臺灣博碩士論文加值系統

在IC封裝的製程當中為了要得到良好的製程結果並避免嚴重的金線問題發生，大多是藉由田口方法或是經驗良好的工程師經由不斷地反覆的實驗找出良好的製程參數，但此方法隨著產品輕薄化的發展趨勢而逐漸受到考驗，本文提出了一套使用模流分析軟體搭配類神經網路及遺傳演算法的方法來找出製程中最佳的製程參數，以避免嚴重的金線偏移問題發生。首先先選出製程中可能影響實驗結果的數個製程參數，接下來藉由模流分析軟體分析出製程中影響力較大的製程參數，並以這些參數作為類神經網路的輸入項及使用模流分析軟體的模擬結果來訓練類神經網路，找出輸入項與輸出項也就是金線偏移結果之間的非線性關係，接下來使用遺傳演算法計算非線性關係中的最佳結果也就是最佳製程參數。此方法的優勢在於可以大幅的縮減所需的實驗次數，並藉由模流分析軟體搭配類神經網路及遺傳演算法迅速的得出製程中最佳的參數設計以避免嚴重的金線問題發生。本文使用BGA(Ball Grid Array)模型作為例子並以Moldex3D驗證最佳化效果，其研究結果在最大金線偏移有明顯的改善。

In electronic packaging, molding encapsulates a package to protect the integrated circuit chips and wires from environmental or mechanical damages. During molding, defects like wire sweep may occur. Gold wires are common components in integrated circuit IC packages to transfer electronic signals between the die and the lead-frame contacts. Number-increased I/Os are built with more wires and smaller wire gaps increasing wire sweep problems. Earlier, experienced engineers solved these problems through trial and error requiring a large number of molding experiments. In this study, a process optimization approach with CAE tools, neural network, and genetic algorithm is proposed for preventing serious wire sweep. The approach determines the optimal process parameter settings for transfer molding electronic packages. The proposed method eliminates the need to perform a large number of experiments, and even improves the experimental parameter settings if those experiments were performed.

目錄
摘要 I
Extended abstract II
致謝 XXIII
目錄 XXIV
表目錄 XXVII
圖目錄 XXIX
符號說明 XXXII
第一章 緒論 1
1-1前言 1
1-2 研究動機 1
1-3 文獻回顧 3
1-4 論文架構 6
第二章 理論背景 7
2-1 模流分析 7
2-1-1 Kinetics Equation 9
2-1-2 Viscosity Equation 14
2-1-3 Drag Force Model 16
2-2 類神經網路 18
2-2-1 倒傳遞類神經網路 19
2-2-2 基本架構 20
2-2-3 Levenberg-Marquard演算法 23
2-2-4 Bayesian Regularization 25
2-3 遺傳演算法 27
第三章 Moldex3D模流分析 32
3-1 模流分析 32
3-2 金線密度分析 32
3-3 研究分析模型建立 33
3-4 Moldex3D模流分析流程 37
3-5 EMC初始轉化率計算 44
3-6 金線偏移量定義 45
3-7 分析結果 45
3-8 驗證分析 49
3-8-1 進澆速度驗證實驗 50
3-8-2 Single die structure模型驗證分析 52
第四章 製程最佳化流程 58
4-1 田口方法 58
4-2 製程最佳化步驟 59
4-3 類神經網路建立 60
4-4 遺傳演算法建立 61
第五章 製程參數最佳化實例 62
5-1製程參數範圍設計 62
5-2 田口實驗 63
5-3 類神經網路 67
5-4 遺傳演算法 69
5-5 最佳化製程參數結果與原始製程參數結果比較 71
第六章 結論與未來展望 74
6-1 結論 74
6-2 未來展望 76
參考資料 77
索引 80

[1]L. Nguyen, Wire bond behavior during molding operations of electronic packages, Polymer Engineering ＆ Science, vol. 28, no. 14, pp. 926-943, 1988.
[2]F. S. Sherman, Viscous flow. McGraw-Hill New York, 1990.
[3]A. Tay, K. Yeo, J. Wu, and T. Lim, Wirebond deformation during molding of IC packages, Journal of Electronic Packaging, vol. 117, no. 1, pp. 14-19, 1995.
[4]S. Han and K. Wang, A study on wire sweep in encapsulation of semiconductor chips using simulated experiments, Journal of Electronic Packaging, vol. 117, no. 3, pp. 178-184, 1995.
[5]J. Su, S.-J. Hwang, F. Su, and S. K. Chen, An efficient solution for wire sweep analysis in IC packaging, Journal of Electronic Packaging, Transactions of the ASME, vol. 125, no. 1, pp. 139-143, 2003.
[6]S.-Y. Teng and S.-J. Hwang, Simulations and experiments of three-dimensional paddle shift for IC packaging, Microelectronic Engineering, vol. 85, no. 1, pp. 115-125, 2008.
[7]D. Ramdan, Z. M. Abdullah, M. A. Mujeebu, W. K. Loh, C. K. Ooi, and R. C. Ooi, FSI simulation of wire sweep PBGA encapsulation process considering rheology effect, IEEE Transactions on Components, Packaging and Manufacturing Technology, vol. 2, no. 4, pp. 593-603, 2011.
[8]W. Jong, Y. Chen, and T. Kuo, Wire density in CAE analysis of high pin-count IC packages： Simulation and verification, International Communications in Heat and Mass Transfer, vol. 32, no. 10, pp. 1350-1359, 2005.
[9]C.-C. Pei, Wire Sweep and Paddle Shift Modeling of IC Packages During Encapsulation Process, PhD Thesis, Department of Mechanical Engineering, National Cheng Kung University, 2005.
[10]C.-C. Pei and S.-J. Hwang, Prediction of wire sweep during the encapsulation of IC packaging with wire density effect, Journal of Electronic Packaging, vol. 127, no. 3, pp. 335-339, 2005.
[11]N. Rochester, J. Holland, L. Haibt, and W. Duda, Tests on a cell assembly theory of the action of the brain, using a large digital computer, IRE Transactions on information Theory, vol. 2, no. 3, pp. 80-93, 1956.
[12]P. Werbos, Beyond Regression： New Tools for Prediction and Analysis in the Behavioral Sciences, Ph. D. dissertation, Harvard University, 1974.
[13]D. E. Rumelhart, G. E. Hinton, and R. J. Williams, Learning representations by back-propagating errors, Cognitive modeling, vol. 5, no. 3, p. 1, 1988.
[14]C.-T. Su and T.-L. Chiang, Optimizing the IC wire bonding process using a neural networks/genetic algorithms approach, Journal of Intelligent Manufacturing, vol. 14, no. 2, pp. 229-238, 2003.
[15]B. Ozcelik and T. Erzurumlu, Comparison of the warpage optimization in the plastic injection molding using ANOVA, neural network model and genetic algorithm, Journal of materials processing technology, vol. 171, no. 3, pp. 437-445, 2006.
[16]B. M. Wilamowski and H. Yu, Improved computation for Levenberg–Marquardt training, IEEE transactions on neural networks, vol. 21, no. 6, pp. 930-937, 2010.
[17]F. D. Foresee and M. T. Hagan, Gauss-Newton approximation to Bayesian learning, in Proceedings of International Conference on Neural Networks (ICNN'97), 1997, vol. 3： IEEE, pp. 1930-1935.
[18]C. Khor, M. A. Mujeebu, M. Z. Abdullah, and F. C. Ani, Finite volume based CFD simulation of pressurized flip-chip underfill encapsulation process, Microelectronics Reliability, vol. 50, no. 1, pp. 98-105, 2010.
[19]H.-L. Lee, S.-J. Chang, S.-J. Hwang, F. Su, and S. Chen, Computer-aided design of a TSOP II LOC package using Taguchi's parameter design method to optimize mold-flow balance, Journal of Electronic Packaging, vol. 125, no. 2, pp. 268-275, 2003.
[20]K. Taki, N. Shoji, M. Kobayashi, and H. Ito, A kinetic model of viscosity development for in situ ring-opening anionic polymerization of ϵ-caprolactam, Microsystem Technologies, vol. 23, no. 5, pp. 1161-1169, 2017.
[21]J. M. Kenny, Determination of autocatalytic kinetic model parameters describing thermoset cure, Journal of Applied Polymer Science, vol. 51, no. 4, pp. 761-764, 1994.
[22]M. Kamal and S. Sourour, Kinetics and thermal characterization of thermoset cure, Polymer Engineering ＆ Science, vol. 13, no. 1, pp. 59-64, 1973.
[23]C. Khor and M. Z. Abdullah, Optimization of IC encapsulation considering fluid/structure interaction using response surface methodology, Simulation Modelling Practice and Theory, vol. 29, pp. 109-122, 2012.
[24]H. Wang, H. Zhou, Y. Zhang, and D. Li, Stabilized filling simulation of microchip encapsulation process, Microelectronic Engineering, vol. 87, no. 12, pp. 2602-2609, 2010.
[25]C. Ayela and L. Nicu, Micromachined piezoelectric membranes with high nominal quality factors in newtonian liquid media： A Lamb's model validation at the microscale, Sensors and Actuators B： chemical, vol. 123, no. 2, pp. 860-868, 2007.
[26]H.-K. Kung, The analysis of the bending and twisting moments induced sweep deflection for semiconductor package applications, in 2005 7th Electronic Packaging Technology Conference, 2005, vol. 1： IEEE, p. 6 pp.
[27]S.-i. Amari, Backpropagation and stochastic gradient descent method, Neurocomputing, vol. 5, no. 4-5, pp. 185-196, 1993.
[28]J. H. Holland, Adaptation in natural and artificial systems： an introductory analysis with applications to biology, control, and artificial intelligence. MIT press, 1992.