Method of diagnosing malfunctions in semiconductor manufacturing equipment

A diagnostic machine for manufacturing equipment comprising:

• a sensor for sensing output values from said manufacturing equipment;
• a source of equipment models that describe predicted relationships between possible manufacturing equipment inputs and predicted output values of said manufacturing equipment;
• means coupled to said source and said sensor and responsive to said equipment models and said sensor output values sensed by said sensor for backsolving actual equipment inputs without linearizing to explain output values from said equipment to derive fault hypotheses;
• and
• means coupled to said backsolving means for ranking said fault hypotheses based on their quality and consistency with respect to sensed output from said sensor.

A computer implemented method of diagnosing malfunctions in equipment for plasma enhanced processes in a vacuum chamber as used in semiconductor manufacturing comprising the steps of:

• a) sensing effects of the process on the semiconductor wafers in said vacuum chamber to provide sensed signals;
• b) providing equipment models that describe predicted relationship between input gases, radio frequency power used to create the plasma and the pressure in said vacuum chamber and predicted effects on the semiconductor wafer;
• c) backsolving actual vacuum chamber inputs in a processor under program control in the complete range of validity of the equipment models for which a response surface model value predicts values closest to values associated with said sensed signals without linearizing said response surface models to provide signals representing multiple fault hypotheses;
• and
• d) selecting a signal representing a fault hypothesis for which the differences between values associated with said sensed and predicted values is the least.