This paper presents a tool for the real-time diagnosis ofintegrated circuit fabrication equipment. The approach focuses onintegrating neural networks into a knowledge-based expert system. Thesystem employs evidential reasoning to identify malfunctions bycombining evidence originating from equipment maintenance history,on-line sensor data, and in-line past-process measurements. Neuralnetworks are used in the maintenance phase of diagnosis to approximatethe functional form of the failure history distribution of eachcomponent. Predicted failure rates are then converted to belief levels.For on-line diagnosis in the case of previously unencountered faults, aCUSUM control chart is implemented on real sensor data to detect verysmall process shifts and their trends. For the known fault case,hypothesis resting on the statistical mean and variance of the sensordata is performed to search for similar data patterns and assign belieflevels. Finally, neural process models of process figures of merit (suchas etch uniformity) derived from prior experimentation are used toanalyze the in-line measurements, and identify the most suitablecandidate among faulty input parameters (such as gas flow) to explainprocess shifts. A working prototype for this hybrid diagnostic system isbeing implemented on the Plasma Therm 700 series reactive ion etcherlocated in the Georgia Tech Microelectronic Research Center
展开▼