Productivity Throughput Prediction Model of Ion Implant Machine by Artificial Neural Network

• Main Authors: PEI YIN CHU, 朱倍吟
• Other Authors: S. W. Lin
• Format: Others
• Published: 2010
• Online Access: http://ndltd.ncl.edu.tw/handle/05530323384163923437

碩士 === 長庚大學 === 資訊管理學系 === 98 === Semiconductor industry is a capital and technology intensive industry. In order to maintain its competitiveness, semiconductor fabrication companies must continuously invest in advanced processes and equipments to pursue customer demands. How to increase the utility ratio of the ion implantation equipment is a key issue to increase the product competitiveness under the pressure of the expensive equipment cost. Therefore, an efficient and systematic throughput prediction model must be constructed to improve the machine utilization. In this thesis, a back-propagation artificial neural network and multiple regression analysis were used to develop a product throughput prediction model for ion implantation equipments. The differences between the predicted performance and the real data was compared and discussed. The throughput prediction model constructed by the back-propagation neural network has more precise result compared with the model calculated from the multiple regression analysis in this study. By understanding the weighting of those influence factors for the throughput prediction model, engineers can quickly and accurately adjust the process parameters to correct the throughput deviation.