超薄氧化铝层的电场调制效应（英文）

研究了基于高k介质材料的阻变存储器的写入/擦除(SET/RESET)特性和物理机制.研究发现基于NbAlO材料的阻变存储器SET/RESET电压具有较大波动性,通过结构优化,在Al_2O_3/NbAlO/Al_2O_3纳米薄片堆垛结构器件中获得高度稳定性的可重复的阻变特性.基于电场调制效应,提出了一种统一的电阻开关模型去模拟阻变存储器的SET/RESET行为,并探讨了单层阻变薄膜的阻变存储器中由导电单元形成和湮灭的巨大随机性引起的阻变特性分布.当在NbAlO基阻变存储器中嵌入超薄Al_2O_3膜后,阻变存储器的SET/RESET电压稳定性将显著提升,其原因在于采用堆垛结构的阻变器件中各介质层中的电场重新分布并精确可控,因此导电细丝的导通/断裂通过电场调制作用稳定均匀地在发生在具有高电场的薄缓冲层介质层中.     

超薄氧化铝层的电场调制效应

研究了基于高k介质材料的阻变存储器的写入/擦除 (SET/RESET) 特性和物理机制.研究发现基于NbAlO材料的阻变存储器SET/RESET电压具有较大波动性, 通过结构优化, 在Al2O3/NbAlO/Al2O3纳米薄片堆垛结构器件中获得高度稳定性的可重复的阻变特性.基于电场调制效应, 提出了一种统一的电阻开关模型去模拟阻变存储器的SET/RESET行为, 并探讨了单层阻变薄膜的阻变存储器中由导电单元形成和湮灭的巨大随机性引起的阻变特性分布.当在NbAlO基阻变存储器中嵌入超薄Al2O3膜后, 阻变存储器的SET/RESET电压稳定性将显著提升, 其原因在于采用堆垛结构的阻变器件中各介质层中的电场重新分布并精确可控, 因此导电细丝的导通/断裂通过电场调制作用稳定均匀地在发生在具有高电场的薄缓冲层介质层中.     

一种基于HfO2材料的RRAM波动性模型

基于HfO₂材料制作了一种具有良好非易失性的阻变存储器(RRAM)。根据ECM导电细丝机制,建立了动态的Verilog-A模型,该模型包含了RRAM的波动特性。对模型进行直流电压扫描验证,与实际器件的电流-电压特性曲线进行拟合。验证结果表明,该模型具备RRAM器件优良的电学特性,波动特性的加入对电路的前期设计具有指导意义。对模型进行脉冲仿真,仿真结果表明高、低阻态之间的鉴别窗口大于100倍,转变时间小于30 ns。     

新型阻变存储器内的电压解决方案

相对于现在流行的FLASH型存储器,新型阻变存储器(resistive-RAM,RRAM)有很多优势,比如较高的存储密度和较快的读写速度。而针对RRAM的读写操作特性,提出了一种适用于新型阻变存储器的提供操作电压的电路。该方案解决了新型存储器需要外部提供高于电源电压的操作电压的问题,使得阻变存储器能应用于嵌入式设备。同时,对工艺波动和温度波动进行补偿,从而降低了阻变存储器的读写操作在较差的工艺和温度环境下的失败概率,具有很强的实际应用意义。该设计采用0.13μm标准CMOS 6层金属工艺在中芯国际(SMIC)流片实现,测试结果表明,采用此电路的RRAM能正确地进行数据编程和擦除等操作,测试结果达到设计要求。     

基于I-V特性的阻变存储器的阻变机制研究

随着器件尺寸的缩小,阻变存储器(RRAM)具有取代现有主流Flash存储器成为下一代新型存储器的潜力。但对RRAM器件电阻转变机制的研究在认识上依然存在很大的分歧,直接制约了RRAM的研发与应用。通过介绍阻变存储器的基本工作原理、不同的阻变机制以及基于阻变存储器所表现出的不同I-V特性,研究了器件的阻变特性;详细分析了阻变存储器的五种阻变物理机制,即导电细丝(filament)、空间电荷限制电流效应(SCLC)、缺陷能级的电荷俘获和释放、肖特基发射效应(Schottky emission)以及普尔-法兰克效应(Pool-Frenkel);同时,对RRAM器件的研究发展趋势以及面临的挑战进行了展望。     

还原时间对AlOx/WOx RRAM开关速度的调制作用

基于128 kbit AlOx/WOx双层结构阻变存储器(RRAM)芯片,提出并验证了还原时间对RRAM开关速度的调制作用,同时设计了一种固定电压幅值逐步增大脉宽的算法用于RRAM阵列中速度的测试.还原处理的时间越长,AlOx层的厚度越薄,同时氧空位的含量增多,可加快导电细丝的形成、断裂和重新连接,进而提升芯片的开关速度.测试结果表明,还原时间由10 min增加至30 min,在4V和4.5V操作电压下,FORMING速度分布的均值分别由200 ns减小至120 ns和由100 ns减小至60 ns;在4V和4.5V操作电压下,RESET速度分布的均值分别由160 ns减小至120 ns和由120 ns减小至100 ns;SET速度分布的均值在4V电压下可由120 ns减小至80 ns.此外,还原时间的增长可以改善速度分布的一致性,减小速度的波动.     

基于低功函数铪电极的Si纳米空气沟道二极管研究

基于电子无散射弹道输运的纳米空气沟道二极管具有响应速度快和截止频率高等特点,在毫米波与太赫兹领域中有着巨大的应用潜力。本文研究了以低功函数金属Hf为电极的Si基垂直结构纳米空气沟道二极管的I-V与时间响应特性,并与相同结构的Au电极器件进行了对比。研究发现,相比于先前文献报道较多的截面发射结构,面内发射构型的Hf电极纳米空气沟道二极管发射电流增强了近2倍,5 V电压下达到13.5 mA,并且其I-V曲线线型从符合空间电荷限制电流的V^3/2变成了符合F-N场发射的规律。相比于以Au为电极的Si基纳米空气沟道二极管,以Hf为电极的器件发射电流显著提升了近4倍,且器件响应上升沿仅为2 ns,具有超快响应的特性,通过降低器件面积和电容还可进一步提升响应速度。     

RRAM的氧空位与金属细丝机制SPM的比较

应用扫描探针显微镜(SPM)技术实现了氧化物阻变薄膜局部区域高低阻态的互相转变。通过电激励、编程和擦除等操作,控制细丝的产生和断裂,实现了阻变薄膜局域的重复编程/擦除操作。用该方法分别研究了氧空位机制与金属导电细丝机制的氧化物薄膜的阻变特性,对两种机制做了对比研究。结果表明:在阻变存储器(RRAM)中氧空位机制在导电细丝和数据密度方面要高于金属细丝机制。同时,金属细丝机制阻变薄膜部分区域因编程/擦除操作发生了永久性形貌变化,可能对阻变器件的电极产生永久性破坏,这说明氧空位机制阻变薄膜在未来的高密度存储上具有较好的应用前景。     

阻变存储阵列的自动化测试系统

阻变存储器(RRAM)是一种新型的不挥发存储技术,研究阻变存储器阵列规模的存储性能以及可靠性问题是推进RRAM实用化的关键.目前通用的基于微控探针台的半导体参数分析的常规测量系统无法完成对阵列的自动化测试.利用半导体参数分析仪(4200-SCS)、开关矩阵以及相关外围电路搭建了一套针对阻变存储阵列的自动测试系统,实现了1MbitRRAM芯片的初始阻态分布的读取、初始化测试、存储单元的自动化编程/擦除操作.测试结果表明,该测试系统可以实现阻变存储阵列的自动化测试,为进一步工艺参数和编程算法的优化设计奠定基础.     

相变存储单元RESET电流与电极尺寸关系的数值模拟

应用数值模拟计算方法对相变存储器单元RESET电流和加热电极尺寸关系进行了研究,建立了二维存储单元模型,模拟了电脉冲作用下不同电极尺寸相变存储单元的RESET过程,包括不同电极尺寸相变单元RESET过程的温度场分布和电阻-电流关系。模拟结果表明,RESET电流随着电极尺寸减小而急剧减小。     

High κ for MIM and RRAM applications: Impact of the metallic electrode and oxygen vacancies

Influence of metallic electrode and oxygen vacancies in MIM capacitors and MIM RRAM high κ based devices is studied. For both MIM capacitors and MIM RRAM it is shown that the electrode composition strongly influences the overall behavior of the devices and more precisely, the capacitance–voltage curve (nonlinearities) for MIM capacitors, and the switching mechanism (SET/RESET) for MIM RRAM. Best results for HfO₂ RRAM are obtained with Pt as bottom electrode instead of TiN while very low capacitance variations are observed for high work function electrodes, or more precisely electrodes with low oxygen affinity. These evolutions are related to the oxygen vacancies concentration and migration to the cathode electrode/high κ interface.     

Identification of the generation/rupture mechanism of filamentary conductive paths in ReRAM devices using oxide failure analysis

Constant (CVS) and ramped (RVS) voltage stress data are combined with the aim of identifying the acceleration law that drives the generation and rupture of filamentary conductive paths in HfO₂-based ReRAM devices. The acceleration factor integral (AFI) method is used to find the equivalency between RVS and CVS in order to compare the SET and RESET events statistics and determine the adequacy of different degradation models frequently considered in oxide failure analysis: voltage power-law, E-model, and 1/E-model. The obtained results indicate that the E-model, with E the local electric field, exhibits the lowest dispersion in the acceleration factor values both for the SET and RESET transitions as well as the best overall consistency.     

Impact of Ni Concentration on the Performance of Ni Silicide/HfO2/TiN Resistive RAM (RRAM) Cells

We present a study of Ni silicide as the bottom electrode in HfO₂-based resistive random-access memory cells. Various silicidation conditions were used for each device, yielding different Ni concentrations within the electrode. A higher concentration of Ni in the bottom electrode was found to cause a parasitic SET operation during certain RESET operation cycles, being attributed to field-assisted Ni cation migration creating a Ni filament. As such, the RESET is affected unless an appropriate RESET voltage is used. Bottom electrodes with lower concentrations of Ni were able to switch at ultralow currents (RESET current <1 nA) by using a low compliance current (<500 nA). The low current is attributed to the tunneling barrier formed by the native SiO₂ at the Ni silicide/HfO₂ interface.     

Resistive switching effects of HfO2 high-k dielectric

Resistive switching behavior of HfO₂ high-k dielectric has been studied as a promising candidate for emerging non-volatile memory technology. The low resistance ON state and high resistance OFF state can be reversibly altered under a low SET/RESET voltage of ±3 V. The memory device shows stable retention behavior with the resistance ratio between both states maintained greater than 103. The bipolar nature of the voltage-induced hysteretic switching properties suggests changes in film conductivity related to the formation and removal of electronically conducting paths due to the presence of oxygen vacancies induced by the applied electric field. The effect of annealing on the switching behavior was related to changes in compositional and structural properties of the film. A transition from bipolar to unipolar switching behavior was observed upon O₂ annealing which could be related to different natures of defect introduced in the film which changes the film switching parameters. The HfO₂ resistive switching device offers a promising potential for high density and low power memory application with the ease of processing integration.     

Application of the defect clustering model for forming,SET and RESET statistics in RRAM devices

The choice of the right statistical model to describe the distribution of switching parameters (forming, SET and RESET voltages) is a critical requirement for RRAM, as it is used to analyze the worst case scenarios of operation that have to be accounted for while designing the cross-bar array structures, so as to ensure a robust design of the circuit and reliable data storage unit. Several models have been proposed in the recent past to characterize the voltage variations in V_FORM, V_SET and V_RESET using the percolation framework. However, most of these models assume defect generation to be a Poisson process and apply the standard Weibull distribution for parameter extraction and lifetime extrapolation. Recent dielectric breakdown studies both at the front-end as well as back-end have shown that the Weibull statistics does not describe the stochastic trends well enough, more so in downscaled structures at the low and high percentile regions given the possibility of defect clustering which is either physics-driven or process quality-driven. This phenomenon of defect clustering is all the more applicable in the context of resistive random access memory (RRAM) devices, as switching occurs repeatedly at ruptured filament locations where defect clusters pre-exist. This study examines the validity of the clustering model for RRAM switching parameter statistics (time/voltage to FORM, SET and RESET) and presents a physical picture to explain the origin of clustering in RRAM. A large set of data from various published studies has been used here to test the suitability and need for a clustering model based reliability assessment. Dependence of the clustering factor on temperature, voltage, device area, dielectric microstructure and resistance state has also been examined.     

Study of radiation hardness of HfO2-based resistive switching memory at nanoscale by conductive atomic force microscopy

A resistive switching random access memory (RRAM) with an HfO₂/Ti structure grown on a molybdenum (MO) substrate was fabricated, and a gold (Au) conductive atomic force microscopy (CAFM) tip was used as the top electrode such that the cell area of the resulting RRAM device is as small as 3 × 10⁻¹² cm². The pre- and post-irradiated resistive switching behaviors of the RRAM device with various HfO₂ layer thicknesses were investigated after being subjected to Co⁶⁰ γ-ray irradiation with different radiation doses. It is found that the forming voltage (V_forming), set voltage (V_set), resistance of high resistance state (RHRS) and resistance of low resistance state (RLRS) of the RRAM device are all radiation dose-dependent. The V_forming, V_set, RHRS and RLRS all decrease as the radiation dose increases due to increasing radiation-induced oxygen vacancies or defects inside the HfO₂ layer. Our experimental results indicate that the HfO₂-based RRAM cell with an extremely small cell area is not actually radiation hard since the operating voltage will change with V_forming and V_set after irradiation.     

二氧化铪材料中Cu的最佳迁移路径

The movement of Cu in a HfO2-based resistive random access memory （RRAM） device is investigated in depth by first-principle calculations. Thermodynamics analysis shows that the dominant motion of Cu tends to be along the [001] orientation with a faster speed. The migration barriers along different routes are compared and reveal that the [001] orientation is the optimal migration route of Cu in HfO2, which is more favorable for Cu transportation. Furthermore, the preferable HfOz growth orientation along [100], corresponding to Cu migration along [001], is also observed. Therefore, it is proposed that the HfO2 material should grow along [100] and the operating voltage should be applied along [001], which will contribute to the improvement of the response speed and the reduction of power consumption of RRAM.     

缺陷相互作用对HfO2基阻变存储器件均匀性的影响：第一性原理研究

The physical mechanism of doping effects on switching uniformity and operation voltage in Al-doped HfO₂ resistive random access memory（RRAM） devices is proposed from another perspective:defects interactions, based on first principle calculations.In doped HfO₂,dopant is proved to have a localized effect on the formation of defects and the interactions between them.In addition,both effects cause oxygen vacancies（V_O） to have a tendency to form clusters and these clusters are easy to form around the dopant.It is proved that this process can improve the performance of material through projected density of states（PDOS） analysis.For V_O filament-type RRAM devices, these clusters are concluded to be helpful for the controllability of the switching process in which oxygen vacancy filaments form and break.Therefore,improved uniformity and operation voltage of Al-doped HfjO₂ RRAM devices is achieved.     

Physical properties and electrical characteristics of H2O-based and O3-based HfO2 films deposited by ALD

Ozone (O₃) and H₂O are used as the oxidant to deposit hafnium oxide (HfO₂) thin films on p-type Si (1 0 0) wafers by atomic layer deposition (ALD). The physical properties and electrical characteristics of HfO₂ films change greatly for different oxidants and deposition temperature. Compared with O₃ as the oxidant, HfO₂ films grown with H₂O as the oxidant are more consistent in composition and growth rate. The O₃-based HfO₂ films have lower C impurity and higher concentration N impurity than the H₂O-based HfO₂ films. The impact of the annealing process on the electrical properties and stability of HfO₂ films are also investigated. A width step is observed in the O₃-based HfO₂ C–V curves, which disappears after annealing process. It is because the unstable Hf–O–N and Hf–N bonds in O₃-based HfO₂ films are re-bonded with the non-HfO₂ oxygen after annealing process, and the binding energy of N_1s shifts.     

Phase-Change Memory Device Using Si-Sb-Te Film for Low Power Operation and Multibit Storage

The switching characteristics of the electrical phase-change memory device using a SiSbTe film were studied. The SiSbTe film has a wider variation of electrical resistivity (up to 10⁷ times) along with crystallization than that of the conventionally used Ge₂Sb₂Te₅ film, and the SiSbTe film crystallizes predominantly into the hexagonal phase in a manner similar to the Sb₂Te₃ phase. The threshold voltage of the device is 5.87 V. The device was successfully operated with a 100 ns–5.5 V pulse for setting and a 20 ns–3 V pulse for resetting. The RESET current is about 1.37 mA, and the programming energies for resetting and setting are about 110 pJ and 60 pJ, respectively. More than 100 cycles were achieved with a RESET/SET resistance ratio higher than 50. In addition, multiple stable resistance stages can be obtained by adjusting the SET pulse, which makes multibit storage per cell possible.