Qualitatively modeling heterojunction bipolar transistors for optimization: a neural network approach

Abstract

A neural network approach is developed to qualitatively model the relationship between fabrication process parameters and the characteristics of a heterojunction bipolar transistor (HBT). An equivalent circuit model is used as an intermediate representation format for this objective. The goal of this research project is to develop a method that can predict and explain changes in the behavior of a device without the need for precise problem formulations and computationally intensive methods. The primary use of such a neural network model is in a reverse modeling process which performs device optimization.<>