SiC 為第三代半導體的重要原料，因具有寬能隙、耐高溫、耐 高壓與低耗損等特性，是目前產業界主要的發展方向之一。SiC 的 硬度極高，亦造成生產時間及成本的耗費。化學機械研磨為目前半 導體晶圓主要的拋光技術，為有效提升 SiC 晶圓拋光製程之材料移 除率，利用飛秒雷射輔助加工技術進行 SiC 晶圓表面改質及材料移 除，使其有效縮短後續化學機械研磨加工時間，提升產能，增加 SiC 的工業應用。本論文以上述為出發點，目標在建立一 SiC 雷射表面 加工品質預測模型，收集 SiC 雷射表面加工之雷射相關參數與加工 品質，以常見的機器學習演算法如決策樹、隨機森林、Extra Trees、 Gradient Boosting 及 Stacking 建立模型，了解各雷射加工參數與加 工品質之間的關係，穩定加工品質，作為日後與 SiC 相關雷射加工 之研究準備。;SiC is an important raw material for third-generation semiconductors. Because of its wide energy gap, high temperature resistance, high pressure resistance, and low loss, it is one of the main development directions of the current industry. The hardness of SiC is extremely high, which also causes the consumption of production time and cost. Chemical mechanical polishing is currently the main polishing technology for semiconductor wafers. In order to effectively improve the material removal rate of the SiC wafer polishing process, femtosecond laser assisted processing technology is used to modify the surface of SiC wafers and remove materials to make it effective Shorten the follow-up chemical mechanical grinding processing time, increase production capacity, and increase the industrial application of SiC. This paper takes the above as the starting point and aims to establish a SiC laser surface processing quality prediction model. Collect laser-related parameters and processing quality of SiC laser surface processing, and build models with common machine learning algorithms such as decision trees, random forests, Extra Trees, Gradient Boosting and Stacking, Understand the relationship between laser processing parameters and processing quality, stabilize processing quality, and prepare for future research related to SiC laser processing.
