【摘要】 随着深空望远镜、先进红外成像系统、极紫外光刻机等高端装备在航空航天、国防军工与先进民用领域对核心光学系统性能需求的日益提高，决定光学元件面形精度、表面粗糙度与亚表面损伤深度等关键指标的超精密加工技术正面临严峻挑战。超精密磨削作为光学元件加工链中承上启下的关键工序，是实现高效率、低损伤、高精度加工的重要途径。近年来，硬脆光学材料磨削的延性-脆性转变机理、磨削质量预测、关键工艺技术以及超精密磨削机床设计均取得了快速发展。文中从多尺度材料去除机理入手，分析了原子级键断裂、微纳尺度位错滑移、高压相变与裂纹演化等行为对延性去除与亚表面损伤的影响机制；归纳了表面粗糙度与亚表面损伤的预测模型及质量控制方法；梳理了砂轮设计与修整、磨削参数优化、超声振动和激光辅助等关键加工技术在提升延性去除能力与降低损伤方面的优势与应用进展；探讨了高刚性、高静动态特性的机床结构、多种静压支承技术及其在超精密磨削中的性能提升作用。然而，高端光学元件超精密磨削技术在去除机理、质量预测及工艺装备等方面仍面临模型统一性与广泛性不足、过程监测与感知能力有限、工艺参数难以精准调控以及机床多源误差难以全面补偿等重要挑战，文中针对以上研究挑战进行初步总结，并对从经验磨削到机理驱动磨削、智能化技术辅助磨削等方向进行了展望，以期为高端光学元件超精密磨削技术发展提供指导。更多还原

【Abstract】 Significance Ultra-precision grinding is a key enabling technology for manufacturing high-end optical components used in extreme ultraviolet lithography, laser fusion, infrared imaging, and space optics. As optical systems evolve toward larger apertures, higher numerical apertures, and wider spectral bands, the performance margins allowed for surface figure error, surface roughness, and subsurface damage become increasingly stringent. In hard and brittle optical materials such as fused silica, sapphire, silicon carbide, and silicon, the machining window for stable ductile-regime removal is narrow, and the coexistence of plastic flow, phase transformation, and brittle fracture makes process optimization nontrivial. Ultra-precision grinding plays a critical bridging role between coarse shaping and ultra-precision polishing: it is expected to provide high material removal rates while delivering a surface and subsurface state that is “polishing-friendly”, thereby shortening the overall manufacturing chain and improving yield. Therefore, systematically summarizing material removal mechanisms, quality prediction and control, key process technologies, and ultra-precision machine tools is essential for guiding both academic research and industrial deployment of high-efficiency, low-damage optical manufacturing.Progress Recent advances can be summarized from four tightly coupled aspects: mechanism, quality modeling,process technology, and machine tool capability. Multi-scale material removal mechanisms. At the atomic scale, molecular dynamics studies reveal that material removal initiates through collective atomic bond breaking accompanied by severe local shear and layered deformation; chip formation is governed by the sequence of atomic accumulation, energy build-up, bond rupture,and material separation(Fig.3). At the micro/nano scale, dislocation slip, twinning, and pressure-induced phase transformation can accommodate plasticity and promote ductile removal, whereas crack initiation and propagation dominate when the local stress intensity exceeds the fracture resistance. The coupling between abrasive geometry,undeformed chip thickness, strain-rate effects, and thermal-mechanical fields determines the ductile–brittle transition and the evolution of subsurface damage.Grinding Quality Prediction and Evaluation Surface roughness and subsurface damage are the two most critical quality indicators in ultra-precision grinding. For roughness, models have evolved from purely geometric formulations toward hybrid frameworks that incorporate probabilistic grain engagement, elastic recovery, and parameter distributions. Representative theoretical and semi-empirical roughness models, together with their applicability and limitations, are compared to support model selection and engineering use(Tab.2). Meanwhile,data-driven approaches are increasingly used to map process parameters and sensing signals to roughness outcomes, enabling rapid prediction and online monitoring; typical convolutional neural network architectures for parameter–topography mapping and vibration–roughness mapping demonstrate how learning-based models can enhance robustness under complex operating conditions(Fig.6). For subsurface damage, prediction frameworks based on indentation fracture mechanics, crack evolution, and combined mechanism descriptions have been established for different materials and loading conditions. These models relate damage depth to crack systems(median/radial and lateral cracks), surface state, and dynamic effects such as spindle vibration, providing multiple pathways for damage estimation and control(Tab.3). Key process technologies for high efficiency and low damage. Progress has been made in grinding wheel design, precision dressing/truing, and the integration of auxiliary energy fields. Structured and engineered wheels can tailor local chip thickness and improve coolant access, while advanced dressing methods(e.g., laser-or electrical-discharge-related approaches) enhance grain protrusion uniformity and shape accuracy, which are crucial for deterministic optical grinding. In addition, ultrasonic vibration-assisted grinding and laser-assisted grinding have become important routes to enlarge the ductile machining window, reduce grinding forces, and suppress crack formation by modifying contact mechanics and near-surface material response. These technologies collectively aim to improve the “damage-to-removal-rate” tradeoff, especially for high-hardness and high-brittleness substrates. Ultra-precision grinding machines and system-level capability. The achievable surface integrity is ultimately constrained by machine tool motion accuracy, dynamic stiffness, and thermal stability. Recent machines emphasize high-rigidity structural layouts(e.g., T-frame and gantry-type configurations) to enhance dynamic performance and error resistance(Fig.12). Hydrostatic guideways and hydrostatic spindles provide low friction,high damping, and high stiffness, enabling smoother motion and better nano-scale positioning under varying loads. A comparison of key parameters across representative domestic and international ultra-precision grinding machines highlights trends in granite-based beds, multi-axis linear motor drives, and high-end numerical control systems, reflecting the shift toward higher bandwidth and higher stability platforms for optical fabrication(Tab.6).Conclusions and Prospects Substantial progress has been achieved in clarifying the multi-scale removal physics, improving roughness and subsurface damage prediction, developing auxiliary-field-assisted processes,and upgrading machine tool structures and hydrostatic functional units. However, challenges remain in establishing unified predictive models that bridge atomic-to-macro scales, achieving robust controllability across materials and tool states, and compensating thermo-mechanical coupling errors under long-duration, high-stability operation. Future research is expected to focus on mechanism-driven, data-enhanced grinding models; intelligent process regulation through multi-sensor fusion and digital twins; advanced wheel design and high-consistency dressing strategies; and next-generation intelligent ultra-precision grinding machines with stronger thermal management and error compensation. These directions are anticipated to support higher efficiency, lower damage,and higher determinism in the manufacturing of high-end optical components.更多还原