An optical fiber probe based on micro-nanoscale structured film has been verified theoretically and experimentally, enriching the variety of optical fiber magnetic field sensors through a "Lab on Fiber" technology. The micro-nanoscale structure of single-spring-type microcantilever (SSC) is designed and fabricated by femtosecond laser direct writing (FLDW), which solves the hysteresis effect and saturation effect of traditional fiber magnetic field sensor. By modifying graphene oxide (GO) with Fe3O4 nanoparticles, Fe3O4/GO films as magnetic response units are obtained. Through the design and optimization of micro-nano film structure, the magnetic field sensitivity of the prepared SSC is increased by 10.24 times to 189.39 pm/mT, with a fast response time of 217 ms. Compared with the magnetic field sensors reported so far, the proposed probe has broad application prospects due to its advantages of miniaturization (30 mu m x 62.5 mu m, length x radius), high stability (sigma=+/- 0.02732 nm), and low temperature crosstalk (0.13 mT/degrees C). Furthermore, the proposed scheme is universal and not limited to magnetic field measurement. With the rapid development of advanced smart materials, the scheme will be suitable for the preparation of fiber tip microsensors and micro-actuators with various stimulus response characteristics (light, heat, electric, magnetic, gas, liquid).

Miniaturization and multi-functionalization integration of microparticle manipulation devices remain the two major challenges in micro–nano control. Here, we demonstrate a fiber-tip photothermal microparticle controller based on asymmetric thermal effect. The fiber-tip controller is composed of an optical fiber microcavity structure and a GO+AuNBPs/PDMS composite film. The optical fiber microcavity is used to limit the spread of incident light to enhance the interaction between the optical field and the composite film and to provide real-time spectral monitoring of the thermal and deformation states of the composite film. The photothermal conversion efficiency has been enhanced significantly due to the special structure of composite film. Theory and experiments indicate that the target microparticle exhibits stable periodic oscillations and self-rotations manipulated by the controller. The oscillation frequency increased as the power of driven laser. As the power of driven laser increased exceeds 18.6 mW, the composite film undergoes irreversible deformation, which further enhances the asymmetry of the surrounding thermal field. The target microparticle would maintain stable self-rotation in asymmetric thermal field. The proposed fiber-tip controller enables the precise capture of microparticle in liquid environments, providing a reliable tool for the manipulation and selection of microparticle at a microscale.

Micronano capture systems capable of efficiently trapping target particles have been a focus in biomedical engineering, lab-on-a-chip devices, and targeted drug delivery. Here, we propose a miniature photodriven microbubble capture actuator based on an optical fiber microcavity. The actuator consists of an optical fiber microcavity and a graphene oxide/polydimethylsiloxane-gold nanobipyramids (GO/PDMS-AuNBPs) photothermal composite film. The composite film exhibits high photothermal conversion efficiency, enabling the actuator to generate a stable temperature gradient in a liquid environment. The resulting temperature gradient induces a surface tension gradient along the surface of the bubble, thereby driving the bubble toward the trapping actuator. Both theoretical analyses and experimental results confirm that the actuator successfully captures microbubbles with radii of 5 to 200 μm. Compared with conventional particle capture technologies, the proposed actuator offers several advantages, including low power consumption, a compact structural design, and high stability in liquid environments. The system provides a novel platform for biomedical applications such as live cell manipulation and microreactor construction.

Chiral optical field (COF) constitutes a class of structural light characterized by intensity, phase, or polarization chirality. Recently, COFs have attracted considerable interest due to their versatile and tunable degrees of freedom. However, existing COFs typically exhibit a single dominant structure and lack independent and controllable partitions, thereby limiting their information-carrying capacity. In this work, we propose a quasi-singularity line structure to segment the COF, enabling independent control of each partition within the COF. This structure, constructed based on the tangent function, induces rapid phase variations that generate localized zero-intensity boundaries, effectively achieving segmentation of the field. Using this approach, we generate a multi-partition chiral vector vortex field with controllable partition number, partition length, partition spatial position, and partition polarization state. Numerical simulation and experimental results demonstrated the ability to independent control each partition of the field. The proposed method advances the partitioned modulation of COFs and holds promise for applications in optical communications and beyond.

This study proposes a high-performance microforce sensor integrating a film-lollipop structure with an optical fiber microcavity. The sensor consists of two core components: a support structure composed of a segment of single-mode fiber (SMF) and hollow-core fiber (HCF), and a sensing structure comprising a polyvinyl alcohol (PVA) film and a "lollipop" probe. Specifically, the end face of the SMF and PVA film forms a Fabry-Perot Interferometer (FPI), while a colloidal pillar capped with a silica sphere serves as the microforce action point. Within the lollipop structure, the silica sphere enables gentle contact with the measured object, and the ultraviolet (UV) adhesive pillar concentrates the applied force on the force-sensitive area of the PVA film. To clarify the influence of structural parameters on sensing performance, finite element analysis (FEA) simulations and numerical calculations were conducted to analyze the factors that affect the performance of the sensor, with experimental results confirming that reducing the diameter of the colloidal pillar effectively enhances sensor sensitivity. Fabricated via a facile process, the sensor exhibits outstanding comprehensive performance: a maximum microforce sensitivity of up to -3.142 nm/mu N, a detection limit of 65 nN, a maximum bearable force exceeding 1800 mu N, stable repeatability, and near real-time response. This sensor offers a high-performance solution for microforce detection in biomedical engineering, microfluidics, and microelectromechanical systems.

Waveguide-integrated optical metasurfaces, which integrate flexible wavefront-shaping metasurfaces with planar waveguides, enable on-chip multifunctional light field tuning and facilitate the development of highly integrated chip-scale optical applications. Herein, we propose an innovative on-chip metasurface scheme for the generation and manipulation of vectorial structured light. Taking tight focusing as a representative example, this scheme overcomes the traditional limitation of separately implementing vector beam generation and high-numerical-aperture (NA) focusing. By integrating a single metasurface onto a planar waveguide, the proposed design achieves the simultaneous realization of both functionalities within a compact on-chip architecture, representing a significant advancement in miniaturized optical manipulation systems. Various types of vector beams, including cylindrical vector beams and optical skyrmion beams, are validated by using our proposed scheme. Furthermore, we demonstrate the potential of this platform for multifunctional applications, such as optical trapping and dynamic topological structure tuning, thereby highlighting the versatility of this metasurface-based on-chip light control technology.

Unidirectional scattering plays an essential role in controlling light-matter interactions and supports a wide range of applications, including optical sensing, nanoantennas, and biomedical imaging. However, achieving high-intensity, customized unidirectional scattering in a single subwavelength nanoparticle governed by electric modes remains a significant challenge. We propose a classification-assisted transformer-based deep neural network, termed CSSformer, for the intelligent design of strong unidirectional scattering driven solely by electric modes in four-layer core-shell structures (CSSs). By employing a spectrum-splitting scheme, CSSformer overcomes the severe dimensional mismatch between input and output vectors, significantly outperforming traditional multilayer perceptron (MLP) in both forward prediction and inverse design. Furthermore, to simultaneously ensure high directionality and strong scattering intensity, both unidirectional and total scattering spectra are taken into account during the inverse training process, while a classification module is embedded to enable efficient learning across multiple distinct CSS configurations. An on-demand spectral pre-design strategy is introduced, allowing users to define arbitrary spectral features for real-time, tailored design of strong directional scattering. This study provides an efficient deep-learning framework for manipulating scattering directionality, with promising implications for photovoltaics, antenna design, and optical sensing.

In deep-space exploration and remote sensing, characterizing radiative transfer in complex planetary atmospheres is fundamental for robust target detection and optical navigation. On the Martian surface, intense scattering and attenuation by dust aerosols pose severe environmental interference, challenging star sensors used for high-precision navigation. To address this, this study develops a spectral radiative transfer model based on the Null Collision Monte Carlo Method to characterize the optical background of the dusty Martian atmosphere. Mie scattering theory is employed for dust particles, while gas molecular absorption is modeled via line-by-line integration. The simulated sky radiance is validated against Mars rover Navcam observations, yielding an average relative error of 7.83% between the modeled and observed radiance values across scattering angles greater than 5°. Building on this, an imaging link model evaluates surface-based detection performance, including signal-to-noise ratio, detection success probability, and star count. Optical parameters—aperture, field of view, and integration time—are optimized for nighttime and dawn-dusk modes. Spatio-temporal assessments are conducted globally across Martian years, focusing on the Zhurong landing site and Tianwen-3 candidates. Finally, an Earth-environment equivalence experiment using a 60% transmittance filter verifies the design’s robustness. This work confirms the feasibility of star-sensor-based attitude determination on Mars.

Inspired by the otolith structure in biology, an ultraminiature otolith-inspired microcantilever sensor was presented and applied to micro-vibration sensing and vibration mode analysis. By imitating the grafting and transfer processes in phytology, the fabrication method proposed for the first time, to our knowledge, can realize a micron-scale functionalized cantilever using only a common ultraviolet laser. The sensor is tiny in size and has an ultrahigh sensitivity of 7123 mV/g. Due to its outstanding performance, this structure enables the perception of minute vibrations and detailed recognition. Combining the recognition ability of the sensor with the simplest actions of tapping and patting, an ordinary desktop can be transformed into a Morse code transmitter that is workable on the entire area. Furthermore, by integrating the sensor with time-frequency analysis and machine learning techniques, it is possible to classify and identify multiple vibration modes occurring on the ground. The ultraminiature vibration sensor, as well as the fabrication method, provides a novel solution for low-cost and high-performance optical micro-nano probes. (c) 2026 Chinese Laser Press

3D object detection is a critical technology for the advancement of autonomous vehicles, as it plays a fundamental role in ensuring reliable and safe navigation. The integration of multi-modal sensor fusion, specifically combining LiDAR and camera data, significantly improves both detection accuracy and reliability. Nevertheless, challenges such as feature scale mismatch, loss of local details, and extensive computational demands continue to persist. This paper presents the FP-RCNN framework, which utilizes Feature Fusion Pyramid Attention strategy to tackle these challenges effectively. By generating multi-scale feature maps from the backbone network and applying a dual self-attention mechanism, FP-RCNN adeptly aligns image and point cloud data across varying resolutions, filtering out irrelevant features and emphasizing crucial ones. Furthermore, to address issues related to data absence and occlusion, we enhance the point cloud segmentation process during the 3D instance detection stage. Local and global features are merged utilizing shared Multi-Layer Perceptron and an advanced aggregation method known as NetVLAD, which not only improves semantic representation but also reduces sensitivity to feature irregularities. Experiments conducted on the KITTI dataset reveal that FP-RCNN substantially boosts detection accuracy, especially in challenging scenarios. Our findings demonstrate an Average Precision of 78.07% in the difficult category for cars, and 50.09% and 45.77% in the medium and difficult categories for pedestrians, respectively; for cyclists, the model achieves the accuracy of 60.91% in the difficult category. These results surpass those of selected comparative algorithms. This research significantly contributes to the advancement of 3D object detection technology within autonomous driving, offering a robust solution for navigating complex environments.

Accurate micro-force detection is essential for micro-operation and biomedical applications. However, conventional techniques, such as atomic force microscopy (AFM), are bulky and spatially demanding. In this study, we presented micro-spring probes fabricated by femtosecond-laser two-photon polymerization for flexible and precise mechanical sensing. The mechanical stiffness of the probes was tuned by adjusting the spring-wire diameter, with diameters of 2 u00B5m, 2.5 u00B5m, and 3 u00B5m, exhibiting the corresponding micro-force sensitivity of u22120.28 nm/u00B5N, u22120.09 nm/u00B5N, and u22120.07 nm/u00B5N, respectively. The probes enabled accurate measurement of Youngu2019s modulus of the polydimethylsiloxane (PDMS), exhibiting excellent consistency with AFM results. We further proposed a fiber-based self-referenced displacement detection scheme that integrated sensing and reference springs, allowing Youngu2019s modulus characterization without the need for a piezoelectric stage. With the growing development of the u201Clab-on-fiberu201D paradigm, the proposed micro-spring probe provides a promising platform for high-accuracy and compact micro-force sensing.

ABSTRACT Optical skyrmions, as topologically structured optical fields, hold significant potential for advanced optical tweezers and high‐density optical storage. However, most recently proposed schemes for the generation of optical skyrmion are diffraction‐limited or confined in near field, with bulky optical setups. Here, we propose and experimentally realize subwavelength optical skyrmions confined within a 3D dark focus. Our approach is built upon a chip‐scale spin‐decoupled metasurface, enabling simultaneous generation and manipulation of skyrmions. Through superoscillation in the focal region, this metasurface facilitates the formation of optical skyrmions beyond the diffraction limit. The skyrmion size reaches 0.06 λ in theory and 0.15 λ for experimental verification, both significantly smaller than the diffraction limit in this situation. Remarkably, these subwavelength skyrmions are generated approximately 1000 λ from the metasurface plane, under a relatively low numerical aperture of 0.3, underscoring the practicality of our metasurface‐based strategy. Moreover, the skyrmion size can be flexibly tuned by modulating the intensity profile of the incident wave, greatly enhancing the adaptability of the scheme for practical applications. This work establishes a feasible platform for embedding ultra‐compact optical skyrmions into 3D dark traps and opens new avenues toward multifunctional cold‐atom manipulation and topological optical information technologies.

All-dielectric metasurfaces that support resonators with multiple dipole resonances offer a versatile and effective approach for controlling electromagnetic fields at the nanoscale. Toroidal dipoles, as a peculiar component of multipole resonances, have attracted significant attention due to their unique ability to facilitate light-matter interactions, enabling enhanced absorption, amplified nonlinear responses, advanced data processing and storage capabilities, and highly sensitive sensing applications. The active control of toroidal dipoles in all-dielectric metasurfaces holds great significance for the development of adaptable photonic devices. However, achieving tunable toroidal dipoles in all-dielectric metasurfaces remains challenging due to the volumetric mode property of toroidal dipoles. To address this challenge, we propose a solution by integrating a thin layer of indium tin oxide (ITO) into the all-dielectric metasurfaces. Active modulation of the transmittance amplitude in toroidal dipole-based all-dielectric metasurfaces is achieved by precisely tuning the coupling between the metasurface modes and the ITO thin film. Furthermore, a tunable diffractive grating has been developed based on these findings, demonstrating its prospects for beam direction control applications.

Optical encryption has attracted considerable attention in information security due to its low loss and parallel processing capability. Perfect vector vortex beams (PVVBs), featuring a topological charge (TC)-independent ring radius and non-uniform polarization, offer rich degrees of freedom for encoding information, making them ideal encryption carriers. However, simultaneously controlling beam shape, multiple TCs, initial phase, and wavelength remains a fundamental challenge, which limits the capacity and security of PVVB-based encryption. We propose a single dielectric metasurface to generate multiple high-dimensional grafted perfect vector vortex beams (GPVVBs), enabling large-capacity and high-security optical encryption. By employing a spin-multiplexed and grafted strategy, a GPVVB with predesigned ellipticity and spatially variant polarization distribution is realized, and its polarization distribution can be dynamically tuned by rotating the incident linear polarization angle. Furthermore, by incorporating spatial and wavelength multiplexing, multiple multispectral GPVVBs with distinct polarization distributions are generated at customized positions, substantially expanding the encoding spaces and enhancing key complexity. This work provides a promising strategy for miniaturized, high-security optical encryption and holds significant potential for the applications of high-dimensional GPVVBs in information storage, optical communications, and anti-counterfeiting.

The integration of low-loss phase-change materials (PCMs) with silicon photonics has attracted increasing interest for reconfigurable and non-volatile photonic devices. However, most existing approaches selectively pattern PCM regions through additional lithography and lift-off processes. Here, we propose a whole-circuit integration approach to simplify the fabrication of PCM-based silicon photonic devices, where a PCM thin film is deposited on a silicon-on-insulator (SOI) wafer and co-etched with a silicon layer. Using low-loss Sb2Se3, the additional propagation loss is limited to ∼0.78 dB/mm. Meanwhile, whole-circuit integration provides a modulation region far exceeding selective integration. A full 2π multi-level phase modulation is experimentally demonstrated in an unbalanced Mach-Zehnder interferometer (UMZI). These results establish whole-circuit PCM integration as a scalable and foundry-compatible route toward programmable silicon photonic circuits.

Abstract Sb 2 Se 3 has established itself as a leading phase-change material for reconfigurable photonics, exhibiting broadband transparency and reversible optical transitions. However, previous metasurface implementations remain constrained by offline thermal or optical control mechanisms. We develop an electrically driven platform featuring monolithic integration of Sb 2 Se 3 nanostructures with addressable microheater arrays, achieving localized phase transitions at microsecond timescales. This hybrid architecture enables selective excitation of distinct coupled resonant modes in the near-infrared spectrum, delivering electrically controlled amplitude modulation exceeding 80 percent and phase modulation approaching 2π coverage. Advancing beyond unit-cell demonstrations, we implement a 6 × 6 electrically addressable metasurface array that yields an efficient spectral transmission matrix. Here, we show that integrating this system with neural-network-assisted computational methodologies achieves high-precision spectral reconstruction across a 500-nanometer short-wave infrared bandwidth, establishing a robust framework for computational spectroscopy and intelligent sensing in reconfigurable photonics.

Dimethyl carbonate (DMC) plays an important role in the electrolyte of new energy, so the importance of DMC gas detection is self-evident. This article reports a fiber-optic DMC gas sensor based on metal organic frameworks (MOFs). The fiber structure consists of two sections of four mode optical fiber (FMF) and one section of splicing misaligned centerless optical fiber (COF). The sensing arm of Mach-zehnder interferometer (MZI) is constructed by filling Cu-MOFs/polyacrylonitrile (PAN) by electrospinning in the misalignment gap between COF and FMF, with the COF serving as the reference arm. At the same time, surface plasmon resonance (SPR) is excited by depositing a Ag/zinc oxide (ZnO) film on the other side of the COF, and PAN is deposited as a temperature sensitive layer by electrospinning. The research results indicate that the DMC sensitivity of MZI is 93.25 pm/ ppm, and the temperature sensitivity of SPR is 596.2 pm/degrees C. The DMC gas concentration and temperature errors based on dual parameter matrix demodulation are E (Delta C) = 1.38 % and E (Delta T) = 1.80 % respectively. Finally, the changes of DMC gas concentration and temperature during the leakage of lithium battery are successfully analyzed. This sensor has the advantages of simple manufacturing, small size, and high sensitivity, and has potential application prospects in terms of lithium battery monitoring.

Recently, chiral optical fields (COFs) have garnered significant attention due to their multiple controllable degrees of freedom (DOFs), enabling applications in diverse areas such as optical tweezers, manufacturing, and holographic encryption. However, existing schemes fail to achieve precise control over certain aspects, particularly the fine‐tuning of sidelobes, including their overall shape and structural characteristics, which limits their practical applications. Herein, an approach to achieve fine sculpting of COFs using a modular multilayer annular phase plate (MMAPP) is proposed. By adjusting the number and mode of the annular spiral phase in the two modules of the MMAPP, as well as the axicon phase, the high‐order cross phase, and the low‐order cross phase, the experimental manipulation of nine DOFs of COFs is demonstrated, including the chirality, size, sidelobe number, sidelobe distortion degree, sidelobe segment length, sidelobe segment rotation direction, overall polygonal shape, ellipticity, and rotation angle. The proposed method enhances the modulation capabilities of COFs and gives rise to potential applications in particle manipulation and information encryption.

Metasurface, spatiotemporal metasurface and tunable metasurface have attracted more and more attention in the fields of communication, polarization image and terahertz detection. Numerous tunable metasurfaces have been proposed recently for generating a diverse array of applications. However, many proposed tunable metasurfaces have been limited to a single state switching function, which has severely hampered their development. In this paper, eight meta-atoms were designed by employing the InSb, that dielectric constant could be modulated by temperature. Consequently, the phase delay of the meta-atoms is various under different temperature. Thus, two thermally tunable metasurfaces were proposed. At 260 K, the InSb-based dynamic vortex beam generator can generate a vortex beam with topological charge of l = 1 at 2.25 THz. Correspondingly, at 180 K, the metasurface exhibits a phase coverage of 4 pi and the topological charge of generated beam is converted from l = 1 to l = 2. With decreasing of the temperature, the phase gradient between the proposed meta-atom increases. Therefore, a phase gradient metasurface was proposed to generate light beam with adjustable direction. This proposed metasurface could provide a valuable scheme for developing tunable devices and terahertz metasurfaces.