颠簸
晶片切割
材料科学
倒装芯片
薄脆饼
胶粘剂
晶圆级封装
进程窗口
晶片测试
固化(化学)
光电子学
复合材料
机械工程
平版印刷术
图层(电子)
工程类
作者
Yuhu Xia,Biyun Ling,Xiaoyue Wang,Minli Cai,Biqing Zhou,Yaming Wu
出处
期刊:IEEE Transactions on Components, Packaging and Manufacturing Technology
[Institute of Electrical and Electronics Engineers]
日期:2024-02-06
卷期号:14 (3): 359-367
标识
DOI:10.1109/tcpmt.2024.3362882
摘要
Temporary protective packaging is a critical enabler to the development of micromirror array (MMA) module. Herein, aimed at the vertical integration of optical-window-free through-silicon-via-based (TSV-based) MMA and its corresponding application specific integrated circuits (ASICs), this paper proposes a temporary protective packaging method that is compatible with a series of back-end processing, including under-bumping-metallization (UBM), dicing, solder bumping and flip-chip bonding. An ultraviolet-curing (UV-curing) strippable adhesive with tolerance to water, organic solvent and high temperature is carefully selected and introduced into this temporary protective packaging method to bond caps to MMA wafer. A programmable dispenser is employed to inject the chosen adhesive into pre-fabricated trenches on MMA wafer, which enables patterning of this adhesive. In response to its contradiction between detachability and sufficient adhesion, a double-level temporary protective packaging is proposed, which is composed of wafer-level cap and die-level cap. Wafer-level cap not only focuses on adhesion to guarantee MMA wafer being sealed in UBM fabrication and to keep die-level cap free of mechanical influence before it comes into use, but also guarantees wafer profile quality in specific processing. Die-level cap focuses on its detachability, and comes into play after dicing to protect MMA in processing of solder bumping and flip-chip bonding. The effectiveness of the chosen adhesive is verified experimentally. Finally, an intact optical-window-free vertical integrated MMA module is achieved after all the back-end process and removal of die-level cap.
科研通智能强力驱动
Strongly Powered by AbleSci AI