A neural network method for restoring the initial impurity concentration distribution from data of ion sputter depth profiling

Abstract

A new approach to solving the problem of restoring the initial impurity concentration distribution from data of ion sputter depth profiling is proposed. The algorithm of impurity profile restoration is based on using an artificial neural network with the input signals representing surface concentrations of impurity determined at sequential moments of sputter depth profiling. The artificial neural network is trained for various depths and thicknesses of the impurity-containing layer and various values of parameters of the adopted model equation of diffusion-like ion mixing.