摘要
Most integrated circuit chips are constructed on single crystal silicon wafers. Silicon wafers in fabricating microchips must be very flat such that circuits can be printed on the wafers via lithographic processes. Wafer flatness directly impacts device line-width capability, process latitude, yield, and throughput. Thus, grinding to minimize roughness of wafers is considered to be a bottleneck when manufacturing silicon wafers. Since grinding is such important in the wafer manufacturing processes, any technique which can improve wafer quality, cannot be out of consideration in the grinding process. The purpose of this study is to provide a practical method for handling the grinding wheel wear problem. Since measurement items of process quality and the parameters of quality control are lack of precision in measuring flatness, this study utilizes a set of confidence intervals to generate triangular fuzzy numbers for estimating uncertain capability. Once the grinding process capability is obtained by regression method, the appropriate time to replace a worn grinding wheel can be determined by decision rules.