在ＩＣ載板產業裡良率損失不只是製程參數問題還與製程環境及生產週期相關,一方面載板在較長的生產週期中產品會被製程環境落塵及長生產週期間接造成良率損失，因此生產週期及環境環境條件都可能造成良率損失的直接或間接因子 良率資料分析屬於高度複雜的計算，現今載板製程多達百道步驟，分別落在不同無塵室落塵環境中及生產週期計算分別區分在百道步驟間，因此需做龐大的數據資料收集及分析， 為了以上問題本文提出一套基於生產環境條件及生產週期之良率分析建立模型，針對關鍵製程環境及生產週期篩選與良率高度相關步驟，此模型使用皮爾森相關係數之近似值，快速篩選出高度與相關良率相關製程環境與生產週期，再以逐步迴歸方式建立良率預測模型 模擬結果使用迴歸分析模型對生產環境及生產週期進行良率分析，可指出關鍵製程環境及生產週期 ，不但可得到相似辨識率且用有較佳分析效率，找到關係影響良率因子可用於業界良率改善;In the IC substrate industry.Yield loss is not only the process parameter problem. but also related to the process environment and production cycle time .On the one hand, the substrate will be indirectly caused by the process environment dust and long production cycle time,Therefore, the production cycle time and environmental conditions may cause indirect factors of yield loss. Yield data analysis is a highly complex calculation.Nowadays, there are hundreds of steps in the substrate process.Falling in different cleanroom .The production cycle time calculation is divided between hundreds of steps,Therefore, huge data collection and analysis is necessary. In order to solve the above problems, this paper proposes a set of models based on production environmental conditions and production cycle time and yield analysis, and screens processes that are highly correlated with yield in key process about environments and production cycles.This model uses the approximation of the Pearson correlation coefficient to quickly screen out the high correlation yield-related process about environment and production cycle, and then establish the yield prediction model by stepwise regression. The simulation results use the regression model to analyze the production environment and production cycle. It can point out the key process environment and production cycle. Not only can the similar recognition rate be obtained, but also the better analysis efficiency can be obtained. Industry yield improvement
