用神经网络研究大规模集成电路中的电迁移模型(英文)

STUDY ON VLSI ELECTROMIGRATION MODEL WITH NEURAL NETWORKS

本文用ADALine模型来解Black方程并用背传模型预测了中值失效时间 .其它没有包括在Black方程中的参数例如线的宽度也都用在了背传模型中 .本文给出的结果显示人工神经网络是研究金属合金中电迁移特性的非常强有力的手段之一 .

ADALine based model to solve Black's equation is proposed and the Back Propagation (BP) models is adopted to predict median time to failure of the failure distribution. Additional variables such as line width that can not be reflected in the Black's equation are used in this BP model. Results demonstrate that the neural network is one of the powerful tools for study on the properties of electromigration in metal alloys.