Plackett-Burman Design and Artificial Neural Network for Optimization of Electrothermal Response of GAAFET Transistor

In this study, Plackett-Burman design and artificial neural network (ANN) models are used to optimize the thermal and electrical responses of gate all-around field effect transistor (GAAFET) devices. The MLR models derived from the PBD design showed moderate performance in predicting electrical responses, with significant errors and moderate explanatory power. However, the thermal response model showed strong predictive ability with low errors. ANOVA analysis highlighted key factors influencing the electrical and thermal responses, such as gate voltage and drain voltage. Subsequently, ANN models were developed that exhibited superior predictive accuracy for electrical and thermal responses, with high $\mathbf{R}^{\mathbf{2}}$ values during the training, validation, and test phases. The results suggest that while MLR models provide initial information, ANN models provide robust and accurate predictions, thereby improving the design and optimization of GAAFET transistors for high performance applications.