Statistical control of VLSI fabrication processes

The authors present a formal approach to statistical process control of VLSIC fabrication processes. The system, developed at Carnegie Mellon University (CMU), aims at integrating the design (process and layout), wafer processing, testing, diagnosis, and control into a set of strongly interacting procedures that will aim at maximizing the profit from the fabrication line. The authors describe an implementation of this approach within the CMU computer-aided-manufacturing (CAM) framework. The key feature of this methodology is the exploration of the multivariate distribution of parameters measured in the fabrication process. This allows for more accurate decision-making compared to the traditional statistical quality control (SQC) chart-based approaches. The statistical control system presented, while relying on more sophisticated algorithms, looks quite similar to the standard systems from the user standpoint. The system uses the in-process measurements available during manufacturing. The authors focus on the statistical quality control subsystem of the CAM framework