurfaces in the last decade. The unit-cell based metasurface design approach that relies on a library of single element responses is known to result in reduced efficiency attributed to the inadequate accounting of the coupling effects between meta-atoms. Metasurfaces with extended unit-cells containing multiple resonators can improve design outcomes but their design requires extensive numerical computing and optimizations. We report a deep learning based design methodology for the inverse design of extended unit-cell metagratings. In contrast to previous reports, our approach learns the metagrating spectral response across its reflected and transmitted orders. Through systematic exploration, we discover network architectures and training dataset sampling strategies that allow such learning without requiring extensive ground-truth generation. The one-time investment of model creation can then be used to significantly accelerate numerical optimization of multiple functionalities as demonstrated by considering the inverse design of various spectral and polarization dependent splitters and filters. The proposed methodology is not limited to these proof-of-concept demonstrations and can be broadly applied to meta-atom-based nanophotonic system design and in realising the next generation of metasurface functionalities with improved performance.
Highlights
State-of-the-art nanofabrication technologies with extraordinary lateral resolution and high stitching accuracy enable the realization of wide area precision nanostructures
We propose an improved learning based nanophotonic structure discovery methodology for the constrained inverse design of extended unit-cell metagratings
A DE optimization is employed in this work for the inverse design of the metagratings where a randomly initialized set of solutions are taken and nature inspired techniques like mutation and crossover are performed to produce a new set of solutions [87, 88]
Summary
State-of-the-art nanofabrication technologies with extraordinary lateral resolution and high stitching accuracy enable the realization of wide area precision nanostructures. The numerical calculations by Gigli et al show that the efficiency of the Huygens’ elements remains below 40% for monochromatic operation, even for meta-gratings with periods significantly larger than the wavelength This performance reduction is well known in the field and is ascribed to inter-element electromagnetic coupling [20,21,22]. We propose an improved learning based nanophotonic structure discovery methodology for the constrained inverse design of extended unit-cell metagratings. S. Hegde: Learning based design of metagratings | 347 capability of models via surrogate-assisted optimization are discussed in subsection 2.2.4.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.