中国半导体行业协会封装分会会刊

中国电子学会电子制造与封装技术分会会刊

导航

电子与封装

• 综述 •    下一篇

2.5D封装关键技术的研究进展

马千里1,2,3,马永辉1,钟诚2,李晓2,廉重2,3,刘志权3   

  1. 1. 哈尔滨工程大学烟台研究院,山东 烟台  264006;2. 深圳先进电子材料国际创新研究院,广东 深圳  518100;3. 中国科学院深圳先进技术研究院,广东 深圳  518055
  • 收稿日期:2025-02-24 修回日期:2025-03-12 出版日期:2025-04-02 发布日期:2025-04-02
  • 通讯作者: 马永辉

Research Progress on key Technologies of 2.5D Packaging

MA Qianli1,2,3, MA Yonghui1, ZHONG Cheng2, LI Xiao2, LIAN Zhong2,3, LIU Zhiquan3   

  1. 1. Yantai Research Institute of Harbin Engineering University, Yantai 264006, China; 2. Shenzhen Institute of Advanced Electronic Materials, Shenzhen 518103, China; 3. Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences, Shenzhen 518055, China
  • Received:2025-02-24 Revised:2025-03-12 Online:2025-04-02 Published:2025-04-02

摘要: 随着摩尔定律在晶体管微缩领域逼近物理极限,先进封装技术通过系统微型化与异构集成,成为突破芯片性能瓶颈的关键路径。作为先进封装的核心分支,2.5D封装通过硅/玻璃中介层实现高密度互连与多芯片异构集成,兼具高带宽、低延迟和小型化优势,广泛应用于人工智能、高性能计算及移动电子领域。本文系统阐述了2.5D封装的核心结构(如CoWoS、EMIB和I-Cube)及其技术特征,重点剖析了小芯片(Chiplet)模块化设计、硅通孔(TSV)工艺优化、微凸点可靠性提升、铜-铜直接键合界面工程以及再布线层(RDL)多物理场协同设计等关键技术的最新进展。未来研究需聚焦低成本玻璃基板、原子层沉积技术抑制界面氧化以及多物理场协同设计等方面,以突破良率和散热瓶颈,推动2.5D封装在后摩尔时代高算力场景中的广泛应用。

关键词: 2.5D封装, 再布线层, 微凸点, 硅通孔, 铜铜直接键合

Abstract: As Moore’s law approaches its physical limits in transistor scaling, advanced packaging technologies have emerged as a pivotal pathway to overcome performance bottlenecks through system miniaturization and heterogeneous integration. As a critical branch of advanced packaging, 2.5D packaging enables high-density interconnects and multi-chip heterogeneous integration via silicon/glass interposers, offering advantages in high bandwidth, low latency, and compact form factors, which are widely adopted in artificial intelligence (AI), high-performance computing (HPC), and mobile electronics. This paper systematically elaborates the core architectures of 2.5D packaging (e.g., CoWoS, EMIB, and I-Cube) and their technical characteristics, with a focused analysis of the latest advancements in key technologies. These include modular design of chiplets, process optimization of through-silicon vias (TSVs), reliability enhancement of microbumps, interface engineering of Cu-Cu direct bonding, and multi-physics co-design of redistribution layers (RDLs). Future researches should focus on low-cost glass substrates, atomic layer deposition (ALD) enabled interface engineering, and multi-physics co-design methodologies to overcome yield and thermal management bottlenecks, thereby accelerating the adoption of 2.5D packaging in high-performance computing scenarios within the post-Moore era.

Key words: 2.5D packaging, redistribution layer, micro bump, through silicon via, Cu-Cu direct bonding