展期论坛 | 晶圆制造智能化技术论坛，议程及嘉宾阵容抢先看

本报告聚焦AI视觉技术在晶圆检测中的应用与最新进展，重点围绕晶圆表面缺陷检测和内部缺陷检测展开。表面检测主要介绍微小缺陷、低对比度缺陷及复杂背景下的识别、定位和在线检测技术，以及针对未知缺陷的检测思路；内部检测主要介绍X射线、声学等无损成像技术，以及与智能分析相结合的内部缺陷识别和评估方法。在此基础上，进一步介绍多模态检测、视觉大模型等新技术在晶圆缺陷检测中的应用思路，并结合高良率和零缺陷制造需求，讨论晶圆智能检测技术的发展方向。

This report focuses on the application and latest advances of AI-based vision technologies in wafer inspection, with particular emphasis on wafer surface defect detection and internal defect detection. For surface inspection, the report mainly introduces techniques for the recognition, localization, and online detection of tiny defects, low-contrast defects, and defects under complex backgrounds, as well as approaches for detecting unknown defect types. For internal inspection, it presents nondestructive imaging techniques such as X-ray and acoustic imaging, together with intelligent analysis methods for internal defect identification and evaluation. On this basis, the report further discusses emerging technologies, including multimodal inspection and vision-language/foundation models, and their potential applications in wafer defect detection. Finally, in the context of high-yield and zero-defect manufacturing, the report explores future development trends in intelligent wafer inspection technologies.

随着人工智能、机器学习和机器人技术等先进技术的快速发展，制造业正加速向智能化和数字化升级。作为全球领先的半导体设备供应商，泛林集团早已在智能化技术和解决方案的深耕布局。持续推进具有自我感知、自我维护、自我适应和自我工艺开发能力的设备智能（ Equipment Intelligence® ）的发展，使其成为驱动设备实时决策与自适应控制的关键力量。

本次报告将重点介绍 Lam 的一体化 Equipment Intelligence® 技术能力，阐述嵌入式模型、先进传感和数据驱动算法如何协同工作，加速制程开发、提升设备健康度，并改善整体生产力表现。同时，我们也将介绍全新的智能功能模块，展示它们如何作为Equipment Intelligence®的核心组成部分，共同开拓通往未来智能制造的发展道路。通过设备级洞察驱动工厂级优化的深度协同，Equipment Intelligence® 正在塑造更自主、高效、可扩展的新一代半导体制造范式。

With the rapid advancement of artificial intelligence, machine learning, and robotics, the semiconductor industry is accelerating its transition toward intelligent and digitalized manufacturing. As a global leader in semiconductor equipment, Lam Research has spent decades developing intelligent technologies and solutions. Equipped with selfaware, selfmaintained, selfadaptive, and selfdeveloping capabilities, Equipment Intelligence® plays a pivotal role in enabling real‑time decision‑making and adaptive control.

This presentation will introduce Lam’s integrated Equipment Intelligence® capabilities, highlighting how embedded models, advanced sensors, and data‑driven algorithms work together to accelerate process development, enhance tool health, and improve productivity performance. We will also highlight our new smart functions, showing how they serve as the core features of our Equipment Intelligence® for smart manufacturing and drive a unified path toward the Fab of the Future. By strengthening the synergy between equipment‑level insights and fab‑level optimization, Equipment Intelligence® is shaping a more autonomous, efficient, and scalable manufacturing paradigm for the next generation of semiconductor fabs.

在先进工艺生产中，AI已经在算法级别在很多制造类EDA软件中得到了广泛的应用。本文介绍了广立微在利用Agent AI 在工具级别打通制造类EDA，特别是良率提升上的工作。Agentic AI在建立优秀的Skill后，能够自行调用各种软件工具，利用相关数据和Foundry的知识积累，快速完成过去需要多部门协作的良率分析工作，极大提升工作效率。

Presentation Overview and Key Highlights: AI has been widely deployed at the algorithm level in manufacturing EDA for advanced processes. This work introduces Semitronix Agent AI solution for tool-level integration of manufacturing EDA targeting yield improvement. With well-trained skills, Agentic AI can autonomously call software tools, utilize data and foundry know-how, and accelerate  yield analysis which requires multiple teams cooperation in the past.

本报告通过介绍化合物半导体领域射频器件、光通讯及集成电路领域光掩模版等产业的主流工艺，分析了电子束曝光设备主流技术路线及特点、电子束曝光设备的国产化现状、主要技术难点，同时以关键技术难点作为切入点，提出了电科装备的解决方案。

最后，针对电子束曝光设备国产化需求，本报告介绍了电科装备近几年在电子束曝光设备领域所作出的工作，展示了电子束直写曝光设备、可变形电子束曝光设备、电子束量测设备的最新设备研制及工艺应用进展。

本次演讲聚焦新施诺第六代国产OHT天车系统，围绕12英寸晶圆厂对高效率、高稳定性及自主可控的搬运需求，介绍新一代OHT在控制架构、运动控制及整车性能方面的技术升级。第六代OHT采用自研车规级主控、深度定制操作系统及自主运动控制算法，实现软硬件协同优化，运动控制接口响应达到微秒级；同时通过机械结构与控制策略优化，提升夹爪、升降及分岔通行效率。演讲还将结合核心部件自主研发与本土供应链建设，分享国产OHT在12英寸晶圆厂规模化应用中的技术路径与实践。

This presentation focuses on NSYNU’s sixth-generation domestic OHT system for fully automated material handling in 12-inch wafer fabs. It will introduce key technical upgrades in control architecture, motion control, and overall vehicle performance, addressing the industry’s growing requirements for higher efficiency, operational stability, and technological autonomy. The sixth-generation OHT integrates a self-developed automotive-grade master controller, a deeply customized operating system, and proprietary motion-control algorithms, enabling microsecond-level response for motion-control interfaces. Through optimized mechanical design and control strategies, the system also improves gripper operation, hoisting efficiency, and branch-passing performance. The presentation will further discuss the development of self-controlled core components and the localization of the supply chain, sharing NSYNU’s technical approach and practical experience in the large-scale deployment of domestic OHT systems for 12-inch wafer fabs.

聚焦工业智能体在晶圆制造领域的创新应用，探讨 AI 如何从辅助分析走向自主感知、决策与执行，推动晶圆制造从数字化、智能化向自主智能演进。

Focus on the innovative application of Industrial AI Agents in wafer manufacturing, exploring how AI is evolving from assisted analysis toward autonomous perception, decision-making, and execution, driving wafer manufacturing from digitalization and intelligence toward autonomous operations.