應用實驗計畫法到半導體業金屬測定品管回收率之研究

Abstract

在晶圓清洗過程中容易受到周圍環境、操作過程、洗淨的原料(如化學藥品、去離子水、氣體)及機台所誘發而產生微粒及金屬粒子汙染。其中，金屬雜質的汙染，會影響生產出的元件其電氣特性，最終可能會有變質的情形發生。因此在金屬離子檢測之品管格外重要，我們可在採樣介質回收率之穩定性及多寡做評估，確認分析資料的穩定度及可信度，以確定接下來的步驟可以穩定進行。本研究將以實驗計畫法的架構下進行設計，找出最佳參數，將各待測金屬元素(鈉、鋁、鉀、鈣、鉻、鐵、鎳、銅、鋅)之回收率能落在80%~120%間，以確保測試之穩定。而透過我們的實驗計畫法分析結果得知，在可程式自動化控制液滴掃描儀取樣的空白液滴濃度、取樣時之轉速、添加液滴量與添加液滴數，對於回收率有明確的影響。因此這四項參數的有效調整，對於回收率之不穩定與過低現象的情形確有大幅改善，以期改善品管及檢驗的不確定性。

The process of wafer cleaning may be easily influenced by its environment, controlling process, cleaning material (such as chemicals, DI water, gas) and machine and then led to particles and metal particles contamination. The metallic ions may affect the devices and its electrical characteristics which finally caused degradation. Therefore, the metallic ions become extremely important to judge the sample of stability of media recovery and its quality. We have to follow the instructions as followed to make sure the stability and reliability of analysis data. The design of experiment (DOE) will find the best parameters to examine the stability of metallic elements (Na, Al, K, Ca, Cr, Fe, Ni, Cu, Zn) and its recovery rate will be between 80% and 120%. With the analysis of the DOE, we found that the blank concentration of Pad-Scan, its rpm, spike volume and its drop counts influenced the recovery rate. Therefore, with more effective adjustment to these four parameters might improve the stability of recovery rate and hope to better the uncertainties of quality control and its examination.