A ferroelectric memristor (1206.3397v1)

Abstract: Memristors are continuously tunable resistors that emulate synapses. Conceptualized in the 1970s, they traditionally operate by voltage-induced displacements of matter, but the mechanism remains controversial. Purely electronic memristors have recently emerged based on well-established physical phenomena with albeit modest resistance changes. Here we demonstrate that voltage-controlled domain configurations in ferroelectric tunnel barriers yield memristive behaviour with resistance variations exceeding two orders of magnitude and a 10 ns operation speed. Using models of ferroelectric-domain nucleation and growth we explain the quasi-continuous resistance variations and derive a simple analytical expression for the memristive effect. Our results suggest new opportunities for ferroelectrics as the hardware basis of future neuromorphic computational architectures.

• The paper introduces a ferroelectric memristor design using tunnel junctions to achieve tunable resistance states with over 100x contrast and 10 ns switching speed.
• It employs ferroelectric domain nucleation and growth models to explain quasi-continuous resistance variation in BaTiO3/La0.67Sr0.33MnO3 layers.
• The study paves the way for advanced neuromorphic architectures by leveraging enhanced memristive effects for artificial synapses and spike timing-dependent plasticity.

A Ferroelectric Memristor

The paper "A ferroelectric memristor" presents a detailed paper of ferroelectric tunnel junctions (FTJs) demonstrating memristive behavior, a novel step in neuromorphic computing architectures. The research primarily focuses on the utilization and control of ferroelectric domain structures to achieve finely tunable resistance states, which can serve as hardware for future neuromorphic computational systems.

The authors introduce a model that explains the quasi-continuous resistance variations in Ferroelectric Tunnel Barriers (FTBs). These barriers, comprising BaTiO$_3$/La$_{0.67}$Sr$_{0.33}$MnO$_3$ layers, have resistance differences greater than two orders of magnitude with an operational speed of 10 ns. This is a significant improvement over previously developed purely electronic memristors, where the resistance contrast has been modest. By integrating models of ferroelectric-domain nucleation and growth, the authors derive a straightforward expression to describe the memristive effect.

One of the key points of the paper is the explanation of the giant tunnel electroresistance (TER) effect in FTJs, which arises from the asymmetric polarization screening at barrier/electrode interfaces. The paper diverges from prior approaches that only considered FTJs for binary data storage. The authors delve into the intricate dynamics of ferroelectric domain configurations, which are crucial for obtaining a broad range of resistance states.

Experimental observations were conducted using Piezoresponse Force Microscopy (PFM) and showed that domain configurations within a ferroelectric barrier could be manipulated. By adjusting the amplitude and sequence of voltage pulses (ranging from 10 to 200 ns), researchers obtained a virtually continuous range of resistance states—from OFF to ON—which are governed by the dynamics of ferroelectric domains during polarization reversal. The memristive effect, confirmed through resistive switching measurements, holds implications for the advancement of ferroelectrics in neuromorphic computing architectures.

Moreover, the authors explored the dynamics of the resistance switching processes, modeling them based on domain nucleation and propagation concepts governed by the Kolmogorov-Avrami-Ishibashi (KAI) model. The work herein extends to address the potential latency and asymmetry in switching observed in ferroelectric devices, attributed to variations in nucleation and propagation kinetics across the sample.

The implications of this research are profound for next-generation computing. By offering a clear path to the exploitation of ferroelectric domains for memristive behavior, the paper lays a substantial foundation for the integration of ferroelectrics into advanced neuromorphic systems. FTJs thus open new pathways in the design of artificial synapses and improvement of synaptic transmission fidelity through mechanisms like spike timing-dependent plasticity.

Going forward, theoretical and practical explorations could delve into further optimization of ferroelectric film qualities such as epitaxy and thickness to refine domain control and device efficiency. Additionally, while substantial strides have been made in utilizing ferroelectrics for non-volatile memory applications, potential exists for enhancing performance under operational conditions closer to that of biological systems, such as operating temperatures and environment.

In essence, the detailed investigation into FTJs as memristive components underscores their potential in enhancing computational architectures that mimic the neural functionality found in the human brain. The paper provides vital experimental data and theoretical insights intended to aid further advancement in this promising field.