Biorealistic Implementation of Synaptic Functions with Oxide Memristors through Internal Ionic Dynamics

Memristors have attracted broad interest as a promising candidate for future memory and computing applications. Particularly, it is believed that memristors can effectively implement synaptic functions and enable efficient neuromorphic systems. Most previous studies, however, focus on implementing specific synaptic learning rules by carefully engineering external programming parameters instead of focusing on emulating the internal cause that leads to the apparent learning rules. Here, it is shown that by taking advantage of the different time scales of internal oxygen vacancy (V_O) dynamics in an oxide‐based memristor, diverse synaptic functions at different time scales can be implemented naturally. Mathematically, the device can be effectively modeled as a second‐order memristor with a simple set of equations including multiple state variables. Not only is this approach more biorealistic and easier to implement, by focusing on the fundamental driving mechanisms it allows the development of complete theoretical and experimental frameworks for biologically inspired computing systems.     

Coexistence of multiple bifurcation modes in memristive diode-bridge-based canonical Chua’s circuit

Based on a memristive diode bridge cascaded with series resistor and inductor filter, a modified memristive canonical Chua’s circuit is presented in this paper. With the modelling of the memristive circuit, a normalised system model is built. Stability analyses of the equilibrium points are performed and bifurcation behaviours are investigated by numerical simulations and hardware experiments. Most extraordinary in the memristive circuit is that within a parameter region, coexisting phenomenon of multiple bifurcation modes is emerged under six sets of different initial values, resulting in the coexistence of four sets of topologically different and disconnected attractors. These coexisting attractors are easily captured by repeatedly switching on and off the circuit power supplies, which well verify the numerical simulations.     

Anticipation of digital patterns

A memristive device is a novel passive device, which is essentially a resistor with memory. This device can be used for novel technical applications like neuromorphic computation. In this paper, we focus on anticipation—a capability of a system to decide how to react in an environment by predicting future states. Especially, we have designed an elementary memristive circuit for the anticipation of digital patterns, where this circuit is based on the capability of an amoeba to anticipate periodically occurring unipolar pulses. The resulting circuit has been verified by digital simulations and has been realized in hardware as well. For the practical realization, we have used an Ag‐doped TiO_2?x‐based memristive device, which has been fabricated in planar capacitor structures on a silicon wafer. The functionality of the circuit is shown by simulations and measurements. Finally, the anticipation of information is demonstrated by using images, where the robustness of this anticipatory circuit against noise and faulty intermediate information is visualized.     

Crisis‐induced coexisting multiple attractors in a second‐order nonautonomous memristive diode bridge‐based circuit

This paper reports a second‐order nonautonomous memristive diode bridge‐based circuit, upon which a system model is established. The AC and DC equilibrium points and their stability evolutions are theoretically analyzed, and the mechanisms of complex dynamical behaviors are explored in detail. Furthermore, the stimulus‐dependent dynamical behaviors are numerically performed by the single‐parameter bifurcation diagrams, Lyapunov exponents, and phase portraits. Of particular concern, it should be highly emphasized that multiple kinds of crisis scenarios associated with the initial conditions are found in a specified parameter region, resulting in that coexisting multiple attractors under different initial conditions are discovered for the fixed system parameters. Additionally, hardware experiments and PSpice circuit simulations are used to confirm the numerically simulated results.     

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.     

A Synaptic Transistor based on Quasi‐2D Molybdenum Oxide

Biological synapses store and process information simultaneously by tuning the connection between two neighboring neurons. Such functionality inspires the task of hardware implementation of neuromorphic computing systems. Ionic/electronic hybrid three‐terminal memristive devices, in which the channel conductance can be modulated according to the history of applied voltage and current, provide a more promising way of emulating synapses by a substantial reduction in complexity and energy consumption. 2D van der Waals materials with single or few layers of crystal unit cells have been a widespread innovation in three‐terminal electronic devices. However, less attention has been paid to 2D transition‐metal oxides, which have good stability and technique compatibility. Here, nanoscale three‐terminal memristive transistors based on quasi‐2D α‐phase molybdenum oxide (α‐MoO₃) to emulate biological synapses are presented. The essential synaptic behaviors, such as excitatory postsynaptic current, depression and potentiation of synaptic weight, and paired‐pulse facilitation, as well as the transition of short‐term plasticity to long‐term potentiation, are demonstrated in the three‐terminal devices. These results provide an insight into the potential application of 2D transition‐metal oxides for synaptic devices with high scaling ability, low energy consumption, and high processing efficiency.     

Ionically Mediated Mechanical Deformation Associated with Memristive Switching

Ionically mediated phenomena underpin the functioning of various devices, including batteries, solid oxide fuel cells, memristors, and neuromorphic devices. The ionic behavior corresponding to ionically mediated phenomena causes not only variations in the electrical properties but also mechanical deformation, which is crucial for device reliability. However, the interrelation between ionically mediated electrical properties and mechanical deformation has not been elucidated yet. This study investigates ionically mediated mechanical deformation accompanied by memristive switching in a TiO₂ single crystal through simultaneous conductive atomic force microscopy and electrochemical strain microscopy. A comprehensive analysis indicates the existence of a relationship between mechanical deformation and memristive switching based on the ionic behavior. Furthermore, an ionic state variable is used to simplify the interrelation between the electrochemical strain hysteresis and memristive switching associated with applied voltage. This study provides insights on the ionic behavior and can be extended to other systems for the general analysis of the relationship between mechanical deformation and electrical properties.     

Tuning Memristivity by Varying the Oxygen Content in a Mixed Ionic–Electronic Conductor

The rising interest shown for adaptable electronics and brain‐inspired neuromorphic hardware increases the need for new device architectures and functional materials to build such devices. The rational design of these memory components also benefits the comprehension and thus the control over the microscopic mechanisms at the origin of memristivity. In oxide‐based valence‐change memories, the control of the oxygen drift and diffusion kinetics is a key aspect in obtaining the gradual analog‐type change in resistance required for artificial synapse applications. However, only a few devices are designed with this in mind, as they are commonly built around ionic insulating active materials. This shortcoming is addressed by using a mixed ionic–electronic conductor as functional memristive material. This work demonstrates how the oxygen content in La₂NiO_4+δ (L2NO4), tuned through post‐annealing treatments, has a critical influence on the memory characteristics of L2NO4‐based memristive devices. The presence of interstitial oxygen point defects in L2NO4 affects both its structure and electrical properties. High oxygen stoichiometry in the pristine state leads to an increased electrical conductivity, ultimately resulting in an improved memory window with highly multilevel, analog‐type memory programing capabilities, desirable for analog computing and synaptic applications in particular.     

Global Synchronization of Stochastically Disturbed Memristive Neurodynamics via Discontinuous Control Laws

This paper presents the theoretical results on the master-slave (or driving-response) synchronization of two memristive neural networks in the presence of additive noise. First, a control law with a linear time-delay feedback term and a discontinuous feedback term is introduced. By utilizing the stability theory of stochastic differential equations, sufficient conditions are derived for ascertaining global synchronization in mean square using this control law. Second, an adaptive control law consisting of a linear feedback term and a discontinuous feedback term is designed to achieve global synchronization in mean square, and it does not need prior information of network parameters or random disturbances. Finally, simulation results are presented to substantiate the theoretical results.