Abstract
In this thesis,we use Central Composite Design and Orthogonal Array to arrange the experiments.The parameters of etching rate include ICP power、Bias power、flow rate and pressure. Besides,we use spectrometer and RF impedance meter to measure plasma intensity and RF voltage.Before etching process,we have a standard warm-up process to confirm the ICP etcher is correct and stable.We found etch rate increases when ICP power and Bias power increase. When Flow rate increase,etch rate increases slightly.Intensity of Cl atom spectrum and RF voltage increase,etch rate increases.In our system,etch rate and the product of Intensity of Cl atom spectrum and RF voltage have linear relationship. We establish the relationship between etch rate and ICP power、Bias power、flow rate and pressure in regression equation model and neural network model.We found in non-linear systems neural network model has better predictive abilities.