目前，大多數電子元件所使用的基板材料通常為電木板和塑膠板，然而這類材料在高溫高濕的環境下會與材料的結構性質與熱應力交互作用而發生脆性老化之現象，在高溫下，基板材料可能會失去其機械強度和穩定性，而在高濕度環境下，材料可能會吸收水分導致尺寸膨脹或變形。進而造成線路受損並使得元件需要進行更換。因此，為解決這一問題並提高電子元件的可靠性，利用陶瓷作為基板取代原有之材料。因陶瓷本身之特性具有高溫高溼穩定性、耐化學性、絕緣性以及高機械強度，使電子元件能在較為苛刻的環境下保持良好之性能，此外陶瓷材料的尺寸穩定性也能夠使其在微細加工與高精度製造方面具有極大的優勢。
在陶瓷互連元件中為了製作出高精度且微細尺寸的雕刻線路，通常會使用雷射進行加工，但因材料特性與雷射加工參數之間的交互作用影響是難以用固定公式進行推算，為獲得微米等級之線路凹槽尺寸，並能夠針對各式各樣之狀況提供對應之加工參數。本研究加入人工智慧之模型根據過往經驗進行學習，以達到改善加工品質之目的。
本研究將整合雷射雕刻技術、影像擷取系統、影像處理以及強化學習四個部分達到上述目標。第一部分使用波長532 nm之綠光雷射雕刻機於陶瓷基板上製作出雕刻線；第二部分透過光學顯微鏡擷取影像並透過雲端傳送至伺服器；第三部分藉由OpenCV函式庫對雕刻線進行影像處理包括：影像除噪、影像校準、邊緣檢測以及計算寬度、深度與截面積計算獎勵分數；第四部分則是利用強化學習之模型對當前狀態預測下一回合之加工參數進行學習。
本研究建立之加工參數優化策略可在陶瓷材料上加工出凹槽，其尺寸可達寬度25 ± 2 μm、深度18 μm並以此基礎達到最大截面積。利用訓練完成之策略比訓練前之加工參數加工出的凹槽截面積提升20%，驗證此策略能夠有效提升加工品質之能力。

Currently, fiberglass boards and plastic boards are the most frequently utilized substrate materials in most electronic components. However, due to the combination of the structural characteristics and thermal stress in high-temperature and high-humidity conditions, these materials show brittle aging phenomena. In settings with high humidity, the substrate material may absorb moisture, resulting in dimensional expansion or deformation. High temperatures may cause the substrate material to lose its mechanical strength and stability. As a result, the circuit is damaged, requiring component replacement.
Ceramics are employed as substrates to take care of this problem and increase the dependability of electronic components. Ceramics' great mechanical strength, chemical resistance, insulation, and stability at high temperatures and humidity enable electronic components to continue operating well in harsher conditions. Additionally, ceramic materials' dimensional stability offers important benefits for precise processing and high-precision production.
Laser processing is frequently used in the manufacture of very accurate and compact etched circuits in ceramic connecting devices. However, it is challenging to compute using fixed formulas due to the interaction between material qualities and laser processing factors. This work includes artificial intelligence models to learn from prior experiences, seeking to increase processing quality, to produce micron-level groove dimensions for circuits and give corresponding processing parameters for diverse conditions.
To accomplish the goals, this study combines laser engraving technology, picture capturing systems, image processing, and reinforcement learning into four pieces. In the first section, ceramic substrates are etched with lines using a green laser engraving device with a 532 nm wavelength. The second stage involves the capturing of images using an optical microscope and the cloud-based transmission of those images to a server. In the third stage, the etched lines are processed using the OpenCV library, which also does edge recognition, image denoising, picture calibration, and computations of width, depth, and cross-sectional area to determine reward scores. The processing parameters for the subsequent round are learned in the fourth section using a reinforcement learning model that predicts the present state.
The established parameter optimization strategy in this study enables the production of grooves on ceramic materials, with dimensions reaching a width of 25 ± 2 μm and a depth of 18 μm, thereby achieving maximum cross-sectional area. By applying the trained strategy, the groove's cross-sectional area is improved by 20% compared to grooves produced using the parameters before training. This verification demonstrates the effectiveness of the strategy in enhancing the quality of the laser processing results.

摘要 i
Abstract iii
致謝 v
目錄 iv
圖目錄 viii
表目錄 xii
第一章 緒論 1
1.1 前言 1
1.2 研究動機 2
第二章 文獻回顧 4
2.1 陶瓷互連元件 4
2.1.1 陶瓷互連元件簡介 4
2.1.2 雷射誘導選擇性金屬化 5
2.1.3 化學電鍍 7
2.2 雷射加工 8
2.2.1 雷射簡介 8
2.2.2 雷射光之種類與特性 10
2.2.3 雷射之熱影響區 11
2.2.4 材料與雷射加工參數之交互作用 11
2.2.5 Nd:YVO4雷射加工 14
2.3 影像處理 15
2.3.1 卷積演算法 15
2.3.2 影像除噪 16
2.3.3 影像形態學 20
2.3.4 邊緣檢測 20
2.4 深度學習 23
2.4.1 深度神經網路 24
2.4.2 強化學習 25
2.5 雷射加工與人工智慧應用 29
第三章 研究方法 32
3.1 線路凹槽加工設計 34
3.2 資料收集 36
3.3 影像處理程序開發 37
3.3.1 影像除噪 38
3.3.2 影像校準與邊緣提取 39
3.3.3 影像遮罩 40
3.4 強化學習 41
3.4.1 深度Q網路(Deep Q Network, DQN) 45
3.4.2 離散化動作輸出 47
3.5 實驗儀器與材料簡介 48
3.5.1 雷射雕刻機 48
3.5.2 光學顯微鏡 50
3.5.3 雷射功率計 51
3.5.4 掃描電子顯微鏡 52
3.5.5 實驗材料 53
第四章 結果與討論 55
4.1 加工系統建構 55
4.1.1 設定雷射參數範圍探討 55
4.1.2 資料收集系統探討 56
4.2 影像處理程序分析 58
4.2.1 邊緣校準探討 59
4.2.2 邊緣提取探討 60
4.2.3 獲取凹槽寬度、深度以及截面積 64
4.3 強化學習 65
4.3.1 訓練過程探討 65
4.3.2 獎勵函數設計探討 67
4.3.3 訓練結果探討 74
4.4 優化結果探討 76
第五章 結論與未來展望 80
5.1 結論 80
5.2 研究貢獻 81
5.3 未來展望 82
參考文獻 84

[1] C. B. Carter and M. G. Norton, Ceramic materials: science and engineering. Springer, 2007.
[2] L. Levinson, Electronic Ceramics: Properties: Devices, and Applications. CRC Press, 2020.
[3] M. T. Sebastian and H. Jantunen, "Low loss dielectric materials for LTCC applications: a review," International Materials Reviews, vol. 53, no. 2, pp. 57-90, 2008.
[4] E. Ermantraut et al., "Laser induced selective metallization of 3D ceramic interconnect devices," in 2018 13th International Congress Molded Interconnect Devices (MID), 2018: IEEE, pp. 1-5.
[5] M. Tang et al., "4, 6-Dimethyl-2-mercaptopyrimidine as a potential leveler for microvia filling with electroplating copper," Rsc Advances, vol. 7, no. 64, pp. 40342-40353, 2017.
[6] 丁勝懋, 雷射工程導論. 中央圖書出版社, 1995.
[7] D. J. Bergman and M. I. Stockman, "Surface plasmon amplification by stimulated emission of radiation: quantum generation of coherent surface plasmons in nanosystems," Physical Review Letters, vol. 90, no. 2, p. 027402, 2003.
[8] T. H. Maiman, "Stimulated Optical Radiation in Ruby," vol. 187, pp. 493-394, 1960.
[9] A. L. Schawlow and C. H. Townes, "Infrared and Optical Masers," Physical Review, vol. 112, no. 6, pp. 1940-1949, 1958.
[10] W. M. Steen and J. Mazumder, Laser Material Processing. Springer Science & Business Media, 2010.
[11] A. Voevodin and J. Zabinski, "Laser Surface Texturing for Adaptive Solid Lubrication," Wear, vol. 261, pp. 1285-1292, 2006.
[12] A. H. Hamad et al., "Laser Ablation in Different Environments and Generation of Nanoparticles," Applications of Laser Ablation-Thin Film Deposition, Nanomaterial Synthesis and Surface Modification, pp. 177-196, 2016.
[13] B. Bachy et al., "Novel Ceramic‐Based Material for the Applications of Molded Interconnect Devices (3D‐MID) Based on Laser Direct Structuring," Advanced Engineering Materials, vol. 20, no. 7, p. 1700824, 2018.
[14] R. Fields et al., "Highly efficient Nd: YVO4 diode‐laser end‐pumped laser," Applied physics letters, vol. 51, no. 23, pp. 1885-1886, 1987.
[15] S. Vitabile et al., "Road Signs Recognition Using a Dynamic Pixel Aggregation Technique in the HSV Color Space," in International Conference on Image Analysis and Processing(ICIAP 2001), 2001, p. 572.
[16] K. Simonyan and A. Zisserman, "Very Deep Convolutional Networks for Large-Scale image Recognition," Computer Vision and Pattern Recognition(CVPR), 2014.
[17] A. Chaudhuri et al., "Optical Character Recognition Systems," in Optical Character Recognition Systems for Different Languages with Soft Computing: Springer, 2017, pp. 9-41.
[18] M. Idesawa, "High-Precision Image Position Sensing Methods Suitable for 3-D Measurement," Optics and Lasers in Engineering, vol. 10, no. 3-4, pp. 191-204, 1989.
[19] D. Haussler, "Convolution Kernels on Discrete Structures," Technical report, Department of Computer Science, University of California 1999.
[20] Y.-L. Liu et al., "A Robust and Fast Non-Local Means Algorithm for Image Denoising," 10th IEEE International Conference on Computer-Aided Design and Computer Graphics(CAD/Graphics), vol. 23, pp. 270-279, 2008.
[21] J. Canny, "A Computational Approach to Edge Detection," IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. PAMI-8, no. 6, pp. 679-698, 1986.
[22] S. Gupta and S. G. Mazumdar, "Sobel Edge Detection Algorithm," International Journal of Computer Science and Management Research, vol. 2, no. 2, pp. 1578-1583, 2013.
[23] J. McCarthy et al., "A Proposal For The Dartmouth Summer Research Project On Artificial Intelligence," AI Magazine, vol. 27, no. 4, pp. 12-12, 2006.
[24] C. J. Watkins and P. Dayan, "Q-learning," Machine Learning, vol. 8, no. 3-4, pp. 279-292, 1992.
[25] V. Mnih et al., "Human-Level Control Through Deep Reinforcement Learning," nature, vol. 518, no. 7540, pp. 529-533, 2015.
[26] H. Van Hasselt et al., "Deep Reinforcement Learning with Double Q-learning," in Proceedings of the AAAI Conference on Artificial Intelligence, 2016, vol. 30, no. 1, pp. 2094-2100.
[27] Z. Wang et al., "Dueling Network Architectures for Deep Reinforcement Learning," in International Conference on Machine Learning(ICML), 2016: PMLR, pp. 1995-2003.
[28] J. Schulman et al., "Trust Region Policy Optimization," in International Conference on Machine Learning(ICML), 2015: PMLR, pp. 1889-1897.
[29] V. Mnih et al., "Asynchronous Methods for Deep Reinforcement Learning," in International Conference on Machine Learning(ICML), 2016, pp. 1928-1937.
[30] T. Haarnoja et al., "Soft Actor-Critic: Off-Policy Maximum Entropy Deep Reinforcement Learning with a Stochastic Actor," in International Conference on Machine Learning(ICML), 2018, pp. 1861-1870.
[31] C.-S. Wang et al., "Process Parameter Prediction and Modeling of Laser Percussion Drilling by Artificial Neural Networks," Micromachines, vol. 13, no. 4, p. 529, 2022.
[32] W. Cai et al., "Application of sensing techniques and artificial intelligence-based methods to laser welding real-time monitoring: A critical review of recent literature," Journal of Manufacturing systems, vol. 57, pp. 1-18, 2020.
[33] C.-S. Lin et al., "The application of deep learning and image processing technology in laser positioning," Applied Sciences, vol. 8, no. 9, p. 1542, 2018.
[34] B. Acherjee et al., "Application of artificial neural network for predicting weld quality in laser transmission welding of thermoplastics," Applied soft computing, vol. 11, no. 2, pp. 2548-2555, 2011.
[35] K. Wasmer et al., "Laser processing quality monitoring by combining acoustic emission and machine learning: a high-speed X-ray imaging approach," Procedia Cirp, vol. 74, pp. 654-658, 2018.
[36] Z. Zhang et al., "Real-time penetration state monitoring using convolutional neural network for laser welding of tailor rolled blanks," Journal of Manufacturing Systems, vol. 54, pp. 348-360, 2020.