Abstract High power laser processing is a cornerstone of modern manufacturing and advanced scientific research, where parallelization is critical for increasing throughput and enabling new capabilities. However, simultaneously achieving high power handling, multi-beam parallelization, and compact optical integration remains a long standing challenge. This trade-off is addressed by establishing a wafer scale silicon carbide (SiC) meta-optic platform that combines three key elements: the thermomechanical robustness of SiC, an aperture matched architecture for dense multi-beam integration, and a scalable semiconductor compatible manufacturing process. The platform integrates a SiC diffractive optical element (DOE) with a wafer scale 26 element SiC metalens array. The array consists of metalenses with a diameter of 1.15 cm and a numerical aperture (NA) of 0.5 and exhibits high uniformity across the array, with a 3σ focal length deviation of 10.73 μm, corresponding to 0.11% of the designed focal length, and an average image normalized focusing efficiency of 93.90%. This architecture converts a large aperture input beam into multiple focusing channels compatible with millimeter to centimeter scale industrial laser beams, with 4, 9, and 26 beam configurations fabricated and characterized. As the front end component exposed to the full incident beam before power division, the SiC DOE experiences the most stringent power handling condition in the architecture and exhibits a continuous wave laser induced damage threshold of 6.11 kW/cm 2 , confirming its suitability for high power beam splitting. These results establish a SiC meta-optic platform for robust high power multi-beam photonic systems and suggest a scalable route toward next generation parallel laser processing.
Abstract Femtosecond laser processing provides a versatile route for high-precision material structuring, yet the dynamic modulation of nanoscale feature sizes remains constrained by the far-field diffraction limit and insufficient control over local energy deposition. To overcome these limitations, we report a nanofabrication method based on binary pupil phase modulation of a radially polarized beam, introducing optical-field structural parameters as an active dimension for tuning the feature size of laser-processed structures. This method enables controllable modulation of single-pulse laser-processed features over a range of approximately 50–250 nm. Under an optimized phase state and energy window, this method further pushes the lower limit of size modulation to 14 nm (λ/57). Mechanistic analysis shows that this limiting size arises from the combined action of focal-field reshaping, tunnelling-ionization-dominated nonlinear material response, and threshold selection. We further extend this method to large-area nanopatterning, where the fabricated nanohole arrays and nanogratings can extend to the centimeter scale while maintaining feature sizes at the hundred-nanometer level. This method enables controllable modulation of laser-processed structures at the nanoscale and offers a route for next-generation programmable and highly controllable nanomanufacturing platforms.
Abstract Metasurfaces are progressively reshaping traditional optical paradigms and pushing the boundaries in complex applications where compact designs are essential. However, the design of metasurfaces demands substantial computational resources to numerically solve Maxwell's equations—particularly for large-scale photonic systems. Conventional forward design using electromagnetic solvers is based on specific approximations that may not effectively address complex problems. In contrast, existing inverse design methods are a stepwise process that is often time-consuming. Here, we overcome these challenges by presenting MetaE-former architecture, a Neighborhood Attention Transformer-based transfer learning framework that enables customized surrogate solver development through fine-tuning of pre-trained neural networks with only thousands of data, facilitating highly efficient task-adaptable inverse design of metasurfaces. Moreover, this method achieves global solutions for hundreds of nanostructures simultaneously, providing up to a 250,000-fold speedup during the optimization stage compared with solving for individual meta-atoms based on the FDTD method. As examples, we demonstrate a binarized high-numerical-aperture (~ 1.31) metalens and several optimized structured-light meta-generators. Our method significantly improves the beam shaping adaptability with metasurfaces and paves the way for quick design of large-scale metadevices with high accuracy.
Abstract Fluorescence microscopy has emerged as an indispensable tool in neuroscience research, enabling subcellular-resolution imaging of neural circuits and monitoring of brain function in vivo. However, physiological motion artifacts which include both rigid displacements and non-rigid deformations induced by respiration and tissue dynamics, significantly compromise imaging fidelity and quantitative analysis reliability. In this paper, we propose DeepMoCo, an innovative deep learning-based framework that integrates a graph neural network (GNN) to establish spatiotemporal feature correlations between biological motion patterns and imaging artifacts. This architecture enables rapid tracking and adaptive correction of complex motion patterns, achieving 95% improvement in correlation coefficients and 77% enhancement in peak signal-to-noise ratio compared to existing methods. The proposed method presents a robust and universal solution for high-precision in vivo fluorescence imaging studies.
Abstract Inspired by the key elements and principles in the brain, photonic neuromorphic computing shows great potential for building next-generation intelligent processing systems with high parallelism, low latency, low power consumption, and self-learning capabilities. While summarizing significant advances in this field, this review offers insights into how photonic neuromorphic computing systems support next-generation intelligent information processing. Specifically, we discuss emerging materials and devices, which support more compact integration of efficient physical architectures. Architectures such as photonic spiking neural networks and reservoir computing, together with associated learning paradigms, show a trend of collaborative development. Then we present promising applications, where neuromorphic computing systems provide broadband and multi-domain perception and processing. Finally, we discuss key challenges and future directions. This review aims to offer a clear and comprehensive overview for researchers across a broad community. We hope to present these insights in a “neuromorphic” manner, to reveal why learning from the brain has become increasingly important, especially for overcoming the efficiency bottleneck in traditional von Neumann architecture and data-intensive artificial neural networks.
