Quantum frequency conversion (QFC) plays a pivotal role in enabling coherent photonic interfaces across disparate quantum systems, serving as a foundation for scalable and interoperable quantum networks. Among various approaches, four-wave mixing (FWM) offers a promising path toward efficient, broadband QFC. In this work, we investigate the feasibility of phase-preserving frequency conversion, a necessary condition for coherent quantum state transfer, by characterizing the spectral phase mapping from input to output fields. We model nonlinear propagation through a gas-filled hollow-core capillary fiber, to demonstrate highly phase-coherent frequency conversion across a broad spectral range, including the infrared to ultraviolet, such as telecom (1550 nm) to visible (516 nm) and deep-UV (308 nm). We further examine how phase-mapping quality and conversion efficiency depend on key system parameters, including the intermediate pulse bandwidth and input pulse energy. Our results reveal optimal operating conditions for achieving future efficient and coherent QFC and provide valuable insights into the nonlinear dynamics underlying FWM-based transduction. These findings lay the groundwork for integrating heterogeneous quantum platforms across widely separated optical bands.
Leveraging the natural axial confinement of coherent extreme ultraviolet (EUV) light generated via the high-order harmonic generation process, we demonstrate the spatial separation of EUV and the driving infrared (IR) beams through hollow-core microchannels embedded in a laser machine glass device. This structure enables broadband EUV transmission while attenuating the collinear IR by 2 orders of magnitude. In addition, we explore the potential of integrated photonic architectures based on EUV-guiding hollow structures, laying the foundation for a new class of compact, palm-top devices for EUV and soft X-ray applications.
Two-photon interference is a fundamental resource for quantum technologies and optical quantum computing, underpinning precision measurements, scalable entanglement distribution, and the operation of photonic circuits and quantum network protocols. Here, we report the first demonstration of massively parallel, wavelength-resolved photon bunching, revealing Hanbury Brown–Twiss correlations across 70 independent spectral channels. These observations are enabled by a fast, data-driven single-photon spectrometer that achieves 48 pm spectral and 48 ps temporal resolution over a 10 nm bandwidth, providing simultaneous access to spectro-temporal photon correlations without the need for narrowband filtering. This approach enables high-dimensional quantum interference measurements across a broad spectrum. Our results establish frequency-multiplexed two-photon interference as a scalable and throughput-efficient platform for quantum-enhanced photonic technologies, offering a practical route toward room-temperature architectures that overcome loss limitations and advance the scalability for a variety of applications.
To address the challenge of complex mapping between process parameters and optical performance in smoothing by spectral dispersion (SSD) gratings, we propose an inverse design and process optimization framework based on ensemble learning. By constructing a weighted-average ensemble learning model, this framework establishes a complete mapping chain of “target performance–predicted structure–executable process,” realizing a closed-loop optimization workflow of “design–prediction–fabrication–verification.” The coefficient of determination for its structural parameter predictions is above 0.955, with a mean absolute error of less than 0.85%. Through systematic comparison of the correlations between model predictions and measured data, we confirm that the framework captures genuine dependencies among process parameters, providing further insights for achieving more reliable process mapping. Experimental validation shows that SSD gratings fabricated based on this framework exhibit a deviation of less than 1.5% between the measured diffraction efficiency and the design target. We not only verify the effectiveness of the closed-loop optimization workflow but also mark a paradigm shift in process development from empirical trial-and-error to data-driven intelligent design. The validated interpretable inverse design framework provides a pathway for the intelligent and controllable fabrication of gratings and other micro-nano optical components.
APNexus editors Linjie Zhou and Xianshu Luo present an overview of the articles in the CPO theme issue.
Most existing optical convolutional neural networks (OCNNs) are primarily limited to implementing basic convolution functions, with little focus on leveraging the color dimension. As a result, they can typically only process single-channel or grayscale images, lacking the ability to utilize multichannel information such as RGB. In this work, we propose a multichannel optical convolutional neural network (MOCNN), which is capable of processing both RGB and hyperspectral images using only grayscale sensors by integrating trainable color filters into the optical path. Through simulations and experiments, we demonstrate that the performance of optical convolution is significantly enhanced by incorporating color information, enabling the system to handle complex color-related tasks. Furthermore, we design a loss function tailored to the physical properties of quantum dots. We demonstrate that this training strategy can be extended to other types of filtering materials. Moreover, the proposed technique can serve as an OCNN-based feature-map acquisition camera without additional imaging components, offering a potential route toward compact, low-cost, and privacy-preserving image capture with limited task-performance degradation.
Microlasers emitting multiple coherent orbital angular momentum (OAM) states provide a powerful platform for precisely controlling high-dimensional optical fields on a chip. However, existing coupled-OAM microlasers typically require complex structures and large footprints, hindering large-scale on-chip integration and coordinated manipulation of multiple optical degrees of freedom (DoFs). Herein, we demonstrate an ultra-high-dimensional microlaser array by manipulating coupled OAM modes within a microcavity. By integrating a microring cluster within a Fabry–Pérot microcavity, a pair of coupled elliptical OAM modes with six tunable DoFs are generated: coupling strength, azimuthal orders of each mode, long-axis directions of each mode, and coupling direction. Exploiting these six independent DoFs, we design eight multiplexing channels and implement them within a microcavity array. This study demonstrates high-dimensional optical-field manipulation within a compact microlaser array, offering potential for high-security information storage, advanced optical communications, and emerging quantum technologies.
This study explores co-packaged optics (CPO) using through-silicon via (TSV) and through-glass via (TGV) interposers with 2.5D/3D integration, offering superior performance over conventional 2D integration. The fabricated TSV and TGV interposers demonstrate 3 dB bandwidths exceeding 67 and 110 GHz, respectively, supporting 128 Gbaud signal transmission. CPO solutions are proposed based on TSV and TGV interposers, respectively. Structural design, packaging, and simulation of the CPO transceivers demonstrate that the proposed architecture can support optical engines operating at 112 GBaud, highlighting their potential to enhance integration density, reduce power consumption, and enable next-generation high-speed optical interconnects for artificial intelligence and high-performance computing.
As co-packaged optics (CPO) integrates photonic chips and electronic dies together, accurate and non-destructive post-fabrication characterization of waveguide parameters-width, thickness, and the refractive indices of both core and cladding-becomes critical, because even minute index deviations are amplified in dense interconnects and directly impair the signal integrity and energy efficiency of CPO modules. We take the material refractive index as an example and demonstrate an effective approach for non-destructive parameter extraction of the as-fabricated optical waveguide with a single photonic crystal ring resonator. By lifting the degeneracy of the symmetric and anti-symmetric resonant peak, this single structure suffices to determine the core and clad indices rapidly with high precision. We successfully applied the extracted parameter of the refractive indices to validate the optical transmission of the index-sensitive wavelength division multiplexer device. Our approach has provided a facile solution for post-fabrication device parameter evaluation and is readily extendable to geometric parameters.
The merging of thin-film photovoltaic (PV) technologies with tandem architectures, such as perovskite-on-silicon double-junction solar cells, offers avenues to expand solar electricity. Their combination of flexibility, affordability, low weight, and high efficiency enables applications ranging from portable electronics to building-integrated and vehicle-integrated PV, or solar-powered space systems. Nevertheless, the efficiency of perovskite-on-silicon tandem PV is often constrained by suboptimal optical management, particularly in ultra-thin designs where light trapping (LT) and current/voltage matching are critical. Here, we develop an optoelectronic framework to optimize 2- and 4-terminal perovskite-silicon tandem cells, featuring 1 mu m-thick crystalline silicon absorbers with front-integrated photonic structures. To maximize power conversion efficiency (PCE), both the LT geometry and perovskite thickness were systematically optimized. In addition, indium tin oxide (ITO)-based and optically engineered interlayers are shown to exhibit similar optical performance, reinforcing ITO as a choice for 2-terminal tandems. The ultra-thin photonic-enhanced 2-terminal tandem achieves a PCE of 23.8%, corresponding to a 21.8% relative improvement over its planar counterpart, whereas the 4-terminal configuration reached a combined efficiency of 26.7%, primarily from LT-improved silicon photocurrent. These findings highlight the role of opto-electronically optimized light-management solutions in unlocking the potential of ultra-thin tandem solar cells for flexible, high-efficiency, and energy harvesting, paving the way for next-generation photovoltaics.
We demonstrate a 336 Gbps reconfigurable silicon Ge-Si photodetector (PD) based on low-loss fan-out wafer-level packaging (FOWLP). Silicon-based micro-electro-mechanical system (MEMS) photosensitive composite film is employed to enable seamless integration of electronic integrated circuits and photonic integrated circuits. This approach utilizes benzocyclobutene-based transmission lines, achieving a high interconnect density of >10(2)/mm and low insertion loss of <0.3 dB/mm@100 GHz. By demonstrating a 336 Gbps/lambda transmission, we highlight the potential of low-loss FOWLP heterogeneous integration for seamlessly assembled, high bandwidth-density optical interconnects in next-generation artificial intelligence clusters.
Dielectric metasurfaces and other resonant nanophotonic systems have transformed light–matter interactions by providing exact control over electromagnetic fields. Despite the fact that these systems frequently span multiple coupling regimes and may thus exhibit rich intrinsic temporal dynamics, characterizing them has primarily relied on steady-state frequency-domain analysis. To close this gap, we present a thorough time-domain mode-retrieval framework. The decoupling of resonant modes from the background continuum is made possible by systematically extracting the complex resonant poles of a nanophotonic system directly from its transient response using the vector fitting technique and the Prony method. Specifically, we investigate resonances supported by a silicon metasurface that are quasi-bound states in the continuum (quasi-BICs). This method effectively separates radiative quasi-BIC modes and identifies their fundamental properties, such as the Q-factors. In addition, our approach uncovers a clear temporal beating behavior associated with transient mode interference that is not visible in steady-state spectral measurements. Lastly, this method is expanded to the nonlinear regime to see the third-harmonic generation signal’s temporal evolution. Our results open up possibilities for ultrafast all-optical devices with customized temporal dynamics by establishing a potent semi-analytical tool for time-resolved investigations of ultrafast dynamics in resonant nanophotonic systems.
We propose and experimentally demonstrate the first mid-infrared (IR) fiber-optic surface plasmon resonance (SPR) sensor operating at a resonance wavelength of 4004 nm. The sensor was fabricated by depositing a 120 nm thick indium tin oxide film on a D-shaped (ZrF4−BaF2−LaF3−AlF3−NaF) ZBLAN fiber. For refractive indices in the range of 1.33 to 1.39, the device achieves a refractive index (RI) sensitivity of 23,465.23 nm per RI unit (nm/RIU), which is at least five times higher than that of previous experimentally reported visible and near-IR fiber-optic SPR sensors. To validate the capability of mid-IR SPR for detecting relatively large analytes, standard silica particles with a diameter of 1.5 μm were employed as representative test objects. Simulations and experiments confirm that mid-IR SPR provides substantially deeper field penetration and enables more effective interaction with such micron-scale objects compared with visible SPR, as reflected by consistently higher effective RIs measured at 4004 nm than at 628 nm when sensing 1.5 μm silica particles. These findings highlight the strong potential of mid-IR fiber-optic SPR sensors for microparticle detection, particularly in applications involving bioparticles, synthetic polymers, and complex nanostructured materials, where deeper light–matter interaction is critical for achieving high sensitivity and reliability.
The advent of artificial intelligence, cloud services, and big data applications has propelled the evolution of next-generation high-capacity datacenters. It is highly anticipated that coherent detection will penetrate further into datacenters in the next decade. However, the large number of tunable lasers in data centers makes the optical module structure complex and adds additional thermal power consumption. Meanwhile, optical frequency combs serve as high-precision frequency resolution and wide spectral coverage lasers, holding tremendous potential for the field of wavelength division multiplexing optical communication. Here, we present an innovative global frequency-synchronous optical network (global-FSON) architecture for coherent short-reach optical interconnects by synchronizing optical frequency combs to a global positioning system disciplined oscillator (GPSDO) and distributing them. The global-FSON architecture can share GPS-referenced optical frequency combs as master lasers among different servers, which eliminates the need for a massive number of tunable lasers in datacenter optical interconnects. On this basis, we achieve a reset-free carrier phase recovery analog coherent receiver in the optical domain and demonstrate dual-polarization coherent signals demultiplexing without coherent silicon application-specific integrated circuits. We introduce the global-FSON architecture that provides a laser source with absolute stability for all transponders in datacenter coherent optical interconnects. Its implementation would bolster the potential applicability of coherent optical communication in next-generation datacenter coherent optical interconnects.
Integrated optical phased arrays (OPAs), owing to their high integration level and wide-angle beam-steering capability, are a key promising component for 3D/4D sensing and free-space optical communication. However, integrated OPAs typically employ a single-wavelength laser as the input light source, which limits their application in spectral imaging and secure communication. In this work, we propose and experimentally demonstrate a dual-wavelength coherent beam combining (CBC) scheme using a 64-element integrated OPA. A multiwavelength CBC model is established and validated with two wavelength pairs (1545/1555 and 1535/1565 nm). Both simulation and experimental results confirm that the beam dispersion intensifies with the steering angle and wavelength separation. To overcome this inherent dispersion issue, we employ a stochastic parallel gradient descent algorithm with two avalanche photodiodes as feedback sensors, achieving an arbitrarily configurable angular separation between the two wavelength beams. Finally, we demonstrate the wide field of view beam-steering capability of the combined beam across a 60 deg (±30 deg) range. We provide a foundational framework for advancing multiwavelength OPA toward applications in spectral imaging and secure optical communications.
