High-performance neuromorphic computing based on ferroelectric synapses with excellent conductance linearity and symmetry

Shu-Ting  Yang; Xingyu Li; Tongliang Yu; Jie Wang; Hong Fang; Fang Nie; Bing He; Le Zhao; Weimin Lu; Shishen Yan; Alain Nogaret; Gang Liu; Limei Zheng

doi:10.1002/adfm.202202366

High-performance neuromorphic computing based on ferroelectric synapses with excellent conductance linearity and symmetry

Shu-Ting Yang, Xingyu Li, Tongliang Yu, Jie Wang, Hong Fang, Fang Nie, Bing He, Le Zhao, Weimin Lu, Shishen Yan, Alain Nogaret, Gang Liu, Limei Zheng

Research output: Contribution to journal › Article › peer-review

113 Citations (SciVal)

281 Downloads (Pure)

Abstract

Artificial synapses can boost neuromorphic computing to overcome the inherent limitations of von Neumann architecture. As a promising memristor candidate, ferroelectric tunnel junctions (FTJ) enable the authors to successfully emulate spike-timing-dependent synapses. However, the nonlinear and asymmetric synaptic weight update under repeated presynaptic stimulation hampers neuromorphic computing by favoring the runaway of synaptic weights during learning. Here, the authors demonstrate an FTJ whose conductivity varies linearly and symmetrically by judiciously combining ferroelectric domain switching and oxygen vacancy migration. The artificial neural network based on this FTJ-synapse achieves classification accuracy of 96.7% during supervised learning, which is the closest to the maximum theoretical value of 98% achieved to date. This artificial synapse also demonstrates stable unsupervised learning in a noisy environment for its well-balanced spike-timing-dependent plasticity response. The novel concept of controlling ionic migration in ferroelectric materials paves the way toward highly reliable and reproducible supervised and unsupervised learning strategies.

Original language	English
Article number	2202366
Journal	Advanced Functional Materials
Volume	32
Issue number	35
Early online date	22 Jun 2022
DOIs	https://doi.org/10.1002/adfm.202202366
Publication status	Published - 25 Aug 2022

Bibliographical note

Funding Information:
S.‐T.Y. and X.‐Y.L. contributed equally to this work. The authors acknowledge the support from the National Key Research & Development Program of China (No. 2021YFB3601504), the National Natural Science Foundation of China (Nos. 52072218 and 12074149), the Primary Research & Development Plan of Shandong Province (No. 2019JZZY010313), and the Natural Science Foundation of Shandong province (No. ZR2020KE019 and ZR2020ZD28). The authors would like to thank the Analytical Center for Structural Constituent and Physical Property of Core Facilities Sharing Platform, Shandong University for XRD and PFM analysis. The idea and project was conceived by L.Z., W.‐M.L., and L.‐M.Z.; S.‐T.Y. and X.‐Y.L. characterized the electric transport properties under the supervision of L.‐M.Z.; T.‐L.Y. carried out the artificial neural network simulation under the supervision of L.Z.; J.W., H.F., F.N., and B.H. fabricated the devices and performed the PFM measurements; S.‐T.Y., L.Z., L.‐M.Z., and A.N. wrote the manuscript and all the authors contributed to the discussion of the manuscript.

Keywords

electronic synapses
ferroelectric tunnel junctions
linear and symmetric weight changes
spike-timing-dependent plasticity

ASJC Scopus subject areas

General Chemistry
General Materials Science
Condensed Matter Physics

Access to Document

10.1002/adfm.202202366

AFM manuscript -2022.5.1.PDF
This is the peer reviewed version of the following article: Yang, S.-T., Li, X.-Y., Yu, T.-L., Wang, J., Fang, H., Nie, F., He, B., Zhao, L., -M. Lü, W., Yan, S.-S., Nogaret, A., Liu, G., Zheng, L.-M., High-Performance Neuromorphic Computing Based on Ferroelectric Synapses with Excellent Conductance Linearity and Symmetry. Adv. Funct. Mater. 2022, 2202366., which has been published in final form at https://doi.org/10.1002/adfm.202202366. This article may be used for non-commercial purposes in accordance with Wiley Terms and Conditions for Use of Self-Archived Versions. This article may not be enhanced, enriched or otherwise transformed into a derivative work, without express permission from Wiley or by statutory rights under applicable legislation.
Accepted author manuscript, 953 KB

Cite this

@article{25ad0e07f0d94142a54c6a2658ad6155,

title = "High-performance neuromorphic computing based on ferroelectric synapses with excellent conductance linearity and symmetry",

abstract = "Artificial synapses can boost neuromorphic computing to overcome the inherent limitations of von Neumann architecture. As a promising memristor candidate, ferroelectric tunnel junctions (FTJ) enable the authors to successfully emulate spike-timing-dependent synapses. However, the nonlinear and asymmetric synaptic weight update under repeated presynaptic stimulation hampers neuromorphic computing by favoring the runaway of synaptic weights during learning. Here, the authors demonstrate an FTJ whose conductivity varies linearly and symmetrically by judiciously combining ferroelectric domain switching and oxygen vacancy migration. The artificial neural network based on this FTJ-synapse achieves classification accuracy of 96.7% during supervised learning, which is the closest to the maximum theoretical value of 98% achieved to date. This artificial synapse also demonstrates stable unsupervised learning in a noisy environment for its well-balanced spike-timing-dependent plasticity response. The novel concept of controlling ionic migration in ferroelectric materials paves the way toward highly reliable and reproducible supervised and unsupervised learning strategies.",

keywords = "electronic synapses, ferroelectric tunnel junctions, linear and symmetric weight changes, spike-timing-dependent plasticity",

author = "Shu-Ting Yang and Xingyu Li and Tongliang Yu and Jie Wang and Hong Fang and Fang Nie and Bing He and Le Zhao and Weimin Lu and Shishen Yan and Alain Nogaret and Gang Liu and Limei Zheng",

note = "Funding Information: S.‐T.Y. and X.‐Y.L. contributed equally to this work. The authors acknowledge the support from the National Key Research & Development Program of China (No. 2021YFB3601504), the National Natural Science Foundation of China (Nos. 52072218 and 12074149), the Primary Research & Development Plan of Shandong Province (No. 2019JZZY010313), and the Natural Science Foundation of Shandong province (No. ZR2020KE019 and ZR2020ZD28). The authors would like to thank the Analytical Center for Structural Constituent and Physical Property of Core Facilities Sharing Platform, Shandong University for XRD and PFM analysis. The idea and project was conceived by L.Z., W.‐M.L., and L.‐M.Z.; S.‐T.Y. and X.‐Y.L. characterized the electric transport properties under the supervision of L.‐M.Z.; T.‐L.Y. carried out the artificial neural network simulation under the supervision of L.Z.; J.W., H.F., F.N., and B.H. fabricated the devices and performed the PFM measurements; S.‐T.Y., L.Z., L.‐M.Z., and A.N. wrote the manuscript and all the authors contributed to the discussion of the manuscript. ",

year = "2022",

month = aug,

day = "25",

doi = "10.1002/adfm.202202366",

language = "English",

volume = "32",

journal = "Advanced Functional Materials",

issn = "1616-301X",

publisher = "Wiley-VCH Verlag",

number = "35",

}

TY - JOUR

T1 - High-performance neuromorphic computing based on ferroelectric synapses with excellent conductance linearity and symmetry

AU - Yang, Shu-Ting

AU - Li, Xingyu

AU - Yu, Tongliang

AU - Wang, Jie

AU - Fang, Hong

AU - Nie, Fang

AU - He, Bing

AU - Zhao, Le

AU - Lu, Weimin

AU - Yan, Shishen

AU - Nogaret, Alain

AU - Liu, Gang

AU - Zheng, Limei

N1 - Funding Information: S.‐T.Y. and X.‐Y.L. contributed equally to this work. The authors acknowledge the support from the National Key Research & Development Program of China (No. 2021YFB3601504), the National Natural Science Foundation of China (Nos. 52072218 and 12074149), the Primary Research & Development Plan of Shandong Province (No. 2019JZZY010313), and the Natural Science Foundation of Shandong province (No. ZR2020KE019 and ZR2020ZD28). The authors would like to thank the Analytical Center for Structural Constituent and Physical Property of Core Facilities Sharing Platform, Shandong University for XRD and PFM analysis. The idea and project was conceived by L.Z., W.‐M.L., and L.‐M.Z.; S.‐T.Y. and X.‐Y.L. characterized the electric transport properties under the supervision of L.‐M.Z.; T.‐L.Y. carried out the artificial neural network simulation under the supervision of L.Z.; J.W., H.F., F.N., and B.H. fabricated the devices and performed the PFM measurements; S.‐T.Y., L.Z., L.‐M.Z., and A.N. wrote the manuscript and all the authors contributed to the discussion of the manuscript.

PY - 2022/8/25

Y1 - 2022/8/25

N2 - Artificial synapses can boost neuromorphic computing to overcome the inherent limitations of von Neumann architecture. As a promising memristor candidate, ferroelectric tunnel junctions (FTJ) enable the authors to successfully emulate spike-timing-dependent synapses. However, the nonlinear and asymmetric synaptic weight update under repeated presynaptic stimulation hampers neuromorphic computing by favoring the runaway of synaptic weights during learning. Here, the authors demonstrate an FTJ whose conductivity varies linearly and symmetrically by judiciously combining ferroelectric domain switching and oxygen vacancy migration. The artificial neural network based on this FTJ-synapse achieves classification accuracy of 96.7% during supervised learning, which is the closest to the maximum theoretical value of 98% achieved to date. This artificial synapse also demonstrates stable unsupervised learning in a noisy environment for its well-balanced spike-timing-dependent plasticity response. The novel concept of controlling ionic migration in ferroelectric materials paves the way toward highly reliable and reproducible supervised and unsupervised learning strategies.

AB - Artificial synapses can boost neuromorphic computing to overcome the inherent limitations of von Neumann architecture. As a promising memristor candidate, ferroelectric tunnel junctions (FTJ) enable the authors to successfully emulate spike-timing-dependent synapses. However, the nonlinear and asymmetric synaptic weight update under repeated presynaptic stimulation hampers neuromorphic computing by favoring the runaway of synaptic weights during learning. Here, the authors demonstrate an FTJ whose conductivity varies linearly and symmetrically by judiciously combining ferroelectric domain switching and oxygen vacancy migration. The artificial neural network based on this FTJ-synapse achieves classification accuracy of 96.7% during supervised learning, which is the closest to the maximum theoretical value of 98% achieved to date. This artificial synapse also demonstrates stable unsupervised learning in a noisy environment for its well-balanced spike-timing-dependent plasticity response. The novel concept of controlling ionic migration in ferroelectric materials paves the way toward highly reliable and reproducible supervised and unsupervised learning strategies.

KW - electronic synapses

KW - ferroelectric tunnel junctions

KW - linear and symmetric weight changes

KW - spike-timing-dependent plasticity

UR - http://www.scopus.com/inward/record.url?scp=85132354915&partnerID=8YFLogxK

U2 - 10.1002/adfm.202202366

DO - 10.1002/adfm.202202366

M3 - Article

SN - 1616-301X

VL - 32

JO - Advanced Functional Materials

JF - Advanced Functional Materials

IS - 35

M1 - 2202366

ER -

High-performance neuromorphic computing based on ferroelectric synapses with excellent conductance linearity and symmetry

Abstract

Bibliographical note

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this