MOS器件开启电压值的贝叶斯统计推断  

Bayesian Statistical Inference for the Threshold Voltage of MOSFET

作　　者：严利人[1] 刘道广[2] 刘志弘[1] 梁仁荣[1] YAN Liren;LIU Daoguang;LIU Zhihong;LIANG Renrong(Institute of Microelectronics,Tsinghua University,Beijing 100084,P.R.China;Institute of Nuclear and New Energy Technology,Tsinghua University,Beijing 100084,P.R.China)

机构地区：[1]清华大学微电子学研究所,北京100084 [2]清华大学核能技术研究院,北京100084

出　　处：《微电子学》2020年第5期659-663,共5页Microelectronics

基　　金：广东省重点领域研发计划项目(2019B010143002)。

摘　　要：在实践中准确地测定和表征出MOS器件样管的开启电压,对于大规模集成电路的设计以及将MOS器件作为分立器件的在电路应用,均是至关重要的。文章采用贝叶斯统计推断工具用于器件输入曲线的处理,从中提取出与器件开启有关的更为精准和深入的信息。建立合适的分层模型,应用基于马尔科夫链蒙特卡罗(MCMC)算法的最大后验估计(MAP),求取目标量的后验分布。这类算法为目前概率与统计领域的最高级算法。将该先进算法引入到IC领域来分析处理所获取的大数据是后摩尔时代的一个发展方向。In practice,accurately extracting the threshold voltage of a MOS device acts as a very important role for the further VLSI designing,as well as the in-circuit utilizing of the MOS as a discrete device.How to obtain the detail information from measured MOSFET input curves by using the Bayesian statistical inference method were described in this paper.After setting up a hierarchical linear model for measured data,posterior distribution of the target variable were calculated,and the posterior distribution of the target quantity was obtained by using maximum a posteriori(MAP)based on Markov Chain Monte Carlo-based(MCMC)algorithm.These algorithms were the state-of-art algorithm in the field of probability and statistics.It was a development direction in the post-Moore era to introduce the advanced algorithm into IC industry to analyze and process the big data obtained.

关 键 词：贝叶斯统计推断 开启电压 最大后验估计 后验分布 

分 类 号：TN307[电子电信—物理电子学]