Prediction of Silicon Direct Nitridation Kinetic By An Efficient and Simple Predictive Model Based on Group Method of Data Handling

Abstract

In the present study, a soft computing method namely the group method of data handling (GMDH) is applied to develop a new and efficient predictive model for prediction of conversion percentage of silicon. A comprehensive database is obtained from experimental studies in literature. Several effective parameters like time, temperature, nitrogen percentage, pellet size and silicon particle size are considered. The performance of the model is evaluated through statistical analysis. Moreover, the silicon nitridation was performed in 1573 k and results were evaluated against model results for validation of the model. Furthermore, the performance and efficiency of the GMDH model is confirmed against the two most common analytical models. The most effective parameters in estimating the conversion percentage are determined through sensitivity analysis based on the Gamma Test. Finally, the robustness of the developed model is verified through parametric analysis.