Automated fault detection in microfabrication

A tool system comprising:

• a processor;
• and
• a non-transitory memory coupled to the processor and comprising a program to be executed in the processor, the program comprising instructions for:
• collecting first processing tool machine data from a first processing tool while treating semiconductor substrates, the first processing tool machine data comprising process data mapped to operational codes associated with one or more discrete intervals of time during the treating, wherein the operational codes comprise data generated by a state detector coupled to the first processing tool during the treating and being indicative of an operational status of the first processing tool, the operational status comprising information indicating a fault state of the first processing tool and a normal operation of the first processing tool;
• training a first neural network with the first processing tool machine data, the training being performed by including the information indicating the fault state of the first processing tool and the normal operation of the first processing tool during the treating;
• and
• generating a first output indicative of a fault of the first processing tool from the first neural network, based, at least in part, on applying unmapped machine data from at least one processing tool.

A method comprising:

• initializing a neural network on a processor of a first machine learning system coupled to a first processing tool;
• collecting first processing tool machine data from the first processing tool while treating semiconductor substrates, wherein the first processing tool machine data comprise process data mapped to operational codes associated with one or more discrete intervals of time during the treating, wherein the operational codes comprise information indicating a fault state of the first processing tool and a normal operation of the first processing tool;
• training the neural network with the first processing tool machine data, the training being performed by including the information indicating the fault state of the first processing tool and the normal operation of the first processing tool;
• saving the neural network to a memory of the first machine learning system;
• and
• implementing the first machine learning system on a second processing tool, including generating an output indicative of a fault of the second processing tool from the neural network, based, at least in part, on applying unmapped machine data from at least one processing tool.