Resistive Random Access Memory Research Articles

Oxygen-Plasma-Treated Al/TaOX/Al Resistive Memory for Enhanced Synaptic Characteristics.

In this study, we investigate the impact of O2 plasma treatment on the performance of Al/TaOX/Al-based resistive random-access memory (RRAM) devices, focusing on applications in neuromorphic systems. Comparative analysis using scanning electron microscopy and X-ray photoelectron spectroscopy confirmed the differences in chemical composition between O2-plasma-treated and untreated RRAM cells. Direct-current measurements showed that O2-plasma-treated RRAM cells exhibited significant improvements over untreated RRAM cells, including higher on/off ratios, improved uniformity and distribution, longer retention times, and enhanced durability. The conduction mechanism is investigated by current-voltage (I-V) curve fitting. In addition, paired-pulse facilitation (PPF) is observed using partial short-term memory. Furthermore, 3- and 4-bit weight tuning with auto-pulse-tuning algorithms was achieved to improve the controllability of the synapse weight for the neuromorphic system, maintaining retention times exceeding 103 s in the multiple states. Neuromorphic simulation with an MNIST dataset is conducted to evaluate the synaptic device.

Influence of electrodes on the resistive switching characteristics of Al/Gd2Zr2O7/E (E=Al or ITO) RRAM devices

Sol-gel derived amorphous Gd2Zr2O7 (GZO) thin films with Al/GZO/E (E = ITO or Al) structures were prepared to demonstrate the effect of electrode material on the switching behavior of resistive random access memory (RRAM) devices. All devices exhibit bipolar resistive switching (BRS) mode. After the post-metal annealing (PMA) treatment, the AlOx interfacial layer contributes to enhancing oxygen ion storage capacity and limiting the electromigration of aluminum metal. The forming-free Al/GZO/ITO device exhibits a switching cycle of 3027, a Vset/Vreset ratio of −1.83 V/0.56 V, a Ron/Roff ratio of approximately 10, and a retention time of 104 s. And the Al/GZO/Al device features a dual-layer AlOx structure, significantly improving the integrity of filament rupture to reduce leakage current and increase the switching window. Its switching cycle is 1946, with a Vset/Vreset ratio of −3.47 V/0.29 V, a Ron/Roff ratio of approximately 102, and a retention time of 104 s. The oxygen storage capacity of the dual-layer AlOx interfacial layer in the Al/GZO/Al device surpasses that of the Al/GZO/ITO device, allowing for a reduction in the reset voltage and ensuring more complete filament rupture, thereby enhancing the RHRS. Accordingly, GZO thin films can be applied to achieve the desired RS characteristics on different electrodes, making them promising candidates for RRAM applications.

Convolutional neural network for high-performance reservoir computing using dynamic memristors

In the rapidly advancing field of neuromorphic computing, W/ZnO/TiN resistive random-access memory (RRAM) devices have emerged as a next-generation computational building block. Our findings reveal the significant role played by the thickness of the ZnO layer in determining the electrical properties essential for data storage and neuromorphic applications. The short-term memory (STM) capabilities, which are critical for processing temporal information, are closely examined alongside their potential to simulate biological synaptic functions through multilevel conductance states and synaptic behaviors such as paired-pulse facilitation. Integrating these devices into reservoir computing systems enhances pattern recognition and accelerates learning, which demonstrates their utility in sequential data processing. In addition, conductance modulation via pulse width adjustment is a novel strategy to optimize memory device performance. By showcasing the effectiveness of W/ZnO/TiN devices in neuromorphic computing through high-accuracy image recognition tasks, our study highlights their foundational role in advancing neuromorphic computing technologies. The adaptability, learning capabilities, and efficiency of these devices underscore their potential for developing hardware-based neuromorphic systems that are capable of complex data processing.

Resistive random access memory based on organic-metallic hybrid polymer

Abstract Resistive random access memories (ReRAMs) based on an organic-metallic hybrid polymer, Poly(Fe-btpyb) Purple, were fabricated. This material was synthesized by complexation of metal ion and organic ligand. When applying forward bias, an abrupt resistance change from a low resistance state (LRS) to a high resistance state (HRS), which is known as reset process, was observed. In contrast, a reverse bias switched the resistance from HRS to LRS (set process). The resistive switching phenomenon is probably caused by the electrochemical oxidation-reduction reaction of the metal ion (Fe(II)/Fe(III)). The nonvolatile memory characteristics were measured with data-retention tests, showing no significant degradation over 105 s. The endurance characteristics exhibited sufficient long-term durability, due to no conformational change of the organic ligand. It is proposed that the difference in charge-transfer efficiency between the reduced state (Fe(II)) and oxidized state (Fe(III)) might be the physical mechanism of the resistive switching.

Vanadium Dioxide by Atomic Layer Deposition: A Promising Material for Next-Generation Memory Devices.

The synthesis of a vanadium dioxide (VO2) film using atomic layer deposition (ALD) with vanadium tetrachloride (VCl4) as a precursor for the realization of programmable memory devices is reported. X-ray diffraction analysis revealed the epitaxial growth of VO2 on c-Al2O3. The phase transition was monitored using resistivity measurements across varying temperatures, demonstrating a decrease of >4 orders of magnitude at the transition temperature, thereby confirming the high quality of the material. From this material, memristive devices are fabricated as resistive random-access memory (RRAM). On the basis of spiking voltage inputs, these RRAM exhibited cycle stability over 512 cycles and state retention stability for >450 s, showing <2% drift. With respect to synaptic-like applications, the RRAM devices were piloted through step patterns to enable multilevel memory states. These ALD-grown VO2-based devices demonstrate potential for use as synaptic connections with multiweight synapses, advancing scalability in neuromorphic applications.

Enhancing reliability in oxide-based memristors using two-dimensional transition metal dichalcogenides

Oxide-based memristor is an attractive candidate for future nonvolatile resistive random access memory (RRAM) devices. However, it suffers from insufficient reliability, owing to the randomness of the conductive filaments, hindering the practical use of the memristor for future RRAM applications. Here, we propose harnessing the two-dimensional (2D) transition metal dichalcogenides (TMDs) on oxide memristor to achieve high device reliability by controlling oxygen vacancy-based filaments near the TMDs/oxide interface. By forming the Pt/WSe2/HfxZr1-xO2 (HZO)/TiN structure, the fabricated memristor exhibits high reliability with good cyclic endurance (over 2,000 cycles), retention (104 s), and low cycle-to-cycle variability. Surface chemical analysis reveals the abundant oxygen vacancies induced by forming WSe2/HZO interface are the source of filamentary switching. By incorporating 2D materials and oxides, the practical application of memristor to future information processing devices can be boosted by the enhanced device reliability.

ReSG: A Data Structure for Verification of Majority-based In-memory Computing on ReRAM Crossbars

Recent advancements in the fabrication of Resistive Random Access Memory (ReRAM) devices have led to the development of large-scale crossbar structures. In-memory computing architectures relying on ReRAM crossbars aim to mitigate the processor-memory bottleneck that exists with current complementary metal-oxide semiconductor technology. With this motivation, several synthesis and mapping approaches focusing on the realizations of Boolean functions in the ReRAM crossbars have been proposed earlier. Thus far, the verification of the designs realized on ReRAM crossbars is done either through manual inspection or using simulation-based approaches. Since manual inspections and simulation-based approaches are limited to smaller designs, they cannot be applied to the verification of complex designs on large-scale ReRAM crossbars. Motivated by this, we propose, for the first time, an automatic equivalence checking flow that determines the equivalence between the original function specification (e.g., Majority-inverter Graph ) and the crossbar micro-operations file formats. We consider two crossbar structures, zero-transistor, one-memristor (0T1R) and one-transistor, one-memristor (1T1R) to implement the micro-operations. While the micro-operations file format exists for 0T1R crossbar structures, no representations for micro-operations to be executed in 1T1R crossbars exist yet. In this work, we introduce the micro-operation file format for 1T1R crossbar structures to efficiently represent the micro-operations as ReRAM crossbar netlists. Afterwards, we introduce two intermediate data structures, ReRAM Sequence Graph for 0T1R crossbars (ReSG-0T1R) and for 1T1R crossbars (ReSG-1T1R) , that are derived from the 0T1R and 1T1R crossbar micro-operations file formats, respectively. These ReSGs are then translated into Boolean Satisfiability (SAT) formula, and then the verification is done by checking the generated SAT formulae against the golden functional specification (represented in Verilog) using Z3 Satisfiability solver. Experimental evaluations confirm the effectiveness of the proposed verification methodology on MCNC and ISCAS benchmarks.

Flexible memristors with low-operation voltage and high bending stability based on Cu2AgBiI6 perovskite

Cu2AgBiI6 films were prepared by a one-step spin coating method, and flexible memristors with an Ag/PMMA/Cu2AgBiI6/ITO structure were constructed. The devices showed a bipolar resistive switching behavior with low switching voltage, which is beneficial for reducing energy consumption. Furthermore, this study found that the device exhibits an endurance of about 900 cycles, a higher ON/OFF ratio of over 103, a long retention time (∼104 s), and high stabilities against mechanical stress. Remarkably, the present flexible memristor displayed extraordinary flexibility and stability, with no significant change for the resistive switching behavior even at various bending angles or after undergoing 900 bending cycles. This study establishes that the lead-free halide perovskite Cu2AgBiI6 can be used for the resistive random-access memory of flexible electronics.

Study of resistive properties and neural response of ZrO2/TiO2 heterojunction nanowire array (NWA) RRAM

Nowadays, the researcher has an eye on enhancing the performance of resistive random-access memory (RRAM) and exploring new demands and applications, particularly in the field of artificial intelligence and neural computing. It is crucially important for accurately replicating the observed plasticity within synapses to achieve superior RRAM performance in low-dimensional nanomaterials. In this work, we focus on addressing certainly unsatisfactory electrical properties of RRAM based on titanium oxide nanowire array (TiO2 NWA), for example, short retention time (∼103 s). Theoretically and experimentally, a novel and solution-based RRAM structure, namely ZrO2/TiO2(NWA), has been conducted. With the help of the ZrO2 insertion layer, the RRAM device shows impressive resistive switching characteristics, including 200 current–voltage(I-V) sweeps without degradation, increased cycling endurance (>105 cycles), and a long retention time (∼9 × 104 s). These findings hold significant implications for digital computing systems. Moreover, the successful implementation of the proposed device enables the emulation of fundamental synaptic biological features such as non-linear transmission properties, learning experiences behavior, short-term potentiation (STP) and long-term potentiation (LTP). This research provides evidence of the immense potential of the Ag/ZrO2/TiO2(NWA)/FTO RRAM device in bipolar non-volatile memory (NVM) and biomimetic neuromorphic systems.

Complementary resistive switching characteristics of solid electrolyte chalcogenide AgxTe nanoparticles for high-density crossbar random access memory

Silver telluride (AgxTe) is a member of the chalcogenide family that comprises materials extensively used as solid electrolytes. Because of its high-ionic conductivity, low-optical bandgap, and excellent thermoelectric properties, AgxTe has been studied in many research fields, including optoelectronics and energy harvesting. Herein, AgxTe is proposed as the active channel for resistive random access memory (RRAM) showing complementary resistive switching (CRS) characteristics. AgxTe-based RRAM devices with an Ag/AgxTe/Au structure are fabricated on a glass substrate. AgxTe nanoparticles are synthesized using the colloidal method, and AgxTe thin films are prepared via spin coating of the synthesized nanoparticles dispersed in deionized water. The fabricated AgxTe-based RRAM device exhibits CRS characteristics without any additional built-in selectors or antiserial arrangement. This is attributed to the formation of the inversion of CF geometry and allows the fabrication of high-density crossbar arrays. The AgxTe RRAM device annealed at 200 °C exhibits a resistance on/off ratio of approximately 102 as well as stable retention (∼104 s) and endurance (∼103 cycles). This investigation proposes a new application of AgxTe, as a solid electrolyte, and a new strategy for the development of high-density crossbar RRAM architectures, for the first time.

Analysis of conductive filament heat transfer in TiO2-based RRAM device

Abstract Resistive random access memory (RRAM) has developed into a new type of non-volatile memory that has attracted much attention for its high density and low power; TiO2 has been studied to make RRAM because of its high conductivity. Based on the COMSOL Multiphysics finite element method and domain decomposition method, an electrically-thermally coupled model of TiO2-based RRAM with oxygen vacancy (Vo) conduction mechanism is constructed, the formation and cutting process of RESET and SET conductive filament (CF) under different voltages is simulated. Moreover, the characteristics of the internal temperature distribution of the CF are explored. The results show that the RESET process is more sensitive to heat changes, the transverse direction thermal value of the CF is more obvious than the longitudinal direction, and the rate of change is faster. Additionally, the electric field induces the migration of the Vo in filament, which affects the enthalpy change of the device’s heat transfer and resistive properties. This is an important reference for a deeper understanding of the switching behavior of RRAM devices and the control mechanisms for thermal studies.

Bipolar resistive switching behavior of PVA/g-C3N4 quantum dots hybrid thin films

Graphite-phase carbon nitride quantum dots (g-C3N4 QDs) have garnered widespread attention and considerable research interest as active media for resistive random-access memory (RRAM) devices owing to their exceptional physical properties. In this study, g-C3N4 QDs were synthesized via ultrasonic-assisted liquid-phase exfoliation. Subsequently, the obtained g-C3N4 QDs with uniform size were embedded into polyvinyl alcohol (PVA) for preparing PVA/g-C3N4 QDs hybrid, and the RRAM devices with Al/g-C3N4 QDs-PVA/ITO/PET structure were fabricated. The obtained memory devices exhibit bipolar resistive switching (BRS) behavior and are primarily influenced by the space charge limited conduction (SCLC) mechanism. The data retention time exceeds 104 s at a read voltage of 0.1 V, accompanied by an impressively low set voltage. In addition, the device exhibits a surprising temperature dependence of the resistive switching performance at different temperatures. The aforementioned properties demonstrate the significant potential of g-C3N4 QDs in RRAM for data storage applications.

Controlled Fabrication of Native Ultra‐Thin Amorphous Gallium Oxide From 2D Gallium Sulfide for Emerging Electronic Applications

AbstractOxidation of 2D layered materials has proven advantageous in creating oxide/2D material heterostructures, opening the door for a new paradigm of low‐power electronic devices. Gallium (II) sulfide (β‐GaS), a hexagonal phase group III monochalcogenide, is a wide bandgap semiconductor with a bandgap exceeding 3 eV in single and few‐layer form. Its oxide, gallium oxide (Ga2O3), combines a large bandgap (4.4–5.3 eV) with a high dielectric constant (≈10). Despite the technological potential of both materials, controlled oxidation of atomically‐thin β‐GaS remains under‐explored. This study focuses on the controlled oxidation of β‐GaS using oxygen plasma treatment, addressing a significant gap in existing research. The results demonstrate the ability to form ultrathin native oxide (GaSxOy), 4 nm in thickness, upon exposure to 10 W of O2, resulting in a GaSxOy/GaS heterostructure where the GaS layer beneath remains intact. By integrating such structures between metal electrodes and applying electric stresses as voltage ramps or pulses, their use for resistive random‐access memory (ReRAM) is investigated. The ultrathin nature of the produced oxide enables low operation power with energy use as low as 0.22 nJ per operation while maintaining endurance and retention of 350 cycles and 104 s, respectively. These results show the significant potential of the oxidation‐based GaSxOy/GaS heterostructure for electronic applications and, in particular, low‐power memory devices.

Antiferroelectric Heterostructures Memristors with Unique Resistive Switching Mechanisms and Properties.

A novel antiferroelectric material, PbSnO3 (PSO), was introduced into a resistive random access memory (RRAM) to reveal its resistive switching (RS) properties. It exhibits outstanding electrical performance with a large memory window (>104), narrow switching voltage distribution (±2 V), and low power consumption. Using high-resolution transmission electron microscopy, we observed the antiferroelectric properties and remanent polarization of the PSO thin films. The in-plane shear strains in the monoclinic PSO layer are attributed to oxygen octahedral tilts, resulting in misfit dislocations and grain boundaries at the PSO/SRO interface. Furthermore, the incoherent grain boundaries between the orthorhombic and monoclinic phases are assumed to be the primary paths of Ag+ filaments. Therefore, the RS behavior is primarily dominated by antiferroelectric polarization and defect mechanisms for the PSO structures. The RS behavior of antiferroelectric heterostructures controlled by switching spontaneous polarization and strain, defects, and surface chemistry reactions can facilitate the development of new antiferroelectric device systems.

High-Performance Memristors Based on Imine-Linked Covalent Organic Frameworks Obtained by Using a Protonation Modification Strategy.

Imine-linked covalent organic frameworks (COFs) are garnering substantial interest in resistive random-access memory, attributed to their superior crystallinity, excellent chemical and thermal stability, and modifiable molecular structures. However, the development of high-performance COF-based memristors impeded by challenges such as low conjugation degree of imine bonds and poor electron delocalization ability. Herein, we report a protonation strategy to modify the imine bonds of donor-acceptor (D-A) type COFs. This modification significantly enhances the electron delocalization capability of imine bonds, lowers the energy barriers for electron injection from electrodes, and stabilizes the conductive charge transfer state, thus markedly improving device performance. The protonated COF-BTT-BPy and COF-BTT-TAPT thin films-based memristors show remarkable device performance with a high ON/OFF current ratio of 105, a low driving voltage, and outstanding endurance exceeding 600 and 1300 cycles, respectively, which is nearly twice the durability of analogous non-protonated COFs-based memristors. Notably, the protonated COF-BTT-TAPT-based memristor exhibit the highest number of cycles reported at present. This work not only unprecedentedly enhances the performance of COF-based memristors, but also provides a universal and promising approach for the molecular design and potential application of D-A type imine-linked COFs.

Improved resistive switching characteristics observed in amorphous boron nitride-based RRAM device via oxygen doping: A study based on bulk and interface traps analysis

In this paper, we propose an oxygen-doped amorphous boron nitride (O-doped a-BN)-based resistive random-access memory (RRAM) to address the low on/off ratio and stability concerns of conventional a-BN-based RRAM. Initially, comparative analysis with a-BN based RRAM device revealed that the O-doped a-BN based RRAM exhibited 2.5-fold increase in the on/off ratio, enhanced endurance with stable operation over 100 DC cycles, and improved retention stability lasting for 104 s. Then, on the basis of both interface and bulk trap analysis using impedance spectroscopy and the Terman method, the bulk trap density in the O-doped a-BN and a-BN films acts as a key parameter in RS properties, which results in an increase of its on/off ratio and its stable retention without any degradation. These findings show that O-doping in a-BN films can be a technique to improve the on/off ratio and reliability of a-BN-based RRAM devices.

HfO2-based resistive random access memory with an ultrahigh switching ratio

Resistive Random Access Memory (RRAM) is considered one of the most promising candidates for big data storage. By using atomic layer deposition and magnetron sputtering, HfO2 thin films were prepared on ITO first, which exhibited good resistive switching (RS) characteristics in the structure of Ag/HfO2/ITO. By analyzing the RS mechanism, it is found that both metal conductive filaments and oxygen vacancy conductive filaments coexisted and Sn ion in ITO can influence the retention of RRAM. Furthermore, a device in the structure of Ag/HfO2/Pt was proposed and prepared, which exhibited excellent RS characteristics, including an ultrahigh switching ratio averaging up to 108 and low operating voltage. It is concluded that the difference in the work function between the top and bottom electrodes contributes to improving the switching ratio, reducing the operating voltage. In addition, the Ag/HfO2/Pt device is similar to the Ag/HfO2-based threshold switching selector in the structure and in characteristics of high switching ratio, besides non-volatile memory. Hence, the device is functionally equivalent to the combination of an RRAM and a threshold switching selector. It is the potential way to replace the conventional 1S1R structure memory.

TEFLON: Thermally Efficient Dataflow-aware 3D NoC for Accelerating CNN Inferencing on Manycore PIM Architectures

Resistive random-access memory (ReRAM)-based processing-in-memory (PIM) architectures are used extensively to accelerate inferencing/training with convolutional neural networks (CNNs). Three-dimensional (3D) integration is an enabling technology to integrate many PIM cores on a single chip. In this work, we propose the design of a t hermally e fficient data flo w-aware monolithic 3D (M3D) N oC architecture referred to as TEFLON to accelerate CNN inferencing without creating any thermal bottlenecks. TEFLON reduces the Energy-Delay-Product (EDP) by 42%, 46%, and 45% on an average compared to a conventional 3D mesh NoC for systems with 36-, 64-, and 100-PIM cores, respectively. TEFLON reduces the peak chip temperature by 25 K and improves the inference accuracy by up to 11% compared to sole performance-optimized SFC-based counterpart for inferencing with diverse deep CNN models using CIFAR-10/100 datasets on a 3D system with 100-PIM cores.

An ab-initio investigation of physical characteristics of GdSc1-xTMxO3(TM=Ti; x=0.25, 0.5, 0.75) for photo influenced (NIR)memory storage and allied devices

In the present study, we have computed the structural, electronic, and optical characteristics of pristine GdScO3 and GdSc1-xTMxO3 (TM = Ti; x = 0.25, 0.5, 0.75) for improved performance of Resistive Random Access Memory (RRAM) devices. All calculations have been performed using the Tran Balha-modified Becke Johnson (TB-mBJ) approximation within the framework of the WEIN2K simulation package. Herein, the effect of Ti-dopant (transition metal, i.e., TM) with varying concentrations has been observed, especially in tuning the energy band gap. The structural analysis of composites reveals structural stability and a significant reduction in bulk modulus with increasing dopant concentration, leading to enhanced electrical conductivity. The band structure analysis reveals that the band gap tends to decrease with increasing concentrations of Ti-dopant.As regards total density of states (TDOS), it has been found that with increasing concentrations of Ti, dopant extra states appear in the conduction band. This is the reason behind the formation of the conducting filaments (CF) within the active layer of RRAM devices. The partial density of states (PDOS) elaborates that for all studied composites, the formation of the conduction band (C.B.) and valence band (V.B.) is a consequence of the hybridization of Sc-3d, Gd-5d, and Ti-4d states, along with the minute contribution of O-2p states. Optical characteristics disclose the photoresponse of all studied composites, wherein GdSc0.25Ti0.75O3 exhibits wide range absorption, henceforth making it a making it a more suitable candidate for photoinfluencing, especially in near infrared (NIR) RRAM devices.

Data Pruning-enabled High Performance and Reliable Graph Neural Network Training on ReRAM-based Processing-in-Memory Accelerators

Graph Neural Networks (GNNs) have achieved remarkable accuracy in cognitive tasks such as predictive analytics on graph-structured data. Hence, they have become very popular in diverse real-world applications. However, GNN training with large real-world graph datasets in edge-computing scenarios is both memory- and compute-intensive. Traditional computing platforms such as CPUs and GPUs do not provide the energy efficiency and low latency required in edge intelligence applications due to their limited memory bandwidth. Resistive random-access memory (ReRAM)-based processing-in-memory (PIM) architectures have been proposed as suitable candidates for accelerating AI applications at the edge, including GNN training. However, ReRAM-based PIM architectures suffer from low reliability due to their limited endurance, and low performance when they are used for GNN training in real-world scenarios with large graphs. In this work, we propose a learning-for-data-pruning framework, which leverages a trained Binary Graph Classifier (BGC) to reduce the size of the input data graph by pruning subgraphs early in the training process to accelerate the GNN training process on ReRAM-based architectures. The proposed light-weight BGC model reduces the amount of redundant information in input graph(s) to speed up the overall training process, improves the reliability of the ReRAM-based PIM accelerator, and reduces the overall training cost. This enables fast, energy-efficient, and reliable GNN training on ReRAM-based architectures. Our experimental results demonstrate that using this learning for data pruning framework, we can accelerate GNN training and improve the reliability of ReRAM-based PIM architectures by up to 1.6×, and reduce the overall training cost by 100× compared to state-of-the-art data pruning techniques.