HfOx-based Resistive Random Access Memory Research Articles

Interface characteristic and performance optimization mechanism for HfOx-based RRAM devices

The HfOx-based resistive random access memory (RRAM) is extensively researched for next-generation nonvolatile memory because of its high integration and excellent compatibility with CMOS technology. The interfacial characteristics, including interface state and series resistance, which impact the HfOx-based RRAM performance. However, a quantitative analysis about interface state and series resistance at metal/hafnium oxide (HfOx) interface is still scarce. The interface state and series resistance are related to the interfacial layer and oxygen vacancy concentration. In this work, we use the Ti/HfOx/Pt structural RRAM device with different oxygen contents to control the interfacial layer and oxygen vacancy concentration at metal/HfOx interface. The series resistance at the HfOx/Pt interface in the low resistance state (LRS) is obtained using the Schottky model, while the interface state density and time constant at the Ti/HfOx interface in the high resistance state (HRS) are extracted by frequency dependence of conductance. The results indicate that the device uniformity is improved by reducing the LRS series resistance and HRS interface state density. Furthermore, the device endurance in the HRS deteriorates with the decrease of the interface state time constant. Overall, the RRAM device performance is optimized by modulating the series resistance, interface state density, and time constant.

Modeling of Self-Aligned Selector Based on Ultra-Thin Metal Oxide for Resistive Random-Access Memory (RRAM) Crossbar Arrays.

Resistive random-access memory (RRAM) is a crucial element for next-generation large-scale memory arrays, analogue neuromorphic computing and energy-efficient System-on-Chip applications. For these applications, RRAM elements are arranged into Crossbar arrays, where rectifying selector devices are required for correct read operation of the memory cells. One of the key advantages of RRAM is its high scalability due to the filamentary mechanism of resistive switching, as the cell conductivity is not dependent on the cell area. Thus, a selector device becomes a limiting factor in Crossbar arrays in terms of scalability, as its area exceeds the minimal possible area of an RRAM cell. We propose a tunnel diode selector, which is self-aligned with an RRAM cell and, thus, occupies the same area. In this study, we address the theoretical and modeling aspects of creating a self-aligned selector with optimal parameters to avoid any deterioration of RRAM cell performance. We investigate the possibilities of using a tunnel diode based on single- and double-layer dielectrics and determine their optimal physical properties to be used in an HfOx-based RRAM Crossbar array.

Enhanced resistive switching characteries in HfOx memory devices by embedding W nanoparticles

Resistive random access memory (RRAM) has lots of advantages that make it a promising candidate for ultra-high-density memory applications and neuromorphic computing. However, challenges such as high forming voltage, low endurance, and poor uniformity have hampered the development and application of RRAM. To improve the uniformity of the resistive memory, this paper systematically investigates the HfOx-based RRAM by embedding nanoparticles. In this paper, the HfOx-Based RRAM with and without tungsten nanoparticles (W NPs) is fabricated by magnetron sputtering, UV lithography, and stripping. Comparing the various resistive switching behaviors of the two devices, it can be observed that the W NPs device exhibits lower switching voltage (including a 69.87% reduction in Vforming and a reduction in Vset/Vreset from 1.4 V/-1.36 to 0.7 V/-1.0 V), more stable cycling endurance (>105 cycles), and higher uniformity. A potential switching mechanism is considered based on the XPS analysis and the research on the fitting of HRS and LRS: Embedding W NPs can improve the device performance by inducing and controlling the conductive filaments (CFs) size and paths. This thesis has implications for the performance enhancement and development of resistive memory.

Modeling the conduction mechanisms of intrinsic multi-level states in HfOx-based resistive random access memory

Multi-level cell storage technology based on resistive random access memory (RRAM) with multi-level state characteristics is more attractive in achieving low-cost ultra-high-density nonvolatile memory. Although a large number of literatures have reported the multi-level state characteristics of RRAM, so far there is no unified model that can well explain the intrinsic reasons for the existence of intermediate resistance state (IRS) and the switching mechanism between different resistance states. Multi-level state characteristics are observed by I–V characteristic measurements on RRAM with TiN/HfOx/barrier layer/TiN stacks fabricated using a commercialized 28 nm CMOS process. Compared to other published resistive switching models, the proposed model based on trap-assisted tunneling is more consistent with the measured. The model can reproduce the multi-level state characteristics based on the mechanism that interaction between defects and oxygen vacancies at the interface of HfOx and TiN electrode, resulting in the formation of multiple weak conductive filaments. Furthermore, the wide spread of high resistance state (HRS) and the switching between HRS and IRS are determined by the distance of tunneling gap. As HfOx-based RRAM will soon be commercialized, it is becoming very urgent to clarify the switching mechanisms of multi-level state characteristics and propose a universal model. Consequently, this work satisfied the current demand and significantly advanced the understanding and development of commercialized, cost-effective, high-density multi-bit HfOx-based RRAM technology.

HfTaOx Rectifying Layer for HfOx-Based RRAM for High-Accuracy Neuromorphic Computing Applications

In this study, a Ta2O5-doped HfOx (HfTaOx) thin film was deposited by cosputtering to serve as the rectifying layer for HfOx-based resistive random-access memory (RRAM) with a final structure of Pt/HfOx/HfTaOx/TiN/SiO2/Si. Incorporating the appropriate proportion of lattice and nonlattice O in the rectifying layer enabled forming-free RRAM operation. Moreover, by modifying the compliance current and making use of the deep reset operation, multilevel resistance states were realized. In neuromorphic computing, when mimicking artificial synapses, potentiation and depression were successfully induced, and low nonlinearity was demonstrated, implying efficient weight modulation and reduced energy and time for neural network training. Software-comparable Modified National Institute of Standards and Technology (MNIST) handwritten digit database inference accuracy (97.54%) was achieved for an RRAM-based fully connected neural network with the HfTaOx rectifying layer.

Forming-Free HfOx-Based Resistive Memory With Improved Uniformity Achieved by the Thermal Annealing-Induced Self-Doping of Ge

In this work, a forming-free HfOx-based resistive random access memory (RRAM) with improved uniformity is successfully demonstrated without an additional doping process, ionic injection, or special layer deposition. The significantly enhanced performances can be attributed to the self-doping of Ge atoms from the Ge bottom electrode (BE) into the HfO2 layer during the postdeposition annealing (PDA) process. Besides, the RRAM provides a lower reset current and a higher ON/OFF ratio than the reported forming-free RRAM. These excellent properties are beneficial to the application of power-saving RRAM and its circuit operation.

Improved gradual resistive switching range and 1000× on/off ratio in HfOx RRAM achieved with a Ge2Sb2Te5 thermal barrier

Gradual switching between multiple resistance levels is desirable for analog in-memory computing using resistive random-access memory (RRAM). However, the filamentary switching of HfOx-based conventional RRAM often yields only two stable memory states instead of gradual switching between multiple resistance states. Here, we demonstrate that a thermal barrier of Ge2Sb2Te5 (GST) between HfOx and the bottom electrode (TiN) enables wider and weaker filaments, by promoting heat spreading laterally inside the HfOx. Scanning thermal microscopy suggests that HfOx + GST devices have a wider heating region than control devices with only HfOx, indicating the formation of a wider filament. Such wider filaments can have multiple stable conduction paths, resulting in a memory device with more gradual and linear switching. The thermally enhanced HfOx + GST devices also have higher on/off ratio (>103) than control devices (<102) and a median set voltage lower by approximately 1 V (∼35%), with a corresponding reduction of the switching power. Our HfOx + GST RRAM shows 2× gradual switching range using fast (∼ns) identical pulse trains with amplitude less than 2 V.

Improvement of Resistive Switching Performance in Sulfur-Doped HfOx-Based RRAM.

In order to improve the electrical performance of resistive random access memory (RRAM), sulfur (S)-doping technology for HfOx-based RRAM is systematically investigated in this paper. HfOx films with different S-doping contents are achieved by atmospheric pressure chemical vapor deposition (APCVD) under a series of preparation temperatures. The effect of S on crystallinity, surface topography, element composition of HfOx thin films and resistive switching (RS) performance of HfOx-based devices are discussed. Compared with an undoped device, the VSET/VRESET of the S-doped device with optimal S content (~1.66 At.%) is reduced, and the compliance current (Icc) is limited from 1 mA to 100 μA. Moreover, it also has high uniformity of resistance and voltage, stable endurance, good retention characteristics, fast response speed (SET 6.25 μs/RESET 7.50 μs) and low energy consumption (SET 9.08 nJ/RESET 6.72 nJ). Based on X-ray photoelectron spectroscopy (XPS) data and fitting of the high/low resistance state (HRS/LRS) conduction behavior, a switching mechanism is considered to explain the formation and rupture of conductive filaments (CFs) composed of oxygen vacancies in undoped and S-doped HfOx-based devices. Doping by sulfur is proposed to introduce the appropriate concentration oxygen vacancies into HfOx film and suppress the random formation of CFs in HfOx-based device, and thus improve the performance of the TiN/HfOx/ITO device.

Total Ionizing Dose Effects on Multistate HfOₓ-Based RRAM Synaptic Array

Emerging nonvolatile memories (eNVMs) have demonstrated satisfactory accuracy on various applications in deep learning. Characterized by high density and low leakage power consumption, resistive random access memory (RRAM) becomes very attractive in synaptic devices for deep neural networks (DNNs). RRAM-based synaptic devices include both analog and discrete versions. Unlike analog RRAM synapses which suffer from nonlinearity, discrete but multistate RRAM synapses are better suited for neural network hardware implementation. In this article, the multistate operation in RRAM arrays has been proposed as a synaptic device for DNN inference. Four-state conductance has been achieved in HfO <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">x</sub> -based RRAM synaptic arrays. The impact of total ionizing dose (TID) on the multistate behavior of HfO <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">x</sub> -based RRAM is investigated by irradiating a one-transistor-one-resistor (1T1R) 64-kb array with CMOS peripheral decoding circuitry fabricated at the 90-nm technology node with Co-60 gamma rays ( <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">60</sup> Co γ-ray irradiation).

A Study of the Relationship Between Endurance and Retention Reliability for a HfOₓ-Based Resistive Switching Memory

This study determines the relationship between retention and endurance reliability for a HfOx-based resistive random access memory (ReRAM). A TiN (15 nm) / HfOx (6 nm) / Ti (10 nm) / TiN (40 nm) stacked structure is fabricated and tested to verify its basic characteristics and reliability. The high resistance states (HRS) retention behavior is characterized and is found to degrade over 100x on the endured bits because there is a sequential high temperature procedure. The degradation is reduced slightly to a ~30x drop for the endured devices with one single refresh cycle. During the endurance and retention test procedures, the HRS resistance decreases because neutral oxygen vacancy filaments grow and this cannot be reversed. The I-V characteristics for endured devices are also determined. The results show that isothermal treatment causes a gradual SET and RESET process with multiple feasible states. The thermally induced filament degradation model (isolated filament vs. continuous filament) is verified by the relationship between retention and endurance reliability. Design guidance is recommended for an improvement in ReRAM reliability.

Low power and high uniformity of HfOx-based RRAM via tip-enhanced electric fields

In this paper, the HfOx-based resistive random access memory (RRAM) devices with sub-100 nm pyramid-type electrodes were fabricated. With the help of tip-enhanced electric field around the pyramid-type electrodes, it was experimentally demonstrated that the novel devices have better cycle-to-cycle variation control, lower forming/set voltage (1.97/0.7 V) and faster switching speed (⩽ 30 ns under 0.9 V pulse) as well as better endurance reliability than conventional flat electrode resistive memory devices. In addition, the novel RRAM is experimentally demonstrated with independence of electrode number for the first time to present great potential in scaling down. This novel structure metal-oxide based RRAM will be suitable for the future low-power non-volatile memory application.

Sub-nanosecond pulse programming and device design strategy for analog resistive switching in HfOx-based resistive random access memory

The analog switching behavior of resistive random access memory (RRAM) is crucial for its application in neuromorphic computing. This paper investigates the switching mechanism of RRAM under sub-nanosecond and nanosecond pulse programming, with selected device materials and structures, using the kinetic Monte Carlo simulation method. The microscopic distribution of oxygen vacancies is simulated in all three spatial dimensions and in the time dimension (four-dimensional). According to the simulation results, thermal effects are a critical factor affecting the switching behavior. The thermal effects inside the HfOx switching layer can be almost completely eliminated by using sub-nanosecond pulses with low voltage, finally leading to good analog behavior. When nanosecond pulses are applied, effective heat insulation is the key to realizing analog characteristics. In general, the switching process is proposed to involve three stages. Having the lowest energy consumption, the first stage shows the greatest potential for achieving analog behavior. The use of heat-isolating structures, like capping layers and side wall materials with lower thermal conductance, could be a solution to improve the analog behavior of RRAM at the first stage when down-scaling the device size.

Impact of AlOx layer on resistive switching characteristics and device-to-device uniformity of bilayered HfOx-based resistive random access memory devices

An AlOx layer was deposited on HfOx, and bilayered dielectric films were found to confine the formation locations of conductive filaments (CFs) during the forming process and then improve device-to-device uniformity. In addition, the Ti interposing layer was also adopted to facilitate the formation of oxygen vacancies. As a result, the resistive random access memory (RRAM) device with TiN/Ti/AlOx (1 nm)/HfOx (6 nm)/TiN stack layers demonstrated excellent device-to-device uniformity although it achieved slightly larger resistive switching characteristics, which were forming voltage (VForming) of 2.08 V, set voltage (VSet) of 1.96 V, and reset voltage (VReset) of −1.02 V, than the device with TiN/Ti/HfOx (6 nm)/TiN stack layers. However, the device with a thicker 2-nm-thick AlOx layer showed worse uniformity than the 1-nm-thick one. It was attributed to the increased oxygen atomic percentage in the bilayered dielectric films of the 2-nm-thick one. The difference in oxygen content showed that there would be less oxygen vacancies to form CFs. Therefore, the random growth of CFs would become severe and the device-to-device uniformity would degrade.

Influneces of different oxygen partial pressures on switching properties of Ni/HfOx/TiN resistive switching devices

The HfOx-based resistive random access memory (RRAM) has been extensively investigated as one of the emerging nonvolatile memory (NVM) candidates due to its excellent memory performance and compatibility with CMOS process. In this study, the influences of deposition ambient, especially the oxygen partial pressure during thin film sputtering, on the resistive switching characteristics are discussed in detail for possible nonvolatile memory applications. The Ni/HfOx/TiN RRAMs are fabricated, and the HfOx films with different oxygen content are deposited by a radio frequency magnetron sputtering at room temperature under different oxygen partial pressures. The oxygen partial pressures in the sputter deposition process are 2%, 4% and 6% relative to engineer oxygen content in the HfOx film. Current-voltage (I-V) measurements, X-ray photoelectron spectroscopy, and atomic force microscopy are performed to explain the possible nature of the stable resistive switching phenomenon. Through the current-voltage measurement, typical resistive switching behavior is observed in Ni/HfOx/TiN device cells. It is found that with the increase of the oxygen partial pressure during the preparation of HfOx films, the stoichiometric ratio of O in the film is improved, the root mean square (RMS) of the surface roughness of the film slightly decreases due to the slower deposition rate under a higher oxygen partial pressure, and the high resistance state (HRS) current decreases. In addition, by controlling the oxygen content of the device, the endurance performance of the device is improved, which reaches up to 103 under a 6% oxygen partial pressure. The HfOx films prepared at a higher oxygen partial pressure supply enough oxygen ions to preserve the switching effect. As the oxygen partial pressure increases, the uniformity of the switching voltage is improved, which can be attributed to the fact that better oxidation prevents the point defects (oxygen vacancies) from aggregating into extended defects. Through the linear fitting and temperature test, it is found that the conduction mechanism of Ni/HfOx/TiN RRAM device cells in low resistance state is an ohmic conduction mechanism, while in high resistance state it is a Schottky emission mechanism. The interface between TE and the oxide layer (HfOx) is expected to influence the resistive switching phenomenon. The activation energy of the device is investigated based on the Arrhenius plots in HRS. A switching model is proposed according to the theory of oxygen vacancy conductive filament. Furthermore, the self-compliance behavior is found and explained.

Ultrafast Accelerated Retention Test Methodology for RRAM Using Micro Thermal Stage

We proposed an ultrafast accelerated retention test methodology for resistive random access memory (RRAM) using micro thermal stage (MTS). For accelerated retention test using chuck heating, the two major factors that limit the test speed are the attainable temperature range and thermal time constant. To extend these limits, we built MTS with extended temperature range (~800 °C) and short thermal time constant ( $\sim 10\mu \text{s}$ ). Using high resistance state retention failure of HfOx-based RRAM as an example, we demonstrated that the proposed methodology can reduce accelerated retention test time to <10 ms, achieving a boost of up to six orders of magnitude compared with chuck heating.

Effects of post-metal annealing on the electrical characteristics of HfOx-based resistive switching memory devices

Post-metal annealing (PMA) has been adopted to reduce the operation voltages of HfOx-based resistive random access memory (RRAM) devices, especially the forming voltage (VForming). TiN/Ti/HfOx/TiN stack structures were fabricated and annealed via rapid thermal annealing (RTA) and furnace annealing (FA) to investigate the annealing effects. The result of X-ray photoelectron spectroscopy (XPS) analysis indicates that the distribution of oxygen towards the interposing Ti layer increases after the annealing process, which facilitates the formation of conductive filaments in the dielectric layer. As a result, VForming can be decreased from 4.60 to 3.24 and 3.36 V via RTA for 30 s at 400 °C and via FA for 60 min at 300 °C, respectively, as compared with that without PMA. However, the VForming of the device annealed via FA for 60 min at 400 °C was higher than that at 300 °C. This turn-around phenomenon of the forming voltages of RRAM devices annealed via FA was found. It was attributed to the conversion of the interposing Ti layer into a highly resistive TiO2 film.

Resistive switching characteristic and uniformity of low-power HfOx-based resistive random access memory with the BN insertion layer**Project supported by the National Natural Science Foundation of China (Grant Nos. 61274113, 11204212, 61404091, 51502203, and 51502204), the Tianjin Natural Science Foundation, China (Grant Nos. 14JCZDJC31500 and 14JCQNJC00800), and the Tianjin Science and Technology Developmental Funds of Universities and Colleges, China (Grant No.

In this letter, the Ta/HfOx/BN/TiN resistive switching devices are fabricated and they exhibit low power consumption and high uniformity each. The reset current is reduced for the HfOx/BN bilayer device compared with that for the Ta/HfOx/TiN structure. Furthermore, the reset current decreases with increasing BN thickness. The HfOx layer is a dominating switching layer, while the low-permittivity and high-resistivity BN layer acts as a barrier of electrons injection into TiN electrode. The current conduction mechanism of low resistance state in the HfOx/BN bilayer device is space-charge-limited current (SCLC), while it is Ohmic conduction in the HfOx device.

Advances in RRAM Technology: Identifying and Mitigating Roadblocks

Superior scalability, endurance, low power operation, retention, and operating speed of filamentary crystalline HfOx-based resistive random access memory (RRAM) makes this technology promising for implementation in exascale neuromorphic computing systems. Challenges, roadblocks for the implementation, and possible resolutions are discussed. Technological solutions to overcome RRAM variability (both device-to-device and cycle-to-cycle) and read instability are discussed. Major material properties and operation conditions controlling performance of the crystalline HfOx-based RRAM devices are linked to physical processes determining RRAM characteristics.

Improvement of switching uniformity in HfOx-based resistive random access memory with a titanium film and effects of titanium on resistive switching behaviors

In this study, the effects of the Ti layer on resistive switching behaviors of HfOx-based resistive switching random access memory (ReRAM) were investigated. After introducing a thin Ti layer (∼10 nm) between the Ta and HfOx layers, both cycle-to-cycle uniformity within one cell and cell-to-cell uniformity were significantly improved. In addition, the higher ON/OFF ratio was obtained owing to the resistance increase in the high-resistance state, and a high device yield (>90%) also was achieved. The improvement of switching uniformity can be explained by the Ti doping effect, which is caused by the diffusion of Ti into the HfOx layer. In addition, the Ti layer had the effect of generating more oxygen vacancies in the HfOx layer, which led to the lowering of forming voltage (∼2.6 V). To confirm the change in the amount of oxygen vacancies, X-ray photoelectron spectroscopy analysis was performed. The formation of the TiOx layer and the Ti doping effect are considered to contribute to the generation of more oxygen vacancies in the HfOx layer.

Nanometer-Scale ${\rm HfO}_{x}$ RRAM

HfOx-based resistive random access memory with an active area down to few nanometers in diameter is fabricated and characterized. Scaling trends for forming and switching characteristics are presented. For the smallest device with an active area of few nanometers in diameter, ac switching endurance of 10 <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">8</sup> cycles with more than 100 × resistance window is demonstrated. In addition, multiple resistance states are shown to be stable after 10 <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">5</sup> read cycles and 10 <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">5</sup> s baking at 150 <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">°</sup> C.