Xumeng Zhang

Emulating Short-Term and Long-Term Plasticity of Bio-Synapse Based on Cu/a-Si/Pt Memristor

Short-term plasticity and long-term plasticity of bio-synapse are thought to underpin critical physiological functions in neural circuits. In this letter, we vividly emulated the short-term and long-term synaptic functions in a single Cu/a-Si/Pt memristor. By controlling the injection quantity of Cu cations into the a-Si layer, the device showed volatile and non-volatile resistive switching behaviors. Owing to the unique characteristics of Cu/a-Si/Pt device, the short-term synaptic functions, i.e., short-term potentiation, pair-pulse facilitation, and long-term functions, i.e., long-term potentiation/depression, spike-timing-dependent plasticity, were mimicked in the memristor successfully. Furthermore, the transition from short-term memory to long-term memory of the device was also observed under repeated stimuli. The experimental results confirm that the Cu/a-Si/Pt memristor with various synaptic behaviors has a potential application in the brain-inspired computing systems.

Breaking the Current‐Retention Dilemma in Cation‐Based Resistive Switching Devices Utilizing Graphene with Controlled Defects

Cation-based resistive switching (RS) devices, dominated by conductive filaments (CF) formation/dissolution, are widely considered for the ultrahigh density nonvolatile memory application. However, the current-retention dilemma that the CF stability deteriorates greatly with decreasing compliance current makes it hard to decrease operating current for memory application and increase driving current for selector application. By centralizing/decentralizing the CF distribution, this current-retention dilemma of cation-based RS devices is broken for the first time. Utilizing the graphene impermeability, the cation injecting path to the RS layer can be well modulated by structure-defective graphene, leading to control of the CF quantity and size. By graphene defect engineering, a low operating current (≈1 µA) memory and a high driving current (≈1 mA) selector are successfully realized in the same material system. Based on systematically materials analysis, the diameter of CF, modulated by graphene defect size, is the major factor for CF stability. Breakthrough in addressing the current-retention dilemma will instruct the future implementation of high-density 3D integration of RS memory immune to crosstalk issues.

An Artificial Neuron Based on a Threshold Switching Memristor

Artificial neurons and synapses are critical units for processing intricate information in neuromorphic systems. Memristors are frequently engineered as artificial synapses due to their simple structures, gradually changing conductance and high-density integration. However, few studies have designed memristors as artificial neurons. In this letter, we demonstrate an integration-and-fire artificial neuron based on a Ag/SiO2/Au threshold switching memristor. This neuron displays four critical features for action-potential-based computing: the all-or-nothing spiking of an action potential, threshold-driven spiking, a refractory period, and a strength-modulated frequency response. As a post-synaptic neuron, the designed neuron was demonstrated to be applicable to digit recognition. These results demonstrate that the developed artificial neuron can realize the basic functions of spiking neurons and has great potential for neuromorphic computing.

Improvement of Device Reliability by Introducing a BEOL-Compatible TiN Barrier Layer in CBRAM

Negative-SET behavior, induced by nano-filament overgrowth phenomenon, takes major responsibility to the reset failure phenomenon in conductive bridge random access memory (CBRAM). The unexpected negative-SET behavior in CBRAM devices can result in serious reliability issues and has been an obstacle on the way to mass production. In this letter, we have proposed a back-end-of-line (BEOL) compatible TiN barrier layer to improve the device reliability in CBRAM devices by eliminating the nano-filament overgrowth phenomenon and negative-SET behavior. Thus, a higher reset voltage can be applied to the TiN barrier layer devices to achieve more complete reset process and obtain better resistive switching performance. The results show that the Cu/HfO2/TiN/Ru device with one transistor structure has excellent comprehensive memory properties, including high reliability, fast switching speed, high resistance state uniformity, high endurance, long retention, and multi-level storage ability.

Bipolar Analog Memristors as Artificial Synapses for Neuromorphic Computing

Synaptic devices with bipolar analog resistive switching behavior are the building blocks for memristor-based neuromorphic computing. In this work, a fully complementary metal-oxide semiconductor (CMOS)-compatible, forming-free, and non-filamentary memristive device (Pd/Al2O3/TaOx/Ta) with bipolar analog switching behavior is reported as an artificial synapse for neuromorphic computing. Synaptic functions, including long-term potentiation/depression, paired-pulse facilitation (PPF), and spike-timing-dependent plasticity (STDP), are implemented based on this device; the switching energy is around 50 pJ per spike. Furthermore, for applications in artificial neural networks (ANN), determined target conductance states with little deviation (<1%) can be obtained with random initial states. However, the device shows non-linear conductance change characteristics, and a nearly linear conductance change behavior is obtained by optimizing the training scheme. Based on these results, the device is a promising emulator for biology synapses, which could be of great benefit to memristor-based neuromorphic computing.