Single Conductive Filament Research Articles

Improving the Recognition Accuracy of Memristive Neural Networks via Homogenized Analog Type Conductance Quantization.

Conductance quantization (QC) phenomena occurring in metal oxide based memristors demonstrate great potential for high-density data storage through multilevel switching, and analog synaptic weight update for effective training of the artificial neural networks. Continuous, linear and symmetrical modulation of the device conductance is a critical issue in QC behavior of memristors. In this contribution, we employ the scanning probe microscope (SPM) assisted electrode engineering strategy to control the ion migration process to construct single conductive filaments in Pt/HfOx/Pt devices. Upon deliberate tuning and evolution of the filament, 32 half integer quantized conductance states in the 16 G0 to 0.5 G0 range with enhanced distribution uniformity was achieved. Simulation results revealed that the numbers of the available QC states and fluctuation of the conductance at each state play an important role in determining the overall performance of the neural networks. The 32-state QC behavior of the hafnium oxide device shows improved recognition accuracy approaching 90% for handwritten digits, based on analog type operation of the multilayer perception (MLP) neural network.

Design of wafer-scale uniform Au nanotip array by ion irradiation for enhanced single conductive filament resistive switching

Here, we report a simple and low-cost method to fabricate wafer-scale uniform and flexible nanotip array substrates via ion irradiation and replica-molding process. The diameter of the nanotip can be accurately modulated, and the minimum tip size can reach approximately 50 nm. Then, we exploit this nanotip array to fabricate high-performance single conductive filament (CF) resistive switching memory. As a result, the tailored Ag/HfO2/Au nanotip resistive random access memory (RRAM) devices on polyethylene terephthalate (PET) substrates show good flexibility, a large switching window over 108, an ultralow off-state current of 10−12 A, and low and uniform programming voltages. It is believed that the randomness of the CFs is depressed by the enhanced local electric field near the Au nanotips, contributing to the reliable performance of the tailored Ag/HfO2/Au nanotip/PET-based RRAM devices. This novel method paves the way for fabrication of large single-filament flexible RRAM device arrays.

Reliable Multistate Data Storage with Low Power Consumption by Selective Oxidation of Pyramid-Structured Resistive Memory.

Multilevel data storage using resistive random access memory (RRAM) has attracted significant attention for addressing the challenges associated with the rapid advances in information technologies. However, it is still difficult to secure reliable multilevel resistive switching of RRAM due to the stochastic and multiple formation of conductive filaments (CFs). Herein, we demonstrate that a single CF, derived from selective oxidation by a structured Cu active electrode, can solve the reliability issue. High-quality pyramidal Cu electrodes with a sharp tip are prepared via the template-stripping method. Morphology-dependent surface energy facilitates the oxidation of Cu atoms at the tip rather than in other regions, and the tip-enhanced electric fields can accelerate the transport of the generated Cu ions. As a result, CF growth occurs mainly at the tip of the pyramidal electrode, which is confirmed by high-resolution electron microscopy and elemental analysis. The RRAM exhibits highly uniform and low forming voltages (the average forming voltage and its standard deviation for 20 pyramid-based RRAMs are 0.645 and 0.072 V, respectively). Moreover, all multilevel resistance states for the RRAMs are clearly distinguished and show narrow distributions within 1 order of magnitude, leading to reliable cell-to-cell performance for MLC operation.

A numerical study of multi filament formation in metal-ion based CBRAM

This study investigates the underlying mechanisms of multiple conductive filaments (CF) creation in metal-ion based conductive bridge RRAM (CBRAM) by using the Metropolis Monte Carlo algorithm and suggests a possible explanation for this phenomenon. The simulation method is demonstrated over a Cu/HfO2 structure, starting from a random initial distribution of oxygen vacancies (OV) defects in the resistive switching layer, to a formed CF and ending in a ruptured state. the results indicate that “Hot Spots” (HS), where agglomeration of OV trap like states for electron hopping based conduction induce local heating, create favorable energy conditions to attract diffused metal species originating from the top electrode. While HS may be created and annihilated by random OV generation and recombination processes, the precipitated metal forms a stem out of which a CF could evolve. The CF stem’s final growth stage is mainly driven by drift and diffusion. This process may lead to the formation of one or more CFs as a function of the forming bias voltage. This bias dependence is demonstrated over a large range, where the creation of a single, double and multiple CFs are shown. In addition, the reset process of the multi CF device is presented, and the experimentally observed, step like, gradual CBRAM reset is verified. The simulated results are in good agreement with experimental data and promote the idea that OV defect engineering may be used to improve CBRAM performance.

Imaging the Three-Dimensional Conductive Channel in Filamentary-Based Oxide Resistive Switching Memory.

Filamentary-based oxide resistive memory is considered as a disruptive technology for nonvolatile data storage and reconfigurable logic. Currently accepted models explain the resistive switching in these devices through the presence/absence of a conductive filament (CF) that is described as a reversible nanosized valence-change in an oxide material. During device operation, the CF cycles billion of times at subnanosecond speed, using few tens of microamperes as operating current and thus determines the whole device's performance. Despite its importance, the CF observation is hampered by the small filament size and its minimal compositional difference with the surrounding material. Here we show an experimental solution to this problem and provide the three-dimensional (3D) characterization of the CF in a scaled device. For this purpose we have recently developed a tomography technique which combines the high spatial resolution of scanning probe microscopy with subnanometer precision in material removal, leading to a true 3D-probing metrology concept. We locate and characterize in three-dimensions the nanometric volume of the conductive filament in state-of-the-art bipolar oxide-based devices. Our measurements demonstrate that the switching occurs through the formation of a single conductive filament. The filaments exhibit sizes below 10 nm and present a constriction near the oxygen-inert electrode. Finally, different atomic-size contacts are observed as a function of the programming current, providing evidence for the filament's nature as a defects modulated quantum contact.

Thermal analysis for observing conductive filaments in amorphous InGaZnO thin film resistive switching memory

Local heat produced by an electrical path inside the memory was detected and imaged by the method “Thermal Analysis.” It turned out that the visualized heat spots were conductive filaments (CFs) formed between interlayers of Pt/amorphous InGaZnO (a-IGZO). By using the thermal analysis, the location of CFs and their surface temperature was detected. This method indicated that there was a lot of emitted heat when the memory cell was switched off. It is thought to be accumulated heat causing disruption of the CFs. With great range of measurement, it was found that some memory cells drive with a single CF and others drive with multiple CFs. For the formation of CFs, it is assumed that there are CFs formation sites such as oxygen-related defects, roughness of the layer of a-IGZO, and so on. This method “Thermal analysis” can contribute to detection of the CF's location, the number of CFs, and thermal activity inside the memory devices.

Anisotropic Magnetoresistance of Ferromagnetic Conductive Filament in Resistive Switching Memory

A resistive switching (RS) memory has been expected to be a next-generation memory owing to its high density, low power consumption, and high switching speed. The mechanism of the reversible RS phenomenon has been considered in terms of the formation and rupture of conductive filaments (CF). Recently, it has been reported that a ferromagnetic CF was formed in NiO-based RS memory. This suggested the possibility of the new device that has both RS memory and spintronics device functions. The presence of a ferromagnetic CF in the NiO layer was confirmed by measuring the magnetization of the large number of CFs at room temperature using a SQUID magnetometer. Although, it is not yet confirmed that a single ferromagnetic CF is formed by the SET process. The purpose of this study is to understand the physical properties of a single ferromagnetic CF by focusing on magnetoresistance (MR). We prepared Ni/HfO2/Pt memory device, having ferromagnetic Ni electrode and the oxide of a non-magnetic metal. The fabricated RS memory devices have a cross-bar structure. The crossed electrode size is 1×1 mm2. RS characteristic was measured with the bipolar operation mode. When current limit is 5 mA, the anisotropic magnetoresistance (AMR) was observed in the low resistance state. The direction of the external magnetic field is in-plane to the substrate, and the current is perpendicular to the magnetic field. The AMR showed a convex behavior with magnetic field, and the MR ratio was approximately 0.15%. It is strongly suggested that AMR came from a ferromagnetic CF formed in the oxide layer. Further discussions will be shown in the workshop.

Elucidating the Origin of Resistive Switching in Ultrathin Hafnium Oxides through High Spatial Resolution Tools

Resistive random access memories have raised as one of the most promising devices to store information due to their great yield and easy fabrication. In these cells, the memory concept is based on tuning the electrical resistance of the insulating layer in a metal-insulator-metal structure by applying electrical signals, so that a high resistive state (HRS or “0”) and a low resistive state (LRS or “1”) can be reached. Among all insulator candidates, hafnium oxides are the most preferred due to their good performance and compatibility with the existent microelectronic technology. It is widely known that the resistance switching (RS) mechanism in HfO2 stacks is based on the reversibility of the dielectric breakdown (BD), which is a local phenomenon that takes place in areas of ~100 nm2. Therefore, the use of electrical characterization tools with high spatial resolution is necessary to fully understand the conductivity changes in the dielectric. In this work, we use advanced nanoscale characterization tools to elucidate the origin of resistive switching in hafnium-based oxides.Hafnium oxide thin films with different thicknesses and morphologies are analyzed using a Semiconductor Parameter Analyzer (SPA) connected to a Conductive Atomic Force Microscope (CAFM) working in controlled atmospheres (dry Nitrogen and a vacuum of 10-7 torr). Ramped Voltage Stresses have been performed at single locations of the samples to induce the formation of Conductive Filaments (CFs) - reach the dielectric breakdown (BD), or forming - . We analyze one-by-one the reversibility of those CFs, and we discern those locations where the BD could be recovered. Our experiments indicate that RS is only reachable at those filaments formed at the grain boundaries (GBs) of pseudo-amorphous and poly-crystalline samples (induced by thermal annealing), while the CFs induced at the nanocrystals or through the amorphous samples became irreversible (even using a current limitation). Kelvin Probe Force Microscope maps reveal that GBs are rich in defects, which correlates to larger leakage currents in (current maps) and lower BD voltages (forming, in IV curves). The lower BD voltages observed at the GBs (which, interestingly, coincide to those observed in MIM capacitors) may produce a less dramatic BD event, leading to narrower filaments easier to reoxidize. Moreover, the absolute value of the currents observed in LRS though a CF and a MIM structure are similar, indicating that RS in HfO2 stacks is a local phenomenon. This hypothesis is further supported by the fact that the currents through a CF in HRS and LRS can be fitted to the soft- and hard-BD equations. Finally we also show that, apart from creating/analyzing single CFs, the CAFM can also be used to perform statistical analyzes of the density, current and size of the CFs. In this case, two experiments are suitable to provide such information. The first consists of inducing the stresses in real capacitors, and scanning the surface of the insulator after etching the top electrode; and the second consists of collecting sequences of current maps at the same location of the sample using different voltages to observe a complete set-reset-set cycle. Since the maps cover larger areas and generate many filaments, this second test can be even combined with chemical analyses of lower spatial resolution.In summary, CAFM experiments allow the correlation between nanoscale morphological features and electrical signals, and they can provide essential information to understand the origin of RS. In this work, CAFM studies experimentally in-situ proved, for the first time, that resistive switching in HfO2 stacks only takes place at the grain boundaries in polycrystalline samples due to an unusual high concentration of defects that minimize the violence of the BD event.

Simulation of thermal reset transitions in resistive switching memories including quantum effects

An in-depth study of reset processes in RRAMs (Resistive Random Access Memories) based on Ni/HfO2/Si-n+ structures has been performed. To do so, we have developed a physically based simulator where both ohmic and tunneling based conduction regimes are considered along with the thermal description of the devices. The devices under study have been successfully fabricated and measured. The experimental data are correctly reproduced with the simulator for devices with a single conductive filament as well as for devices including several conductive filaments. The contribution of each conduction regime has been explained as well as the operation regimes where these ohmic and tunneling conduction processes dominate.

Resistive memory devices with high switching endurance through single filaments in Bi-crystal CuO nanowires

We propose a simple system to investigate the influence of microstructure on the resistive switching behavior via bi-crystal CuO nanowires. CuO nanowires are prepared by thermally oxidizing transmission electron microscopy copper grids in air. Single-crystal and bi-crystal CuO nanowires can be selectively obtained by adjusting the temperature. The devices made of single-crystal nanowires follow Ohm’s law, with a high resistance, within the sweeping voltage range of 0–4V, whereas those made of bi-crystal nanowires exhibit threshold and memory resistive switching behaviors, which are due to the enrichment of copper ions in the grain boundaries of bi-crystal CuO nanowires providing sources for the formation of conductive filaments. Moreover, the bi-crystal nanowires with higher defect densities in grain boundaries result in lower threshold voltages of switching from high to low resistance states. The threshold resistive switching behavior can be turned into memory resistive switching behavior by increasing the thickness of the device electrodes or reducing the compliance current. The endurance of memory resistive switching through the pre-defined conduction paths in the single grain boundaries of bi-crystal CuO nanowires is at least 1000 cycles without any performance deterioration. This high reliability is ascribed to the single conductive filaments.

A comprehensive analysis on progressive reset transitions in RRAMs

Reset processes in resistive random-access memory devices have been studied in depth. In particular, progressive transitions, where no clear current reduction steps are seen, are analysed by using a previously developed simulator and by comparing with experimental data of devices based on HfO2 oxides. It has been reported that the characterization of progressive reset processes can be performed by separately considering devices with a single conductive filament or devices with more than one conductive filament. In addition, making use of the experimental measurements shown, different numerical methods are proposed to extract the reset voltage. These methods are applied to different I–V reset curves and discussed.

Nanoscale observations of resistive switching high and low conductivity states on TiN/HfO2/Pt structures

Resistive Switching (RS) phenomenon in Metal–Insulator–Metal (MIM) structures with polycrystalline HfO2 layers as dielectric has been studied at the nanoscale using Conductive Atomic Force Microscope (CAFM). The CAFM measurements reveal that (i) the conductive filaments (CFs) created at very small areas are the origin of the RS phenomenon observed at device level and (ii) RS conductive filaments are primarily formed at the grain boundaries, which exhibit especially low breakdown voltage. CAFM images obtained on MIM structures at the Low and High Resistive states also show that, although the current in the Low Resistive State is mainly driven by a completely formed single CF, the cell area dependence of the conductivity in the High Resistive State could be explained by considering the presence of multiple partially formed CFs.

Improved Switching Variability and Stability by Activating a Single Conductive Filament

We successfully fabricated a single-filament-activated resistive memory having limited conductive filament (CF) source through a novel fabrication technique. The formation of a single filament was confirmed using conductive atomic force microscopy. Compared to conventional programmable metallization cell devices having an unlimited CF source, our single-filament device showed a significantly improved switching variability by minimizing the probability of randomly formed CFs. The elimination of unlimited CF sources changed the rate-limiting parameter that determines the retention and disturbance properties and improved thermal and electrical stability.

Programmable Resistance Switching in Nanoscale Two-Terminal Devices

We show that in nanoscale two-terminal resistive switches the resistance switching can be dominated by the formation of a single conductive filament. The probabilistic filament formation process strongly affects the device operation principle, and can be programmed to facilitate new functionalities such as multibit switching with partially formed filaments. In addition, the nanoscale switches exhibit excellent performance metrics making them well suited for memory or logic operations using conventional or emerging hybrid nano/CMOS architectures.