Bio-mimetic synaptic plasticity and learning in a sub-500 mV Cu/SiO2/W memristor

Abstract

The computational efficiency of the human brain is believed to stem from the parallel information processing capability of neurons with integrated storage in synaptic interconnections programmed by local spike triggered learning rules such as spike timing dependent plasticity (STDP). The extremely low operating voltages (approximately 100 mV) used to trigger neuronal signaling and synaptic adaptation is believed to be a critical reason for the brain's power efficiency. We demonstrate the feasibility of spike triggered STDP behavior in a two-terminal Cu/SiO 2 /W memristive device capable of operating below 500 mV. We analyze the state-dependent nature of conductance updates in the device to develop a phenomenological model. Using the model, we evaluate the potential of such devices to generate precise spike times under supervised learning conditions and classify handwritten digits from the MNIST dataset in an unsupervised learning setting. The results form a promising step towards creating a low power synaptic device capable of on-chip learning.

Reversible Transition of Volatile and Nonvolatile Switching in Ag–In–Zn–S Quantum Dot-Based Memristors with Low Power Consumption for Synaptic Applications

“…Neuromorphic computing, a new computing paradigm, has been widely investigated for its capability to confront the bottlenecks of classical von-Neumann computers and meet the needs of future computing system. , Neuromorphic electronic synapses, one of the basic functional elements for constructing highly efficient neuromorphic computing systems, have been employed to emulate the essential biosynaptic functions by adjusting synaptic strength (i.e., conductance). , As of now, metal/insulator/metal (MIM) nanoscale memristors, promising devices suitable for human brain-inspired computing programs with many merits of simple cell structure, low power consumption, fast response time, and excellent CMOS compatibility, have been significantly engineered as artificial synapses to realize advanced information storage and computation. − Distinguished resistive switching (RS) behaviors have been extensively reported in various memristors based on diverse oxides (SiO 2 , TaO x , HfO x , TiO 2 , etc. ), where the device can be reconfigured repeatedly between high-resistance state (HRS) and low-resistance state (LRS) via imposing external electrical input. − Among them, the well-known TiO 2 material has become a superior contender for designing memristors due to its low cost, thermal stability, good resistive switching properties, and compatibility with the CMOS process . The physical characteristics of the RS involving the formation/rupture of conductive filaments (CFs) have been investigated in a TiO 2 -based memristor via in situ probing transmission electron microscopy (TEM) observations .…”

Inkjet Printing: A Cheap and Easy‐to‐Use Alternative to Wire Bonding for Academics

“…Each memristor consists of a metal/insulator/metal structure in which the conductance of the insulator can be set to different values by applying electrical stresses. [16][17][18] In this type of device, the conductance change is driven by the formation and disruption of a conductive nanofilament across the insulating film, [19][20][21][22][23][24] and in many cases the disruption of the filament (and therefore the reset event) is a thermal effect. [25,26] Therefore, studying the temperature of the memristor with a high lateral resolution (below 100 nm) is very important to understand the functioning of these devices.…”

Probing switching mechanism of memristor for neuromorphic computing

“…By applying external voltages, the resistance state will change corresponding to the plasticity of synapses, such as short-term synaptic plasticity (STSP) mainly including short-term potentiation (STP) and short-term depression (STD), long-term synaptic plasticity (LTSP) including long-term potentiation (LTP) and long-term depression (LTD), etc, depending that the change of resistance is non-volatile or volatile [35,36]. So far, a lot of work has realized synaptic function [2,9,12,35,[37][38][39][40][41][42][43][44][45][46]. For example, Zhou et al reported a Pd/MoO x /indium tin oxide (ITO) memristor mimicking the transition from STSP to LTSP using a repeated pulse stimulation, which showed great prospects in the application of a neuromorphic visual system [9].…”