Search Ctrl+K
Change language Switch ThemeSign In
[](https://www.bestblogs.dev/en/explore/brief "Daily Brief")[](https://www.bestblogs.dev/en/explore/newsletter "Weekly Picks")[](https://www.bestblogs.dev/en/explore/topics "Topics")
[](https://www.bestblogs.dev/en/settings "Settings")[](https://www.bestblogs.dev/en/docs "Help Center")
Narrow Mode
82
下一代 EUV 光刻机，巨头有分歧
The article compares Intel and TSMC's different paths in high-NA EUV lithography technology, analyzing each company's technology choices, cost risks, and strategic considerations. Follow·
Today·2011 words (about 9 min)
AI Summary & Key Points
Summary
The article points out that both Intel and TSMC pursue smaller, faster, and more energy-efficient semiconductors, but they adopt different strategies in advanced lithography: Intel is the first to introduce ASML's high-numerical-aperture (High-NA) EUV system, accumulating technical experience early and hoping to simplify critical-layer processes through single exposure; TSMC maintains the traditional 0.33 NA EUV platform, enhancing performance through innovations in photoresist, masks, overlay precision, computational lithography, nanosheet transistors, and backside power delivery, and will consider High-NA volume production only when its economic benefits become clearer. The article also notes that High-NA brings about a resolution improvement of roughly 1.7×, while facing challenges such as cost, field-of-view limitations, depth of focus, photoresist performance, masks, inspection, metrology, and yield, as well as the companies' collaboration at the R&D level. Overall, it presents the trade-offs between the two technology roadmaps and the possibility of future convergence.
Main Points
* 1. Intel adopts High-NA EUV early to gain a technological edge and simplify critical-layer processes.
Intel is the first to introduce ASML's High-NA system, planning to use it in the Intel 14A process, hoping to reduce mask count via single exposure, shorten manufacturing cycle, and lower alignment errors, thereby obtaining a differentiation advantage in the foundry business.
* 2. TSMC takes a cautious strategy, continuing to optimize traditional EUV while combining other innovations to boost performance.
TSMC believes the existing 0.33 NA EUV platform is sufficient for its A16/A14 products, and therefore enhances performance through photoresist, masks, overlay precision, computational lithography, nanosheet transistors, and backside power delivery, and will consider High-NA volume production only when its economic benefits become clearer.
* 3. While High-NA technology improves resolution by about 1.7×, it comes with high cost, halved field of view, and multiple technical challenges.
The High-NA system's cost is significantly higher than that of traditional EUV, with an exposure field only half as large, which may complicate manufacturing of large processors and AI accelerators; additionally, issues such as depth of focus, photoresist performance, masks, inspection, metrology, and yield need to be addressed.
Sign in to highlight text and take notes as you read. Sign in now
半导体行业观察 2026-08-08 10:01 安徽
半导体行业观察：英特尔和台积电的目标一致——制造更小、更快、更节能的半导体——但它们在先进光刻技术方面采取了不同的策略。
公众号记得加星标⭐️，第一时间看推送不会错过。
英特尔和台积电的目标一致——制造更小、更快、更节能的半导体——但它们在先进光刻技术方面采取了不同的策略。英特尔率先开发了高数值孔径极紫外光刻技术（High-NA EUV），而台积电则继续在其最新的生产工艺中扩展传统的极紫外光刻技术。它们的选择反映了各自不同的技术路线图、制造优先级以及对成本和风险的评估。
传统极紫外光刻技术使用波长为13.5纳米的光源和数值孔径为0.33的投影光学系统。由于其能够打印比早期深紫外系统更小的结构，因此已成为制造先进逻辑芯片的关键技术。高数值孔径极紫外光刻技术使用相同的波长，但将数值孔径提高到0.55。这使得分辨率提高了约1.7倍，并可能使制造商能够通过一次曝光打印一些关键图案，而无需使用多次曝光步骤。
英特尔成为首家引进ASML商用高数值孔径（High-NA）开发系统TWINSCAN EXE:5000的芯片制造商。该系统安装在英特尔位于俄勒冈州希尔斯伯勒的研发中心，用于开发未来制造技术的工艺、材料、掩模和设计规则。继Intel 18A工艺之后，英特尔计划在其Intel 14A工艺中引入高数值孔径极紫外（High-NA EUV）技术，同时在经济效益更佳的情况下继续采用传统的极紫外光刻（EUV）和其他光刻方法。
这项早期投入将助力英特尔重夺半导体工艺领先地位并拓展其代工业务。高数值孔径极紫外光刻技术（High-NA EUV）使英特尔有机会在该技术广泛应用之前积累专业知识，同时有望简化其关键芯片层的生产流程。用单次曝光取代多次曝光序列，可以减少掩模数量和加工步骤，缩短制造周期，并减少因对准多个图案而导致的误差。因此，早期采用这项技术既能为英特尔晶圆代工带来技术优势，也能为其打造重要的差异化优势。
台积电采取了更为谨慎的策略。该公司认为，无需立即将高数值孔径极紫外光刻技术（High-NA EUV）投入量产，即可生产A16和A14系列产品。因此，台积电持续改进其成熟的0.33数值孔径极紫外光刻平台，通过改进光刻掩模、光刻胶、套刻精度、计算光刻、工艺优化以及设计技术协同优化等手段来实现。此外，纳米片晶体管、背面供电以及更灵活的标准单元架构等创新技术也带来了性能和密度的提升，而这些提升并不完全依赖于光刻分辨率。
经济因素是台积电决策的核心。高数值孔径（High-NA）系统比传统的极紫外（EUV）扫描仪成本高得多，而且需要一套全新的配套生态系统。此外，其变形光学系统产生的曝光场只有传统EUV曝光场的一半。这种限制可能会使大型处理器和人工智能加速器的制造变得复杂，甚至可能需要将两个图案拼接在一起。高数值孔径系统在景深、光刻胶性能、掩模、检测、计量和良率等方面也面临着诸多挑战。
台积电大规模运营传统极紫外光刻（EUV）设备，并积累了丰富的经验，最大限度地提高了生产效率和可靠性。继续使用这套成熟的基础设施可以降低生产风险，并使公司能够从现有设备中获得更高的收益。
对于服务众多客户的超大批量代工厂而言，在经济效益明确之前就引入最高分辨率的工具，可能比采用成熟可靠、良率稳定的工艺更有价值。
这并不意味着台积电放弃了高数值孔径极紫外光刻技术。该公司已购置相关设备用于研发，并已开始为未来的工艺开发高数值孔径光刻技术。台积电是ASML最大的客户，台积电CEO魏振昌也曾多次表示，两家公司正在高数值孔径极紫外光刻技术方面紧密合作。
台积电表示，是否采用这项技术将取决于可衡量的制造效益、技术成熟度和成本。英特尔也同样不会用高数值孔径光刻技术（High-NA）取代所有传统的极紫外光刻技术（EUV）；它将选择性地在内部产品中分辨率能够带来足够价值的层上使用这项新技术。
归根结底：因此，二者的主要区别在于时机。英特尔愿意承担早期采用者的成本和风险，以换取更早的经验积累和可能的工艺领先地位。台积电则在等待高数值孔径极光刻技术展现出更强的生产经济性的同时，也在扩展其成熟的技术。两种策略最终都可能实现高数值孔径制造，但它们代表了通往下一代半导体微缩技术的不同路径。
（来源：编译自 semiwiki）
*免责声明：本文由作者原创。文章内容系作者个人观点，半导体行业观察转载仅为了传达一种不同的观点，不代表半导体行业观察对该观点赞同或支持，如果有任何异议，欢迎联系半导体行业观察。
_END_ 今天是《半导体行业观察》为您分享的第 4492 内容，欢迎关注。 推荐阅读
加星标⭐️第一时间看推送
求点赞
求分享
求推荐
Key Quotes
> Intel pioneered the development of high-numerical-aperture extreme ultraviolet lithography (High-NA EUV), while TSMC continues to extend traditional extreme ultraviolet lithography in its latest production processes.
> High-numerical-aperture extreme ultraviolet lithography (High-NA EUV) gives Intel the opportunity to accumulate expertise before the technology sees widespread adoption, and is expected to simplify the production flow of its critical chip layers. Replacing multiple exposure sequences with a single exposure can reduce mask count and processing steps, shorten the manufacturing cycle, and mitigate errors arising from aligning multiple patterns.
Tags
Semiconductor
EUV lithography
High-NA
Foundry
Manufacturing process
Related Articles
* Nvidia Buys 5 Million Optical Modules a Year, Then Dismantles Them Itself * Intel: How Far from a Trillion-Dollar Market Cap? * The next bottleneck for AI may not be GPU and storage, but semiconductor equipment. * Broadcom Told the Truth, But No One Noticed * The Missing Link in Co-Packaged Optics, emphasizing that merely improving photonic device performance cannot achieve scale; instead, package PDK must enable multi-domain system-level integration—covering optoelectrical, thermal, mechanical, and manufacturing aspects—and ensure manufacturability.") * Advanced Packaging: Entering a New Battlefield, exploring its advantages in addressing cost and size bottlenecks, TSMC and domestic vendors' layout, and the process challenges faced.") * The Rise of SK Hynix: An Absurd, Melodramatic Chaebol Tale — How It Strangled Global AI's Neck
Make your daily reading actually fit you.A daily brief built from the sources you follow. Get started free HomeDiscoverSettings