Light, science & applications最新文献

Deployment of terahertz communication and spectroscopy systems relies on the availability of low-noise and fast detectors, with plug-and-play capabilities. However, most current technologies are stand-alone, discrete components. They are often slow or susceptible to temperature drifts and require tight beam focusing to maximize the signal-to-noise of the detector. Here, we demonstrate an integrated photonic architecture in thin-film lithium niobate that addresses these challenges by exploiting the electro-optic modulation induced by a terahertz signal onto an optical beam at telecom frequencies. Leveraging on the low optical losses provided by this platform, we integrate a double array of up to 18 terahertz antennas within a Mach-Zehnder interferometer, considerably extending the device collection area and boosting the interaction efficiency between the terahertz signal and the optical beam. We show that the double array coherently builds up the probe modulation through a mechanism of quasi-phase-matching, driven by a periodic terahertz near-field pattern, circumventing physical inversion of the crystallographic domains. This provides means to fully custom-tailor the frequency response of the device, limit it to a desired frequency band and effectively suppress out-of-band signals. The large detection area ensures correct operation with diverse terahertz beam settings. Furthermore, we show that the antennas act as pixels that allow reconstruction of the terahertz beam profile impinging on the detector area. Our on-chip design in thin-film lithium niobate overcomes the detrimental effects of two-photon absorption and fixed phase-matching conditions, which have plagued previously explored electro-optic detection systems, especially in the telecom band, paving the way for more advanced on-chip terahertz systems.

Photonics is promising to handle extensive vector multiplications in artificial intelligence (AI) techniques due to natural bosonic parallelism and high-speed information transmission. However, the dimensionality of current photonic linear operation is limited and tough to improve due to the complex beam interaction for implementing optical matrix operation and digital-analog conversions. Here, we propose a programmable and reconfigurable photonic linear vector machine with extreme scalability formed by a series of emitter-detector pairs as the independent basic computing units. The elemental values of two high-dimensional vectors are prepared on emitter-detector pairs by bit encoding and analog detecting method without requiring large-scale analog-to-digital converter or digital-to-analog converter arrays. Since there is no interaction among light beams inside, extreme scalability could be achieved by simply multiplicating the independent emitter-detector pair. The proposed architecture is inspired by the traditional Chinese Suanpan or abacus, and thus is denoted as photonic SUANPAN. Experimentally, the computing fidelities for vector inner products could achieve >98% in our implementation with an 8 × 8 vertical cavity surface emission laser (VCSEL) array and an 8 × 8 MoTe₂ two-dimensional material photodetector array. Furthermore, such implementation is applied on two typical AI tasks as 1024-dimensional optimization problem is successfully solved and competitive classification accuracy of 88% is achieved for handwritten digit dataset. We believe that the photonic SUANPAN could serve as a fundamental linear vector machine and enhance various future AI applications.

Single-pixel imaging (SPI) has long been recognized for its potential in spectral regions where conventional imaging sensors fall short, such as the near-infrared spectrum. Yet, despite its sensitivity, SPI and its complex-field variants have faced critical bottlenecks in speed and throughput, hindering their adoption for real-time applications. A recently proposed approach-frequency-comb acousto-optic coherent encoding (FACE)-places an important step in overcoming these barriers, delivering an unprecedented space-bandwidth-time product. By showcasing its versatility through several compelling proof-of-concept demonstrations in real-time complex-field microscopy, this advance paves the way for transformative progress in optical imaging beyond the visible spectrum. We discuss here advantages, challenges and potential future directions for scaling up this technology.

Gas sensing technology is widely applied in various fields, including environmental monitoring, industrial process control, medical diagnostics, safety warnings, and more. As a detection element, the quartz tuning fork (QTF) offers advantages such as high-quality factor (Q-factor), strong noise immunity, compact size, and low cost. Notably, its resonant characteristics significantly enhance system signal strength. Two spectroscopic techniques based on QTF detection, Quartz-enhanced photoacoustic spectroscopy (QEPAS) and light-induced thermoelastic spectroscopy (LITES), are currently research hotspots in the field of spectral sensing. This paper provides a comprehensive and detailed review and highlights pivotal innovations in these two QTF-based spectroscopic techniques. For QEPAS, these encompass high-power excitation methods, novel excitation sources, advanced QTF detection elements, and acoustic wave amplification strategies. Regarding LITES, the researches on optical cavity-enhanced approaches, modified QTF improvement mechanisms, integration with heterodyne demodulation technique, and combination with QEPAS were analyzed. These advances have enabled quartz-enhanced laser spectroscopy to achieve detection limits ranging from parts-per-billion (ppb) to parts-per-trillion (ppt) levels for trace gases such as methane (CH₄), acetylene (C₂H₂), carbon monoxide (CO), and so on. Additionally, prospects for future technological developments are also discussed in the concluding section.

Optical computing holds promise for high-speed, energy-efficient information processing, with diffractive optical networks emerging as a flexible platform for implementing task-specific transformations. A challenge, however, is the effective optimization and alignment of the diffractive layers, which is hindered by the difficulty of accurately modeling physical systems with their inherent hardware imperfections, noise, and misalignments. While existing in situ optimization methods offer the advantage of direct training on the physical system without explicit system modeling, they are often limited by slow convergence and unstable performance due to inefficient use of limited measurement data. Here, we introduce a model-free reinforcement learning approach utilizing Proximal Policy Optimization (PPO) for the in situ training of diffractive optical processors. PPO efficiently reuses in situ measurement data and constrains policy updates to ensure more stable and faster convergence. We validated our method through both simulations and experiments across a range of in situ learning tasks, including targeted energy focusing through a random diffuser, image generation, aberration correction, and optical image classification, demonstrating in each task better convergence and performance. Our strategy operates directly on the physical system and naturally accounts for unknown real-world imperfections, eliminating the need for prior system knowledge or modeling. By enabling faster and more accurate training under realistic experimental constraints, this in situ reinforcement learning approach could offer a scalable framework for various optical and physical systems governed by complex, feedback-driven dynamics.

X-ray scattering has been an indispensable tool in advancing our understanding of matter, from the first evidence of the crystal lattice to recent discoveries of nuclei's fastest dynamics. In addition to the lattice, ultrafast resonant elastic scattering of soft X-rays provides a sensitive probe of charge, spin, and orbital order with unparalleled nanometre spatial and femto- to picosecond temporal resolution. However, the full potential of this technique remains largely unexploited due to its high demand on the X-ray source. Only a selected number of instruments at large-scale facilities can deliver the required short-pulsed and wavelength-tunable radiation, rendering laboratory-scale experiments elusive so far. Here, we demonstrate time-resolved X-ray scattering with spectroscopic contrast at a laboratory-based instrument using the soft-X-ray radiation emitted from a laser-driven plasma source. Specifically, we investigate the photo-induced response of magnetic domains emerging in a ferrimagnetic FeGd heterostructure with 9 ps temporal resolution. The achieved sensitivity allows for tracking the reorganisation of the domain network on pico- to nanosecond time scales in great detail. This instrumental development and experimental demonstration break new ground for studying material dynamics in a wide range of laterally ordered systems in a flexible laboratory environment.

Free-space optical communication (FSOC) enables high-speed, secure, and scalable data transmission, with great potential for space-ground networks. However, existing FSOC systems predominantly employ point-to-point transmitters, each requiring bulky beam steering devices with complex control mechanisms, which severely limits their feasibility for multi-node micro-platform applications. Here, to address such a challenge, we propose a novel point-to-multipoint FSOC scheme based on reconfigurable SiC gratings, which are directly fabricated in stretchable PDMS films via femtosecond laser-induced carbide precipitation. The reconfigurable SiC transmission gratings are with good transparency (~91.9% at 1550 nm), dynamic beam steering capability (hundred-milliradian level), and an ultralightweight design (single grating: 0.4 g). The SiC fringes are specially fabricated within the internally symmetric region of the PDMS film to mitigate the structure distortion during stress regulation, significantly enhancing the long-range transmission capability in degraded atmospheric channels. The system supports 1-to-7 and 1-to-9 dynamic optical communication for 1D and 2D configurations, respectively. In a state-of-the-art 225-meter outdoor experiment, the system achieves reliable 10 Gbps transmission for each node. This portable FSOC system overcomes the limitations of conventional systems, enabling scalable and flexible multibeam steering. This approach establishes a robust foundation for long-range, multinode, and high-capacity FSOC networks among spatial micro-platforms such as unmanned aerial vehicles and micro-satellites.

Cavity electro-optic (EO) modulation plays a pivotal role in optical pulse and frequency comb synthesis, supporting a wide range of applications including communication, computing, ranging, and quantum information. The ever-growing demand for these applications has driven efforts in enhancing modulation coupling strength and bandwidth towards advanced pulse-comb synthesis. However, the effects of strong-coupling and high-bandwidth cavity EO modulation remain underexplored, due to the lack of a general, unified model that captures this extreme condition. In this work, we present a universal framework for pulse-comb synthesis under cavity EO modulation, where coupling strength and modulation bandwidth far exceed the cavity's free spectral range (FSR). We show that, under such intense and ultrafast driving conditions, EO-driven frequency combs and pulses exhibit rich higher-order nonlinear dynamics, including temporal pulse compression and comb generation with arbitrary pump detuning. Leveraging this framework, we reveal a direct link between the higher-order dynamics of EO pulse-comb generation and the band structure of synthetic dimension. Furthermore, we demonstrate arbitrary comb shaping via machine-learning-based inverse microwave drive design, achieving a tenfold enhancement in cavity EO comb flatness by exploring the synergistic effects of high-bandwidth driving and detuning-induced frequency boundaries. Our findings push cavity EO modulation into a new frontier, unlocking significant potential for universal and machine-learning-programmable EO frequency combs, topological photonics, as well as photonic quantum computing in the strong-coupling and high-bandwidth regimes.

Building a sensitive magnetic field sensor is non-trivial; building a more sensitive one by adding extra loss to the sensor is counterintuitive, but with innovative ideas from non-Hermitian physics like an exceptional point, a new magnetic field sensor first of its kind paves the way for broader applications of similar techniques.