Search

Neuromorphic nanowire networks are of broad interest for applications in burgeoning memristive devices and neuromorphic computing areas due to their interesting features such as neural-like topology and nonlinear dynamics. However, the complexity of the neuromorphic nanowire network's behavior and in materia reservoir computing with imperfect device performance still hampers a straight transfer into emerging computing applications. Herein, reliable memristive devices based on unique necklace-like structure Ag@TiO2 nanowire networks are demonstrated for neuromorphic learning and reservoir computing. The memristive devices utilizing necklace-like structure Ag@TiO2 nanowire networks exhibit stable volatile threshold switching characteristics, with a ratio of up to 105, low threshold voltage (<1 V), good endurance, and high uniformity. Besides, the devices have been successfully used to emulate diverse functions of synapses by exploiting the Ag filament dynamics within the nanowire network, including short-term plasticity, and transition from short-term plasticity to long-term plasticity. The nanowire networks that offer nonlinear and short-term dynamics are further harnessed to build a reservoir computing system for the waveform classification task, manifesting its great potential for the development of next-generation reservoir hardware. [ABSTRACT FROM AUTHOR]

The development of two-dimensional (2D) materials has gained significant attention due to their unique properties and potential applications in advanced electronics. This study investigates the fabrication and characterization of Fe-doped SnSe semiconductors using an optimized chemical vapor deposition (CVD) method. Fe doping was achieved by dissolving FeCl3 in deionized water, applying it to SnSe powder, and conducting vacuum drying followed by high-temperature CVD at 820 °C. Structural and morphological properties were characterized using optical microscopy, scanning electron microscopy (SEM), and energy dispersive X-ray spectroscopy (EDX). Results revealed differently shaped flakes, including rectangles, discs and wires, influenced by Fe content. Micro-Raman spectroscopy showed significant vibrational mode shifts, indicating structural changes. X-ray photoelectron spectroscopy (XPS) confirmed the presence of Sn-Se and Fe-Se bonds. Electrical characterization of the memristive devices showed stable switching between high- and low-resistance states, with a threshold voltage of 1.6 V. These findings suggest that Fe-doped SnSe is a promising material for non-volatile memory and neuromorphic computing applications. [ABSTRACT FROM AUTHOR]

Abstract Blooming and pruning is one of the most important developmental mechanisms of the biological brain in the first years of life, enabling it to adapt its network structure to the demands of the environment. The mechanism is thought to be fundamental for the development of cognitive skills. Inspired by this, Chialvo and Bak proposed in 1999 a learning scheme that learns from mistakes by eliminating from the initial surplus of synaptic connections those that lead to an undesirable outcome. Here, this idea is implemented in a neuromorphic circuit scheme using CMOS integrated HfO2-based memristive devices. The implemented two-layer neural network learns in a self-organized manner without positive reinforcement and exploits the inherent variability of the memristive devices. This approach provides hardware, local, and energy-efficient learning. A combined experimental and simulation-based parameter study is presented to find the relevant system and device parameters leading to a compact and robust memristive neuromorphic circuit that can handle association tasks.

Driven by the explosive development of data-centric computation applications, it is becoming urgent to develop in-memory computing devices that are beyond the von Neumann architecture with an arrangement of separated logic and memory components. The transistor-type solid-state non-volatile memories, such as ferroelectric field-effect transistors (FeFETs), have long been regarded as a competitive candidate for future in-memory computing architectures. However, the density scaling towards high-density arrays would require advanced FeFETs with reduced footprints, which remains a great challenge so far. Here, a vertical-transport (VT) FeFET that flips the charge transport channel perpendicular to the substrate plane is proposed, in which a ferroelectric gate and a van der Waals (vdW) heterojunction channel are vertically integrated, effectively reducing the device footprints. The proposed VT-FeFET shows not only the robust binary non-volatile memory states but also several key synaptic functionalities at the device level. An artificial neural network with supervised learning was simulated based on the device conductance switching properties, showing excellent classification accuracy for the MNIST handwritten digits. These findings suggest that the proposed VT-FeFET could offer a new solution for future non-volatile memories as well as more advanced neuromorphic systems. [ABSTRACT FROM AUTHOR]

Resistive switching devices based on metal oxides are candidates for nonvolatile memory storage. They often rely on the valence change mechanism, the field‐induced movement of donor ions leading to nanoscale conductive paths in filamentary‐type devices. Devices usually consist of a transition metal oxide like Ta2O5 sandwiched between two metal electrodes. Critical parameters of the devices, such as cycle‐to‐cycle variability, Roff/Ron ratio, and endurance depend on the morphology and composition of the filaments. However, investigating filaments on the nanoscale is cumbersome, and commonly applied techniques such as conductive atomic force or transmission electron microscopy require delaminating the metal top electrode, inhibiting in operando investigations over many switching cycles. Here, the authors use infrared scattering‐type scanning near‐field optical microscopy (s‐SNOM) to investigate resistive switching in Ta2O5 films with a graphene top electrode in operando and reveal individual filaments on the device level. By selecting an appropriate illumination frequency, the authors can trace the evolution of filaments and the joule heating‐induced retraction of the top electrode until device failure. s‐SNOM promises a deeper understanding of resistive switching devices' microscopic switching behavior and applies to a wide range of resistive switching oxides, such as HfO2, SrTiO3, and SiO2. [ABSTRACT FROM AUTHOR]

Two‐dimensional layered transition metal dichalcogenides (TMDCs) are promising memristive materials for neuromorphic computing systems. Despite extensive experimental work, the underlying switching mechanisms are still not understood, impeding progress in material and device functionality. This study reveals the dominant role of defect dynamics in the switching process of 2D TMDC materials. The switching process is governed by the formation and annihilation dynamics of a local vacancy depletion zone. It explains the distinct features of the device characteristics observed experimentally, including fundamentally different device behavior previously thought to originate from multiple mechanisms. Key influence factors are identified and discussed with a fully coupled and dynamic charge transport model for electrons, holes, and ionic point defects, including image‐charge‐induced Schottky barrier lowering (SBL). Thermal effects and local Joule heating are considered by coupling the transient heat transfer equation to the electronic properties. The model is validated with hysteresis and pulse measurements for various lateral 2D MoS2‐based devices, strongly corroborating the relevance of vacancy dynamics in TMDC devices and offering a new perspective on the switching mechanisms. The insights gained from this study can be used to extend the functional behavior of 2D TMDC memristive devices in future neuromorphic computing applications. [ABSTRACT FROM AUTHOR]

In this study, we report on a memristive device structure wherein monolayers of two-dimensional (2D) molybdenum disulfide (MoS2) are integrated with an ultrathin yttrium oxide (Y2O3) layer to simulate artificial synapses functionality. The proposed physical simulation methodology is implemented in COMSOL Multiphysics tool and is based on the minimization of free energy of the used materials at the applied input voltage. The simulated device exhibits a stable bipolar resistive switching and the switching voltages is significantly reduced by increasing the number of MoS2 layers, which is key to conventional low-power computing and neuromorphic applications. The device is shown to perform synaptic functionalities under various applied bias conditions. The resulting synaptic weight decreases almost linearly with the increasing number of MoS2 layers due to the increase in the device thickness. The simulation outcomes pave the way for the development of optimised metal oxide-based memristive devices through their integration with semiconducting 2D materials. Also, the 2D MoS2 integration can enable the optoelectronic operation of this memory device.

The development of two-dimensional (2D) materials has gained significant attention due to their unique properties and potential applications in advanced electronics. This study investigates the fabrication and characterization of Fe-doped SnSe semiconductors using an optimized chemical vapor deposition (CVD) method. Fe doping was achieved by dissolving FeCl3 in deionized water, applying it to SnSe powder, and conducting vacuum drying followed by high-temperature CVD at 820 °C. Structural and morphological properties were characterized using optical microscopy, scanning electron microscopy (SEM), and energy dispersive X-ray spectroscopy (EDX). Results revealed differently shaped flakes, including rectangles, discs and wires, influenced by Fe content. Micro-Raman spectroscopy showed significant vibrational mode shifts, indicating structural changes. X-ray photoelectron spectroscopy (XPS) confirmed the presence of Sn-Se and Fe-Se bonds. Electrical characterization of the memristive devices showed stable switching between high- and low-resistance states, with a threshold voltage of 1.6 V. These findings suggest that Fe-doped SnSe is a promising material for non-volatile memory and neuromorphic computing applications.

Abstract Memristive hardware with reconfigurable conductance levels are leading candidates for achieving artificial neural networks (ANNs). However, owing to difficulties in device character design and circuit combination, the ability to perform complicated online‐learning tasks on a memristive network is not well understood. Here, tandem (T) material states are harnessed in a phase‐change memory (PCM) element, i.e., the primed‐amorphous state and the partial‐crystallized state, by utilizing an impetus‐and‐consequent pair pulse through a large degree of configurational ordering, and illustrate the development of an integrated system for achieving in‐memory computing and neural networks (NNs). A correct classification of 96.1% of 10,000 separate test images from the conventional Modified‐National‐Institute‐of‐Standards‐and‐Technology (MNIST) database in the tandem neural‐network (T‐NN) model is achieved, as well as image recognition for 28×28‐pixel pictures. The T‐NN configuration exhibits an in situ learning, with 50% of the elements stuck in the low‐conductance state, and at the same time, maintains an identification accuracy of ≈90%. The structural origin of the large degree of configurational‐ordering‐enhanced improvement in the extent of the conductance uniformity in the T‐based memristive element is revealed by theoretical studies. This work opens the door for attaining a widely relevant hardware system capable of performing artificial intelligence tasks with a large power‐time efficacy.

Abstract Two‐dimensional layered transition metal dichalcogenides (TMDCs) are promising memristive materials for neuromorphic computing systems. Despite extensive experimental work, the underlying switching mechanisms are still not understood, impeding progress in material and device functionality. This study reveals the dominant role of defect dynamics in the switching process of 2D TMDC materials. The switching process is governed by the formation and annihilation dynamics of a local vacancy depletion zone. It explains the distinct features of the device characteristics observed experimentally, including fundamentally different device behavior previously thought to originate from multiple mechanisms. Key influence factors are identified and discussed with a fully coupled and dynamic charge transport model for electrons, holes, and ionic point defects, including image‐charge‐induced Schottky barrier lowering (SBL). Thermal effects and local Joule heating are considered by coupling the transient heat transfer equation to the electronic properties. The model is validated with hysteresis and pulse measurements for various lateral 2D MoS2‐based devices, strongly corroborating the relevance of vacancy dynamics in TMDC devices and offering a new perspective on the switching mechanisms. The insights gained from this study can be used to extend the functional behavior of 2D TMDC memristive devices in future neuromorphic computing applications.

Arrays of memristive devices coupled with photosensors can be used for capturing and processing visual information, thereby realizing the concept of "in-sensor computing". This is a promising concept associated with the development of compact and low-power machine vision devices, which is crucial important for bionic prostheses of eyes, on-board image recognition systems for unmanned vehicles, computer vision in robotics, etc. This concept can be applied for the creation of a memristor based neuromorphic analog machine vision systems, and here, we propose a new architecture for these systems in which captured visual data are fed to a spiking artificial neural network (SNN) based on memristive devices without analog-to-digital and digital-to-analog conversions. Such an approach opens up the opportunities of creating more compact, energy-efficient visual processing units for wearable, on-board, and embedded electronics for such areas as robotics, the Internet of Things, and neuroprosthetics, as well as other practical applications in the field of artificial intelligence. [ABSTRACT FROM AUTHOR]

Inspired by the principles of the biological nervous system, neuromorphic engineering has brought a promising alternative approach to intelligence computing with high energy efficiency and low consumption. As pivotal components of neuromorphic system, artificial spiking neurons are powerful information processing units and can achieve highly complex nonlinear computations. By leveraging the switching dynamic characteristics of memristive device, memristive neurons show rich spiking behaviors with simple circuit. This report reviews the memristive neurons and their applications in neuromorphic sensing and computing systems. The switching mechanisms that endow memristive devices with rich dynamics and nonlinearity are highlighted, and subsequently various nonlinear spiking neuron behaviors emulated in these memristive devices are reviewed. Then, recent development is introduced on neuromorphic system with memristive neurons for sensing and computing. Finally, we discuss challenges and outlooks of the memristive neurons toward high-performance neuromorphic hardware systems and provide an insightful perspective for the development of interactive neuromorphic electronic systems. [ABSTRACT FROM AUTHOR]

ABSTRACTInspired by the principles of the biological nervous system, neuromorphic engineering has brought a promising alternative approach to intelligence computing with high energy efficiency and low consumption. As pivotal components of neuromorphic system, artificial spiking neurons are powerful information processing units and can achieve highly complex nonlinear computations. By leveraging the switching dynamic characteristics of memristive device, memristive neurons show rich spiking behaviors with simple circuit. This report reviews the memristive neurons and their applications in neuromorphic sensing and computing systems. The switching mechanisms that endow memristive devices with rich dynamics and nonlinearity are highlighted, and subsequently various nonlinear spiking neuron behaviors emulated in these memristive devices are reviewed. Then, recent development is introduced on neuromorphic system with memristive neurons for sensing and computing. Finally, we discuss challenges and outlooks of the memristive neurons toward high-performance neuromorphic hardware systems and provide an insightful perspective for the development of interactive neuromorphic electronic systems.

Memristive devices are promising candidates for analog computing applications such as neuromorphic computation. Larger dynamic ranges and more sufficient multilevel states can enable the significant development of memristor‐based utilizations. Herein, a method to improve the analog switching performance of memristors through a hybrid tuning (coarse and fine tuning) of two sub‐filaments is demonstrated. The creation of sub‐filaments inside the dielectric switching layer is realized by deploying Pt metal islands in the switching layer. Given the different material stack configurations of the two sub‐filaments, they exhibit different switching properties to play the roles of coarse and fine tuning respectively in the memristor. Based on the above mechanism, a Pt/Ta/Al2O3/Pt island/Al2O3‐x/TiOy/Al2O3‐x/Pt memristor is proposed and fabricated. Through the hybrid tuning of two sub‐filaments, a combined dynamic range of 600 Ω to 50 kΩ is achieved. Compared to the reference Pt/Ta/Al2O3/Pt memristors (dynamic range: 600 Ω to 8 kΩ), both dynamic range and multilevel resistance states are increased significantly. Meanwhile, the energy efficiency is improved because the resistance of tunable states can be set to larger values. Furthermore, this mechanism can be incorporated into various existing memristors to improve their dynamic range and multilevel states, which extensively enriches the applications of memristors. [ABSTRACT FROM AUTHOR]

Spiking neural networks (SNNs) provide an efficient way to apply artificial intelligence systems on edge devices. Memristive devices with tunable conductance states can be used to emulate the functions of biological neurons and synapses and to build SNNs. In this work, fully-connected SNNs based on various memristive devices are constructed, and a hardware-compatible spike-timing-dependent plasticity (STDP) learning rule is applied to train the SNNs. Strategies are designed to suppress the overfitting problem and improve the performance of the SNNs in the case of a small training set. However, the properties of memristive devices are never perfect. The effects of imperfect device properties, e.g., asymmetric weight update, insufficient number of conductance states, and low on/off ratio, on the performance of the SNNs are elaborated. [ABSTRACT FROM AUTHOR]

Valence change memories are novel data storage devices in which the resistance is determined by a reversible redox reaction triggered by voltage. The oxygen content and mobility within the active materials of these devices play a crucial role in their performance. Therefore, materials which present fast oxygen migration properties and can accommodate variable oxygen stoichiometry are promising candidates. In this work, the perovskite La0.5Sr0.5MnO3‐δ (LSM50) as memristive material is studied, which presents a more facile oxygen vacancy formation and faster oxygen migration compared to other strontium‐substituted manganites. For the first time reproducible resistive switching is reported in epitaxial LSM50‐based devices with active Ti electrodes, which show large operating window and stable multilevel states. Based on the structural, chemical, and electrical results, a simple phenomenological description of the resistive switching phenomena taking place in these novel LSM50‐based memristive devices is proposed. [ABSTRACT FROM AUTHOR]

The great demand of multifunctional portable electronic products in daily life and the need of a large integration of memories with logic devices and sensors, have increased the interest in the identification of suitable materials for neuromorphic computing applications. Major innovations in this direction have been achieved by exploring materials belonging to different fields of applications and taking advantage of already consolidated deposition methods. Despite the great interest in the field and the large use in complementary applications such as sensing electrodes, neural and cellular interfaces, the use of diamond‐like materials in neuromorphic applications is still limited to a few examples. Herein, the development of a synaptic element based on high‐quality polycrystalline diamond layers containing Ti inclusions showing a marked and reproducible resistance switching behavior is reported. Realized by means of a hybrid chemical vapor deposition‐powder flowing technique, this titanium doped diamond shows a 3D polycrystalline organization that is characterized by globular grains of a few microns. The coupling of Raman spectroscopy, X‐ray photoelectron spectroscopy, and X‐ray diffraction analyses confirms the good quality of diamond phase and convincingly points out the inclusion of the titanium species within the diamond lattice. [ABSTRACT FROM AUTHOR]

Computation-in-Memory accelerators based on resistive switching devices represent a promising approach to realize future information processing systems. These architectures promise orders of magnitudes lower energy consumption for certain tasks, while also achieving higher throughputs than other special purpose hardware such as GPUs, due to their analog computation nature. Due to device variability issues, however, a single resistive switching cell usually does not achieve the resolution required for the considered applications. To overcome this challenge, many of the proposed architectures use an approach called bit slicing, where generally multiple low-resolution components are combined to realize higher resolution blocks. In this paper, we will present an analog accelerator architecture on the circuit level, which can be used to perform Vector-Matrix-Multiplications or Matrix-Matrix-Multiplications. The architecture consists of the 1T1R crossbar array, the optimized select circuitry and an ADC. The components are designed to handle the variability of the resistive switching cells, which is verified through our verified and physical compact model. We then use this architecture to compare different bit slicing approaches and discuss their tradeoffs. [ABSTRACT FROM AUTHOR]

This work is devoted to providing an approximate analytical method to analyze memristor devices. One of basic property of the memristor is pinched hysteresis, considered to be a signature of the existence of memristance. The presence of hysteresis defines the material implementation of its memristive effects. This it its fundamental property, which looks more like a nonlinear anomaly. The memristor technology offers lower heat generation as it utilizes less energy. Optimal auxiliary functions method (OAFM) is implemented to find an approximate solution for the state variable in memristor with a very high accuracy. The presence of the auxiliary functions and some optimal convergence-control parameters Ci assure a fast convergence of the solutions. [ABSTRACT FROM AUTHOR]

Resistively switching electrochemical metallization memory cells are gaining huge interest since they are seen as promising candidates and basic building blocks for future computation-in-memory applications. However, especially filamentary-based memristive devices suffer from inherent variability, originating from their stochastic switching behavior. A variability-aware compact model of electrochemical metallization memory cells is presented in this study and verified by showing a fit to experimental data. It is an extension of the deterministic model. Since this extension consists of several different features allowing for a realistic variability-aware fit, it depicts a unique model comprising physics-based, stochastically and experimentally originating variabilities and reproduces them well. In addition, a physics-based model parameter study is executed, which enables a comprehensive view into the device physics and presents guidelines for the compact model fitting procedure.

Abstract Valence change memories are novel data storage devices in which the resistance is determined by a reversible redox reaction triggered by voltage. The oxygen content and mobility within the active materials of these devices play a crucial role in their performance. Therefore, materials which present fast oxygen migration properties and can accommodate variable oxygen stoichiometry are promising candidates. In this work, the perovskite La0.5Sr0.5MnO3‐δ (LSM50) as memristive material is studied, which presents a more facile oxygen vacancy formation and faster oxygen migration compared to other strontium‐substituted manganites. For the first time reproducible resistive switching is reported in epitaxial LSM50‐based devices with active Ti electrodes, which show large operating window and stable multilevel states. Based on the structural, chemical, and electrical results, a simple phenomenological description of the resistive switching phenomena taking place in these novel LSM50‐based memristive devices is proposed.

Memristors are expected to be one of the key building blocks for the development of new bio-inspired nanoelectronics. Memristive effects in transition metal oxides are usually linked to the electromigration at the nanoscale of charged oxygen vacancies (OV). In this paper we address, for Pt/TiO x /TaO y /Pt devices, the exchange of OV between the device and the environment upon the application of electrical stress. From a combination of experiments and theoretical simulations we determine that both TiO x and TaO y layers oxidize, via environmental oxygen uptake, during the electroforming process. Once the memristive effect is stabilized (post-forming behavior) our results suggest that oxygen exchange with the environment is suppressed and the OV dynamics that drives the memristive behavior is restricted to an internal electromigration between TiO x and TaO y layers. Our work provides relevant information for the design of reliable binary oxide memristive devices. [ABSTRACT FROM AUTHOR]

Arrays of memristive devices coupled with photosensors can be used for capturing and processing visual information, thereby realizing the concept of “in-sensor computing”. This is a promising concept associated with the development of compact and low-power machine vision devices, which is crucial important for bionic prostheses of eyes, on-board image recognition systems for unmanned vehicles, computer vision in robotics, etc. This concept can be applied for the creation of a memristor based neuromorphic analog machine vision systems, and here, we propose a new architecture for these systems in which captured visual data are fed to a spiking artificial neural network (SNN) based on memristive devices without analog-to-digital and digital-to-analog conversions. Such an approach opens up the opportunities of creating more compact, energy-efficient visual processing units for wearable, on-board, and embedded electronics for such areas as robotics, the Internet of Things, and neuroprosthetics, as well as other practical applications in the field of artificial intelligence.

Analog compute schemes and compute-in-memory (CIM) have emerged in an effort to reduce the increasing power hunger of convolutional neural networks (CNNs), which exceeds the constraints of edge devices. Memristive device types are a relatively new offering with interesting opportunities for unexplored circuit concepts. In this work, the use of memristive devices in cascaded time-domain CIM (TDCIM) is introduced with the primary goal of reducing the size of fully unrolled architectures. The different effects influencing the determinism in memristive devices are outlined together with reliability concerns. Architectures for binary as well as multibit multiply and accumulate (MAC) cells are presented and evaluated. As more involved circuits offer more accurate compute result, a tradeoff between design effort and accuracy comes into the picture. To further evaluate this tradeoff, the impact of variations on overall compute accuracy is discussed. The presented cells reach an energy/OP of 0.23 fJ at a size of $1.2~{\mu{ }}\text{m}^{2}$ for binary and 6.04 fJ at $3.2~\mu \text{m}^{2}$ for $4\times 4$ bit MAC operations.

Valence change memories, in which internal redox reactions control the change in resistance are promising candidates for resistive random access memories (ReRAMs) and neuromorphic computing elements. In this context, La2NiO4+δ (L2NO4), a mixed ionic‐electronic conducting oxide, well known for its highly mobile oxygen interstitial ions, emerges as a potential switching material for novel L2NO4‐based memristive devices. However, their integration in complementary metal oxide semiconductor (CMOS) technology still has to be demonstrated, as the major focus of previous studies has been carried out on epitaxial films grown on single crystals. In this work, the optimization of the deposition temperature and precursor solution composition is presented, allowing to obtain high‐quality polycrystalline L2NO4 thin films grown by metal organic chemical vapor deposition on a platinized silicon substrate, and to use these films to build memristive devices in vertical configuration with Ti top electrodes. A bipolar analog‐type transition in resistance can be achieved in Ti/L2NO4/Pt memristive devices. While the "forming" process required for the devices based on nonoptimized L2NO4 thin films is considered as a drawback, the Ti/optimized L2NO4/Pt devices are forming‐free and exhibit a good cyclability. These results prove the switching response of L2NO4‐based devices in a vertical configuration for the first time. [ABSTRACT FROM AUTHOR]

Interface‐type valence change memories (VCMs) are exciting candidates for multilevel storage in resistive random access memories (RRAM) and as artificial synapses for neuromorphic computing. Several materials have been proposed as VCM candidates and, depending on the materials and electrodes of choice, different switching mechanisms take place leading to the change in resistance. Here, the focus is on La0.8Sr0.2MnO3–δ (LSM) perovskite and, particularly, the role of its nanostructure on the memristive device performance. The nanostructural details of the layers are modified by growing LSM epitaxial thin films on different substrates, i.e., SrTiO3 (STO) and LaAlO3 (LAO), by metal‐organic chemical vapor deposition (MOCVD). An interface‐type memristive response is observed using Ti as active electrode and Pt as inert electrode. The modifications in the nanostructure of LSM (strain and dislocations) determine the memristive performance, leading to differences in cycle to cycle reproducibility and multilevel capabilities by the modification of the LSM's oxygen migration properties. The results show that nanostructure engineering is a promising approach for optimizing the performance of memristive devices, an approach which can also be extended and applied to other nanoionic electrochemical devices. [ABSTRACT FROM AUTHOR]

Artificial neural networks (ANNs) are attracting attention for their high performance in various fields, because increasing the network size improves its functioning. Since large-scale neural networks are difficult to implement on custom hardware, a two-dimensional (2D) structure is applied to an ANN in the form of a crossbar. We demonstrate a synapse crossbar device from recent research by applying a memristive system to neuromorphic chips. The system is designed using two-dimensional structures, graphene quantum dots (GQDs) and graphene oxide (GO). Raman spectrum analysis results indicate a D-band of 1421 cm-1 that occurs in the disorder; band is expressed as an atomic characteristic of carbon in the sp2 hybridized structure. There is also a G-band of 1518 cm-1 that corresponds to the graphite structure. The G bands measured for RGO-GQDs present significant GQD edge-dependent shifts with position. To avoid an abruptly-formed conduction path, effect of barrier layer on graphene/ITO interface was investigated. We confirmed the variation in the nanostructure in the RGO-GQD layers by analyzing them using HR-TEM. After applying a negative bias to the electrode, a crystalline RGO-GQD region formed, which a conductive path. Especially, a synaptic array for a neuromorphic chip with GQDs applied was demonstrated using a crossbar array. [ABSTRACT FROM AUTHOR]

The multidisciplinary field of memristors calls for the necessity for theoretically‐inclined researchers and experimenters to join forces, merging complementary expertise and technical know‐how, to develop and implement rigorous and systematic techniques to design variability‐aware memristor‐based circuits and systems. The availability of a predictive physics‐based model for a memristor is a necessary requirement before commencing these investigations. An interesting dynamic phenomenon, occurring ubiquitously in non‐volatile memristors, is fading memory. The latter may be defined as the appearance of a unique steady‐state behavior, irrespective of the choice of the initial condition from an admissible range of values, for each stimulus from a certain family, for example, the DC or the purely‐AC periodic input class. This paper first provides experimental evidence for the emergence of fading memory effects in the response of a TaOx redox‐based random access memory cell to inputs from both of these classes. Leveraging the predictive capability of a physics‐based device model, called JART VCM v1, a thorough system‐theoretic analysis, revolving around the Dynamic Route Map graphic tool, is presented. This analysis allows to gain a better understanding of the mechanisms, underlying the emergence of history erase effects, and to identify the main factors, that modulate this nonlinear phenomenon, toward future potential applications. [ABSTRACT FROM AUTHOR]

The advent of the intelligent society leads to the exponential growth of information, imposing urgent requirements in a time‐ and energy‐efficient way to process information where data are generated. This issue can be addressed by the neuromorphic paradigm of computing inspired by biological sensory systems that build up the association between external stimuli and the response of an organism in real‐time; in the paradigm, a neuromorphic system is integrated with sensors to form an artificial sensory system. Herein, a neuromorphic sensory system with integrated capabilities of gas sensing, data storage, and processing is demonstrated. Leaky integrate‐and‐fire (LIF) neurons, the basic computing units in the system, are realized with volatile memristive device Pt/Ag/TaOx/Pt; sensory neurons, i.e., the LIF neurons connected with an array of gas sensors, detect gases and convert the chemical information of gases into neural spikes; synapses based on nonvolatile memristive device Pt/Ta/TaOx/Pt transmit the signals from sensory neurons to relay neurons according to synaptic weights, which are trained by the supervised spike‐rate dependent plasticity; relay neurons then process the signals from the synapses and classify gases. The approach of this work can also be applied to emulate other biological perceptions through the integration with different sensors. [ABSTRACT FROM AUTHOR]

The advent of the intelligent society leads to the exponential growth of information, imposing urgent requirements in a time‐ and energy‐efficient way to process information where data are generated. This issue can be addressed by the neuromorphic paradigm of computing inspired by biological sensory systems that build up the association between external stimuli and the response of an organism in real‐time; in the paradigm, a neuromorphic system is integrated with sensors to form an artificial sensory system. Herein, a neuromorphic sensory system with integrated capabilities of gas sensing, data storage, and processing is demonstrated. Leaky integrate‐and‐fire (LIF) neurons, the basic computing units in the system, are realized with volatile memristive device Pt/Ag/TaOx/Pt; sensory neurons, i.e., the LIF neurons connected with an array of gas sensors, detect gases and convert the chemical information of gases into neural spikes; synapses based on nonvolatile memristive device Pt/Ta/TaOx/Pt transmit the signals from sensory neurons to relay neurons according to synaptic weights, which are trained by the supervised spike‐rate dependent plasticity; relay neurons then process the signals from the synapses and classify gases. The approach of this work can also be applied to emulate other biological perceptions through the integration with different sensors.

• Homogeneous pulsed laser deposition of PCMO on 4″ wafer. • Addressing the issue of plume broadening by a slit system. • Deposition strategy development with simulation software. • Area-dependent memristive switching of PCMO /AlO x heterostructures. • Process temperature is compatible with back end of line integration into standard CMOS technology. Heterostructures of the perovskite Pr 0.7 Ca 0.3 MnO 3 (PCMO) and a tunnel oxide such as AlO x are highly interesting memristive devices for emulating synaptic properties in neuromorphic circuits. Future chip generations with these memristive elements requires PLD systems which enables PCMO growth on a full wafer. To address the issue of plume broadening in this study, a slit system was used to localize the deposition region, thereby eliminating the need for excessive scanning. This study investigates the effect of the slit system and process gas pressure on plume broadening. This investigation has been used to parameterize a simulation software to assess the effectiveness of different movement speeds and to develop strategies for homogeneous deposition, adjustable to a chosen film thickness in the order of twenty nanometers. We used these strategies to fabricate area dependent switching memory cells on a standard 4″ Si wafer using the material combination of Pr 0.7 Ca 0.3 MnO 3 and AlO x. The influence of the aluminum oxide thickness on the switching shows that the IV loop starts to exhibit a hysteresis for AlO x thicknesses ≥ 3 nm. An increase in AlO x thickness leads to an increase in resistance and capacitive charging. An additional thermal treatment reduces the resistance and the switching voltage and increases the ratio between low resistive and high resistive state. Devices without thermal treatment of the PCMO are compatible for back end of line processing of standard CMOS technology. [ABSTRACT FROM AUTHOR]

Memristive devices based on the Valence Change Mechanism (VCM) are promising devices for storage class memory, neuromorphic computing and logic-in-memory (LIM) applications. They are suited for such a wide range of applications, due to their possibility for extreme dense integration, low power consumption and multilevel capabilities. Through LIM concepts, Boolean logic operations can be performed directly in memory. In many of these concepts, the resistance state of the device is interpreted as the logical input and output of the logic function, which is why these concepts are called ‘stateful’ logic. Most of the proposed ideas, however, are defined based on only binary switching VCM devices and neglect their multi-level capabilities. Extending LIM concepts towards multinary logic, e.g. a ternary logic, would increase the data density inside the memory array and reduce the number of devices required to perform a certain operation. In this work, we discuss two possibilities of realizing a ternary logic based on the analog switching in the RESET or in the SET direction. For both directions we verify the logic functionality by showing the basic operations of implication, negation and false operation, which together form a functionally complete logic. Additionally, for both switching directions, we discuss a 41-Trit ( ${\approx}$ 64-Bit) addition. For all investigations the physics-based compact model JART VCM v1b is used, which has been verified on the RESET direction multilevel properties of the TaO $_{\mathrm{x}}$ devices.

Brain-inspired computing proposes a set of algorithmic principles that hold promise for advancing artificial intelligence. They endow systems with self learning capabilities, efficient energy usage, and high storage capacity. A core concept that lies at the heart of brain computation is sequence learning and prediction. This form of computation is essential for almost all our daily tasks such as movement generation, perception, and language. Understanding how the brain performs such a computation is not only important to advance neuroscience, but also to pave the way to new technological brain-inspired applications. A previously developed spiking neural network implementation of sequence prediction and recall learns complex, high-order sequences in an unsupervised manner by local, biologically inspired plasticity rules. An emerging type of hardware that may efficiently run this type of algorithm is neuromorphic hardware. It emulates the way the brain processes information and maps neurons and synapses directly into a physical substrate. Memristive devices have been identified as potential synaptic elements in neuromorphic hardware. In particular, redox-induced resistive random access memories (ReRAM) devices stand out at many aspects. They permit scalability, are energy efficient and fast, and can implement biological plasticity rules. In this work, we study the feasibility of using ReRAM devices as a replacement of the biological synapses in the sequence learning model. We implement and simulate the model including the ReRAM plasticity using the neural network simulator NEST. We investigate two types of ReRAM memristive devices: (i) a gradual, analog switching device, and (ii) an abrupt, binary switching device. We study the effect of different device properties on the performance characteristics of the sequence learning model, and demonstrate that, in contrast to many other artificial neural networks, this architecture is resilient with respect to changes in the on-off ratio and the conductance resolution, device variability, and device failure.

An effective synthesis of conductive polymer brushes, i.e., self-templating surface-initiated copolymerization (ST-SICP), is developed. It proceeds through copolymerization of pendant thiophene groups in the precursor multimonomer poly(3-methylthienyl methacrylate) (PMTM) brushes with free 3-methylthiophene (3MT) monomers leading to PMTM-co-P3MT brushes. This approach leads to improved conformational freedom of generated conjugated poly(thiophene)-based chains and their higher share in the brushes with respect to conjugation of pendant thiophene groups only. As a result, best performing conjugated PMTM-co-P3MT brushes demonstrate high ohmic conductivity in both out-of-plane and in-plane direction. Furthermore, thanks to the covalent anchoring as well as intra- and intermolecular connections, highly stable and mechanically robust nanocoatings are produced which can survive mechanical cleaning and long-term storage under ambient conditions. Grafting of ionic poly(sodium 4-styrenesulfonate) (PSSNa) in between PMTM-co-P3MT chains brings new properties to such binary mixed brushes that can operate as thin-film memristive coating with switchable conductance. It is worth mentioning that the crucial synthetic steps, i.e., grafting of precursor PMTM brushes by surface-initiated organocatalyzed atom transfer radical polymerization (SI-O-ATRP) and PSSNa chains by surface-initiated photoiniferter-mediated polymerization (SI-PIMP) are conducted under ambient conditions using only microliter volumes of reagents providing methodology that can be considered for use beyond the laboratory scale., (© 2024 Wiley‐VCH GmbH.)

Summary: There are two ideal memristive devices, first‐order memristor and second‐order memristor, which could be widely applied in nonlinear electronic circuits. The main difference between them is their ϕ − q characteristics. However, there are only a few studies available in the literature about the different dynamic behavior that can be observed when they are connected in the same circuit in turns. In this paper, to demonstrate the different consequences for the above situation, the complex dynamics of a memristive band‐pass filter (BPF) are analyzed in detail as one of the classical first‐order circuits. Also, the existence of coexisting attractors, hysteretic dynamics, and antimonotonicity are confirmed. Subsequently, the impact of memristor coefficients on the working dynamics of BPF is revealed. The conclusion is that a higher‐order memristor can lead to a greater number of chaotic attractors and more complex nonlinearity, which is the basis to determine that substitutability and interchangeability of equivalent memristance do not apply for memristor devices with the different ϕ − q relationship when analyzing memristive circuits. Finally, the experimental simulations are presented to further prove the above theoretical analysis. Additionally, other circuits, such as the Chua circuit and the Jerk circuit, are also investigated in order to verify our analysis. In particular, a full Feigenbaum tree is demonstrated in the Chua circuit. [ABSTRACT FROM AUTHOR]

Neuromorphic computing based on spiking neural networks has the potential to significantly improve on-line learning capabilities and energy efficiency of artificial intelligence, specially for edge computing. Recent progress in computational neuroscience have demonstrated the importance of heterosynaptic plasticity for network activity regulation and memorization. Implementing heterosynaptic plasticity in hardware is thus highly desirable, but important materials and engineering challenges remain, calling for breakthroughs in neuromorphic devices. In this mini-review, we propose an overview of the latest advances in multi-terminal memristive devices on silicon with tunable synaptic plasticity, enabling heterosynaptic plasticity in hardware. The scalability and compatibility of the devices with industrial complementary metal oxide semiconductor (CMOS) technologies are discussed.