超声激振的倒装芯片焊球缺陷检测方法研究  

Defect inspection of flip chip solder balls based on ultrasonic excitation

作　　者：宿磊 胡啸 顾杰斐 李可 SU Lei;HU Xiao;GU Jiefei;LI Ke(School of Mechanical Engineering,Jiangnan University,Wuxi 214122,Jiangsu China)

机构地区：[1]江南大学机械工程学院,江苏无锡214122

出　　处：《华中科技大学学报（自然科学版）》2023年第12期117-122,共6页Journal of Huazhong University of Science and Technology(Natural Science Edition)

基　　金：中国博士后科学基金资助项目(2021T140279);国家自然科学基金资助项目(51705203,11902124,52175096)。

摘　　要：针对倒装芯片内部焊球缺陷难以检测的问题,提出了基于声固多物理场耦合的超声激振检测方法.首先,建立倒装芯片超声激振检测仿真模型,模拟了含有典型缺陷(缺球和虚焊)的倒装芯片在超声激励下的振动响应,仿真结果显示焊球缺失和虚焊会引起倒装芯片振动速度的明显变化;其次,搭建超声激振倒装芯片检测系统,对倒装芯片进行检测,获取含不同缺陷倒装芯片的实际振动信号;最后,利用核主成分分析算法改进多粒度级联森林网络,对含典型缺陷的倒装芯片振动信号进行识别与分类.仿真和实验结果验证了基于多粒度级联森林网络和超声激振技术检测倒装芯片焊球缺陷的有效性.To solve the problem of inspecting the solder ball defect in the flip chips,an ultrasonic vibration detection method based on acoustic and solid physical fields was proposed.Firstly,the simulation model of ultrasonic exciting inspection of the flip chip was established,and the vibration responses with typical defect of solder balls(missing and open)under ultrasonic excitation was simulated.The simulation results show that the missing and open of solder balls will cause significant changes in the vibration speed of the flip chip.Then,the system of ultrasonic excitation flip chip detection was developed to detect the flip chips and obtain the actual vibration signals of flip chips with different defects.The improved multi-grained cascaded forest(MGS-KPCA-CF)network optimized by the kernel principal component analysis(KPCA)was used to identify and classify the vibration signals of flip chips with typical defects.The simulation and experimental results validate the effectiveness of MGS-KPCA-CF network and ultrasonic excitation to detect solder ball defects of flip chips.

关 键 词：倒装芯片 缺陷检测 超声激振 核主成分分析 多粒度级联森林 

分 类 号：TH89[机械工程—仪器科学与技术]