No abstract

Neuromorphic engineering is a promising technology for developing new computing systems owing to the low-power operation and the massive parallelism similarity to the human brain. Optimal function of neuronal networks requires interplay between rapid forms of Hebbian plasticity and homeostatic mechanisms that adjust the threshold for plasticity, termed metaplasticity. Metaplasticity has important implications in synapses and is barely addressed in neuromorphic devices. An understanding of metaplasticity might yield new insights into how the modification of synapses is regulated and how information is stored by synapses in the brain. Here, we propose a method to imitate the metaplasticity inhibition of long-term potentiation (MILTP) for the first time based on memristors. In addition, the metaplasticity facilitation of long-term potentiation (MFLTP) and the metaplasticity facilitation of long-term depression (MFLTD) are also achieved. Moreover, the mechanisms of metaplasticity in memristors are discussed. Additionally, the proposed method to mimic the metaplasticity is verified by three different memristor devices including oxide-based resistive memory (OxRAM), interface switching random access memory, and conductive bridging random access memory (CBRAM). This is a further step toward developing fully bio-realistic artificial synapses using memristors. The findings in this study will deepen our understanding of metaplasticity, as well as provide new insight into bio-realistic neuromorphic engineering.

The sensory nervous system (SNS) builds up the association between external stimuli and the response of organisms. In this system, habituation is a fundamental characteristic that filters out irrelevantly repetitive information and makes the SNS adapt to the external environment. To emulate this critical process in electronic devices, a LixSiOy‐based memristor (TiN/LixSiOy/Pt) is developed where the temporal response under repetitive stimulation is similar to that of habituation. By connecting this synaptic device to a leaky integrate‐and‐fire neuron based on a Ag/SiO2:Ag/Au memristor, a fully memristive SNS with habituation is experimentally demonstrated. Finally, a habituation spiking neural network based on the SNS is built and its application in obstacle avoidance for robot navigation is successfully presented. The results provide that a direct emulation of the biologically inspired learning process by memristors could be a sound choice for neuromorphic hardware implementation.

Neuromorphic devices are among the most emerging electronic components to realize artificial neural systems and replace traditional complementary metal-oxide semiconductor devices in recent times. In this work, tri-layer HfO 2 /BiFeO 3 (BFO)/HfO 2 memristors are designed by inserting traditional ferroelectric BFO layers measuring ≈4 nm after thickness optimization. The novel designed memristor shows excellent resistive switching (RS) performance such as a storage window of 10 4 and multi-level storage ability. Remarkably, essential synaptic functions can be successfully realized on the basis of the linearity of conductance modulation. The pattern recognition simulation based on neuromorphic network is conducted with 91.2% high recognition accuracy. To explore the RS performance enhancement and artificial synaptic behaviors, conductive filaments (CFs) composed of Hafnium (Hf ) single crystal with a hexaganal lattice structure are observed using high-resolution transmission electron microscopy. It is reasonable to believe that the sufficient oxygen vacancies in the inserting BFO thin film play a crucial role in adjusting the morphology and growth of Hf CFs, which leads to the promising synaptic and enhanced RS behavior, thus demonstrating the potential of this memristor for use in neuromorphic computing.

The scaling down of switching media encounters high leakage current in the traditional oxide material based memristors, resulting in high power consumption of chips. Two-dimensional (2D) materials promise an ultimate device scaling down to atomic layer thickness. Herein, black phosphorus (BP) and its self-assembly phosphorous oxide (BP) memristors are constructed, which leverages the high on/off ratio operation of oxides and low leakage current of 2D materials with high performance. The memristors exhibit reproducible and reliable switching characteristics with the on/off ratio >107 and data retention >104 s. Depending on the high reproducibility, basic “AND” and “OR” gates have been constructed on flexible substrates. Moreover, on the basis of the symmetry and linearity of conductance in the devices, the neural network simulation for supervised learning presents an online learning accuracy of 91.4%. This work opens an avenue for future flexible electronics.

No abstract

Conductive‐bridging random access memory (CBRAM), dominated by conductive filament (CF) formation/rupture, has received much attention due to its simple structure and outstanding performances for nonvolatile memory, neuromorphic computing, digital logic, and analog circuit. However, the negative‐SET behavior can degrade device reliability and parameter uniformity. And large RESET current increases power consumption for memory applications. By inserting 2D material, molybdenum disulfide (MoS2), for interface engineering with the device configuration of Ag/ZrO2/MoS2/Pt, the negative‐SET behavior is eliminated, and the RESET current is reduced simultaneously. With the ion barrier property of MoS2, the CF can probably not penetrate the MoS2 layer, thus eliminating the negative‐SET behavior. And with the low thermal conductivity of MoS2, the internal temperature of the device would be relatively high at RESET, accelerating probably redox reactions. As a result, the RESET current is reduced by an order of magnitude. This interface engineering opens up a way in improving the resistive switching performances of CBRAM, and can be of great benefit to the potential applications of MoS2 in next‐generation data storage.

No abstract