Reference: Murdock, L. & Hayes-Roth, B. Intelligent Monitoring and Control of Semiconductor Manufacturing Equipment. December, 1991.
Abstract: This paper describes efforts to apply AI methods to enhance the quality and efficiency of semiconductor manufacturing in a state-of-the-art fabrication device called the "rapid thermal multiprocessor(RTM)". Semiconductor fabrication involves many complex processing steps with limited opportunities for measurement of process and product properties. By applying more knowledge to that limited data, AI monitoring and control methods augment classical control methods through detection of abnormalities and trends, prediction of failures, diagnosis, planning of corrective action sequences, explanation of diagnoses or predictions, and reaction to anomalous conditions that classical control systems typically would not correct. An architecture for AI control is described that adapts to complex changing environments such as that found in fabrication facilities. We illustrate architectural capabilities and our research efforts directed at reasoning about physical components of the RTM with scenarios from RTM wafer fabrication as well as from our parallel effort in monitoring intensive care patients.