A self-rectifying TaO y/nanoporous TaO x memristor synaptic array for learning and energy-efficient neuromorphic systems

Sanghyeon Choi, Seonghoon Jang, Jung Hwan Moon, Jong Chan Kim, Hu Young Jeong, Peonghwa Jang, Kyoung Jin Lee, Gunuk Wang

Research output: Contribution to journalArticle

3 Citations (Scopus)

Abstract

The human brain intrinsically operates with a large number of synapses, more than 1015. Therefore, one of the most critical requirements for constructing artificial neural networks (ANNs) is to achieve extremely dense synaptic array devices, for which the crossbar architecture containing an artificial synaptic node at each cross is indispensable. However, crossbar arrays suffer from the undesired leakage of signals through neighboring cells, which is a major challenge for implementing ANNs. In this work, we show that this challenge can be overcome by using Pt/TaOy/nanoporous (NP) TaOx/Ta memristor synapses because of their self-rectifying behavior, which is capable of suppressing unwanted leakage pathways. Moreover, our synaptic device exhibits high non-linearity (up to 104), low synapse coupling (S.C, up to 4.00 × 10−5), acceptable endurance (5000 cycles at 85 °C), sweeping (1000 sweeps), retention stability and acceptable cell uniformity. We also demonstrated essential synaptic functions, such as long-term potentiation (LTP), long-term depression (LTD), and spiking-timing-dependent plasticity (STDP), and simulated the recognition accuracy depending on the S.C for MNIST handwritten digit images. Based on the average S.C (1.60 × 10−4) in the fabricated crossbar array, we confirmed that our memristive synapse was able to achieve an 89.08% recognition accuracy after only 15 training epochs.

Original languageEnglish
JournalNPG Asia Materials
DOIs
Publication statusAccepted/In press - 2018 Jan 1

Fingerprint

Memristors
synapses
Synapse
Energy Efficient
learning
Neural networks
Leakage
Plasticity
Artificial Neural Network
Brain
Durability
leakage
time measurement
spiking
Sweeping
energy
digits
endurance
Cell
Sweep

ASJC Scopus subject areas

  • Modelling and Simulation
  • Materials Science(all)
  • Condensed Matter Physics

Cite this

A self-rectifying TaO y/nanoporous TaO x memristor synaptic array for learning and energy-efficient neuromorphic systems. / Choi, Sanghyeon; Jang, Seonghoon; Moon, Jung Hwan; Kim, Jong Chan; Jeong, Hu Young; Jang, Peonghwa; Lee, Kyoung Jin; Wang, Gunuk.

In: NPG Asia Materials, 01.01.2018.

Research output: Contribution to journalArticle

Choi, Sanghyeon ; Jang, Seonghoon ; Moon, Jung Hwan ; Kim, Jong Chan ; Jeong, Hu Young ; Jang, Peonghwa ; Lee, Kyoung Jin ; Wang, Gunuk. / A self-rectifying TaO y/nanoporous TaO x memristor synaptic array for learning and energy-efficient neuromorphic systems. In: NPG Asia Materials. 2018.
@article{1a1b839572714764bf81fdbd7de73eff,
title = "A self-rectifying TaO y/nanoporous TaO x memristor synaptic array for learning and energy-efficient neuromorphic systems",
abstract = "The human brain intrinsically operates with a large number of synapses, more than 1015. Therefore, one of the most critical requirements for constructing artificial neural networks (ANNs) is to achieve extremely dense synaptic array devices, for which the crossbar architecture containing an artificial synaptic node at each cross is indispensable. However, crossbar arrays suffer from the undesired leakage of signals through neighboring cells, which is a major challenge for implementing ANNs. In this work, we show that this challenge can be overcome by using Pt/TaOy/nanoporous (NP) TaOx/Ta memristor synapses because of their self-rectifying behavior, which is capable of suppressing unwanted leakage pathways. Moreover, our synaptic device exhibits high non-linearity (up to 104), low synapse coupling (S.C, up to 4.00 × 10−5), acceptable endurance (5000 cycles at 85 °C), sweeping (1000 sweeps), retention stability and acceptable cell uniformity. We also demonstrated essential synaptic functions, such as long-term potentiation (LTP), long-term depression (LTD), and spiking-timing-dependent plasticity (STDP), and simulated the recognition accuracy depending on the S.C for MNIST handwritten digit images. Based on the average S.C (1.60 × 10−4) in the fabricated crossbar array, we confirmed that our memristive synapse was able to achieve an 89.08{\%} recognition accuracy after only 15 training epochs.",
author = "Sanghyeon Choi and Seonghoon Jang and Moon, {Jung Hwan} and Kim, {Jong Chan} and Jeong, {Hu Young} and Peonghwa Jang and Lee, {Kyoung Jin} and Gunuk Wang",
year = "2018",
month = "1",
day = "1",
doi = "10.1038/s41427-018-0101-y",
language = "English",
journal = "NPG Asia Materials",
issn = "1884-4049",
publisher = "Nature Publishing Group",

}

TY - JOUR

T1 - A self-rectifying TaO y/nanoporous TaO x memristor synaptic array for learning and energy-efficient neuromorphic systems

AU - Choi, Sanghyeon

AU - Jang, Seonghoon

AU - Moon, Jung Hwan

AU - Kim, Jong Chan

AU - Jeong, Hu Young

AU - Jang, Peonghwa

AU - Lee, Kyoung Jin

AU - Wang, Gunuk

PY - 2018/1/1

Y1 - 2018/1/1

N2 - The human brain intrinsically operates with a large number of synapses, more than 1015. Therefore, one of the most critical requirements for constructing artificial neural networks (ANNs) is to achieve extremely dense synaptic array devices, for which the crossbar architecture containing an artificial synaptic node at each cross is indispensable. However, crossbar arrays suffer from the undesired leakage of signals through neighboring cells, which is a major challenge for implementing ANNs. In this work, we show that this challenge can be overcome by using Pt/TaOy/nanoporous (NP) TaOx/Ta memristor synapses because of their self-rectifying behavior, which is capable of suppressing unwanted leakage pathways. Moreover, our synaptic device exhibits high non-linearity (up to 104), low synapse coupling (S.C, up to 4.00 × 10−5), acceptable endurance (5000 cycles at 85 °C), sweeping (1000 sweeps), retention stability and acceptable cell uniformity. We also demonstrated essential synaptic functions, such as long-term potentiation (LTP), long-term depression (LTD), and spiking-timing-dependent plasticity (STDP), and simulated the recognition accuracy depending on the S.C for MNIST handwritten digit images. Based on the average S.C (1.60 × 10−4) in the fabricated crossbar array, we confirmed that our memristive synapse was able to achieve an 89.08% recognition accuracy after only 15 training epochs.

AB - The human brain intrinsically operates with a large number of synapses, more than 1015. Therefore, one of the most critical requirements for constructing artificial neural networks (ANNs) is to achieve extremely dense synaptic array devices, for which the crossbar architecture containing an artificial synaptic node at each cross is indispensable. However, crossbar arrays suffer from the undesired leakage of signals through neighboring cells, which is a major challenge for implementing ANNs. In this work, we show that this challenge can be overcome by using Pt/TaOy/nanoporous (NP) TaOx/Ta memristor synapses because of their self-rectifying behavior, which is capable of suppressing unwanted leakage pathways. Moreover, our synaptic device exhibits high non-linearity (up to 104), low synapse coupling (S.C, up to 4.00 × 10−5), acceptable endurance (5000 cycles at 85 °C), sweeping (1000 sweeps), retention stability and acceptable cell uniformity. We also demonstrated essential synaptic functions, such as long-term potentiation (LTP), long-term depression (LTD), and spiking-timing-dependent plasticity (STDP), and simulated the recognition accuracy depending on the S.C for MNIST handwritten digit images. Based on the average S.C (1.60 × 10−4) in the fabricated crossbar array, we confirmed that our memristive synapse was able to achieve an 89.08% recognition accuracy after only 15 training epochs.

UR - http://www.scopus.com/inward/record.url?scp=85058467347&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85058467347&partnerID=8YFLogxK

U2 - 10.1038/s41427-018-0101-y

DO - 10.1038/s41427-018-0101-y

M3 - Article

AN - SCOPUS:85058467347

JO - NPG Asia Materials

JF - NPG Asia Materials

SN - 1884-4049

ER -