電子封裝中，為滿足現今市場需求，傾向於更小、更輕以及更高效能的方向發展，從早期的DIP(Dual in-line Packaging)、SOP(Small Outline Packaging )以及BGA(Ball Grid Array)到更多引腳且體積更小的覆晶封裝(Flip Chip)以及本研究所探討的晶圓級封裝(Wafer Level Packaging, WLP)等。在電子封裝問世前，會經過一系列的實驗確保其可靠，而有許多因素皆會影響其壽命如:結構幾何之尺寸、材料的性質以及製程等。加速熱循環(Thermal Cycling Test)即為確保產品可靠度的實驗方法。然而該實驗的缺點是其花費太多時間以及成本，且無法符合市場的需求導致產品上市延宕，企業的競爭力下降。
　　為縮短研發時間，有限單元分析已被廣泛應用於可靠度之測試與模擬上，透過以與實驗驗證過之模型即可以得到可靠度壽命，以此減少做時間所需要的時間成本，本研究以ANSYS模擬晶圓級封裝體在加速熱循環負載下整體之應力分佈並將錫球所產生的等效塑性應變代入Coffin-Manson應變式預估錫球之疲勞可靠度壽命，同時鎖定最常發生失效與破裂之位置的最佳網格大小，以穩定預估結果並與實驗之數據相互驗證以驗證該模型的可行性。
　　而有限單元分析的缺點為不同的研究人員因為能力以及探討因素不同而導致模擬結果不盡相同。為消除該模擬誤差，本研究引入了人工智慧之機器學習的概念應用於封裝體的可靠度評估，除此之外，本研究結合了有限單元分析以及機器學習兩者，以經過與實驗比對認證後的有限單元模型建構不同的資料集合，輸入循環神經網路做訓練以此得到一穩定模型，可以更快速且大量的預估不同尺寸下的電子封裝可靠度壽命，加速壽命預估流程，節省成本。
　　本研究之機器學習演算法選用循環神經網路(Recurrent Neural Network)，並應用了不同的資料量、不同運算單元、不同結構以及不同運算環境探討其影響，透過與有限單元模擬結果比較，評斷合適的條件、資料量、運算單元以及結構和運算環境等，以此改善機器學習模型的預估結果以及訓練時間，加速可靠度壽命預估以及減少研發所需的成本。

To fulfill the current marketing demand, in the field of electronic Packaging, the trend to toward lighter, smaller and higher efficiency. From DIP (Dual in-line Packaging), SOP (Small Outline Packaging) and BGA (Ball Grid Array) to more lead number but smaller volume and higher density packaging such as Flip Chip and Wafer Level Packaging which is apply in this study. Before the product come out, it will undergo a series of reliability testing, there are a lot of factors may affect the reliability such as, packaging geometry, material property, process technology, etc. One of experiment method to get the reliability life is Thermal Cycling Test to ensure the quality of product before it sell. However, the disadvantage of experiment is that it takes too much time to afford marketing demand, resulting in delayed product launch time.
In order to reduce the research time, Finite Element Methods has been used in the reliability simulation and assessment widely. Using the model has already validated by experiment to get the product’s performance can reduce the research time. In this study, ANSYS simulation software has used to simulate the stress distribution and the equivalent plastic strain increment of Wafer Level Packaging under thermal cycling test condition. The Coffin-Manson Method has used to estimate the reliability life, at the same time, fixing the optical mesh size of position which always occur failure and crack situation to make the simulation result stable. The simulation result has compared with the experiment result to verify the feasible of the finite element model.
Although simulation can reduce research time, however, the problem is that the result obtained by different researchers will not the same due to the ability and knowledge. To eliminate the difference, the propose of the study is apply Artificial Intelligence so called Machine Learning in electronic packaging reliability assessment and make the research at the same basic point. Besides, this study combined finite element methods and machine learning together. Using the finite element model which successfully validated by experiment to build different data set. With the data input the Recurrent Neural Network model can get the robust and general model, therefor can predict a lot of different packaging geometry’s reliability at the same time. Save the time from experiment and finite element analysis.
In this study, Recurrent Neural Network has been used, at the same time, apply different data size, different calculation unit, different structure and different calculation environment in the machine learning model to discuss the influence. Using the comparison of predicting reliability and simulation result to judge the suitable condition and improve the performance the prediction. Therefore, it can accelerate the prediction of electronic packaging reliability and save the money from research.

[1] C. M. Liu, C. C. Lee, and K. N. Chiang, “Enhancing the Reliability of Wafer Level Packaging by using Solder Joints Layout Design,” IEEE Transactions on Component and Packaging Technologies, vol. 29, pp. 877-885, 2006.
[2] K. N. Chiang and W. L. Chen, “Electronic Packaging Reflow Shape Prediction for the Solder Mask Defined Ball Grid Array,” Transactions on Journal of Electronic Packaging, Vol. 120, No. 2, pp.175-178, 1998.
[3] C. Y. Tsou, T. N Chang, K. C. Wu, P. L. Wu and K.N. Chiang, “Reliability Assessment using Modified energy based model for WLCSP Solder Joints”, International Conference on Electronics Packaging (ICEP), Yamagata, Japan, April 19-22, 2017.
[4] L. F. Coffin, "A study of the effects of cyclic thermal stresses on a ductile metal”, Tarnsactions-American Society of Mechanical Engineers Journal of Electronic Packaging, vol. 76, pp. 931-950, 1954.
[5] S. S. Manson, “Behavior of materials under conditions of thermal stress”, National Advisory Committee for Aeronautics-Techncal Note 2933, pp. 317-350, 1953.
[6] R. Darveaux, K. Banerji, A Mawer, and G. Dody, “Reliability of plastic ball grid array assembly”, Ball Grid Array Technology, New York, 1995.
[7] R. Darveaux, “Effect of simulation methodology on solder joint crack growth correlation”, 2000 Proceedings. 50th Electronic Components and Technology Conference (Cat. No.00CH37070), Las Vegas, NV, USA, pp. 1048-1058, May 21-24, 2000
[8] W. R. Jong, H.C. Tsai, H.T. Chang, S.H. Peng, “The effects of temperature cyclic loading on lead-free solder joints of wafer level chip scale package by Taguchi method.” J. Electron. Packag. 130(1), pp.1–10, 2008
[9] R. J. Solomonoff, “An inductive inference machine,” IRE Convention Record, pt. 2, pp. 56-62, 1957
[10] W. S. McCulloch, and W. Pitts, “A logical calculus of the ideas immanent in nervous activity”, The bulletin of mathematical biophysics, Vol.5, Issue 4, pp 115–133, Dec., 1943.
[11] 李育承, “以循環神經網路回歸模型評估晶圓級封裝之可靠度”, 碩士論文, 國立清華大學動力機械工程學系, 2019
[12] C. Cortes, and V. Vapnik, “Support-Vector Networks”, Machine Learning, Vol. 20, Issue 3, pp. 273–297, 1995.
[13] P.J. Phillips, "Support Vector Machines Applied to Face Recognition", National Institute of Standards and Technology, 1998
[14] 沈奕廷, “以支援向量迴歸模型評估晶圓級封裝之可靠度”, 碩士論文, 國立清華大學動力機械工程學系, 2019
[15] J. R. Quinlan, “Induction of Decision Trees”, Machine Learning, Vol.1, Issue 1, pp. 81–106, 1986.
[16] T. K. Ho, “Random Decision Forest”, 1995 3rd International Conference on Document Analysis and Recognition, Montreal, Quebec., Canada, pp. 278-282 vol.1, 1995
[17] 蕭翔云,“以隨機森林回歸模型評估晶圓級封裝之可靠度”, 碩士論文, 國立清華大學動力機械工程學系, 2019
[18] N. S. Altman, “An introduction to kernel and nearest-neighbor nonparametric regression,” The American Statistician., vol. 46, no. 3, pp. 175-185, 1992.
[19] Sarkar M, Leong TY,”Application of K-nearest neighbors algorithm on breast caner diagnosis problem.” Proc AMIA Symp. pp.759-763, 2000.
[20] W. N. Findley, J. S. Lai, and K. Onaran, “Creep and relaxation of nonlinear viscoelastic materials, with an introduction to linear viscoelasticity. Amsterdam” ; New York : North-Holland Pub. Co, 1976.
[21] George E. Dieter , Mechanical Metallurgy. McGraw Hill Book Company, pp. 236-237, 1988
[22] N. E. Dowling, Mechanical Behavior of Materials: Engineering Methods for Deformation, Fracture, and Fatigue. Upper Saddle River, New Jersey: Prentice-Hall, 1999.
[23] J. L. Chaboche, “Constitutive Equations for Cyclic Plasticity and Cyclic Viscoplasticity,” International Journal of Plasticity, Vol. 5, pp. 247-302, 1989.
[24] J. L. Chaboche, “On Some Modifications of Kinematic Hardening to Improve the Description of Ratchetting Effects,” International Journal of Plasticity, Vol. 7, pp. 661-678, 1991.
[25] W. F. Chen, D. J. Han, Plasticity for structural engineers. Springer-Verlag, New York, pp.606, 1988.
[26] L. S. Goldmann, “Geometry optimization of controlled collapse interconnections,” IBM Journal of Research and Development, Vol 13, pp. 251-265, May, 1969.
[27] S. M. Heinrich, M. Schaefer, S. A. Schroeder, P. S. Lee, “Prediction of solder joint geomertry on array-type interconnections,” American Society of Mechanical Engineers Journal of Electronic Packaging, Vol. 118, pp. 114-121, 1996.
[28] K. A. Brakke, Surface Evolver Manual. version 2.01 Minneapolis, MN: The Geometry Center, 1996.
[29] L. F. Coffin, “A Study of the Effects of Cyclic Thermal Stress on a Ductile Metal,” Transactions ASME, Vol. 76, pp. 931-950, 1954.
[30] S. S. Manson, “Thermal stress and low cycle fatigue”, New York: McGraw-Hill, 1966.
[31] Y. J. Xu, L. Q. Wang, F. S. Wu, W. S. Xia, H. Liu, “Effect of Interface Structure on Fatigue Life under Thermal Cycle with SAC305 Solder Joints,” 2013 14th International Conference on Electronic Packaging Technology, Dalian, China, Aug 11-14, 2013.
[32] 吳凱強,“先進封裝錫球接點於不同溫度循環負載速率下之可靠度評估”, 博士論文, 國立清華大學動力機械工程學系, 2015
[33] Andreas Kaplan, Michael Haenlein,”Siri, Siri in my Hand, who's the Fairest in the Land? On the Interpretations, Illustrations and Implications of Artificial Intelligence”, Business Horizons, 62(1),p.p. 15-25, 2019
[34] S. Raschka, Python Machine Learning. Packt Publishing, 2015.
[35] F. Rosenblatt, “The Perceptron: A Probabilistic Model for Information Storage and Organization in the Brain,” Psychological Review, 65(6), pp. 386-408, 1958.
[36] Rumelhart, E. David , Hinton, E. Geoffrey and Williams, J. Ronald “Learning Representations by Back-propagating Error,” Nature, 323(6088), pp. 533-536, 1986
[37] W. H. Press, S. A. Teukolsky, T. W. Vetterling, and B. P. Flannery, Numerical Recipes: The Art of Scientific Computing, 3rd ed. New York: Cambridge University Press, 2007.
[38] J. Duchi, E. Hazan, and Y. Singer, “Adaptive Subgradient Methods for Online Learning and Stochastic Optimization.” J. Mach. Learn. Res. 12, pp. 2121-2159, 2011.
[39] T. Tieleman and G. Hinton. “Lecture 6.5-rmsprop, coursera: Neural networks for machine learning,” University of Toronto, Technical Report, 2012.
[40] D. P. Kingma and J. Ba, "Adam: A method for stochastic optimization," 2014 International Conference on Learning Representations, 2014.
[41] S. Hochreiter and J. Schmidhuber, “Long Short-Term Memory,” Neural Computation, Vol. 9, pp. 1735-1780, 1997.
[42] F. A. Gers and J. Schmidhuber and F. Cummins, “Learning to forget: continual prediction with LSTM,” 1999 Ninth International Conference on Artificial Neural Networks (ICANN 99), Edinburgh, UK, Sep. 7-10, 1999.
[43] F. A. Gers and E. Schmidhuber, “LSTM recurrent networks learn simple context-free and context-sensitive languages,” IEEE Transactions on Neural Networks, Vol. 12, no. 6, pp. 1333-1340, Nov. 2001.
[44] K. Cho and B. van Merrienboer, Ç. Gülçehre, D. Bahdanau, F. Bougares, H. Schwenk, and Y. Bengio, “Learning Phrase Representations using RNN Encoder-Decoder for Statistical Machine Translation,” Conference on Empirical Methods in Natural Language Processing (EMNLP), Doha, Qatar, Oct. 25-29, 2014.
[45] B. Rogers and C. Scanlan, “Improving WLCSP Reliability Through Solder Joint Geometry Optimization,” International Symposium on Microelectronics, Orlando, USA, September 30 – October 3, 2013 Vol.1, pp. 546-550.
[46] M. Motalab, M. Mustafa, J. C. Suhling, J. Zhang, J. Evans, M. J. Bozack, & P. Lall, “Thermal Cycling Reliability Predictions for PBGA Assemblies That Include Aging Effects,” ASME 2013 International Technical Conference and Exhibition on Packaging and Integration of Electronic and Photonic Microsystems, San Francisco, USA , July 16-18, 2013.
[47] M. C. Hsieh and S. L. Tzeng, “Solder joint fatigue life prediction in large size and low cost wafer-level chip scale packages,” 2014 15th International Conference on Electronic Packaging Technology, Chengdu, China, August 12-15, 2014.
[48] M. C. Hsieh, “Modeling correlation for solder joint fatigue life estimation in wafer-level chip scale packages,” 2015 10th International Microsystems, Packaging, Assembly and Circuits Technology Conference (IMPACT), Taipei, Taiwan, Oct 21-23, 2015.
[49] 曾驛捷, “不同數據於循環神經網路模型對晶圓級封裝可靠度預估影響研究”, 碩士論文, 國立清華大學動力機械工程學系, 2020