TCAD simulations are crucial for semiconductor device development, predicting performance before costly fabrication. They involve process (fabrication steps) and device (electrical characteristics) simulations, both computationally intensive. Traditional numerical methods are limited in accuracy and speed, hindering optimization. This project aims to use machine learning to accelerate TCAD simulations. The student will: – Develop machine learning models to speed up TCAD. – Optimize a CMOS sensor prototype using these models. This involves: – Generating training data from TCAD simulations. – Training and evaluating neural network models. – Applying the model to design a novel CMOS sensor with a multiplication layer for improved timing. – Comparing simulated results with measurements to calibrate the model.

The project involves collaboration with Liverpool and FBK. The student will receive TCAD training.