Constructing and Application of Running Efficiency Indices of Processing Tools in an IC Fab

• Main Authors: Wang, Yu-Chih, 王有志
• Other Authors: Tong, Lee-Ing
• Format: Others
• Language: zh-TW
• Published: 2010
• Online Access: http://ndltd.ncl.edu.tw/handle/60995344028955226375

博士 === 國立交通大學 === 工業工程與管理學系 === 98 === Overall equipment efficiency (OEE) is a composite metric, which includes quality efficiency (QE), available efficiency (AE), operational efficiency (OE) and rate efficiency (RE). OEE is widely adopted in semiconductor manufacturing to assess and enhance the productivities of processing tools. The RE is a very popular index for fab managers to evaluate or monitor whether the processing tools keep their maximum production speeds. However, the RE is easily influenced by the production environment, and thus it can not be utilized to evaluate the actual production speeds of the processing tools. Consequently, the improvement opportunities and processing tool capacities are missed due to the biased measurement of RE. Therefore, this study develops a running efficiency (RUNE) to replace RE to help fab managers to keep the processing tools in their maximum production speeds. The RUNE is not affected by the production environment, and can be employed to identify the sources of equipment’s variations. Additionally, a RUNE management procedure is also proposed. The proposed procedure incorporates an automatic target-setting scheme to set the target production time of each recipe on every motion in equipment to obtain the updated RUNE value of equipment. An exponentially weighted moving average (EWMA) control chart is then utilized in the proposed procedure to monitor the RUNE value of equipment. A case study of a 300m IC foundry in Taiwan is utilized to obtain the RUNE values of tools and EWMA control charts of RUNE values to evalluate and monitor the production rate of tools. Finally, a survey is conducted to study the potential capacity losses of tools due to the biased RE measurement result. The RUNE is then proved to be more robust than RE, and thus can evaluate the true production speeds of the processing tools more accurately.