Modeling of the MEMS Reactive Ion Etching Process Using Neural Networks

Reactive ion etch (RIE) is commonly used in microelectromechanical systems (MEMS) fabrication as plasma etching method, where ions react with wafer surface substrate in plasma environment. Due to the importance of RIE in the MEMS field, two prediction models are established to predict the wafer status in reactive ion etching process: back-propagation neural network (BPNN) and principle component analysis BPNN (PCABPNN). These models have the potential to reduce the overall cost of ownership of MEMS equipment by increasing the wafer yield, and not depend upon monitoring wafers or expensive metrology rather it will enable inexpensive real-time wafer-to-wafer control applications in RIE. The artificial neural net (ANN) is trained with historical available input-output process data. Once trained, the ANN forecasts the process output rapidly if given the input values.