A Novel Inversion Approach for Predicting Critical Etch Dimensions Based on Electrical Parameters

With the ongoing miniaturization of semiconductor devices, the precise observation of etch profiles has become indispensable for optimizing the etching process. This paper proposes a novel method for predicting the critical dimension of etch profile based on electrical parameters. Utilizing simulated electrical parameters and critical etch profile dimensions obtained from identical etch process recipes, this study employ machine learning algorithms to establish a predictive model mapping electrical parameters to the critical dimensions of etch trenches. The model exhibits a coefficient of determination (R-squared) of 0.9841 and a mean squared error (MSE) of 0.0016, thereby validating the feasibility of this approach.