Learning through ferroelectric domain dynamics in solid-state synapses

• Published in: Nature communications Vol. 8; no. 1; pp. 14736 - 7
• Main Authors: Boyn, Sören, Grollier, Julie, Lecerf, Gwendal, Xu, Bin, Locatelli, Nicolas, Fusil, Stéphane, Girod, Stéphanie, Carrétéro, Cécile, Garcia, Karin, Xavier, Stéphane, Tomas, Jean, Bellaiche, Laurent, Bibes, Manuel, Barthélémy, Agnès, Saïghi, Sylvain, Garcia, Vincent
• Format: Journal Article
• Language: English
• Published: London Nature Publishing Group UK 03.04.2017; Nature Publishing Group; Nature Portfolio
• Subjects: 60 APPLIED LIFE SCIENCES; 639/301/1005/1007; 639/766/119/996; 639/925/927/1007; Electricity; electronic devices; Engineering Sciences; Ferroelectrics; ferroelectrics and multiferroics; Humanities and Social Sciences; Iron - chemistry; Micro and nanotechnologies; Microelectronics; multidisciplinary; Neural Networks (Computer); Plasticity; Science; Science (multidisciplinary); Time Factors
• ISSN: 2041-1723; 2041-1723
• DOI: 10.1038/ncomms14736

In the brain, learning is achieved through the ability of synapses to reconfigure the strength by which they connect neurons (synaptic plasticity). In promising solid-state synapses called memristors, conductance can be finely tuned by voltage pulses and set to evolve according to a biological learning rule called spike-timing-dependent plasticity (STDP). Future neuromorphic architectures will comprise billions of such nanosynapses, which require a clear understanding of the physical mechanisms responsible for plasticity. Here we report on synapses based on ferroelectric tunnel junctions and show that STDP can be harnessed from inhomogeneous polarization switching. Through combined scanning probe imaging, electrical transport and atomic-scale molecular dynamics, we demonstrate that conductance variations can be modelled by the nucleation-dominated reversal of domains. Based on this physical model, our simulations show that arrays of ferroelectric nanosynapses can autonomously learn to recognize patterns in a predictable way, opening the path towards unsupervised learning in spiking neural networks. Accurate modelling of memristor dynamics is essential for the development of autonomous learning in artificial neural networks. Through a combined theoretical and experimental study of the polarization switching process in ferroelectric memristors, Boyn et al . establish a model that enables learning and retrieving patterns in a neural system.