一种基于斯坦福RRAM模型的参数提取方法  

A Parameter Extraction Method Based on Stanford RRAM Model

作　　者：孙海燕[1] 张硕[1] 张晓波[1] 戴澜[1] Sun Haiyan;Zhang Shuo;Zhang Xiaobo;Dai Lan(College of Information,North China University of Technology,Beijing 100144,China)

机构地区：[1]北方工业大学信息学院,北京100144

出　　处：《半导体技术》2022年第8期670-675,共6页Semiconductor Technology

摘　　要：阻变随机存取存储器(RRAM)是一种新型非易失性存储器件,比主流的闪存(Flash)器件具有更快的读/写速度、更低的编程电压和功耗。然而由于工艺不成熟等因素,RRAM准确的器件模型需要在已有的斯坦福物理模型基础上,对多达6个曲线拟合参数反复进行调整,具有一定的建模难度。提出了一种简并参数的建模方法,通过器件实测数据计算出其中4个曲线拟合参数I_(0)、g_(0)、γ_(0)、β。该方法大大降低了将物理模型适配为实际模型的建模难度,且由于基于实测数据,该建模方法理论上对不同工艺具有适配性。最后在不同工艺条件下制作了基于HfO_(x)材料的RRAM器件,并验证了该方法的有效性、先进性和准确性。Resistive random access memory(RRAM)is a new type of non-volatile memory device with faster read/write speed,lower programming voltage and power consumption than mainstream Flash devices.However,due to the process immaturity and other factors,the accurate device model of RRAM requires repeated adjustment of up to six curve-fitting parameters based on the existing Stanford physical model,which has certain modeling difficulties.A modeling method of simplifying and merging parameters was proposed,in which four curve-fitting parameters I_(0),g_(0),γ_(0) andβwere calculated from the measured data of the device.This method greatly reduces the modeling difficulty of adapting the physical model to the actual model,and since it is based on the measured data,the modeling method is theoretically adaptable to different processes.Finally,RRAM devices based on HfO_(x) materials were fabricated under different process conditions,and the validity,advancement and accuracy of this method were verified.

关 键 词：阻变随机存取存储器(RRAM) 非易失性存储器 器件模型 HfOx材料 导电细丝理论 

分 类 号：TP333.8[自动化与计算机技术—计算机系统结构]