基于声显微成像的集成电路封装内部缺陷诊断方法研究

Scanning Acoustic Microscopy (SAM) is an important tool in reliability testing of IC packages. The tendency of packaging technology towards 3D integrated system solutions and miniaturization, brings a big challenge for non-destructive testing (NDT) of advanced IC packages. In this proposal, a few techniques are proposed to solve some key issues to meet the challenge. Firstly, to develop a sparse signal representation (SSR) technique specifically for acoustical imaging of IC packages by utilizing the a-priori knowledge of the package and the imaging system. The developed SSR technique is used to separate the heavily overlapped echoes, achieving sub-wavelength resolution SAM imaging; Secondly, to develop a SSR technique on the basis of the Bernoulli-Gaussian model to solve the stability and reliability issues of using SSR to decompose ultrasonic signal; Thirdly, to develop a flaw diagnostic method by the use of geometrical and statistical features extracted from an ultrasonic C-scan image for quantitative detection of solder joint flaws inside a package with smaller pitch and high pin counts. Fourthly, to carry out through-life nondestructively monitoring and develop a life prediction model; Finally, to carry out finite elemental modelling of acoustical imaging of advance packages such as stacked die, SiP and TSV, and investigate the fundamental acoustical phenomena inside these IC packages. The simulated research provides a theoretical guidance for the development of the above techniques.

扫描声显微镜是集成电路封装可靠性测试的一种重要无损检测仪器。随着封装技术全面向系统集成、小型化和超薄化方向发展，对扫描声显微镜的检测能力带来前所未有的挑战。本项目利用封装和成像系统的先验信息，开发超声信号稀疏分解技术，分离高度重叠的超声回波，实现亚波长超声成像；提出基于Bernoulli-Gaussian分布模型的稀疏信号分解技术，解决稀疏分解在扫描声显微成像应用中的稳定性和可靠性问题；提出基于超声C扫描图像的几何与统计特征的缺陷诊断方法，解决高密度超细间距倒装焊内部的微小缺陷定量无损检测与评价的国际难题；在此基础上提出封装内部缺陷全寿命跟踪以及寿命预测方法。最后，提出通过有限元建模研究超声波在堆叠封装、SiP、TSV等新型复杂封装内部的传播机制，为上述关键缺陷检测技术的开发提供理论指导。

通过有限元建模研究了超声波在芯片封装内部的传播机理以及超声C扫描图像中的边缘效应形成机制。利用封装和成像系统的先验信息，研究了超声信号稀疏分解理论与技术，开发了多种超声信号稀疏分离算法，对高度重叠的超声回波进行回波分离与回波估计，实现了亚波长超声成像。研究了稀疏分解在扫描声显微成像应用中的稳定性和可靠性问题，并提出了几种解决方法；提出了几种超声C扫描图像特征量提取方法；提出了两种高密度超细间距倒装焊内部的微小缺陷定量无损检测与评价方法；通过联合加速热循环测试、加速振动测试、扫描声显微成像，实现了加速可靠性测试中的微电子器件与体统的倒装焊点无损缺陷全寿命监控，采集得到超过1TB的超声图像与三维超声数据。基于这些实验获得的大数据，开发了一种基于超声显微成像的倒装焊点寿命预测方法。该项目产生的科技成果在集成电路芯片可靠性领域具有很大的学术价值和潜在的应用价值。