扇出型面板級封裝(Fan-out Panel Level Packaging, FO-PLP)在半導體產業是一個較新的趨勢，其製程與晶圓級封裝(Wafer Level Packaging, WLP)相似，面板級封裝會先在矩形載板上進行封裝製程，結束後再切割成單顆封裝，相較於晶圓級封裝，面板級封裝載板與晶片形狀皆為矩形，所以晶片在載板上覆蓋率較高，此外，面板級封裝載板尺寸可達晶圓級封裝載板的數倍，因此在材料使用效率及成本上，面板級封裝都有著一定的優勢。
在製程中，翹曲現象是一個常見的問題。面板級封裝製程中的封膠製程，將結構進行加熱再降回室溫接續下個製程，溫度變化會造成翹曲的發生。此時，造成翹曲的原因主要有兩個，一是因不同材料間的熱膨脹係數不匹配，使得在溫度變化下，不同材料收縮量不同而導致翹曲的發生；二是在封膠製程中，環氧樹酯材料在升溫時發生固化反應，造成體積收縮，進一步導致結構翹曲。
翹曲量過大會使後續製程進行困難，有限元素法(Finite Element Method, FEM )是一個常用來預估翹曲量的一種方法，使用模擬來預估實驗可能會發生的情況，可在開發產品時，漸少大量的時間成本，但若要對封膠製程中環氧樹酯的固化收縮進行完整的預估，需要經過大量實驗及計算才能取得。因此，本文會利用等效熱膨脹係數的方法，來簡化與計算封膠製程後的翹曲量。之後在利用同樣的建模方式，建立在不同的幾何結構下所造成的翹曲量資料庫，使用機器學習的方法訓練電腦學習面板級封裝之幾何結構與翹曲量之關係，本文也會使用卷積層、池化層之概念，以及邊緣檢測的技術，來增快學習的速率，並比較其中的差異與討論學機器學習的成果。往後只要知道面板級封裝之幾何結構，就能利用訓練好的機器學習之模型，快速預估面板級封裝之結構翹曲量。

FO-PLP (Fan-out Panel Level Packaging) is one of the latest trends in the semiconductor industry. PLP conducts the molding process on the glass panel followed by debonding the molded panel and then divides into singular package. The process is similar with WLP (Wafer Level Packaging). However, PLP carrier used for packaging provides over two times platform compared to WLP carrier. Besides, because of the same shape between panel and chip, the usage ratio of PLP can achieve more than 95% and bring cost benefits in contrast to WLP, which usage ratio is less than 85%.
Warpage is a typical technical problem on large wafer or panel platforms. During the molding process, the whole structure will be heated, then cooling down to room temperature and cause structure warpage. Meanwhile, structure warpage is mainly attributed to two reason. The first one is thermal induced warpage. CTE(Coefficient of Thermal Expansion) mismatch lead to different shrinkage amount between different materials and cause warpage. Second one is during the heating, molding material, epoxy, will occur curing reaction and cause volume shrinkage.
Excessive warpage may make subsequent process difficult to execute. Nowadays, a widely used method to analysis warpage phenomenon is Finite Element Method (FEM). Use simulation method to correctly predict the situation in the real experiment can save lots of developmental time and cost. But in the molding process, if simulation wants to describe the molding material, epoxy, from start to finish, it need to take huge number of experiments and calculation first. So in this paper, will use Finite Element Method software ANSYS® to build the PLP model with equivalent CTE of molding material and predict the final amount of warpage in the end of molding process. Use the same simulation model and process to establish a data set of different geometric model related to amount of warpage. Apply machine leaning method, neural network, to train the computer to learn the relation between geometry and the amount of warpage by this data set. Convolution, pooling concept and edge detection technique are used to enhance the training process. The difference and result of training method are also discussed. Next time if we want to know the amount of warpage under specific geometry, just put the key information into the well-trained machine learning model, then we can get the warpage result without a second.

[1] T. Braum, S. Voges, M. Töpper, M. Wilke, M. Wöhrmann, U. Maaß, M. Huhn, K.-F. Becker, S. Raatz , J. U. Kim , R. Aschenbrenner, K. D. Lang , C. O’Connor , R. Barr , J. Calvert , M. Gallagher , E. Iagodkine , T. Aoude , and A. Politis. “Material and Process Trends for Moving From FOWLP to FOPLP ” Electronics Packaging and Technology Conference (EPTC), Singapore, Singapore Dec 2-4,2015.
[2] H. w. Liu, Y. W. Liu, J. J. Liao, A. Chen, Y. H. Chen, N. Kao, and Y. C. Lai. “Warpage characterization of panel fan-out (P-FO) package.” Electronic Components and Technology Conference (ECTC), Orlando, FL, USA, May 27-30, 2014.
[3] H. J. Borchardt and F. Daniels “The Application of Differential Thermal Analysis to the Study of Reaction Kinetics” J. Am. Chem. SOC., pp 41-46, 1957.
[4] K. E. J. Barrett. “Determination of Rates of Thermal Decomposition of Polymerization Initiators with a Differential Scanning Calorimeter”, Journal of Applied Polymer Science, Vol. 22, No. 9, pp. 1617-1626, 1967.
[5] K. Horie, H. Hiura, M. Sawada, I. Mita and H. Kambe. “CalorimetricInvestigation of Polymerization Reactions. III. Curing Reaction of Epoxides With Amines”, Journal of Polymer Science Part A: Polymer Chemistry, Vol. 8, pp. 1357-1372, 1970.
[6] M. R. Kamal and S. Sourour. “Kinetics and Thermal Characterization of Thermoset Cure”, Polymer Engineering & Science, Vol. 13, pp. 59-64, 1973.
[7] M. R. Kamal and M. E. Ryan. “The Behavior of Thermosetting Compound in Injection Molding Cavities”, Polymer Engineering & Science, Vol. 20, pp. 859-867, 1980.
[8] S. J. Hwang, and Y. S. Chang. “PVTC Equation for Epoxy Molding Compound”, IEEE Transactions on Components and Packaging Technologies, Vol. 29, No. 1, pp. 112-117, 2006.
[9] Tingyu Lin, Fengze Hou, Haiyan Liu, Daicong Pan, Feng Chen, Jun Li, Hengyun Zhang and Liqiang Cao. “Warpage simulation and Experiment for Panel Level Fan-out Package”, 2016 IEEE CPMT Symposium Japan (ICSJ), Kyoto, Japan, Nov. 7-9, 2016
[10] F. Hou, T. lin. L. Cao. F. Liu. J. Li. X. Fan, and G.Q. Zhang. “Experimental Verification and Optimization Analysis of Warpage for Panel-Level Fan-Out Package.” IEEE Transactions on Components, Packaging and Manufacturing Technology, Vol. 7, pp 1721-1728, 2017.
[11] M. Su, L. Cao, T. Lin, F. Chen, J. Li, C. Chen, and G. Tian. “Warpage simulation and experimental verification for 320 mm× 320 mm panel level fan-out packaging based on die-first process.” Microelectronics Reliability, Vol. 83, pp29-38, 2018.
[12] Shai Shalev-Shwartz and Shai Ben-David. “Understanding Machine Learning From Throry to Algorithms” Cambridge University Press, 2014.
[13] S. Aich, K. Younga, K. L. Hui, A. A. Al-Absi and M. Sain, “A nonlinear decision tree based classification approach to predict the Parkinson's disease using different feature sets of voice data”, 2018 20th International Conference on Advanced Communication Technology (ICACT), Korea (South), Feb. 11-14, 2018.
[14] D. Wu, C. Jennings , J. Terpenny, R. X. Gao, S. Kumara, “A Comparative Study on Machine Learning Algorithms for Smart Manufacturing: Tool Wear Prediction Using Random Forests”, Journal of Manufacturing Science and Engineering, Vol.139, Issue 7, 071018, 2017.
[15] R. A. Fisher, “The Use of Multiple Measurements in Taxonomic Problems”, Annals of Eugenics, Vol.7, Issue 2, pp. 179–188, 1936.
[16] D. R. Cox, “The regression analysis of binary sequences (with discussion)”, Journal of the Royal Statistical Society, Vol 20, pp. 215-242, 1958.
[17] F. Rosenblatt "The Perceptron: A Probabilistic Model For Information Storage And Organization In The Brain", Psychological Review, Vol 65, Issue 6, pp. 386–408, 1958.
[18] T. M. Cover and P. E. Hart, “Nearest neighbor pattern classification”, IEEE Transactions on Information Theory, Vol.13, Issue 1, pp. 21–27, 1967.
[19] N. S. Altman, “An introduction to kernel and nearest-neighbor nonparametric regression”, The American Statistician, Vol. 46, Issue 3, 1992.
[20] J. R. Quinlan, “Induction of Decision Trees”, Machine Learning, Vol.1, Issue 1, pp. 81–106, 1986.
[21] T. K. Ho, “Random Decision Forest”, Proceedings of 3rd International Conference on Document Analysis and Recognition, Montreal, Que., Vol 1, pp. 278-282, 1995.
[22] W. S. McCulloch, W. Pitts, “A logical calculus of the ideas immanent in nervous activity”, The bulletin of mathematical biophysics, Vol.5, Issue 4, pp 115–133, 1943.
[23] David E. Rumelhart, Geoffrey E. Hinton, and Ronald J. Williams, “Learning internal representations by back-propagating errors”, Nature, Vol. 323, pp.533-536, 1986.
[24] C. Cortes, V. Vapnik, “Support-Vector Networks”, Machine Learning, Vol. 20, Issue 3, pp. 273–297, 1995.
[25] 蔡佳翰,“使用等校熱膨脹係數於扇出型面板及封裝之翹曲研究”,國立清華大學動力機械工程學系,碩士論文, 2018
[26] R. D. Cook, D. S. Malkus, M. E. Plesha and R. J. Witt, Concepts and Applications of Finite Element Analysis, 4th ed., Wiley, 2002.
[27] M. Sadeghinia, K. M. B. Jansen, and L. J. Ernst. “Characterization and Modeling the Thermo-Mechanical Cure-Dependent Properties of Epoxy Molding Compound”, International Journal of Adhesion and Adhesives, Vol. 32, pp. 82-88, 2012.
[28] S. J. Hwang, and Y. S. Chang. “PVTC Equation for Epoxy Molding Compound”, IEEE Transactions on Components and Packaging Technologies, Vol. 29, No. 1, pp. 112-117, 2006.
[29] K. N. Chiang, H. C. Cheng and W. H. Chen. “Large-Scaled 3-D Area Array Electronic Packaging Analysis”, Journal of Computer Modeling and Simulation in Engineering, Vol. 4, No. 1, pp. 4-11, 1999.
[30] S. Raschka, Python Machine Learning, Packt Publishing, 2015
[31] Tomáš Mikolov, Martin Karafiát, Lukáš Burget, Jan Černocký, Sanjeev Khudanpur, “Recurrent Neural Network Based Language Model” ISCA, Makuhari, Chiba, Japan, Sep. 26-30, 2010.
[32] Hochreiter, Sepp and Schmidhuber, J¨urgen, “Long short-term memory”, Neural computation, Vol 9, issue 8, pp.1735-1780, 1997.
[33] D. H. Hubel and T. N. Wiesil “Receptive Fields, Binocular Interaction and Functional Architecture in the Cat’s Visual Cortex” J. Physiology, Vol. 160, Issue 1 pp. 106-154, Jan 1962.
[34] Kunihiko Fukushima “Neocognitron: A self-organizing neural network model for a mechanism of pattern recognition unaffected by shift in position”, Biolobical Cybernetics, Vol. 36, Issue 4, pp 193-202, April 1980.
[35] Y. Lecun, L. Bottou, Y. Bengio, P. Haffner, ” Gradient-based learning applied to document recognition” IEEE, Vol 86, Issue 11, pp 2278-2324, 1998.
[36] A. Krizhevsky, I. Sutskever, G. K. Hinton “ImageNet classification with deep convolutional neural networks” Advances in Neural Information Processing Systems, Vol. 1, pp 1097-1105,2012.
[37] Jia Deng, Wei Dong, Richard Socher, Li-Jia Li, Kai Li, Li Fei-Fei, “ImageNet: a large-scale hierarchical image database” IEEE Conference on Computer Vision and Pattern Recognition, Miami, FL, USA, June 20-25, 2009.