基于Mix-IS算法的SRAM设计及良率分析

发布时间：2018-03-16 00:19

本文选题：静态随机存储器　切入点：工艺参数变化　出处：《苏州大学》2015年硕士论文　论文类型：学位论文

【摘要】：SRAM是一种重要的存储器,具有速度快、功耗低、可靠性高等优点,被广泛应用于系统级芯片。随着CMOS工艺的不断进步,SRAM的性能不断提高的同时,其生产成本也在不断下降。然而,当工艺尺寸降低到100nm以后,工艺参数的随机变化引起的MOS管阈值电压波动,对SRAM稳定性的影响越来越严重,成为限制SRAM良率提高的重要因素之一。正因为如此,近年来在SRAM,特别是全定制SRAM设计过程中,充分考虑工艺变化的设计思想(variation-aware design)成为业界的一大热点。传统的PVT分析方法已经不能满足先进工艺下对SRAM高良率的要求,采用蒙特卡洛方法分析工艺参数变化对SRAM稳定性及性能的影响已被广泛接纳。本文首先分析了CMOS工艺以及工艺参数变化对SRAM稳定性的影响。随之,鉴于高密度的SRAM对存储单元失效率的严格要求,探讨了蒙特卡罗分析方法存在的收敛速度慢、仿真时间长等问题。为解决这些题,本文基于混合重要性采样算法,实现了一种快速蒙特卡罗方法,该方法在保证仿真精度的基础上,极大地高了仿真速度。将本文所实现的快速蒙特卡罗方法应用到40G工艺32M SRAM测试芯片设计中,通过对SRAM存储单元稳定性的分析,选择稳定性最好的存储单元,并应用到测试片设计中。最后对所设计测试片进行的一系列仿真数据表明,对于工艺参数变化的影响,测试片仍然具有良好的稳定性,并且存储阵列可以保持非常高的良率。
[Abstract]:SRAM is an important memory with the advantages of high speed, low power consumption and high reliability. It is widely used in system-level chips. However, when the process size is reduced to 100nm, the threshold voltage fluctuation of MOS tube caused by the random change of process parameters becomes more and more serious to the stability of SRAM. It is one of the important factors that limit the improvement of SRAM yield. Therefore, in recent years, in the design process of SRAM, especially in fully customized SRAM, Fully considering the design idea of process change, variation-aware) has become a hot spot in the industry. The traditional PVT analysis method can no longer meet the requirements for high yield of SRAM in advanced technology. It is widely accepted to use Monte Carlo method to analyze the influence of process parameters on the stability and properties of SRAM. In this paper, firstly, the effects of CMOS process and process parameters on the stability of SRAM are analyzed. In view of the strict requirements of high density SRAM for the failure rate of memory cells, the problems of slow convergence and long simulation time in Monte Carlo analysis are discussed. In order to solve these problems, the hybrid importance sampling algorithm is used to solve these problems. A fast Monte Carlo method is implemented, which can greatly improve the speed of simulation on the basis of guaranteeing the accuracy of simulation. The fast Monte Carlo method is applied to the design of 32M SRAM test chip for 40G process. Through the analysis of the stability of SRAM memory cell, the best storage cell is selected and applied to the design of test piece. Finally, a series of simulation data for the designed test piece show that the effect on the change of process parameters is obtained. The test chip still has good stability, and the memory array can maintain a very high yield.
【学位授予单位】：苏州大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TP333

【参考文献】