Control of semiconductor processing

A system for semiconductor device manufacturing comprising:

• a metrology element for developing a multiple parameter characterization describing an in-process feature on a semiconductor substrate;
• a process control neural network having the in-process feature multiple parameter characterization as an input and having an estimated process control parameter value as an output;
• and
• a processing element having the estimated process control parameter value as an input for performing a process on the semiconductor substrate to develop a final feature on the semiconductor substrate.

A system for control of a semiconductor manufacturing operation, the system comprising:

• a process control neural network having as an input a characterization of a semiconductor wafer feature prior to processing and having as an output a control parameter value for controlling a manufacturing process, the process control neural network adapted find a mapping from the pre-processing characterization to an estimated control parameter value predicted to achieve a design semiconductor wafer feature after processing;
• a training neural network having as an input a characterization of an actual semiconductor wafer feature after processing through the manufacturing process controlled by the estimated control parameter and having as an output an ideal control parameter value for achieving the actual semiconductor wafer feature;
• and
• a comparator for comparing the estimated control parameter value and the ideal control parameter value to generate an error parameter.