Multichannel meta-imagers for accelerating machine vision.

Rapid developments in machine vision technology have impacted a variety of applications, such as medical devices and autonomous driving systems. These achievements, however, typically necessitate digital neural networks with the downside of heavy computational requirements and consequent high energy consumption. As a result, real-time decision-making is hindered when computational resources are not readily accessible. Here we report a meta-imager designed to work together with a digital back end to offload computationally expensive convolution operations into high-speed, low-power optics. In this architecture, metasurfaces enable both angle and polarization multiplexing to create multiple information channels that perform positively and negatively valued convolution operations in a single shot. We use our meta-imager for object classification, achieving 98.6% accuracy in handwritten digits and 88.8% accuracy in fashion images. Owing to its compactness, high speed and low power consumption, our approach could find a wide range of applications in artificial intelligence and machine vision applications.

Digital Modeling on Large Kernel Metamaterial Neural Network.

Deep neural networks (DNNs) utilized recently are physically deployed with computational units (e.g., CPUs and GPUs). Such a design might lead to a heavy computational burden, significant latency, and intensive power consumption, which are critical limitations in applications such as the Internet of Things (IoT), edge computing, and the usage of drones. Recent advances in optical computational units (e.g., metamaterial) have shed light on energy-free and light-speed neural networks. However, the digital design of the metamaterial neural network (MNN) is fundamentally limited by its physical limitations, such as precision, noise, and bandwidth during fabrication. Moreover, the unique advantages of MNN's (e.g., light-speed computation) are not fully explored via standard 3×3 convolution kernels. In this paper, we propose a novel large kernel metamaterial neural network (LMNN) that maximizes the digital capacity of the state-of-the-art (SOTA) MNN with model re-parametrization and network compression, while also considering the optical limitation explicitly. The new digital learning scheme can maximize the learning capacity of MNN while modeling the physical restrictions of meta-optic. With the proposed LMNN, the computation cost of the convolutional front-end can be offloaded into fabricated optical hardware. The experimental results on two publicly available datasets demonstrate that the optimized hybrid design improved classification accuracy while reducing computational latency. The development of the proposed LMNN is a promising step towards the ultimate goal of energy-free and light-speed AI.

Compound Meta-Optics for Complete and Loss-Less Field Control.

Optical metasurfaces offer a compact platform for manipulation of the amplitude, phase, and polarization state of light. Independent control over these properties, however, is hindered by the symmetric transmission matrix associated with single-layer metasurfaces. Here, we utilize multilayer birefringent meta-optics to realize high-efficiency, independent control over the amplitude, phase, and polarization state of light. High-efficiency control is enabled by redistributing the wavefront between cascaded metasurfaces, while end-to-end inverse design is used to realize independent complex-valued functions for orthogonal polarization states. Based on this platform, we demonstrate spatial mode division multiplexing, optical mode conversion, and universal vectorial holograms, all with diffraction efficiencies over 80%. This meta-optic platform expands the design space of flat optics and could lead to advances in optical communications, quantum entanglement, and information encryption.

Meta-optic accelerators for object classifiers.

Rapid advances in deep learning have led to paradigm shifts in a number of fields, from medical image analysis to autonomous systems. These advances, however, have resulted in digital neural networks with large computational requirements, resulting in high energy consumption and limitations in real-time decision-making when computation resources are limited. Here, we demonstrate a meta-optic-based neural network accelerator that can off-load computationally expensive convolution operations into high-speed and low-power optics. In this architecture, metasurfaces enable both spatial multiplexing and additional information channels, such as polarization, in object classification. End-to-end design is used to co-optimize the optical and digital systems, resulting in a robust classifier that achieves 93.1% accurate classification of handwriting digits and 93.8% accuracy in classifying both the digit and its polarization state. This approach could enable compact, high-speed, and low-power image and information processing systems for a wide range of applications in machine vision and artificial intelligence.

Gold Ion Beam Milled Gold Zero-Mode Waveguides.

Zero-mode waveguides (ZMWs) are widely used in single molecule fluorescence microscopy for their enhancement of emitted light and the ability to study samples at physiological concentrations. ZMWs are typically produced using photo or electron beam lithography. We report a new method of ZMW production using focused ion beam (FIB) milling with gold ions. We demonstrate that ion-milled gold ZMWs with 200 nm apertures exhibit similar plasmon-enhanced fluorescence seen with ZMWs fabricated with traditional techniques such as electron beam lithography.

van der Waals Semiconductor Empowered Vertical Color Sensor.

Biomimetic artificial vision is receiving significant attention nowadays, particularly for the development of neuromorphic electronic devices, artificial intelligence, and microrobotics. Nevertheless, color recognition, the most critical vision function, is missed in the current research due to the difficulty of downscaling of the prevailing color sensing devices. Conventional color sensors typically adopt a lateral color sensing channel layout and consume a large amount of physical space, whereas compact designs suffer from an unsatisfactory color detection accuracy. In this work, we report a van der Waals semiconductor-empowered vertical color sensing structure with the emphasis on compact device profile and precise color recognition capability. More attractive, we endow color sensor hardware with the function of chromatic aberration correction, which can simplify the design of an optical lens system and, in turn, further downscales the artificial vision systems. Also, the dimension of a multiple pixel prototype device in our study confirms the scalability and practical potentials of our developed device architecture toward the above applications.

Collective Phonon-Polaritonic Modes in Silicon Carbide Subarrays.

Localized surface phonon polaritons (LSPhPs) can be implemented to engineer light-matter interactions through nanoscale patterning for a range of midinfrared application spaces. However, the polar material systems studied to date have mainly focused on simple designs featuring a single element in the periodic unit cell. Increasing the complexity of the unit cell can serve to modify the resonant near-fields and intra- and inter-unit-cell coupling as well as to dictate spectral tuning in the far-field. In this work, we exploit more complicated unit-cell structures to realize LSPhP modes with additional degrees of design freedom, which are largely unexplored. Collectively excited LSPhP modes with distinctly symmetric and antisymmetric near-fields are supported in these subarray designs, which are based on nanopillars that are scaled by the number of subarray elements to ensure a constant unit-cell size. Moreover, we observe an anomalous mode-matching of the collective symmetric mode in our fabricated subarrays that is robust to changing numbers of pillars within the subarrays as well as to defects intentionally introduced in the form of missing pillars. This work therefore illustrates the hierarchical design of tailored LSPhP resonances and modal near-field profiles simultaneously for a variety of IR applications such as surface-enhanced spectroscopies and biochemical sensing.

Electrothermoplasmonic Trapping and Dynamic Manipulation of Single Colloidal Nanodiamond.

Low-power trapping of nanoscale objects can be achieved by using the enhanced fields near plasmonic nanoantennas. Unfortunately, in this approach the trap site is limited to the position of the plasmonic hotspots and continuous dynamic manipulation is not feasible. Here, we report a low-frequency electrothermoplasmonic tweezer (LFET) that provides low-power, high-stability and continuous dynamic manipulation of a single nanodiamond. LFET harnesses the combined action of the laser illumination of a plasmonic nanopillar antenna array and low-frequency alternating current (ac) electric field to establish an electrohydrodynamic potential capable of the stable trapping and dynamic manipulation of single nanodiamonds. We experimentally demonstrate the fast transport, trapping, and dynamic manipulation of a single nanodiamond using a low-frequency ac field below 5 kHz and low-laser power of 1 mW. This nanotweezer platform for nanodiamond manipulation holds promise for the scalable assembly of single photon sources for quantum information processing and low noise quantum sensors.

Piezoelectric Actuation of Graphene-Coated Polar Structures.

Ferroelectric materials based on lead zirconate titanate (PZT) are widely used as sensors and actuators because of their strong piezoelectric activity. However, their application is limited because of the high processing temperature, brittleness, lack of conformal deposition, and a limited possibility to be integrated with the microelectromechanical systems (MEMS). Recent studies on the piezoelectricity in the 2-D materials have demonstrated their potential in these applications, essentially due to their flexibility and integrability with the MEMS. In this work, we deposited a few layer graphene (FLG) on the amorphous oxidized Si3N4 membranes and studied their piezoelectric response by sensitive laser interferometry and rigorous finite-element modeling (FEM) analysis. Modal analysis by FEM and comparison with the experimental results show that the driving force for the piezoelectric-like response can be a polar interface layer formed between the residual oxygen in Si3N4 and the FLG. The response was about 14 nm/V at resonance and could be further enhanced by adjusting the geometry of the device. These phenomena are fully consistent with the earlier piezoresponse force microscopy (PFM) observations of the piezoelectricity of the graphene on SiO2 and open up an avenue for using graphene-coated structures in the MEMS.

To switch or not to switch - a machine learning approach for ferroelectricity.

With the advent of increasingly elaborate experimental techniques in physics, chemistry and materials sciences, measured data are becoming bigger and more complex. The observables are typically a function of several stimuli resulting in multidimensional data sets spanning a range of experimental parameters. As an example, a common approach to study ferroelectric switching is to observe effects of applied electric field, but switching can also be enacted by pressure and is influenced by strain fields, material composition, temperature, time, etc. Moreover, the parameters are usually interdependent, so that their decoupling toward univariate measurements or analysis may not be straightforward. On the other hand, both explicit and hidden parameters provide an opportunity to gain deeper insight into the measured properties, provided there exists a well-defined path to capture and analyze such data. Here, we introduce a new, two-dimensional approach to represent hysteretic response of a material system to applied electric field. Utilizing ferroelectric polarization as a model hysteretic property, we demonstrate how explicit consideration of electromechanical response to two rather than one control voltages enables significantly more transparent and robust interpretation of observed hysteresis, such as differentiating between charge trapping and ferroelectricity. Furthermore, we demonstrate how the new data representation readily fits into a variety of machine-learning methodologies, from unsupervised classification of the origins of hysteretic response via linear clustering algorithms to neural-network-based inference of the sample temperature based on the specific morphology of hysteresis.

Large amplitude charge noise and random telegraph fluctuations in room-temperature graphene single-electron transistors.

We analyze the noise in liquid-gated, room temperature, graphene quantum dots. These devices display extremely large noise amplitudes. The observed noise is explained in terms of a charge noise model by considering fluctuations in the applied source-drain and gate potentials. We show that the liquid environment and substrate have little effect on the observed noise and as such attribute the noise to charge trapping/detrapping at the disordered graphene edges. The trapping/detrapping of individual charges can be tuned by gating the device, which can result in stable two-level fluctuations in the measured current. These results have important implications for the use of electronic graphene nanodevices in single-molecule biosensing.

Multifunctional metaoptics based on bilayer metasurfaces.

Optical metasurfaces have become versatile platforms for manipulating the phase, amplitude, and polarization of light. A platform for achieving independent control over each of these properties, however, remains elusive due to the limited engineering space available when using a single-layer metasurface. For instance, multiwavelength metasurfaces suffer from performance limitations due to space filling constraints, while control over phase and amplitude can be achieved, but only for a single polarization. Here, we explore bilayer dielectric metasurfaces to expand the design space for metaoptics. The ability to independently control the geometry and function of each layer enables the development of multifunctional metaoptics in which two or more optical properties are independently designed. As a proof of concept, we demonstrate multiwavelength holograms, multiwavelength waveplates, and polarization-insensitive 3D holograms based on phase and amplitude masks. The proposed architecture opens a new avenue for designing complex flat optics with a wide variety of functionalities.

Single-mode porous silicon waveguide interferometers with unity confinement factors for ultra-sensitive surface adlayer sensing.

Guided wave-optics has emerged as a promising platform for label free biosensing. However, device sensitivity toward surface-bound small molecules is directly limited by the evanescent interaction and low confinement factor with the active sensing region. Here, we report a mesoporous silicon waveguide design and inverse fabrication technique that resolves the evanescent field interaction limitation while achieving maximal transverse confinement factors and preserving single-mode operation. The waveguide sensors are characterized in a Fabry-Perot interferometer configuration and the ultra-high sensitivity to small molecule adlayers is demonstrated. We also identify dispersion to be a promising degree of freedom for exceeding the sensitivity limits predicted by the conventional non-dispersive effective medium theory.

Photonic crystal nanobeam biosensors based on porous silicon.

Photonic crystal (PhC) nanobeams (NB) patterned on porous silicon (PSi) waveguide substrates are demonstrated for the specific, label-free detection of oligonucleotides. These photonic structures combine the large active sensing area intrinsic to PSi sensors with the high-quality (Q) factor and low-mode volume characteristic of compact resonant silicon-on-insulator (SOI) PhC NB devices. The PSi PhC NB can achieve a Q-factor near 9,000 and has an approximately 40-fold increased active sensing area for molecular attachment, compared to traditional SOI PhC NB sensors. The PSi PhC NB exhibits a resonance shift that is more than one order of magnitude larger than that of a similarly designed SOI PhC NB for the detection of small chemical molecules and 16-base peptide nucleic acids. The design and fabrication of PSi PhC NB sensors are compatible with CMOS processing, sensor arrays, and integration with lab-on-chip systems.

Environmental Gating and Galvanic Effects in Single Crystals of Organic-Inorganic Halide Perovskites.

Understanding the impact of environmental gaseous on the surface of organometal halide perovskites (OMHPs) couples to the electronic and ionic transport is critically important. Here, we explore the transport behavior and origins of the gas sensitivity in MAPbBr3 single crystals (SCs) devices using impedance spectroscopy and current relaxation measurements. Strong resistive response occurs when crystals are exposed to different environments. It was shown that SC response to the environment is extremely different at the surface as compared to the bulk due to the disorder surface chemistry. The nonlinear transport properties studied using ultrafast Kelvin probe force microscopy (G-KPFM) to unravel spatio-temporal charge dynamics at SC/electrode interface. The relaxation processes observed in pulse relaxation and G-KPFM measurements along with gas sensitivity of crystals suggest the presence of a triple-phase boundary between environment, electrode, and crystal. Results indicate that the environment is a nontrivial component in the operation of OMHP devices which is reminiscent of fuel cell systems. Furthermore, the triple-phase boundary can play a significant role in the transport properties of OMHPs due to the possibility of the redox processes coupled to the concentration of bulk ionic species. Although instrumental for understanding the device characteristics of perovskites, our studies suggest a new opportunity of coupling the redox chemistry of the Br2-Br- pair that defines the bulk ionic conductivity of MAPbBr3 with the redox chemistry of gaseous (or liquid) environment via a suitable electrocatalytic system to enable new class of energy storage devices and gas sensors.

Surface-Enhanced Raman Scattering (SERS) Studies of Disc-on-Pillar (DOP) Arrays: Contrasting Enhancement Factor with Analytical Performance.

The use of nanomachining methods capable of reproducible construction of nano-arrayed devices have revolutionized the field of plasmonic sensing by the introduction of a diversity of rationally engineered designs. Significant strides have been made to fabricate plasmonic platforms with tailored interparticle gaps to improve their performance for surface-enhanced Raman scattering (SERS) applications. Over time, a dichotomy has emerged in the implementation of SERS for analytical applications, the construction of substrates, optimization of interparticle spacing as a means to optimize electromagnetic field enhancement at the localized surface plasmon level, and the substrate sensitivity over extended areas to achieve quantitative performance. This work assessed the enhancement factor of plasmonic Ag/SiO2/Si disc-on-pillar (DOP) arrays of variable pitch with its analytical performance for quantitative applications. Experimental data were compared with those from finite-difference time-domain (FDTD) simulations used in the optimization of the array dimensions. A self-assembled monolayer (SAM) of benzenethiol rendered highly reproducible signals (RSD ∼4-10%) and SERS substrate enhancement factor (SSEF) values in the orders of 106-108 for all pitches. Spectra corresponding to rhodamine 6G (R6G) and 4-aminobenzoic acid demonstrated the advantages of using the more densely packed DOP arrays with a 160 nm pitch (gap = 40 nm) for quantitation in spite of the strongest SSEF was attained for a pitch of 520 nm corresponding to a 400 nm gap.

Nonlinear light generation in topological nanostructures.

Topological photonics has emerged as a route to robust optical circuitry protected against disorder1,2 and now includes demonstrations such as topologically protected lasing3-5 and single-photon transport6. Recently, nonlinear optical topological structures have attracted special theoretical interest7-11, as they enable tuning of topological properties by a change in the light intensity7,12 and can break optical reciprocity13-15 to realize full topological protection. However, so far, non-reciprocal topological states have only been realized using magneto-optical materials and macroscopic set-ups with external magnets4,16, which is not feasible for nanoscale integration. Here we report the observation of a third-harmonic signal from a topologically non-trivial zigzag array of dielectric nanoparticles and the demonstration of strong enhancement of the nonlinear photon generation at the edge states of the array. The signal enhancement is due to the interaction between the Mie resonances of silicon nanoparticles and the topological localization of the electric field at the edges. The system is also robust against various perturbations and structural defects. Moreover, we show that the interplay between topology, bi-anisotropy and nonlinearity makes parametric photon generation tunable and non-reciprocal. Our study brings nonlinear topological photonics concepts to the realm of nanoscience.

Multilayer Noninteracting Dielectric Metasurfaces for Multiwavelength Metaoptics.

Metasurfaces provide a versatile platform for manipulating the wavefront of light using planar nanostructured surfaces. Transmissive metasurfaces, with full 2π phase control, are a particularly attractive platform for replacing conventional optical elements due to their small footprint and broad functionality. However, the operational bandwidth of metasurfaces has been a critical limitation and is directly connected to either their resonant response or the diffractive dispersion of their lattice. While multiwavelength and continuous band operation have been demonstrated, the elements suffer from either low efficiency, reduced imaging quality, or limited element size. Here, we propose a platform that provides for multiwavelength operation by employing tightly spaced multilayer dielectric metasurfaces. As a proof of concept, we demonstrate a multiwavelength metalens doublet (NA = 0.42) with focusing efficiencies of 38% and 52% at wavelengths of 1180 and 1680 nm, respectively. We further show how this approach can be extended to three-wavelength metalenses as well as a spectral splitter. This approach could find applications in fluorescent microscopy, digital imaging, and color routing.

Quantum metasurface for multiphoton interference and state reconstruction.

Metasurfaces based on resonant nanophotonic structures have enabled innovative types of flat-optics devices that often outperform the capabilities of bulk components, yet these advances remain largely unexplored for quantum applications. We show that nonclassical multiphoton interferences can be achieved at the subwavelength scale in all-dielectric metasurfaces. We simultaneously image multiple projections of quantum states with a single metasurface, enabling a robust reconstruction of amplitude, phase, coherence, and entanglement of multiphoton polarization-encoded states. One- and two-photon states are reconstructed through nonlocal photon correlation measurements with polarization-insensitive click detectors positioned after the metasurface, and the scalability to higher photon numbers is established theoretically. Our work illustrates the feasibility of ultrathin quantum metadevices for the manipulation and measurement of multiphoton quantum states, with applications in free-space quantum imaging and communications.

Nonlinear Wavefront Control with All-Dielectric Metasurfaces.

Metasurfaces, two-dimensional lattices of nanoscale resonators, offer unique opportunities for functional flat optics and allow the control of the transmission, reflection, and polarization of a wavefront of light. Recently, all-dielectric metasurfaces reached remarkable efficiencies, often matching or out-performing conventional optical elements. The exploitation of the nonlinear optical response of metasurfaces offers a paradigm shift in nonlinear optics, and dielectric nonlinear metasurfaces are expected to enrich subwavelength photonics by enhancing substantially nonlinear response of natural materials combined with the efficient control of the phase of nonlinear waves. Here, we suggest a novel and rather general approach for engineering the wavefront of parametric waves of arbitrary complexity generated by a nonlinear metasurface. We design all-dielectric nonlinear metasurfaces, achieve a highly efficient wavefront control of a third-harmonic field, and demonstrate the generation of nonlinear beams at a designed angle and the generation of nonlinear focusing vortex beams. Our nonlinear metasurfaces produce phase gradients over a full 0-2π phase range with a 92% diffraction efficiency.