Table of Contents
Fetching ...

Spatio-temporal evolution of resistance state in simulated memristive networks

Fabrizio Di Francesco, Gabriel A. Sanca, Cynthia P. Quinteros

TL;DR

By comparing the response of homogeneous and heterogeneous networks, this work delineates differences that might be experimentally observed when the number of memristive units is scaled up and disorder arises as an inevitable feature.

Abstract

Originally studied for their suitability to store information compactly, memristive networks are now being analysed as implementations of neuromorphic circuits. An extremely high number of elements is thus mandatory. To surpass the limited achievable connectivity - due to the featuring size - exploiting self-assemblies has been proposed as an alternative, in turn posing more challenges. In an attempt for offering insight on what to expect when characterizing the collective electrical response of switching assemblies, in this work, networks of memristive elements are simulated. Collective electrical behaviour and maps of resistance states are characterized upon different electrical stimuli. By comparing the response of homogeneous and heterogeneous networks, we delineate differences that might be experimentally observed when the number of memristive units is scaled up and disorder arises as an inevitable feature.

Spatio-temporal evolution of resistance state in simulated memristive networks

TL;DR

By comparing the response of homogeneous and heterogeneous networks, this work delineates differences that might be experimentally observed when the number of memristive units is scaled up and disorder arises as an inevitable feature.

Abstract

Originally studied for their suitability to store information compactly, memristive networks are now being analysed as implementations of neuromorphic circuits. An extremely high number of elements is thus mandatory. To surpass the limited achievable connectivity - due to the featuring size - exploiting self-assemblies has been proposed as an alternative, in turn posing more challenges. In an attempt for offering insight on what to expect when characterizing the collective electrical response of switching assemblies, in this work, networks of memristive elements are simulated. Collective electrical behaviour and maps of resistance states are characterized upon different electrical stimuli. By comparing the response of homogeneous and heterogeneous networks, we delineate differences that might be experimentally observed when the number of memristive units is scaled up and disorder arises as an inevitable feature.

Paper Structure

This paper contains 2 sections, 5 equations, 4 figures.

Figures (4)

  • Figure 1: Memristive response of isolated units, defined as $R_{init} = R_{OFF} = 200$ k$\Omega$, $R_{ON} = 2$ k$\Omega$, $V_t = 0.6$ V, upon one sinusoidal cycle of $1$ Hz. (a) Current as a function of voltage (I-V) using fixed $\beta$ ($= 5 \cdot 10^5~\frac{\Omega}{V \cdot s}$) under four different $A$ values. (b) $X$ as a function of $V_{ext}(t)$ corresponding to the same runs shown in (a). (c) I-V obtained upon fixed $A$ ($= 2.0$ V) for two different $\beta$ values.
  • Figure 2: Resistance map of a memristive network comprising 4x4 nodes and 24 memristive units collapsing every device's $X$ in a coloured scale.
  • Figure 3: Collective resistance state (R$_{coll}$) as a function of time (lower panel) of an homogeneous 4x4 network ($R_{OFF} = 200$ k$\Omega = R_{init}$, $R_{ON} = 2$ k$\Omega$, $\beta = 500$$\frac{k\Omega}{V \cdot s}$, $V_t = 0.6$ V) upon ten sinusoidal cycles of $1$ Hz with $A = 12$ V (upper panel). The resistance states associated to a single device are included for reference.
  • Figure 4: Sensitization experiments. Upper panel: $V_{ext} (t)$. Lower panel: R$_{coll}$ as a function of the position of the sensitized unit. The four different symbols (color online) account for R$_{coll}$ quantified at the subsequent remnant conditions (consistently depicted in the upper panel).