Oxide-based Resistive Switching Research Articles

Unraveling the importance of fabrication parameters of copper oxide-based resistive switching memory devices by machine learning techniques

In the present study, various statistical and machine learning (ML) techniques were used to understand how device fabrication parameters affect the performance of copper oxide-based resistive switching (RS) devices. In the present case, the data was collected from copper oxide RS devices-based research articles, published between 2008 to 2022. Initially, different patterns present in the data were analyzed by statistical techniques. Then, the classification and regression tree algorithm (CART) and decision tree (DT) ML algorithms were implemented to get the device fabrication guidelines for the continuous and categorical features of copper oxide-based RS devices, respectively. In the next step, the random forest algorithm was found to be suitable for the prediction of continuous-type features as compared to a linear model and artificial neural network (ANN). Moreover, the DT algorithm predicts the performance of categorical-type features very well. The feature importance score was calculated for each continuous and categorical feature by the gradient boosting (GB) algorithm. Finally, the suggested ML guidelines were employed to fabricate the copper oxide-based RS device and demonstrated its non-volatile memory properties. The results of ML algorithms and experimental devices are in good agreement with each other, suggesting the importance of ML techniques for understanding and optimizing memory devices.

Binary metal oxide-based resistive switching memory devices: A status review

Binary metal oxide-based resistive switching memory devices: A status review

Fabrication of Graphene Oxide-Based Resistive Switching Memory by the Spray Pyrolysis Technique for Neuromorphic Computing

Fabrication of Graphene Oxide-Based Resistive Switching Memory by the Spray Pyrolysis Technique for Neuromorphic Computing

Effect of annealing on forming-free bipolar resistive switching of Gd2O3 thin films

Oxide-based resistive switching devices are a promising candidate for non-volatile memory applications. Here, we present a Gd2O3-based non-volatile memory device deposited using e-beam evaporation. In order to know the effect of annealing on the device performance, the as-grown sample was annealed at 500 °C. The capacitance (C) – voltage (V) and conductance (G) – voltage (V) measurements were investigated in the range of 200 kHz to 1 MHz. It was observed that the C-V and G-V changes with frequency due to the presence of interface states. The effect of series resistance (Rs) and interface state density (Dit) was also investigated for both devices. Furthermore, the resistive switching was recorded at± 5 V. Both devices exhibit bipolar resistive switching with charge conduction mechanism ascribed to ohmic, space charge limited conduction, and trap-filled space charge limited conduction. Consequently, the annealed device displayed an improved resistance ratio of 102 and a stable endurance of up to 1000 cycles with better data retention of over 103 s.

Machine learning-assisted design guidelines and performance prediction of CMOS-compatible metal oxide-based resistive switching memory devices

Machine learning (ML) has accelerated the discovery of new materials and properties of electronic devices, reducing development time and increasing efficiency. In this study, ML was used to provide design guidelines and predict the performance of industry-standard resistive switching (RS) memory devices based on HfO2/x, Ta2O5, and TaOx materials. The model building, analyses, and prediction processes were based on a database of peer-reviewed articles published between 2007 and 2020. More than 15,000 property entries were used for our ML tasks. Moreover, supervised and unsupervised ML techniques were used to provide design guidelines for the categorical and continuous feature sets. In addition, a linear model, artificial neural network, and the random forest algorithm were employed to predict the continuous-type features, and gradient boosting was used to understand how device parameters can affect RS performance. Finally, the ML predictions were validated by fabricating the corresponding RS devices. The results indicated that the ML techniques accelerated the discovery and understanding of different RS properties.

Enhanced resistive switching performance of hafnium oxide-based devices: Effects of growth and annealing temperatures

Although hafnium oxide-based resistive switching (RS) devices have been explored extensively, extremely high operating voltages and currents hinder their application in low power operations. In this study, the growth temperature and annealing conditions have been varied to significantly improve the characteristics of hafnium oxide-based RS devices. Deposited using a rf magnetron sputtering at different temperatures, HfOx films are subsequently subjected to vacuum annealing to create a high density of oxygen vacancies. It is observed that the crystallinity, defect density, and RS device performance of these films are influenced by these thermal treatments. The HfOx film annealed at 450 ºC exhibited bipolar RS with the substantial improvement in the device performance, such as electroforming, set/reset voltages, and operating currents as low as ~1 V, 0.5/− 0.5 V, and ~1 µA, respectively, in addition to large ON/OFF ratio> 103, good endurance (>103 cycles), and retention capability (>104 s). The localized conductive filaments composed of oxygen vacancies are found to be responsible for the RS behavior in Al/HfOx/FTO devices, and the simple vacuum annealing approach considerably increases the proportion of oxygen vacancies. With the insights into the influence of annealing temperatures on the switching parameters, this work paves the way for the design of low-power devices for data storage applications.

Improved threshold switching characteristics of vanadium oxide/oxynitride-based multilayer selector in a cross-point array

We developed a selector device with the bi-directional threshold switching characteristics of vanadium oxide (VOx) by implementing vanadium oxynitride (VOx:Ny) in a sandwich structure. The proposed device with the Pt/VOx:Ny/VOx/VOx:Ny/Pt structure as opposed to a cell with a Pt/VOx/Pt structure exhibits lower off-current and stable switching characteristics. The elements of the device were analyzed using high-resolution transmission electron microscopy and X-ray photoelectron spectroscopy, and its memory characteristics were verified by applying the Pt/VOx:Ny/VOx/VOx:Ny/Pt selector to amorphous indium-gallium-zinc oxide-based resistive switching (RS) devices in the 1 selector-1 ReRAM structure. Our device shows stable RS properties over 100 DC cycles and> 106 AC cycles, an improved ON/OFF ratio (~67), a high switching speed (<110 ns), and a rectification ratio of ~1 × 103 when the read margin is 10%.

Wafer-Scalable Single-Layer Amorphous Molybdenum Trioxide.

Molybdenum trioxide (MoO3), an important transition metal oxide (TMO), has been extensively investigated over the past few decades due to its potential in existing and emerging technologies, including catalysis, energy and data storage, electrochromic devices, and sensors. Recently, the growing interest in two-dimensional (2D) materials, often rich in interesting properties and functionalities compared to their bulk counterparts, has led to the investigation of 2D MoO3. However, the realization of large-area true 2D (single to few atom layers thick) MoO3 is yet to be achieved. Here, we demonstrate a facile route to obtain wafer-scale monolayer amorphous MoO3 using 2D MoS2 as a starting material, followed by UV-ozone oxidation at a substrate temperature as low as 120 °C. This simple yet effective process yields smooth, continuous, uniform, and stable monolayer oxide with wafer-scale homogeneity, as confirmed by several characterization techniques, including atomic force microscopy, numerous spectroscopy methods, and scanning transmission electron microscopy. Furthermore, using the subnanometer MoO3 as the active layer sandwiched between two metal electrodes, we demonstrate the thinnest oxide-based nonvolatile resistive switching memory with a low voltage operation and a high ON/OFF ratio. These results (potentially extendable to other TMOs) will enable further exploration of subnanometer stoichiometric MoO3, extending the frontiers of ultrathin flexible oxide materials and devices.

Impact of Zr top electrode on tantalum oxide-based electrochemical metallization resistive switching memory: towards synaptic functionalities.

Electrochemical metallization memory (ECM) devices have been made by sub-stoichiometric deposition of a tantalum oxide switching film (Ta2O5−x) using sputtering. We investigated the influence of zirconium as the active top electrode material in the lithographically fabricated ECM devices. A simple capacitor like (Pt/Zr/Ta2O5−x/Pt) structure represented the resistive switching memory. A cyclic voltammetry measurement demonstrated the electrochemical process of the memory device. The I–V characteristics of ECMs show stable bipolar resistive switching properties with reliable endurance and retention. The resistive switching mechanism results from the formation and rupture of a conductive filament characteristic of ECM. Our results suggest that Zr can be considered a potential active electrode in the ECMs for the next generation of nonvolatile nanoelectronics. We successfully showed that the ECM device can work under AC pulses to emulate the essential characteristics of an artificial synapse by further improvements.

“Unraveling the Importance of Fabrication Parameters of Copper Oxide-Based Resistive Switching Memory Devices by Machine Learning Techniques”

“Unraveling the Importance of Fabrication Parameters of Copper Oxide-Based Resistive Switching Memory Devices by Machine Learning Techniques”

Investigation of weight updating modes on oxide-based resistive switching memory synapse towards neuromorphic computing applications

Investigation of weight updating modes on oxide-based resistive switching memory synapse towards neuromorphic computing applications

Impact of the Ohmic Electrode on the Endurance of Oxide-Based Resistive Switching Memory

As one of the key aspects in the reliability of redox-based resistive switching memories (ReRAMs), maximizing their endurance is of high relevance for industrial applications. The major limitation regarding endurance is considered the excessive generation of oxygen vacancies during cycling, which eventually leads to irreversible RESET failures. Thus, the endurance could be increased by using combinations of switching oxide and ohmic electrode (OE) metal that provides a high barrier for the generation of oxygen vacancies [defect formation energy (DFE)]. In this work, we present a sophisticated programming algorithm that aims to maximize the endurance within reasonable measurement time. Using this algorithm, we compare ReRAM devices with four different OE metals and confirm the theoretically predicted trend. Thus, our work provides valuable information for device engineering toward higher endurance.

HRS Instability in Oxide-Based Bipolar Resistive Switching Cells

One of the key challenges in the reliability of valence change [valence change-based memory (VCM)] resistive switching random access memories (ReRAMs) is the short-term instability of the programed state. Due to read noise, program verify or shaping algorithms are ineffective and read current (or resistance) distributions always revert to their intrinsic statistics. In this work, we analyze the instability of the high resistive state (HRS) measured on ZrO <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sub> -based devices via Factorial Hidden Markov Models. The extracted current jumps are explained by distinct ionic jumps via physics-based kinetic Monte Carlo (KMC) models. The simulation results reveal jumps of oxygen vacancies from the densely packed filament (plug) region to a sparsely packed gap (disc) region as origin of the most critical, large current jumps. These findings are used to extend our compact model (JART v1b) by a read noise module. We demonstrate simulated HRS instability in excellent agreement with our experimental data. Whereas the KMC approach provides a physical understanding of the processes underlying the HRS instability, the compact model enables the simulation of read noise up to industrially relevant array scales.

Endurance degradation of solution-processed ZnO polycrystalline film-based resistive switching memory

Metal oxide-based resistive switching memory cell has potential applications in the next generation of nonvolatile memory devices due to its simple structure, fast reading/writing speed, and good endurance. Zinc oxide (ZnO) is one of excellent resistive switching candidate materials owing to its low-cost, simple preparation process and significant resistive switching characteristics. In this work, homogeneous ZnO precursor solution is spin-coated on indium tin oxide (ITO)-coated glass substrate to deposit a smooth and dense film composed by ZnO nanoparticles with the hexagonal wurtzite structure. Atomic force microscopy images reveal the average grain size of ZnO polycrystalline nanoparticle is about 23.7 nm. An optical band gap of about 3.3 eV is derivate from the ultraviolet-visible light absorption spectra. Current-voltage ( I - V ) curves of Ag/ZnO/ITO memory cells exhibit reversible low-voltage bipolar resistive switching characteristics. Their switching voltages ( V Set/ V Reset) are lower than ± 0.4 V, and the ON/OFF resistance ratio is 103–104 at the read voltage of −0.1 V, which are superior to resistive switching performance of other ZnO polycrystalline films prepared via similar solution methods. However, the memory cell can only maintain 60 stable I - V cycles. After that, the low resistance state gradually degrades to the high resistance state, and the resistive switching memory window completely disappears after 100 I - V cycles. X-ray photoelectron spectra of Zn 2p and O 1s core-levels for ZnO film before and after I - V measurement demonstrate that as-grown film is oxygen-defective ZnO x I - V measurement. Under sustained electric field stress, some drifted oxygen ions combine with oxygen vacancies and change to localized lattice oxygen atoms, this partially remedies the oxygen-deficiency of as-grown non-stoichiometric ZnO x

Nano-scale single layer TiO2-based artificial synaptic device

Synaptic nano-electronic devices for brain-inspired computing have become increasingly popular because of their biological neuron-like properties such as massive parallelism with lower power consumption. Metal oxide-based resistive switching memory devices for the implementation of synapses are of great interest due to their low cost, easy production and complementary metal-oxide semiconductors (CMOS) compatibility. This study presents a simple, single-layer nano-scale TiO2-based artificial synaptic device for neuromorphic applications. The structural properties of the proposed nano-scale TiO2-based device were confirmed via both X-ray photoelectron spectroscopy (XPS) and energy dispersive X-ray spectroscopy (EDX). The bipolar resistive switching behavior of the device is shown by gradual increases in the low resistance (SET) state and gradual decreases in the high resistance (RESET) state. The synaptic characteristics of the device were resolved by applying voltage pulses. The typical potentiation and depression functions were obtained. Conforming to spike time dependent plasticity (STDP) was also achieved using synaptic weight changes. Homogeneous synaptic behaviors were associated with oxygen vacancies in the TiO2 layer, while abrupt changes in synaptic behavior were ascribed to filamentary transitions resulting from impurities in the metal oxide layer.

(Invited) Artificial Neurons and Synapses with CVD MoS2 Facilitated By Electrode Engineering

The prevalent von Neumann architecture in today’s processors involves memory and processing units to reside in physically separate locations. With memory speeds lagging behind the processor speeds, latency in accessing data from memory has resulted in the “von Neumann bottleneck”. To alleviate this issue, several alternative non-von Neumann architectures have been explored. Neuromorphic computing is one such non-von Neumann approach, inspired by the human brain’s ability of cognitive recognition. The brain operates through a network of neurons that are connected to each other by synapses. In 2008, after the discovery of memristors, the fourth circuit element, researchers have explored the idea of mimicking synaptic behavior with a single memristive device.1-6 Albeit significant advances, synaptic devices using phase change materials-based memristors3,7-9 and metal oxide-based resistive switching devices5,6,10-13 have some limitations. These devices exhibit high programming current in the range of µA to mA. Additionally, the synaptic weight update, i.e. the increase (decrease) of the synapse’s conductance with the application of a continuous stream of identical positive (negative) input voltage pulses, is non-linear. Non-linearity in weight update increases the complexity of using these devices for real-time unsupervised learning. So, it becomes necessary to employ a materials system which exhibits low programming current as well as a linear weight update.14 Recently, two-dimensional (2D) materials are being largely explored to demonstrate their viability as electronic synapses and neurons.15-21 In this talk, we shall discuss the realization of a synaptic device using graphene/MoS2 heterostructures. In these devices, CVD-grown monolayer graphene acts as an electrode to CVD MoS2. These memristive devices exhibit low programming currents and a high dynamic range from 1 nA to 1 mA. In contrast with oxide-based or PCM-based synapses, these devices exhibit a gradual set and reset process when symmetric input voltage pulses are applied, resulting in a near-linear weight update. We shall also present the demonstration of an integrate-and-fire (IF) neuron using Ag/MoS2/Au vertical structures. These devices possess the four crucial features of an IF neuron – all-or-nothing spiking, threshold-driven firing, post-firing refractory period and stimulus strength based frequency response. Realizing neurons and synapses using the same materials system allows the monolithic integration of the essential building blocks of neuromorphic hardware, and bears potential for a highly scalable spiking neural networks suitable for unsupervised learning applications.

Influence of deposition pressure of CuTe electrode on the tantalum oxide-based resistive switching memory

The resistive switching characteristics of Pt/TaOx/CuTe devices were investigated with CuTe bottom electrodes deposited under different working pressure. The CuTe bottom electrode was prepared using a magnetron radio-frequency sputtering. The resistance and distribution of HRS were decreased with increasing deposition pressure of CuTe bottom electrode and then the resistance of LRS was increased however the distribution of LRS was unchanged. The CuTe bottom electrode was systematically analyzed to identify the effect of deposition condition on resistive switching behavior. The intermixing between the TaOx layer and CuTe bottom electrode was clearly identified as an active factor causing the variation of resistive switching parameters.

Selectorless Oxide-based Resistive Switching Memory with Nonuniform Dielectric for Low Power Crossbar Array Applications

Resistive random access memory (RRAM) using various binary metal oxide structures have attracted attentions since the current nonvolatile memory (NVM) has been approaching the scaling limit. RRAMs has excellent scalability, high speed and low power operation ability which is desired for current hardware needs i.e. high density storage memory and in-memory computing architecture. Meanwhile, selector devices are essential to suppress the sneak path issue, which causes reading errors and hinders the implementation of RRAM in high-density cross-bar array configuration. However, additional selector device as commonly proposed one selector-one resistor (1S-1R) design, rises the process complexity and cost (Fig. (a)). In this work, non-uniform dielectric selectorless memristor devices have been demonstrated by utilizing the nonlinear (NL) resistive switching (RS) characteristics with gap design method for RRAM array applications. The device characteristics with current transport mechanisms, and effective trap level below conduction band in stacks have also been investigated. Interestingly, a seasoning effect determination with nonlinearity improvement is also demonstrated in a bilayer stacking architecture. Specifically, the low-k layer quality can be examined by 1st RESET process electrically after electroforming process in non-uniform dielectric constant layers structure, which provides an intrinsically design of layered functionality with mechanism insight deeply. The RRAM devices with various side lengths of 400 nm have been fabricated. The starting substrates were heavily-doped N+ Si wafers. Titanium nitride (TiN) of 200 nm was deposited as bottom electrode. Then, 7 nm of SiOx and followed by 7 nm of HfOx were deposited as resistive switching dielectric layers for realizing the bilayer selectorless structures by radio frequency (RF) sputtering. The graphite oxide was formed after HfOx deposition as reported in our previous work. Platinum of 165 nm was then deposited as top electrodes, as followed by lift-off method for RRAM devices (Fig. (b)). The HfOx (11 nm) single layer devices, SiOx (10 nm) single layer, and HfOx (4 nm)/SiOx (9 nm) bilayer devices are used as references. An Agilent B1500 and Lakeshore probe station were used for electrical characterization of the RRAM devices. To consider the selectorless RRAM stability, Fig. (c) shows the examination process for low-k layer quality determination by analyzing forming and 1st RESET processes in detail (HfOx (4 nm)/ SiOx (9 nm) structure as example). The I-V characteristics with high electroforming voltage and leaky post-forming resistance did not perform the nonlinearity selectorless characteristics as compared to the I-V characteristics with low electroforming voltage and resistive post-forming resistance one, which indicated that overshoot damage by forming process can determine the nonlinearity and selectorless device reliability. Note, although the low forming voltage devices have resistive post-forming resistance, the 1st RESET process at higher voltage side shows a nonlinearity increasing region (read at -0.5V). Transport in this voltage range is found to fit well to the Fowler-Nordheim (F-N) tunneling formula (Red region in Figure and zoom-in plot of Fig. (d) inset) with scientifically accepted accuracy (R2 =99%), which shown that less overshoot damage in lower forming voltage may induce Fowler–Nordheim (F-N) tunneling, as compared to which with higher forming voltage devices (Ohmic behavior in LRS, data not shown here). The results also confirm that the seasoning effect determination with NL improvement by robust low-k layer quality and less forming process damage can improve the nonlinearity and selectorless device reliability. In this study, build-in NL characteristics have been realized for the non-uniform dielectric selectorless memristors without an diode/selector. The seasoning effect determination with nonlinearity NL is examined the low-k layer quality by 1st RESET process after electroforming process, which provides an intrinsically design of layered functionality with mechanism insights deeply. Benchmark our results shown that the achievement of high performance and excellent reliability of built-in nonlinearity in 1R-only selectorless RRAM (NL > 120, memory window > 100, sub-µs switching speed, and NL retention) for the future bilayer selectorless RRAM low-power memory array configurations (Fig. (e)). The non-uniform dielectric memristor are promising for high-density, low-power, selectorless RRAM array applications. The highly NL characteristics observed in non-uniform dielectric bilayer devices are desirable in suppressing sneak path currents in crossbar arrays. Figure 1

Selectorless Oxide-Based Resistive Switching Memory with Nonuniform Dielectric for Low Power Crossbar Array Applications

Selectorless resistive random-access memory (RRAM) with self-rectifying behavior is presented as a way to suppress sneak path current without an additional transistor or switch device integration, which is beneficial for reducing the cost and complexity of crossbar array fabrication. In this work, selectorless 1R-only graphite-based memristor devices have been demonstrated by utilizing the nonlinear (NL) resistive switching (RS) characteristics with well-designed operation conditions. The early stage of cycling observation as seasoning effect is investigated, while the wear-out mechanism is also discussed for the selectorless 1R-only RRAM crossbar array for high storage class memory applications.

A study on mechanism of resistance distribution characteristics of oxide-based resistive memory

Although oxide-based resistive switching memory (OxRAM) is one of the strong next-generation high capacity memory candidates, it has the critical disadvantage that deviations of resistance levels is too severe to be adopted as a high capacity memory device. More specifically, it is known that the larger on/off current ratios in multi-level operated OxRAMs, the greater deviation of resistance levels from the targeted values. However, despite the seriousness of the problem there has been no concrete theoretical study on the underlying mechanisms of the phenomenon. In this paper, we introduce a theoretical model that clearly explain the underlying mechanism of making such characteristics of programmed resistance levels in multi-level OxRAMs. From this model, we can understand why there is a proportional relationship between resistance level and its deviation, and why it has such a specific range of proportionality constant measured experimentally. And this understanding can certainly reveal the true limitations of OxRAMs’s performance.