MODELING OF A PLASMA PROCESSING MACHINE FOR SEMICONDUCTOR WAFER ETCHING USING ENERGY-FUNCTIONS-BASED NEURAL NETWORKS

The complex processing of plasma etching and deposition is highly nonl inear and its modeling is intractable by analytical basic-principles t echniques, Neural network approaches have shown initial success for sp ecific plasma processes in extracting implicit relations/models based on input-output measurements, The resulting modeling techniques natura lly depend on the neural structure, the adopted learning algorithms, a nd the specific plasma process and machine, We describe a plasma proce ssing machine designed and in operation at Michigan State University, East Lansing, which has been equipped with select sensing devices, The machine exhibits a hysteresic nonlinearity in the desirable processin g modes of operation, The experimental data characterize a testbed pla sma etching process using Argon gas with control inputs including inci dent microwave power, pressure, and cavity size, The internal states a nd the outputs include reflected power, electric field, and ion densit y, We employ several tailored networks with novel learning algorithms derived from functions that include the polynomial and the exponential energy functions, It is shown that the learning algorithms enable fas t and satisfactory convergence of parameters (weights and biases) in s everal scenarios of modeling and generalizing the input-state-output r elations of the plasma process.