Lanthanide-based upconversion nanoparticles (UCNPs) have attracted considerable attention in biomedical applications, due to their anti-Stokes shifted emission enabling autofluorescence-free signal detection. However, residual excitation light can still interfere with their relatively weak emission signals. While commonly used lock-in detection can distinguish weak signals from substantial random background, concurrently modulated residual excitation light is not eliminated. This remains a challenge, particularly under demanding experimental conditions. Here, we propose a photophysical lock-in detection (PP-LID) approach based on the discovery that UCNPs can act as frequency mixers in response to intensity-modulated excitation. Particularly, excitation modulated at multiple base frequencies can generate additional spectral components at the beat frequencies (BFs) between the base modulation frequencies. These signals are resolvable by frame-rate-limited cameras, devoid of ambient and residual excitation light, and can be adapted through nanoparticle engineering. Extracting BF signals by PP-LID thus provides a strategy to significantly enhance signal-to-background conditions in UCNP-based bioimaging and biosensing. Upon dual-frequency excitation upconversion nanoparticles generate beat-frequency signals owing to their inherent optical nonlinearity. These low-frequency signals are absent in any excitation and background light, enabling background-free camera-based detection.
Chip-scale soliton microcombs, particularly those operating at low, electronically detectable repetition rates (≤ 26.5 GHz), are highly promising for portable metrology. However, their practical deployment has been critically hindered by poor robustness against intracavity noise and environmental perturbations. Here, we overcome this limitation by proposing a multimodal locking architecture that actively and simultaneously stabilizes all three fundamental parameters in a soliton microcomb: the pump frequency, the cavity resonance, and the repetition rate. This architecture is confirmed both theoretically and experimentally. Implemented on a Si3N4 microresonator with an FSR of 24.96 GHz, this approach enables robust soliton generation and sustains record-long, collapse-free operation for over 48 h. More importantly, it maintains soliton robustness with exceptional resilience to environmental shocks, under temperature variations exceeding 10°C and vibration accelerations beyond ±4 g. The microcomb’s metrological utility as a precise optical ruler is further validated through optical frequency calibration and frequency-sweeping-based absolute ranging demonstration. This work provides a critical solution for robust and field-deployable low-repetition-rate soliton microcombs, paving the way for their use in portable optical clocks, high-precision ranging, time-frequency transfer and spectroscopic sensing. Implemented on a Si₃N₄ microresonator, a multimodal locking architecture simultaneously stabilizes pump frequency, cavity resonance, and repetition rate, enabling robust and day-scale low-repetition-rate soliton microcombs for precision metrology.
Abstract High-resolution three-dimensional imaging is essential for visualizing fine biological details. Unlike conventional optical imaging, photoacoustic (PA) imaging provides volumetric imaging by capturing time-of-flight ultrasound waves generated from optical absorption within tissues. Optical-resolution photoacoustic microscopy achieves micron-scale lateral resolution through tight optical focusing but suffers from a shallow depth of focus. To address this, non-diffracting beams (e.g., Bessel and needle beams) have been explored, but they introduce high side lobes that reduce the signal-to-noise ratio and distribute laser energy over a larger volume, reducing signal intensity. Here, we present an electrically modulated dual-focus metalens-based photoacoustic microscopy (meta-PAM) system designed to enhance axial coverage (1.2 mm) while preserving high lateral resolution (3.7 μm) in three-dimensional imaging. Our system features a silicon nitride nanostructured metalens that switches between two diffraction-limited focal modes at a visible wavelength without mechanical movement, with focal switching voltages of 0.87 V and 1.03 V and a response time of 250 ms. We demonstrate its effectiveness by imaging a chemically induced corneal burn in rat eyes, successfully monitoring neovascularization with high resolution across the entire eye depth range. Our compact yet effective system provides building blocks for future high-resolution volumetric imaging with an improved signal-to-noise ratio.
Scaling up the area of photonic-crystal surface-emitting lasers (PCSELs) to produce higher output power while maintaining single-mode operation has been a longstanding objective in photonics and laser physics. Achieving this goal requires two fundamental physics: the difference between the modal losses of the fundamental and higher-order modes, and the mode competition between them. Both physics contribute to the lasing threshold difference between the fundamental and higher-order modes and have been considered individually in past works. But there has not been a theoretical study that treats both physics together for the design of PCSELs. Here we provide a theoretical framework to elucidate the interplay between differential modal losses and mode competition in PCSELs. Based on this framework, we introduce designs based on spatial engineering of photonic-crystal slabs to simultaneously control the modal properties and mode competition. Compared to designs that focus on only one of these physics, our designs can increase the difference between the lasing thresholds of the fundamental and higher-order modes by orders of magnitude. Our findings provide new insights into realizing scalable PCSELs for large-area single-mode lasing.
Abstract Light inherently carries multidimensional information, including intensity, polarization, and spectrum. Employing a miniaturized device to simultaneously, instantaneously, and accurately capture the multidimensional information of incident light in a single exposure holds significant applications across numerous fields, but remains challenging. Here, we demonstrate high-accuracy hyperspectro-full-Stokes-polarimetric real-time imaging across a broad wavelength range (1150–1650 nm) with 167 spectral channels, via a metasurface integrated near infrared camera empowered by a residual attention network. The metasurface, which simultaneously performs polarization multiplexing and spectral dispersion, functions as a multidimensional encoder. A tailored deep learning network with a residual attention module is established and trained to reconstruct the multidimensional information of incident light with high accuracy. The key performance metrics—including spectral and image resolution, as well as spectral and polarization reconstruction accuracy—all surpass the best-reported specifications of hyperspectro-polarimetric cameras, with enhancements ranging from several-fold to one order of magnitude. Based on this approach, even intricately coupled multidimensional information can be resolved. In addition, the acquisition, processing, and inference time only takes 18 ms, allowing high-dimensional hyperspectral-polarimetric imaging at 55 frames per second in real time. This work offers a promising solution enabling high-accuracy hyperspectro-polarimetric real-time imaging with a compact camera.
Abstract Trivial-nontrivial topological switching provides a distinctive physical pathway for multifunctional electromagnetic systems, yet has never been exploited for integrated sensing and communication (ISAC). Existing ISAC architectures rely almost exclusively on trivial spatial-wave beamforming, constraining near-field sensing robustness and limiting hardware scalability in 6 G scenarios. Here, we propose an intelligent ISAC platform enabled by a programmable topological metasurface (PTM) that dynamically switches between trivial radiation states and non-trivial valley-Hall states, which is achieved through FPGA-controlled symmetry modulation of the PTM’s unit cells. In its non-trivial state, the PTM forms multiple topologically protected domain-wall channels, guiding surface waves with robustness and enabling the extraction of electromagnetic signatures for human localization. A convolutional neural network trained on these signatures achieves a localization accuracy of 99.54%. Upon position recognition, the PTM transitions to its trivial radiation phase, generating spatial phase-gradient beams for directional wireless communication without requiring additional hardware. Experimental results are consistent with theoretical predictions, validating an implementation of topological state switching for dual-mode ISAC functionality. It suggests that topological state programmability could be a potential mechanism for building compact, robust, and intelligent electromagnetic platforms for next-generation wireless systems.
Abstract Light detection and ranging (LiDAR) systems enable three-dimensional (3D) imaging for emerging applications in autonomous driving and mixed reality. The operational wavelength of LiDAR detectors is transitioning from 905 nm to 1550 nm, leveraging the higher safety power threshold and superior coherence of 1550 nm lasers. Colloidal quantum dots (CQDs) photodiodes have shown good performance and low cost for two-dimensional (2D) imaging covering 1550 nm, however, been faced two main challenges, including poor spectral selectivity and slow response speed as a LiDAR detector. Here, we introduce a tandem PbSe CQDs photodiode that functions as an integrated photonic-electronic link, overcoming both challenges by synergistically merging optical pre-processing and electronic post-processing within a single structure. The designed tandem structure leverages a synergistic combination of RC time constant suppression and charge collection narrowing, yielding a record-fast response time of 1.01 ns among the reported CQD devices and a 13-fold suppressed external quantum efficiency (EQE) in the visible range compared to single-junction devices. We further showcase a spectral-selective LiDAR operating at 1550 nm using the tandem CQDs photodiodes, achieving a centimeter-level spatial resolution while maintaining 97.9% of the signal-to-noise ratio under strong background interference.
Abstract Optical neural networks (ONNs) hold substantial potential in artificial intelligence, promising faster processing speed and reduced energy consumption compared to traditional electronic neural networks, by implementing matrix operations with optical computations. Current ONN architectures predominantly rely on single- or multi-channel convolutions to accelerate computing operations. However, high-performance neural networks, such as ResNet-50, SSD, and Transformer, employ residual convolutions for deep feature extraction instead of single- or multi-channel convolutions. To make ONNs widely functioned in deep neural networks, we propose the Non-Volatile All-Optical Residual Convolution Accelerator (NARCA), based on phase change material (PCM), which weight update energy consumption is only 9.8 μW, much smaller than that of thermal-optical unit with about 10 mW. The NARCA remain high convolution precision, and the experimental results show that the NARCA-based architecture outperforms the conventional optical convolution architecture across different neural-network tasks. Moreover, we demonstrate 128 GHz high-speed optical residual convolution, which greatly improves the residual convolution operation speed compared with the electrical architecture, with a relative root mean square error (RMSE) of less than 0.125.
Abstract The angular momentum of light can be elegantly mapped onto high-order Poincaré spheres, providing a powerful framework for describing structured light beams. Such beams have shown extraordinary potential across diverse applications, including high-capacity optical communications, precision metrology, and quantum information processing. While various methods exist for generating structured light beams, the dynamic synthesis and flexible control of arbitrary vectorial states on diverse, multidimensional Poincaré spheres still rely on bulky free-space optical components, posing significant challenges for scalability and integration. To date, a fully tunable solution implemented on a single photonic chip has yet to be realized. Here, we present the first fully tunable on-chip meta-generator capable of dynamically mapping arbitrary scalar, vectorial, and hybrid modes across multiple Poincaré spheres, and extendable to higher-dimensional Poincaré hyperspheres within a four-dimensional Hilbert space. Our device is implemented on an eight-channel space-multiplexed multimode silicon photonic integrated circuit, where densely integrated mode multiplexers, amplitude–phase modulators, and an inverse-designed multimode meta-waveguide together enable compact, precise, and programmable control of structured light. The multimode meta-waveguide directly maps eight on-chip guided modes to orbital angular momentum (OAM), supporting broadband generation of high-purity OAM modes with diverse polarization states and topological charges. By simultaneously engineering amplitude, phase, polarization, and topological charge, we achieve full-field control over OAM mode bases, enabling fully tunable access to arbitrary scalar and vectorial states across more than eight distinct Poincaré spheres. This work represents a significant step toward reconfigurable on-chip manipulation of multidimensional Poincaré spheres, paving the way for advanced applications in optical communications, quantum photonics, and beyond.
Dual-channel luminescent materials are attractive due to their unique electronic structures and potential applications in various fields. Different from traditional dual-channel emitters with closed-shell structure showing singlet fluorescence and/or triplet phosphorescence, we demonstrate herein dual-channel cerium(III) complexes with open-shell structure showing doublet 5d-4f emission from the central Ce(III) ion. By systematically designing and studying a series of cerium(III) complexes Ce(Tp4Me)(Bp4Me)2, Ce(Tp4Me)2(Bp4Me) and Ce(Tp4R)3 (R = Me, Cl, I) with gradually varied coordination numbers from seven to eight and nine, respectively, it is found that the nine-coordinated complexes Ce(Tp4R)3 showed crowded coordination environments and longer coordination bonds, which underwent a coordination structural relaxation after high energy light excitation, resulting in a new channel emission. This work demonstrates an intriguing dual-channel doublet emission phenomenon in lanthanide Ce(III) complexes and will inspire new insights both into the design of novel lanthanide-based luminescent materials and dual-channel luminescent materials.
Abstract High-dimensional quantum entangled states exhibit unique properties. While generalizations like qudit Bell, Greenberger–Horne–Zeilinger (GHZ), and Cluster states have been studied, the qudit W states remain underexplored. We discover that a superposition of qudit Dicke states can serve as high-dimensional generalizations of W states, demonstrating their enhanced robustness against qudit loss, and present linearly scalable quantum circuits for their generation. Using a programmable silicon-photonic entanglement generator, we first theoretically and experimentally demonstrate the generation of 3-ququart GHZ and W states using two photons and one ancillary ququart, with fidelities of 0.932 ± 0.009 and 0.901 ± 0.012, respectively. Genuine multipartite entanglement of the GHZ state and the improved resilience of the W state are confirmed. Our findings provide new insights and efficient approaches for qudit-based quantum science and technologies.
Abstract Achieving higher degrees and dimensions of light is one of the accessible paths for increasing the capacity of optical data-transmission systems. Orbital angular momentum (OAM), owing to its infinite and mutually orthogonal nature, offers a promising carrier for high-capacity and high-dimensional data transmission. However, current OAM-based shift keying and multicasting schemes predominantly rely on single-mode encoding or simple mode complexing, which fails to fully exploit the high-dimensional potential of OAM. This limitation primarily arises from the challenge of directly modulating complex multiplexed OAM states—such as OAM combs—which typically requires bulky optical setups and complicated iteration algorithms. Here, a hybrid intelligent strategy for high-dimensional OAM comb multicasting is demonstrated, enabling simultaneous multi-channel transmission with high-dimensional OAM encoding using a single phase-only hologram. By jointly AI-driven OAM comb generation and physics-guided wave-vector manipulation, a phase-only modulation framework is developed to directly tailor multiple structured OAM combs in a single modulation step, significantly improving the photon efficiency of OAM shift keying. The scalability and fidelity of the proposed approach are experimentally validated through 4, 6, and 8-channel multicasting demonstrations. Furthermore, a mixed encoding protocol is introduced to enhance transmission security in one-to-many multicasting scenarios, enabling parallel delivery of distinct image contents to different users. A six-channel OAM comb communication system achieves real-time transmission with a bit-error rate below 7 × 10⁻⁵. Our proposal offers a compact and scalable solution for high-dimensional OAM-based data transmission and provides a promising pathway toward next-generation large-capacity optical networks.
Abstract Panvascular diseases (PVDs) stand as the leading cause of global mortality, necessitating a paradigm shift from local anatomical repair to the systemic restoration of vascular homeostasis. While intravascular optical imaging has revolutionized diagnosis, it remains a passive observation tool, restricted by “physical bottlenecks” in resolution and “cognitive bottlenecks” in interpretation. To address these challenges, we frame our analysis around “Suitcordance”, a concept aiming to capture the dynamic state of matching between interventional devices and the vascular microenvironment. In this review, we use Suitcordance as a working analytical framework to represent such a clinically-targeted, integrated perspective, and to organize the evidence on intravascular optical imaging and its integration with artificial intelligence. First, we summarize recent advances in intravascular imaging modalities, including micro-OCT, hybrid systems, and emerging detection technologies. Second, we review how AI-based image analysis and image-derived digital twin models are being applied to interpret these data and to support procedural decision-making. On this basis, we discuss how such tools may contribute to a more individualized assessment of device-vessel matching in panvascular disease.
The rapid switching of materials when excited by ultrashort pulses of light is central for many optical technologies, and in particular to the developing field of time-varying metamaterials. These out-of-equilibrium interactions are difficult to capture with traditional theoretical models. Here we combine experiments and theory to unravel different regimes of interactions and a response saturation for a 44 fs, near-infrared pump pulse exciting a switchable doped semiconductor indium tin oxide thin film target. We model this process as a change in plasma frequency due to the excitation of hot electrons in a non-parabolic conduction band, which increases their effective mass. Our calculations show that saturation at high pump intensities arises because the pump heavily depopulates electrons from below the Fermi level. Excellent agreement with values extracted from experimental data confirms our model. For lower pump intensities, a two-temperature model is consistent with our data, but at higher intensities, it is apparent that other processes are at work, which we attribute to Auger transitions from the valence band, which introduce complex structure into the response, due to non-equilibrium rearrangement of energy between electrons and holes.
Abstract The advancement of electromagnetic technologies necessitates strategies to efficiently exploit spectral and spatial resources in increasingly congested and interference-prone environments. This paper demonstrates, for the first time, a fully passive sliding metasurface that unifies sharp frequency filtering and wide-angle beam steering within a single architecture, thereby establishing a two-dimensional frequency–angle selection window. The proposed metasurface replaces local phase reconfiguration with a global geometric displacement between two complementary phase profiles. This sliding strategy produces a controlled translation of the spectral centroid in reciprocal space, enabling continuous beam steering and suggesting extensibility to other frequency regimes. In the demonstrated implementation, a transmission-line-inspired design is adopted, in which stacked layers are strategically reused to form a quasi–Fabry–Pérot cavity within the meta-atom, enabling simultaneous phase modulation and frequency-selective filtering. Experimental results validate sharp spectral selectivity, ultra-wide out-of-band suppression, and stable beam steering over a ± 57° scanning range. The proposed sliding metasurface offers a compact and broadly applicable platform for next-generation electromagnetic systems, enabling a new modality for coordinated spectral–spatial wave control.
Abstract High-speed volumetric stimulated Raman scattering (SRS) microscopy offers unique capabilities for label-free chemical imaging in living systems, yet its performance is fundamentally constrained by the trade-off between imaging speed and signal-to-noise ratio (SNR). At the short pixel dwell times required for three-dimensional dynamic imaging, photon-limited detection leads to severe noise that cannot be effectively mitigated by existing denoising approaches, owing to the lack of ground truth data and temporally redundant measurements in live-cell conditions. Here we present PHYSIQ, a physics-paired in-phase and quadrature SRS imaging framework that fundamentally redefines data acquisition for noise-limited optical microscopy. By exploiting the intrinsic quadrature nature of heterodyne detection, PHYSIQ-SRS simultaneously acquires two spatially co-registered and temporally near-synchronous SRS image channels with statistically independent shot noise. This physics-paired measurement enables fully self-supervised Noise2Noise restoration without requiring ground truth or temporal redundancy. The implementation integrates dual-channel lock-in detection with defocus-corrected spatial co-registration and controlled temporal offset, establishing a robust and generalizable strategy for generating unbiased training pairs directly from physical measurements. This innovative approach achieves an SNR enhancement of ~12.5 dB while preserving quantitative Raman contrast, effectively overcoming the conventional speed-sensitivity limitation in volumetric SRS microscopy. The improved performance enables video-rate volumetric imaging and label-free 3D tracking of lipid droplets (LDs) in living cells. Using this capability, we uncover that LD dynamics are governed by discrete motility states with condition-dependent transitions, including spatial redistribution under nutrient perturbation, selective suppression of long-range transport upon glycolytic inhibition, and phase-dependent reprogramming during mitosis. PHYSIQ-SRS establishes a new paradigm of physics-enabled self-supervised imaging, providing a general solution to shot-noise-limited detection in laser-scanning microscopy. This advance opens new opportunities for high-speed, label-free volumetric imaging and quantitative investigation of live-cell biology, metabolic phenotyping, developmental imaging, and biomedical discovery.
Abstract Photodynamic therapy (PDT) is fundamentally limited by inefficient conversion of photoexcited energy into reactive oxygen species (ROS), especially in hypoxic pathological tissues. Photogenerated radical-ion pairs (PS⁺•/PS⁻•) from photosensitizer could overcome this limitation by broadening photoredox diversity for hypoxia-tolerant ROS production. However, conventional substrate-dependent mechanism typically generates only one radical species, limiting both photoredox diversity and overall ROS production. Here we demonstrate a substrate-free mechanism via noncovalent homodimer fission for efficient generation of PS⁺•/PS⁻• pairs from water-soluble DB-Py, unlocking multipath photoredox for hypoxia-tolerant PDT. Spectroscopic studies and calculations provide experimental evidence for homodimer formation via noncovalent interactions between a photoexcited triplet-state and a ground-state molecule, followed by fission into abundant PS⁺•/PS⁻• pairs. The resulting PS⁺• oxidizes water to generate O 2 in situ, which is subsequently reduced by PS⁻• to produce •OH and O 2 ⁻•, thereby establishing a self‑oxygen‑supplying ROS generation cycle. Consequently, DB-Py achieves 4.9‑fold higher •OH and 2.7‑fold higher O 2 ⁻• compared to commercial Rose Bengal, resulting in 7.9-fold enhanced antibacterial efficacy and 2.2-fold accelerated wound epithelialization. This work establishes a substrate-free radical generation mechanism with broad implications for the development of photodynamic, photocatalytic, and photoredox systems.
Abstract Thermal radiation in the mid-wave infrared (MWIR) and long-wave infrared (LWIR) underpins a wide range of imaging, sensing, and security technologies. While metasurfaces have enabled spectral and angular control of thermal emissivity, achieving spatially resolved and uncorrelated emissivity control across multiple infrared bands on a single platform remains a major challenge. Here we experimentally demonstrate a metasurface-based approach that enables pixel-level control of thermal emissivity in both the MWIR and LWIR bands simultaneously. The platform is based on a cavity-coupled metal-insulator-metal architecture supporting two infrared resonances whose spectral positions and strengths are orthogonally controlled through geometric design parameters. By systematically exploring the design space, we realize a broad emissivity palette and experimentally verify its thermal response using Fourier-transform infrared spectroscopy (FTIR) together with calibrated MWIR and LWIR thermal imaging. This capability allows the encoding of spatial thermal patterns that are either correlated or distinct in the two bands, enabling demonstrations of dual-band thermal camouflage and decoupled thermal image multiplexing on a single chip. The metasurface response is shown to be polarization-insensitive and robust over a wide range of temperatures and viewing angles. These results establish a scalable route toward multi-band, image-level control of thermal emission, with potential applications in thermal imaging, security, and infrared information encoding.
Abstract The vascular network defines the perfusion boundaries that are fundamental to surgical resection and functional preservation. Photoacoustic angiography (PAA) offers stereoscopic visualization and quantification of microvascular anatomy and hemodynamics in vivo, but existing implementations remain constrained in field of view and imaging speed, limiting the intraoperative utility. Here, we introduce a real-time PAA tracking and mapping strategy that estimates the six degree-of-freedom motion of a handheld probe and reconstructs panoramic three-dimensional vascular maps for surgical guidance. By associating geometric and intensity-based hybrid vascular features to prevent degeneracy, the method achieves robust online mapping with 99.67% accuracy, even under dynamic freehand operation. In rat radical gastrectomy, the approach expanded the imaging boundary by 42.3-fold to 50.84 mm³ within 16.7 s, enabling panoramic microvascular visualization beyond the surgical field for preoperative planning, postoperative evaluation, and distal hemodynamic surveillance. This on-the-fly, freehand, and field-of-view–unconstrained PAA establishes a practical framework for vascular-focused surgical interventions, supporting precise and timely decision-making.
