Reconfigurable Synaptic and Neuronal Functions in a V/VO_x/HfWO_x/Pt Memristor for Nonpolar Spiking Convolutional Neural Network

Abstract

The fully memristive neural network consisting of the threshold switching (TS) material-based electronic neurons and resistive switching (RS) one-based synapses shows the potential for revolutionizing the energy and area efficiency in neuromorphic computing while being confronted with challenges such as reliability and process compatibility between memristive synaptic and neuronal devices. Here, a spiking convolutional neural network (SCNN) is constructed with the forming-and-annealing-free V/VO x /HfWO x /Pt memristive devices. Specifically, both highly reliable RS (endurance >10 10 , on-off ratio >10 3 ) and TS (endurance >10 12 ) are found in the same device by setting it at RRAM or selector mode with either the HfWO x or naturally oxidized VO x layers dominating the conductance tuning. Such reconfigurability enables the emulation of both synaptic and nonpolar neuronal behaviors within the same device. A V/VO x /HfWO x /Pt-based hardware system is thus experimentally demonstrated at much simplified process complexity and higher reliability, in which typical neural dynamics including synaptic plasticity and nonpolar neuronal spiking response are imitated. At the network level, a fully memristive SCNN incorporating nonpolar neurons is proposed for the first time. The system level simulation shows competency in pattern recognition with a dramatically reduced hardware consumption, paving the way for implementing fully memristive intelligent systems.

Selective area doping for Mott neuromorphic electronics

“…To avoid detrimental voltage fluctuations due to signal interferences, operational amplifiers (op-amps) are introduced between neural components. Specifically, a voltage follower [i.e., isolator 1 (Iso1)] is placed between the N1 and the synapse ( 38 ), and an inverting amplifier plus a Howland current pump ( 39 , 40 ) for voltage-to-current conversion (i.e., Iso2) are installed between the synapse and the N2 ( 41 ). The synaptic weight ( w ) here is defined as the absolute value of the inverting amplifier’s gain (i.e., the ratio of the resistance across the op-amp over the effective synapse resistance) (fig.…”

Self‐selective memristor‐enabled in‐memory search for highly efficient data mining

“…We built the IMS system using V/VO x /HfO x /Pt SSMs organized in a 32 × 32 crossbar array (Figure 2A). As illustrated in the cross‐sectional transmission electron microscopy image of the device stacked structure (Figure 2B), the interfacial VO x layer (about 5 nm) between V and HfO x severs as a volatile selector with its highly reproducible insulator‐ metal‐transition characteristics 14 (Figure 2C). Figure 2D illustrates the I – V curves of the nonpolar selector for 50 cycles with the memristor in the low resistance state (LRS), indicating highly uniform volatile switching characteristics.…”

Nano t-Se Peninsulas Embedded in Natively Oxidized 2D TiSe2 Enable Uniform and Fast Memristive Switching

“…Memristors (or resistive random access memory, RRAM) have emerged as one of the most promising candidates of the next-generation nonvolatile memory and computing technologies because of their excellent performance such as high speed, low-energy consumption, and down-scaling potential, − which enables wide applications in storage class memory and neuromorphic computing. − Back in 2008, TiO 2 was discovered as the first functional memristive medium in the resistive switching device . After that, a large variety of functional memristive materials have been discovered, normally fabricated by deposition methods .…”