The advantages of metalenses over diffractive lenses

Engelberg, Jacob; Levy, Uriel

doi:10.1038/s41467-020-15972-9

Cited by 149 publications

(84 citation statements)

References 24 publications

(30 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Furthermore, while the miniaturization of D 2 NN systems with the currently available large-area nanofabrication methods is feasible to support an ensemble of diffractive networks that operate at visible wavelengths, high-throughput fabrication and integration of miniaturized optical components such as filters and lenses might be challenging due to the relative bulkiness of such optical components. However, the recently emerging research in meta-surface-based flat optics 47 , 48 has enabled significant miniaturization of traditionally bulky optical components, and this research could be further utilized for practical realizations of miniaturized D 2 NN ensembles.…”

Section: Discussionmentioning

confidence: 99%

Ensemble learning of diffractive optical networks

Rahman

Mengü

et al. 2021

Light Sci Appl

View full text Add to dashboard Cite

A plethora of research advances have emerged in the fields of optics and photonics that benefit from harnessing the power of machine learning. Specifically, there has been a revival of interest in optical computing hardware due to its potential advantages for machine learning tasks in terms of parallelization, power efficiency and computation speed. Diffractive deep neural networks (D2NNs) form such an optical computing framework that benefits from deep learning-based design of successive diffractive layers to all-optically process information as the input light diffracts through these passive layers. D2NNs have demonstrated success in various tasks, including object classification, the spectral encoding of information, optical pulse shaping and imaging. Here, we substantially improve the inference performance of diffractive optical networks using feature engineering and ensemble learning. After independently training 1252 D2NNs that were diversely engineered with a variety of passive input filters, we applied a pruning algorithm to select an optimized ensemble of D2NNs that collectively improved the image classification accuracy. Through this pruning, we numerically demonstrated that ensembles of N = 14 and N = 30 D2NNs achieve blind testing accuracies of 61.14 ± 0.23% and 62.13 ± 0.05%, respectively, on the classification of CIFAR-10 test images, providing an inference improvement of >16% compared to the average performance of the individual D2NNs within each ensemble. These results constitute the highest inference accuracies achieved to date by any diffractive optical neural network design on the same dataset and might provide a significant leap to extend the application space of diffractive optical image classification and machine vision systems.

show abstract

Section: Discussionmentioning

confidence: 99%

Ensemble learning of diffractive optical networks

Rahman

Mengü

et al. 2021

Light Sci Appl

View full text Add to dashboard Cite

show abstract

“…Generally, the wavefront of EM waves is decided both by the intensity and phase at each spatial point, while phase plays a much more important role as suggested by the famous Gerchberg-Saxton algorithm ( Gerchberg and Saxton, 1972 ). Compared with conventional diffractive elements (CDEs), in which the phase of EM waves is accumulated based on the length of the ray path inside the building materials, metasurfaces can realize phase manipulation by tailoring the local resonances and symmetry of the nanostructures ( Banerji et al., 2019 ; Engelberg and Levy, 2020 ; Huang et al., 2018 ). Once the phase accumulation process is achieved, the output light travels according to the diffractive theory for both of the CDE and metasurfaces.…”

Section: Typical Applications Of Dielectric Metasurfacesmentioning

confidence: 99%

“…Recently, researchers also developed a method to directly shape the near-field landscapes of EM waves from far-field with metasurfaces in the microwave region ( Ginis et al., 2020 ). Based on the efficient manipulation of optical fields with nanostructures, metasurfaces have also been demonstrated to be a promising platform for imaging systems ( Aieta et al., 2012 ; Engelberg and Levy, 2020 ), holography ( Huang et al., 2015 ; Wen et al., 2015 ), and quantum applications ( Bekenstein et al., 2020 ; Stav et al., 2018 ).…”

Section: Introductionmentioning

confidence: 99%

Dielectric Resonance-Based Optical Metasurfaces: From Fundamentals to Applications

Liu

Cheng

et al. 2020

iScience

View full text Add to dashboard Cite

“…With a generalization of Snell's law, [ 1 ] metasurfaces can not only substitute bulky optical components with ultra‐thin equivalents, but may also introduce new functionalities (phase discontinuities, anomalous reflection, and refraction, etc.). [ 2,3 ] However, the shape of the wavefront is defined by the metasurface design, including material selection and geometry, and is fixed after fabrication. Active post‐fabrication control requires the tunability of the optical response of each metasurface element.…”

Section: Figurementioning

confidence: 99%

The Potential of Combining Thermal Scanning Probes and Phase‐Change Materials for Tunable Metasurfaces

Michel

Meyer

Essing

et al. 2020

Advanced Optical Materials

View full text Add to dashboard Cite

ultra-thin equivalents, but may also introduce new functionalities (phase discontinuities, anomalous reflection, and refraction, etc.). [2,3] However, the shape of the wavefront is defined by the metasurface design, including material selection and geometry, and is fixed after fabrication. Active postfabrication control requires the tunability of the optical response of each metasurface element. [4,5] One common approach for active metasurfaces is to capitalize on the change of the material polarizabilityeither of the scatterer or its surroundings. This can be achieved, for example, by modulation of the charge density in doped semiconductors (e.g., GaAs) or graphene, [6] by changing the state of liquid crystals adjacent to metallic antennas, [7] or by including phase-transition (e.g., vanadium dioxide VO 2) [8-10] or phase-change [e.g., germanium antimony telluride (GeTe) x (Sb 2 Te 3) 1−x ] [11] materials in the metasurface. Mechanical tuning with stretchable substrates or actuators, [12] chemical reactions at the scatterers, [13] or enhanced optical nonlinearities [14] are other possible approaches for post-fabrication tunability. Phase-change materials (PCMs) are among the best-suited materials to provide tunability of metasurfaces due to the property contrast between their amorphous (A) and crystalline (C) phase. [15] In contrast to phase-transition materials (e.g., VO 2), PCMs feature non-volatile states and thus, no energy is needed to maintain the material properties. The structural change is accompanied with a refractive index change Δn = |n C − n A | Metasurfaces allow for the spatiotemporal variation of amplitude, phase, and polarization of optical wavefronts. Implementation of active tunability of metasurfaces promises compact flat optics capable of reconfigurable wavefront shaping. Phase-change materials (PCMs) are a prominent material class enabling reconfigurable metasurfaces due to their large refractive index change upon structural transition. However, commonly employed laser-induced switching of PCMs limits the achievable feature sizes and restricts device miniaturization. Thermal scanning-probe-induced local switching of the PCM germanium telluride is proposed to realize near-infrared metasurfaces with feature sizes far below what is achievable with diffraction-limited optical switching. The design is based on a planar multilayer and does not require fabrication of protruding resonators as commonly applied in the literature. Instead, it is numerically demonstrated that a broad-band tuning of perfect absorption can be realized by the localized tip-induced crystallization of the PCM. The spectral response of the metasurface is explained using resonance mode analysis and numerical simulations. To facilitate experimental realization, a theoretical description of the tip-induced crystallization employing multiphysics simulations is provided to demonstrate the great potential for fabricating compact reconfigurable metasurfaces. The concept can be applied not only for plasmonic sensing and spatial freq...

show abstract

The advantages of metalenses over diffractive lenses

Cited by 149 publications

References 24 publications

Ensemble learning of diffractive optical networks

Ensemble learning of diffractive optical networks

Dielectric Resonance-Based Optical Metasurfaces: From Fundamentals to Applications

The Potential of Combining Thermal Scanning Probes and Phase‐Change Materials for Tunable Metasurfaces

Contact Info

Product

Resources

About