Heterogeneous Integration Solutions for HPC Application by Using FO-MCM Chip Last Platform

P.J. Su, George Pan, Nistec Chang, Yu-Po Wang
{"title":"Heterogeneous Integration Solutions for HPC Application by Using FO-MCM Chip Last Platform","authors":"P.J. Su, George Pan, Nistec Chang, Yu-Po Wang","doi":"10.23919/IWLPC.2019.8914123","DOIUrl":null,"url":null,"abstract":"For years high performance computing (HPC) products have been integrating more and more functions in the IC. Large die size with high density transistor is still the trend for IC product design, especially networking and CPU product. As Moore's law nears its physical limits, split die and heterogeneous integration in package are the effective solution to increase gross die and wafer yield for cost saving. In order to fulfil electronical performance, line width/ space small than 5/5um is necessary for die to die high density IO interconnection. Currently, 2.5DIC package is one of the mature solutions for fine line interconnection in the industry. Besides, fan-out multi-chip module (FO-MCM) is an alternative technology with low cost benefit that uses redistribution layer (RDL) instead of silicon interposer. For different segment of applications, FO-MCM chip last and chip first platform are widely used in industry. Chip last process has lower thermal budget on die, no die loss due to know good RDL and shorter cycle time advantages, which is ideal for applications requires multi dies as well as fine RDL interconnections. However, it still has the warpage control challenge with bump non-wetting/ bridge concern, due to the mold grinding process on FO module with complex composition (chip with ubump, underfill and compound). In this study, we will address the package warpage performance and the relationship of FO final thickness by FO-MCM chip last process. Finally, we will demonstrated three layer RDL structure and minimum line width/ space down to 2/2um. Package level reliability qualification is also finished with temperature cycle test (TCT), unbiased highly accelerated stress test (uHAST) and high temperature storage life (HTSL) conditions.","PeriodicalId":373797,"journal":{"name":"2019 International Wafer Level Packaging Conference (IWLPC)","volume":"8 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2019-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"2019 International Wafer Level Packaging Conference (IWLPC)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.23919/IWLPC.2019.8914123","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 0

Abstract

For years high performance computing (HPC) products have been integrating more and more functions in the IC. Large die size with high density transistor is still the trend for IC product design, especially networking and CPU product. As Moore's law nears its physical limits, split die and heterogeneous integration in package are the effective solution to increase gross die and wafer yield for cost saving. In order to fulfil electronical performance, line width/ space small than 5/5um is necessary for die to die high density IO interconnection. Currently, 2.5DIC package is one of the mature solutions for fine line interconnection in the industry. Besides, fan-out multi-chip module (FO-MCM) is an alternative technology with low cost benefit that uses redistribution layer (RDL) instead of silicon interposer. For different segment of applications, FO-MCM chip last and chip first platform are widely used in industry. Chip last process has lower thermal budget on die, no die loss due to know good RDL and shorter cycle time advantages, which is ideal for applications requires multi dies as well as fine RDL interconnections. However, it still has the warpage control challenge with bump non-wetting/ bridge concern, due to the mold grinding process on FO module with complex composition (chip with ubump, underfill and compound). In this study, we will address the package warpage performance and the relationship of FO final thickness by FO-MCM chip last process. Finally, we will demonstrated three layer RDL structure and minimum line width/ space down to 2/2um. Package level reliability qualification is also finished with temperature cycle test (TCT), unbiased highly accelerated stress test (uHAST) and high temperature storage life (HTSL) conditions.
基于FO-MCM芯片最后平台的高性能计算应用异构集成解决方案
近年来,高性能计算(HPC)产品在集成电路中集成了越来越多的功能,大尺寸芯片和高密度晶体管仍然是集成电路产品设计的趋势,特别是网络和CPU产品。当摩尔定律接近其物理极限时,分模和封装内异构集成是提高总晶圆和晶圆良率以节约成本的有效解决方案。为了满足电子性能,需要小于5/5um的线宽/空间来实现芯片之间的高密度IO互连。2.5DIC封装是目前业界成熟的细线互连解决方案之一。此外,扇出多芯片模块(FO-MCM)是一种低成本效益的替代技术,它使用再分布层(RDL)代替硅中间层。针对不同的应用领域,FO-MCM芯片后置和芯片先置平台被广泛应用于工业中。芯片最后工艺具有较低的模具热预算,由于知道良好的RDL和较短的周期时间优势,没有模具损失,这对于需要多个模具以及精细RDL互连的应用是理想的。然而,由于FO模块的模具研磨过程具有复杂的成分(带有凸起,下填充和复合的切屑),因此它仍然存在凸起不润湿/桥接问题的翘曲控制挑战。在本研究中,我们将讨论FO- mcm晶片最后制程的封装翘曲性能与FO- mcm晶片最终厚度的关系。最后,我们将演示三层RDL结构和最小线宽/空间低至2/2um。封装级可靠性鉴定也完成了温度循环测试(TCT),无偏高加速应力测试(uHAST)和高温储存寿命(HTSL)条件。
本文章由计算机程序翻译,如有差异,请以英文原文为准。
求助全文
约1分钟内获得全文 求助全文
来源期刊
自引率
0.00%
发文量
0
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
确定
请完成安全验证×
copy
已复制链接
快去分享给好友吧!
我知道了
右上角分享
点击右上角分享
0
联系我们:info@booksci.cn Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。 Copyright © 2023 布克学术 All rights reserved.
京ICP备2023020795号-1
ghs 京公网安备 11010802042870号
Book学术文献互助
Book学术文献互助群
群 号:604180095
Book学术官方微信