Sequential injection of domain walls into ferroelectrics at different bias voltages: Paving the way for “domain wall memristors”

J R Whyte,A Gruverman,C Canalias,J F Einsle,C M Ashcroft,R G P Mcquaid,J M Gregg

doi:10.1063/1.4891347

Abstract

Simple meso-scale capacitor structures have been made by incorporating thin (∼300 nm) single crystal lamellae of KTiOPO4 (KTP) between two coplanar Pt electrodes. The influence that either patterned protrusions in the electrodes or focused ion beam milled holes in the KTP have on the nucleation of reverse domains during switching was mapped using piezoresponse force microscopy imaging. The objective was to assess whether or not variations in the magnitude of field enhancement at localised “hot-spots,” caused by such patterning, could be used to both control the exact locations and bias voltages at which nucleation events occurred. It was found that both the patterning of electrodes and the milling of various hole geometries into the KTP could allow controlled sequential injection of domain wall pairs at different bias voltages; this capability could have implications for the design and operation of domain wall electronic devices, such as memristors, in the future.

Highlights

In 1971, Chua1 stated that the three known basic two terminal components in electrical circuits were insufficient to fully describe the interrelationships between all possible pairs of the four fundamental circuit variables: current (I), charge (q), voltage (V), and flux-linkage (/)
focused ion beam milled holes in the KTP have on the nucleation of reverse domains
two terminal component must exist in order to link

Summary

Introduction

In 1971, Chua stated that the three known basic two terminal components in electrical circuits (resistors, capacitors, and inductors) were insufficient to fully describe the interrelationships between all possible pairs of the four fundamental circuit variables: current (I), charge (q), voltage (V), and flux-linkage (/). He postulated that one further two terminal component must exist in order to link together flux and charge (du 1⁄4 Mdq).. As noted by Mathur, the magnetoelectric coupling coefficient in multiferroics should be inversely proportional to du dq and memristors could have been previously made within the magnetoelectric multiferroic community and overlooked. Since 2008, the creation and investigation of memristors has become a rather active area for research, fuelled by potential use in neuromorphic computer architecture (due to the similarity between memristors and biological synapses) as well as other more conventional logic and memory applications.

Methods

Results

Conclusion