随着倒装焊封装技术的发展，倒装焊芯片的密度越来越高、厚度越来越薄，其内部缺陷的检测成为当今研究的重点。超声成像是检测倒装焊芯片内部缺陷的一种重要技术。本论文重点研究基于SAM的微电子封装缺陷快速成像技术。主要内容包括：

(1)针对超声C扫描成像质量低的问题，提出了一种基于KSVD学习字典的稀疏贝叶斯学习算法的超声成像方法。对超声B扫描图像按照滑动窗口进行迭代分块，将超声B扫描图像块向量化后组成训练样本，使用KSVD算法训练得到含有倒装焊芯片缺陷信息的字典库。通过对稀疏表示矩阵赋予均值为零的高斯独立分布的先验信息，利用期望最大化算法更新迭代稀疏表示矩阵的超参数，直至求得收敛的稀疏表示矩阵的最大后验估计，从而提升超声C扫描图像的质量。实验结果验证了本文提出的KSVD-SBLAMI方法在超声成像上的优越性能。

(2)针对KSVD-SBLAMI方法成像耗时长以及SBL算法估计稀疏表示矩阵质量低的问题，利用基于学习字典的广义近似消息传递-稀疏贝叶斯学习成像方法来提升超声成像的效率和性能。通过构建基于GAMP算法的稀疏表示矩阵统计模型，将高维联合后验概率密度的计算优化为标量计算，降低了成像算法复杂度。在求解模型参数矩阵中，将贝叶斯学习模型和最小均方误差模型进行相互迭代，并采用交替优化方式计算模型中的未知参数，从而提升了稀疏表示矩阵的估计精度。实验结果验证了KSVD-GSBLAMI方法在超声成像的优良性能。

(3)针对超声成像方法耗时的问题，提出了一种基于OpenMP的超声并行成像方法。首先对倒装焊芯片的超声三维体数据进行优化处理，将需要重构的信号集一次性存储在内存中，利用多核多线程技术按照信号集的逻辑结构进行数据块划分，使得每个线程负责处理超声C扫描图像中的一块，最终将各个线程的结果合并形成超声C扫描图像。实验结果表明，在开启多种线程数的情况下，本文所提方法最高可达到4.98倍的加速比，可以有效提升超声C扫描成像的效率。

With the development of flip chip packaging technology, the density and thickness of flip chip are getting higher and thinner. The detection of its internal defects has become the focus of current research. Ultrasonic imaging is an important technology to detect the internal defects of flip chip. This paper focuses on the rapid imaging technology of microelectronic packaging defects based on Sam. The main contents include:

(1) Aiming at the low quality of ultrasonic C-scan imaging, a sparse Bayesian learning algorithm based on ksvd learning dictionary is proposed. The ultrasonic B-scan image is iteratively divided into blocks according to the sliding window. The ultrasonic B-scan image blocks are vectorized to form training samples. The dictionary containing flip chip defect information is trained by ksvd algorithm. By giving the sparse representation matrix a priori information of Gaussian independent distribution with zero mean, the expectation maximization algorithm is used to update the super parameters of the iterative sparse representation matrix until the maximum a posteriori estimation of the convergent sparse representation matrix is obtained, so as to improve the quality of ultrasonic C-scan images. The experimental results verify the superior performance of ksvd-sblami method in ultrasonic imaging.

(2) In view of the long imaging time of ksvd-sblami method and the low quality of SBL algorithm in estimating the sparse representation matrix, the generalized approximate message passing sparse Bayesian learning imaging method based on learning dictionary is used to improve the efficiency and performance of ultrasonic imaging. By constructing the statistical model of sparse representation matrix based on gamp algorithm, the calculation of high-dimensional joint posterior probability density is optimized to scalar calculation, which reduces the complexity of imaging algorithm. In solving the model parameter matrix, the Bayesian learning model and the minimum mean square error model are iterated with each other, and the unknown parameters in the model are calculated by alternating optimization, which improves the estimation accuracy of the sparse representation matrix. The experimental results verify the excellent performance of ksvd-gsblami method in ultrasonic imaging.

(3) Aiming at the time-consuming problem of ultrasonic imaging method, a parallel ultrasonic imaging method based on OpenMP is proposed. Firstly, the ultrasonic 3D volume data of flip chip is optimized, the signal set to be reconstructed is stored in the memory at one time, and the data block is divided according to the logical structure of the signal set by using multi-core and multi thread technology, so that each thread is responsible for processing a piece of the ultrasonic C-scan image, and finally the results of each thread are combined to form the ultrasonic C-scan image. The experimental results show that the acceleration ratio of the proposed method can reach up to 4.98 times when a variety of threads are turned on, which can effectively improve the efficiency of ultrasonic C-scan imaging.

[1] 张晓涛. 全球价值链背景下产业转型升级路径——新加坡半导体产业的发展经验[J]. 国家治理, 2018, 208(40):18-22.

[2] 朱高峰. 全球化时代的中国制造[M]. 北京: 社会科学文献出版社, 2003.

[3] 刁石京. 中国制造 2025 与集成电路产业创新发展[J]. 集成电路应用, 2016, 33(6):4-7.

[4] Li Y, Goyal D. Introduction to 3D microelectronic packaging[M]//3D Microelectronic Packaging. Springer, Singapore, 2021: 1-16.

[5] 石春琦. 浅析我国集成电路产业在全球市场的地位[J]. 集成电路应用, 2016, 33(7):7-9.

[6] 常乾, 朱媛, 曹玉媛. 夹层式叠层芯片引线键合技术及其可靠性[J]. 电子与封装, 2017(2):4-8.

[7] Hassaine N, Villeneuve L, Shen Y, et al. Modeling of Tape Automated Bonding (TAB) for High Performance Integrated Circuits[C]//2006 Canadian Conference on Electrical and Computer Engineering. IEEE, 2006: 2464-2467.

[8] 蒋富裕, 陈亮, 张文林. 倒装LED芯片的焊料互连可靠性评估[J]. 电子工艺技术, 2017(4):187-190.

[9] 夏海梅. Flip chip封装的发展与挑战[J]. 电子制作, 2016(22):51-51.

[10] D. Paehler, D. Schneider, M. Herben. Nondestructive characterization of sub-surface damage in rotational ground silicon wafers by laser acoustics[J]. Microelectroic Engineering, 2007, 84(2):340.

[11] Ghafoor I, Peter W T, Munir N, et al. Non-contact detection of railhead defects and their classification by using convolutional neural network[J]. Optik, 2022, 253: 168607.

[12] Malayappan B, BVVSN P R, Ramaswamy K, et al. Enhancing functionality of free-space and guided-wave optical MEMS devices by integration with photonic circuits[J]. Optik, 2021, 243: 166955.

[13] Li D, Wang C, Cui X, et al. Recent progress and development of interface integrated circuits for piezoelectric energy harvesting[J]. Nano Energy, 2022: 106938.

[14] Z. Fu, B. Zhou, R. Yao and X. Li, "Research on thermal-electric coupling effect of the copper pillar bump in the flip chip packaging," 2016 17th International Conference on Electronic Packaging Technology (ICEPT), 2016, pp. 1377-1380.

[15] Teng P, Wen X, Liu Z, et al. An optical fiber integrated optoelectrode for the photoelectrochemical detection[J]. Optics Communications, 2022, 502: 127436.

[16] Matital R P, Kolymagin D A, Chubich D A, et al. Luminescence confocal microscopy of 3D components of photonic integrated circuits fabricated by two-photon photopolymerization[J]. Journal of Science: Advanced Materials and Devices, 2022, 7(2): 100413.

[17] 高明阳. 电路板元件贴装缺陷视觉检测系统[D]. 华中科技大学，2007.

[18] 杨英豪, 柳青, 崔洁. 机器视觉在焊点检测中的应用[J]. 电子工业专用设备, 2014(11):29-32.

[19] Kim P, Rhee S. Three-dimensional inspection of ball grid array using laser vision system[J]. IEEE Transactions on Electronics Packaging Manufacturing, 1999, 22(2):151-155.

[20] Jung D, Lee H, Kong D, et al. Thermal design and management of micro-pin fin heat sinks for energy-efficient three-dimensional stacked integrated circuits[J]. International Journal of Heat and Mass Transfer, 2021, 175: 121192.

[21] Zha Z, Liao G, Xianglin L U, et al. Research on Defect Inspection of Flip Chip Using Active Thermography[J]. Semiconductor Optoelectronics, 2011, 32(1):56-59.

[22] 班兆伟. 基于红外热成像技术的电子封装缺陷检测方法研究[D]. 北京工业大学，2012.

[23] 曹文浩. 基于红外热成像的内部孔洞缺陷检测方法研究[D]. 中国计量学院, 2013.

[24] 甘贤海, 丁国清, 胡胜华. 主动红外技术在线焊缝检测系统[J]. 激光与红外, 2004, 34(2):94-96.

[25] He Z, Wei L, Shao M, et al. Detection of micro solder balls using active thermography and probabilistic neural network[J]. Infrared Physics & Technology, 2017, 81.

[26] Lee H, Koo B, Chattopadhyay A, et al. Damage detection technique using ultrasonic guided waves and outlier detection: Application to interface delamination diagnosis of integrated circuit package[J]. Mechanical Systems and Signal Processing, 2021, 160: 107884.

[27] S. H. Chiu, C. Chen. Investigation of void nucleation and propagation during electromigration of flip-chip solder joints using x-ray microscopy[J]. Applied Physics letters, 2006, 89(26210626).

[28] Chun-Sam Song, Sung Man C ho, Joon-Hyun K im. Detection of flip-chip bonding error through edge size extraction of X-ray Image. Journal of Institute of Control, Robotics and Systems, 2009, 15(9):916.

[29] Roth H, He Z, Paul T. Inspection of Miniaturised Interconnections in IC Packages with Nanofocus ® X-Ray Tubes and NanoCT[C]. Electronics Packaging Technology Conference. IEEE, 2009:644-649.

[30] Zhang R Q, Zhang X M, Chen Z. 3D X-Ray Based BGA Solder Joints Inspection Algorithm[J]. Applied Mechanics and Materials, 2010, 29-32.

[31] Bauermann L P, Mesquita L V, Bischoff C, et al. Scanning acoustic microscopy as a non-destructive imaging tool to localize defects inside battery cells[J]. Journal of Power Sources Advances, 2020, 6: 100035.

[32] Kozic E, Hammer R, Rosc J, et al. Metallization defect detection in 3D integrated components using scanning acoustic microscopy and acoustic simulations[J]. Microelectronics Reliability, 2018, 88: 262-266.

[33] Reddy B, Sykes J M. Degradation of organic coatings in a corrosive environment: a study by scanning Kelvin probe and scanning acoustic microscope[J]. Progress in Organic Coatings, 2005, 52(4): 280-287.

[34] 李昕昕,党炜,李永正,张泽明.声学扫描显微镜检查标准研究[J].电子产品可靠性与环境试验,2018,36(05):72-76.

[35] E. Grünwald, R. Hammer, J. Rosc, et al. Advanced 3D Failure Characterization in Multi-Layered PCBs[J]. Ndt & E International, 2016, 84:99-107.

[36] 刘晓坤. 基于声扫描显微镜自动聚焦系统设计及其实现[D].北京交通大学,2019.

[37] E. Martin, C. Larato, A.Clementet al. Detection of delaminations in sub-wavelength thick multi-layered packages from the local temporal coherence of ultrasonic signals[J]. NDT & E International, 2008, 41(4):280.

[38] 耿喆，祝海江，杨平．超声C扫描设备定量评价方法研宄［J］．计量学报，2019，40（4）.

[39] Ozkaya U, Melgani F, Bejiga M B, et al. GPR B scan image analysis with deep learning methods[J]. Measurement, 2020, 165: 107770.

[40] 王茹,张磊,吴江涛,刘燕平.爆炸焊接复合板的超声C扫描成像[J].无损检测,2021,43(01):9-11+33.

[41] Song W, Ren J, He P, et al. Quantitative determination of the defects in TC4 diffusion bonded joints via ultrasonic C-scan[J]. Journal of Manufacturing Processes, 2021, 64: 1476-1483.

[42] Patil S, Reddy D M. Impact damage assessment in carbon fiber reinforced composite using vibration-based new damage index and ultrasonic C-scanning method[C]//Structures. Elsevier, 2020, 28: 638-650.

[43] Xiong J T, Sun J R, Wang J C, et al. Evaluation of the bonded ratio of TC4 diffusion bonded joints based on ultrasonic C-scan[J]. Journal of Manufacturing Processes, 2019, 47: 238-243.

[44] Chen D, Xiao H, Xu J. An improved Richardson-Lucy iterative algorithm for C-scan image restoration and inclusion size measurement[J]. Ultrasonics, 2019, 91: 103-113.

[45] Vandendriessche J, Orta A H, Verboven E, et al. Probabilistic ultrasound C-scan imaging of barely visible impact damage in CFRP laminates[J]. Composite Structures, 2022: 115209.

[46] 周正干.复合材料的超声检测技术[J].航空制造技术,2012(8):70-73.

[47] 王茹,张磊,吴江涛,刘燕平.爆炸焊接复合板的超声C扫描成像[J].无损检测,2021,43(01):9-11+33.

[48] 宋国荣,窦致夏,吕炎,马书旺,文硕,张斌鹏,何存富.快速超声C扫描成像中的信号频域分析法及其应用[J].北京工业大学学报,2020,46(12):1315-1322.

[49] 孙珞茗,林莉,马志远.基于声压反射系数幅度谱特征的涂层脱粘超声C扫描成像检测研究[J].机械工程学报,2019,55(12):44-49.

[50] 盛恒,曹丙花,张兴英,范孟豹,叶波.基于涡流C扫描成像的缺陷量化检测技术研究[J].中国科技论文,2018,13(02):121-125.

[51] 申浩. 基于电磁超声的C扫描成像检测研究[D].哈尔滨工业大学,2021.

[52] 曹胜强. 基于 VB.NET 的超声 C 扫特征成像系统研发[D].南昌航空大学,2018.

[53] S. Mallat and Z. Zhang \Matching Pursuit with Time-Frequency Dictionaries", IEEETrans. on Signal Processing, Dec. 1993.

[54] Ai-Ling Qi, Guang-Ming Zhang, Ming Dong, Hong-Wei Ma, David M Harvey. Anartificial bee colony optimization based matching pursuit approach for ultrasonic echoestimation[J].2018.

[55] Wang Y, Liu Y, She B, et al. Data-driven pre-stack AVO inversion method based on fast orthogonal dictionary[J]. Journal of Petroleum Science and Engineering, 2021, 201: 108362.

[56] Luo K, Liu X, Li J, et al. Redundant Gaussian dictionary in compressed sensing for ambulatory photoplethysmography monitoring[J]. Biomedical Signal Processing and Control, 2021, 66: 102479.

[57] Lu W, Li W, Kpalma K, et al. Compressed Sensing Performance of Random Bernoulli Matrices with High Compression Ratio[J]. IEEE Signal Processing Letters, 2015, 22(8): 1074-1078.

[58] 徐立军,杨秋翔,雷海卫.一种基于压缩感知的改进全变分图像去噪方法[J].微电子学与计算机,2016,33(06):100-103.

[59] Wang C, Deng Y, Shen P. A novel convex relaxation-strategy-based algorithm for solving linear multiplicative problems[J]. Journal of Computational and Applied Mathematics, 2022: 114080.

[60] Coropețchi I C, Vasile A, Sorohan Ș, et al. Stiffness Optimization Through a Modified Greedy Algorithm[J]. Procedia Structural Integrity, 2022, 37: 755-762.

[61] Wang Q, Meng C, Wang C, et al. Adaptive Cluster Structured Sparse Bayesian Learning with Application to Compressive Reconstruction for Chirp Signals[J]. Signal Processing, 2022, 190: 108343.

[62] Zhang G M, Harvey D M, Braden D R. Signal denoising and ultrasonic flaw detection via overcomplete and sparse representations[J]. The Journal of the Acoustical Society of America, 2008, 124(5): 2963-2972.

[63] Bomann F , Plonka G , Peter T , et al. Sparse Deconvolution Methods for Ultrasonic NDT[J]. Journal of Nondestructive Evaluation, 2012, 31(3):225-244.

[64] Wipf D P, Rao B D. Sparse Bayesian learning for basis selection[J]. IEEE Transactions on Signal processing, 2004, 52(8): 2153-2164.

[65] Tipping M E. Sparse Bayesian Learning and the Relevance Vector Machine[J]. Journal of Machine Learning Research, 2001, 1(3): 211-244.

[66] Ji S, Xue Y, Carin L. Bayesian Compressive Sensing[J]. IEEE Transactions on Signal Processing, 2008, 56(6): 2346-2356.

[67] Babacan S D, Molina R, Katsaggelos A K. Fast bayesian compressive sensing using Laplace priors[C]// Fast bayesian compressive sensing using Laplace priors. Proceedings of the IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2009, 19-24 April 2009, Taipei, Taiwan.

[68] Zhang L, Zhao Y, Ma Q, et al. A parallel computing-based and spatially stepwise strategy for constraining a semi-distributed hydrological model with streamflow observations and satellite-based evapotranspiration[J]. Journal of Hydrology, 2021, 599: 126359.

[69] Agathos S N, Dimakopoulos V V, Kasmeridis I K. Compiler-assisted, adaptive runtime system for the support of OpenMP in embedded multicores[J]. Parallel Computing, 2022: 102895.

[70] Hori A, Jeannot E, Bosilca G, et al. An international survey on MPI users[J]. Parallel Computing, 2021, 108: 102853.

[71] Delmas V, Soulaïmani A. Multi-GPU implementation of a time-explicit finite volume solver using CUDA and a CUDA-Aware version of OpenMPI with application to shallow water flows[J]. Computer Physics Communications, 2022, 271: 108190.

[72] Murray J, Kreutz-Delgado K. An improved focuss-based learning algorithm for solving sparse linear inverse problems[C]. In IEEE Intl Conf. on Signals, Systems and Computers, 2001, 41: 19-53.

[73] Vidal R, Ma Y, Sastry S. Generalized Principal Component Analysis (GPCA)[J]. IEEE Transactions. Pattern And.Mach.Intell, 2005, 27(12): 1945-1959.

[74] Aharon M, Elad M, Bruckstein A. K-SVD: An algorithm for designing overcomplete dictionaries for sparse representation[J]. IEEE Transactions on signal processing, 2006, 54(11): 4311-4322.

[75] G.-M. Zhang,D.M. Harvey,D.R. Braden. High-resolution AMI technique for evaluation of microelectronic packages[J]. Electronics Letters,2004,40(6).