英文摘要 |
Excellence of products and quickly meet market demand for the semiconductor factory is committed to achieving its goals. In order to enhance the yield of products, keeping process capability stabile and real-time tracing of products qualities is very important. The application of non-product wafers (control wafer and dummy wafer) is to stabilize the machines process capability, and as well as rapid response the process is abnormal or not, then reduce the loss of product wafers. However, the use of a large number of non-production wafers but it will lead to increased costs, so how to find the balance point between enhancing product yield and reducing the amount of control and dummy wafers is the great important issue in semiconductor manufacturing. There are grade distribution of control and dummy wafers, and the different machines and processes need to different wafers to be used. After the wafers have been used by tools, they must enter to reclaim station to clean and re-grade the grading by there own clean degree. The purpose of this research is to develop a Linear Programming Model to optimize the distribution of control and dummy wafers after finishing the clean and regrading of the reclaim station based on the different grade demand of each process. And verifying the model by lingo software can effective minimize the new wafer start quantities and decrease the total cost by case study. |