A neural network approach for classifying test structure results

An approach is described for identifying and classifying semiconductor manufacturing process variation using test structure data. The technique uses a machine-learning algorithm based on neural networks to train computers to detect patterns associated with test structure results. The objective of this work is to develop more reliable machine-learning classification procedures using test structure data from a semiconductor manufacturing environment. An example based on characterizing the performance of a 1- mu m lithography process is presented as well as a description of the test chip.<<ETX>>