Biological Spiking Synapse Constructed from Solution Processed Bimetal Core–Shell Nanoparticle Based Composites

Inspired by the highly parallel processing power and low energy consumption of the biological nervous system, the development of a neuromorphic computing paradigm to mimic brain‐like behaviors with electronic components based artificial synapses may play key roles to eliminate the von Neumann bottleneck. Random resistive access memory (RRAM) is suitable for artificial synapse due to its tunable bidirectional switching behavior. In this work, a biological spiking synapse is developed with solution processed Au@Ag core–shell nanoparticle (NP)‐based RRAM. The device shows highly controllable bistable resistive switching behavior due to the favorable Ag ions migration and filament formation in the composite film, and the good charge trapping and transport property of Au@Ag NPs. Moreover, comprehensive synaptic functions of biosynapse including paired‐pulse depression, paired‐pulse facilitation, post‐tetanic potentiation, spike‐time‐dependent plasticity, and the transformation from short‐term plasticity to long‐term plasticity are emulated. This work demonstrates that the solution processed bimetal core–shell nanoparticle‐based biological spiking synapse provides great potential for the further creation of a neuromorphic computing system.     

Neuromorphic Engineering: From Biological to Spike-Based Hardware Nervous Systems

The human brain is a sophisticated, high-performance biocomputer that processes multiple complex tasks in parallel with high efficiency and remarkably low power consumption. Scientists have long been pursuing an artificial intelligence (AI) that can rival the human brain. Spiking neural networks based on neuromorphic computing platforms simulate the architecture and information processing of the intelligent brain, providing new insights for building AIs. The rapid development of materials engineering, device physics, chip integration, and neuroscience has led to exciting progress in neuromorphic computing with the goal of overcoming the von Neumann bottleneck. Herein, fundamental knowledge related to the structures and working principles of neurons and synapses of the biological nervous system is reviewed. An overview is then provided on the development of neuromorphic hardware systems, from artificial synapses and neurons to spike-based neuromorphic computing platforms. It is hoped that this review will shed new light on the evolution of brain-like computing.     

2D MoS2 Neuromorphic Devices for Brain‐Like Computational Systems

Hardware implementation of artificial synapses/neurons with 2D solid‐state devices is of great significance for nanoscale brain‐like computational systems. Here, 2D MoS₂ synaptic/neuronal transistors are fabricated by using poly(vinyl alcohol) as the laterally coupled, proton‐conducting electrolytes. Fundamental synaptic functions, such as an excitatory postsynaptic current, paired‐pulse facilitation, and a dynamic filter for information transmission of biological synapse, are successfully emulated. Most importantly, with multiple input gates and one modulatory gate, spiking‐dependent logic operation/modulation, multiplicative neural coding, and neuronal gain modulation are also experimentally demonstrated. The results indicate that the intriguing 2D MoS₂ transistors are also very promising for the next‐generation of nanoscale neuromorphic device applications.     

Emerging Memristive Artificial Synapses and Neurons for Energy-Efficient Neuromorphic Computing

Memristors have recently attracted significant interest due to their applicability as promising building blocks of neuromorphic computing and electronic systems. The dynamic reconfiguration of memristors, which is based on the history of applied electrical stimuli, can mimic both essential analog synaptic and neuronal functionalities. These can be utilized as the node and terminal devices in an artificial neural network. Consequently, the ability to understand, control, and utilize fundamental switching principles and various types of device architectures of the memristor is necessary for achieving memristor-based neuromorphic hardware systems. Herein, a wide range of memristors and memristive-related devices for artificial synapses and neurons is highlighted. The device structures, switching principles, and the applications of essential synaptic and neuronal functionalities are sequentially presented. Moreover, recent advances in memristive artificial neural networks and their hardware implementations are introduced along with an overview of the various learning algorithms. Finally, the main challenges of the memristive synapses and neurons toward high-performance and energy-efficient neuromorphic computing are briefly discussed. This progress report aims to be an insightful guide for the research on memristors and neuromorphic-based computing.     

Artificial Synapses Emulated by an Electrolyte‐Gated Tungsten‐Oxide Transistor

Considering that the human brain uses ≈10¹⁵ synapses to operate, the development of effective artificial synapses is essential to build brain‐inspired computing systems. In biological synapses, the voltage‐gated ion channels are very important for regulating the action‐potential firing. Here, an electrolyte‐gated transistor using WO₃ with a unique tunnel structure, which can emulate the ionic modulation process of biological synapses, is proposed. The transistor successfully realizes synaptic functions of both short‐term and long‐term plasticity. Short‐term plasticity is mimicked with the help of electrolyte ion dynamics under low electrical bias, whereas the long‐term plasticity is realized using proton insertion in WO₃ under high electrical bias. This is a new working approach to control the transition from short‐term memory to long‐term memory using different gate voltage amplitude for artificial synapses. Other essential synaptic behaviors, such as paired pulse facilitation, the depression and potentiation of synaptic weight, as well as spike‐timing‐dependent plasticity are also implemented in this artificial synapse. These results provide a new recipe for designing synaptic electrolyte‐gated transistors through the electrostatic and electrochemical effects.     

Reproducible Ultrathin Ferroelectric Domain Switching for High-Performance Neuromorphic Computing

Neuromorphic computing consisting of artificial synapses and neural network algorithms provides a promising approach for overcoming the inherent limitations of current computing architecture. Developments in electronic devices that can accurately mimic the synaptic plasticity of biological synapses, have promoted the research boom of neuromorphic computing. It is reported that robust ferroelectric tunnel junctions can be employed to design high-performance electronic synapses. These devices show an excellent memristor function with many reproducible states (≈200) through gradual ferroelectric domain switching. Both short- and long-term plasticity can be emulated by finely tuning the applied pulse parameters in the electronic synapse. The analog conductance switching exhibits high linearity and symmetry with small switching variations. A simulated artificial neural network with supervised learning built from these synaptic devices exhibited high classification accuracy (96.4%) for the Mixed National Institute of Standards and Technology (MNIST) handwritten recognition data set.     

Flexible Ionic‐Electronic Hybrid Oxide Synaptic TFTs with Programmable Dynamic Plasticity for Brain‐Inspired Neuromorphic Computing

Emulation of biological synapses is necessary for future brain‐inspired neuromorphic computational systems that could look beyond the standard von Neuman architecture. Here, artificial synapses based on ionic‐electronic hybrid oxide‐based transistors on rigid and flexible substrates are demonstrated. The flexible transistors reported here depict a high field‐effect mobility of ≈9 cm² V^?1 s^?1 with good mechanical performance. Comprehensive learning abilities/synaptic rules like paired‐pulse facilitation, excitatory and inhibitory postsynaptic currents, spike‐time‐dependent plasticity, consolidation, superlinear amplification, and dynamic logic are successfully established depicting concurrent processing and memory functionalities with spatiotemporal correlation. The results present a fully solution processable approach to fabricate artificial synapses for next‐generation transparent neural circuits.     

Stimuli‐Enabled Artificial Synapses for Neuromorphic Perception: Progress and Perspectives

Brain‐inspired neuromorphic computing is intended to provide effective emulation of the functionality of the human brain via the integration of electronic components. Recent studies of synaptic plasticity, which represents one of the most significant neurochemical bases of learning and memory, have enhanced the general comprehension of how the brain functions and have thereby eased the development of artificial neuromorphic devices. An understanding of the synaptic plasticity induced by various types of stimuli is essential for neuromorphic system construction. The realization of multiple stimuli‐enabled synapses will be important for future neuromorphic computing applications. In this Review, state‐of‐the‐art synaptic devices with particular emphasis on their synaptic behaviors under excitation by a variety of external stimuli are summarized, including electric fields, light, magnetic fields, pressure, and temperature. The switching mechanisms of these synaptic devices are discussed in detail, including ion migration, electron/hole transfer, phase transition, redox‐based resistive switching, and other mechanisms. This Review aims to provide a comprehensive understanding of the operating mechanisms of artificial synapses and thus provides the principles required for design of multifunctional neuromorphic systems with parallel processing capabilities.     

A multinanodot floating-gate MOSFET circuit for spiking neuron models

Spiking neuron models, which represent information in the form of spatiotemporal patterns in spike pulse trains, have attracted much attention recently in the fields of computational neuroscience and artificial neural networks. The information processing abilities of spiking neuron models have been proven superior to those of the conventional analog-type (rate-coding) neural network models. In particular, the spike response model (SRM), which simplifies the biological neuron operation from the viewpoint of spike response, is important for VLSI implementation and various applications. In the SRM, the generation of post-synaptic potentials (PSPs) is essential. The conventional CMOS devices require complicated circuits in order to realize the function of SRM neurons. In this paper, a new device structure using a MOSFET with multinanodot floating-gate arrays is proposed for the synapse component of SRM neurons. This structure can operate at room temperature, as it utilizes thermal-noise-assisted tunneling between nanodots. The structure generates PSPs by taking advantage of the delay in electron movement due to stochastic tunneling processes. The results of single-electron circuit simulation demonstrate the generation of PSPs. The proposed structure has not yet been fabricated. The aim of this paper is to propose guidelines for the development of new nanoscale devices and fabrication technology for intelligent information processing such as that achieved in the human brain.     

Hybrid discrete-time neural networks

Hybrid dynamical systems combine evolution equations with state transitions. When the evolution equations are discrete-time (also called map-based), the result is a hybrid discrete-time system. A class of biological neural network models that has recently received some attention falls within this category: map-based neuron models connected by means of fast threshold modulation (FTM). FTM is a connection scheme that aims to mimic the switching dynamics of a neuron subject to synaptic inputs. The dynamic equations of the neuron adopt different forms according to the state (either firing or not firing) and type (excitatory or inhibitory) of their presynaptic neighbours. Therefore, the mathematical model of one such network is a combination of discrete-time evolution equations with transitions between states, constituting a hybrid discrete-time (map-based) neural network. In this paper, we review previous work within the context of these models, exemplifying useful techniques to analyse them. Typical map-based neuron models are low-dimensional and amenable to phase-plane analysis. In bursting models, fast-slow decomposition can be used to reduce dimensionality further, so that the dynamics of a pair of connected neurons can be easily understood. We also discuss a model that includes electrical synapses in addition to chemical synapses with FTM. Furthermore, we describe how master stability functions can predict the stability of synchronized states in these networks. The main results are extended to larger map-based neural networks.     

Synaptic Barristor Based on Phase‐Engineered 2D Heterostructures

The development of energy‐efficient artificial synapses capable of manifoldly tuning synaptic activities can provide a significant breakthrough toward novel neuromorphic computing technology. Here, a new class of artificial synaptic architecture, a three‐terminal device consisting of a vertically integrated monolithic tungsten oxide memristor, and a variable‐barrier tungsten selenide/graphene Schottky diode, termed as a ‘synaptic barrister,’ are reported. The device can implement essential synaptic characteristics, such as short‐term plasticity, long‐term plasticity, and paired‐pulse facilitation. Owing to the electrostatically controlled barrier height in the ultrathin van der Waals heterostructure, the device exhibits gate‐controlled memristive switching characteristics with tunable programming voltages of 0.2?0.5 V. Notably, by electrostatic tuning with a gate terminal, it can additionally regulate the degree and tuning rate of the synaptic weight independent of the programming impulses from source and drain terminals. Such gate tunability cannot be accomplished by previously reported synaptic devices such as memristors and synaptic transistors only mimicking the two‐neuronal‐based synapse. These capabilities eventually enable the accelerated consolidation and conversion of synaptic plasticity, functionally analogous to the synapse with an additional neuromodulator in biological neural networks.     

Bio-plausible memristive neural components towards hardware implementation of brain-like intelligence

A memristor can comprehensively emulate the neural components rather than imitating a single characteristic superficially due to its analog and hysteretic resistive switching. Bio-plausible mimicry aims to emulate biological working mechanisms to implement the complicated functional characteristics of a neural network for artificial intelligence (AI). Bio-plausible neuromorphic device using memristor is a direct and efficient approach for the emulation of biological systems, contributing to the realization of brain-like intelligence beyond limited AI applications. In this article, we review recent progress in bio-plausible mimicry of neural components using memristive devices. Memristor-based artificial neurons, synapses, and nerve systems are discussed focusing on the analogy in the operation mechanisms. In addition, we explore the neurological and interdisciplinary approaches in neuronal mapping and brain-computer interfaces to place an emphasis on the relationship between memristive neuromorphic system and biological neural network.     

Memristive Devices with Highly Repeatable Analog States Boosted by Graphene Quantum Dots

Memristive devices, having a huge potential as artificial synapses for low‐power neural networks, have received tremendous attention recently. Despite great achievements in demonstration of plasticity and learning functions, little progress has been made in the repeatable analog resistance states of memristive devices, which is, however, crucial for achieving controllable synaptic behavior. The controllable behavior of synapse is highly desired in building neural networks as it helps reduce training epochs and diminish error probability. Fundamentally, the poor repeatability of analog resistance states is closely associated with the random formation of conductive filaments, which consists of oxygen vacancies. In this work, graphene quantum dots (GQDs) are introduced into memristive devices. By virtue of the abundant oxygen anions released from GQDs, the GQDs can serve as nano oxygen‐reservoirs and enhance the localization of filament formation. As a result, analog resistance states with highly tight distribution are achieved with nearly 85% reduction in variations. In addition the insertion of GQDs can alter the energy band alignment and boost the tunneling current, which leads to significant reduction in both switching voltages and their distribution variations. This work may pave the way for achieving artificial neural networks with accurate and efficient learning capability.     

Instar and outstar learning with memristive nanodevices

The instar and outstar synaptic models are among the oldest and most useful in the field of neural networks. In this paper we show how to approximate the behavior of instar and outstar synapses in neuromorphic electronic systems using memristive nanodevices and spiking neurons. Memristive nanodevices are especially attractive for this application since such devices are tiny, can be densely packed in crossbar-like structures and possess the long time constants, or memory, needed by the synaptic models.     

Synergistic Gating of Electro‐Iono‐Photoactive 2D Chalcogenide Neuristors: Coexistence of Hebbian and Homeostatic Synaptic Metaplasticity

Emulation of brain‐like signal processing with thin‐film devices can lay the foundation for building artificially intelligent learning circuitry in future. Encompassing higher functionalities into single artificial neural elements will allow the development of robust neuromorphic circuitry emulating biological adaptation mechanisms with drastically lesser neural elements, mitigating strict process challenges and high circuit density requirements necessary to match the computational complexity of the human brain. Here, 2D transition metal di‐chalcogenide (MoS₂) neuristors are designed to mimic intracellular ion endocytosis–exocytosis dynamics/neurotransmitter‐release in chemical synapses using three approaches: (i) electronic‐mode: a defect modulation approach where the traps at the semiconductor–dielectric interface are perturbed; (ii) ionotronic‐mode: where electronic responses are modulated via ionic gating; and (iii) photoactive‐mode: harnessing persistent photoconductivity or trap‐assisted slow recombination mechanisms. Exploiting a novel multigated architecture incorporating electrical and optical biases, this incarnation not only addresses different charge‐trapping probabilities to finely modulate the synaptic weights, but also amalgamates neuromodulation schemes to achieve “plasticity of plasticity–metaplasticity” via dynamic control of Hebbian spike‐time dependent plasticity and homeostatic regulation. Coexistence of such multiple forms of synaptic plasticity increases the efficacy of memory storage and processing capacity of artificial neuristors, enabling design of highly efficient novel neural architectures.     

Einfluss der Netztopologie der künstlich neuronalen Netze auf das Ergebnis der Abschätzung zyklischer Kennwerte

Several estimation methods have been developed to estimate the cyclic material parameters out of the static material properties. Most of these methods are based on empirical equations. Increasing numbers of input‐ and influencing parameters lead to an rising effort for determining these equations and the accuracy decreases. For this reason new suitable methods are sought to estimate the cyclic material behaviour. A very promising approach is the application of the artificial neural networks, which can derive self‐depended a relationship between in‐ and output parameters. Static parameters such as yield strength, tensile strength …? etc., which can rapidly be determined used as input parameters. The output parameters are the cyclic material parameters of the strain‐life curve and stress‐strain curve according to the Manson‐Coffin‐Basquin‐ and Ramberg‐Osgood curve. Many different artificial neural networks with different structures and complexity can be applied. In this paper the influence of the topology of an artificial neural network on the estimation accuracy will be investigated. Based on the results of a reference artificial neural network it will be shown, that more complex topologies in the network do not lead inevitably to better estimations.     

The bistability phenomenon in single and coupled oscillators based on VO<Subscript>2</Subscript> switches

New operation regimes of single and coupled oscillators in circuits based on planar VO₂ switches have been studied. The phenomenon of bistability is discovered, which consists in controlled switching of self-sustained oscillations by external pulses, which is a promising basis for the creation of oscillatory memory cells and implementation of pulse coupling regimes in artificial neural networks (ANNs). The duration of switch-on and switch-off pulses is no less that ~20 μs and 30 ms, respectively. It is established that the region of threshold voltages for bistable switching in coupled oscillators is much wider than in a single oscillator and the hysteresis width in the former case can reach 2 V. A regime of initiation of switching packets has been observed that models the ANN packet activity.     

An artificial neural network approach to multiphase continua constitutive modeling

Constitutive equations describe intrinsic relationships among sets of material system parameters. This study utilizes artificial neural networks in place of a traditional micromechanical approach to calculate the global (macroscopic) elastic properties of composite materials given the local (microscopic) properties and local geometry. This approach is shown to be more computationally efficient than conventional numerical micromechanical approaches. An eight sub-celled representative volume element is used for the local geometry. Multi target artificial neural networks (MTANNs) and single target artificial neural networks are studied for applicability in predicting the global properties. The best performing MTANN achieves a precision of 9%. The single target artificial neural networks (STANNs) perform best and predicts the global properties within a target error of 5.3%. The computation time is 1.8 s for all six STANNs to predict six global properties for 19,683 different microstructures.     

Mediating Short‐Term Plasticity in an Artificial Memristive Synapse by the Orientation of Silica Mesopores

Memristive synapses based on resistive switching are promising electronic devices that emulate the synaptic plasticity in neural systems. Short‐term plasticity (STP), reflecting a temporal strengthening of the synaptic connection, allows artificial synapses to perform critical computational functions, such as fast response and information filtering. To mediate this fundamental property in memristive electronic devices, the regulation of the dynamic resistive change is necessary for an artificial synapse. Here, it is demonstrated that the orientation of mesopores in the dielectric silica layer can be used to modulate the STP of an artificial memristive synapse. The dielectric silica layer with vertical mesopores can facilitate the formation of a conductive pathway, which underlies a lower set voltage (≈1.0 V) compared to these with parallel mesopores (≈1.2 V) and dense amorphous silica (≈2.0 V). Also, the artificial memristive synapses with vertical mesopores exhibit the fastest current increase by successive voltage pulses. Finally, oriented silica mesopores are designed for varying the relaxation time of memory, and thus the successful mediation of STP is achieved. The implementation of mesoporous orientation provides a new perspective for engineering artificial synapses with multilevel learning and forgetting capability, which is essential for neuromorphic computing.     

Finite Size Effects on the Switching Dynamics of Spin‐Crossover Thin Films Photoexcited by a Femtosecond Laser Pulse

Using ultrafast optical absorption spectroscopy, the room‐temperature spin‐state switching dynamics induced by a femtosecond laser pulse in high‐quality thin films of the molecular spin‐crossover (SCO) complex [Fe(HB(tz)₃)₂] (tz = 1,2,4‐triazol‐1‐yl) are studied. These measurements reveal that the early, sub‐picosecond, low‐spin to high‐spin photoswitching event, with linear response to the laser pulse energy, can be followed under certain conditions by a second switching process occurring on a timescale of tens of nanoseconds, enabling nonlinear amplification. This out‐of‐equilibrium dynamics is discussed in light of the characteristic timescales associated with the different switching mechanisms, i.e., the electronic and structural rearrangements of photoexcited molecules, the propagation of strain waves at the material scale, and the thermal activation above the molecular energy barrier. Importantly, the additional, nonlinear switching step appears to be completely suppressed in the thinnest (50 nm) film due to the efficient heat transfer to the substrate, allowing the system to retrieve the thermal equilibrium state on the 100 ns timescale. These results provide a first milestone toward the assessment of the physical parameters that drive the photoresponse of SCO thin films, opening up appealing perspectives for their use as high‐frequency all‐optical switches working at room temperature.