People productivity improvement via cloud machine monitor

To maintain high stability and production yield of production equipment in a semiconductor fab, on-line quality monitoring of wafers is required. In current practice, physical metrology is performed only on monitor wafers that are periodically added in production equipment for processing with production wafers. In addition to control wafers usage and loss of tool availability, however, routine monitoring does result in a huge cost of manual operation loading. This is equivalent to about 15% loss of people productivity. To give consideration to quality control and people productivity improvement, the system of Cloud Monitor (CM) is proposed based on stepwise regression and principle component analysis (PCA). The CM is verified by test-runs on the chemical vapor deposition (CVD) and chemical mechanical polishing (CMP) processes. Eight monitor items are considered. The CM is effective to construct forecast models with 1.34% mean absolute prediction errors (MAPE) and 100% OOC catch rate (OCR). The experimental results indicate that the CM is capable of predicting quality of production wafers using real-time sensor data from production equipment. Its performance abnormality or drift can be detected timely as well as improving people productivity.