Algorithmic Optimization of Transistors Applied to Silicon LDMOS

Abstract

We propose a pioneering approach that integrates optimization algorithms and technology computer-aided design to automatically optimize laterally-diffused metal-oxide-semiconductors (LDMOS) with a field-oxide structure. We define the ratio of the square of the breakdown voltage divided by the specific on-resistance as the figure-of-merit (FOM) and the objective function of our optimization. We compare the performance of three different algorithms: Nelder-Mead, Powell, and Bayesian Optimization. We show how the LDMOS performance evolves as each of the three optimization algorithms reach their optimized structure. We show that a straightforward Nelder-Mead optimization leads to a local optimum when optimizing over six input parameters. We find that Bayesian Optimization is the most data-efficient method to find the global optimized structure in the multi-domain design space.INDEX TERMS Bayesian optimization, LDMOS, nelder-mead algorithm, power semiconductor device, powell algorithm, step gate field oxide structure.

An Ultra-Fast and Precise Automatic Design Framework for Predicting and Constructing High-Performance Shallow-Trench-Isolation LDMOS Device

“…9 (a), the proposed automatic design framework that integrates BO with DNN model, while in Fig. 9 (b), it's traditional BO coupled with TCAD simulation tool [22]. The central advantage of proposed framework lies in the speed improvement.…”