Instant ebooks textbook Memristive Devices for Brain Inspired Computing From Materials Devices and Circuits to Applications Computational Memory Deep Learning and Series in Electronic and Optical Materials 1st Edition Sabina Spiga (Editor) download all chapters
Instant ebooks textbook Memristive Devices for Brain Inspired Computing From Materials Devices and Circuits to Applications Computational Memory Deep Learning and Series in Electronic and Optical Materials 1st Edition Sabina Spiga (Editor) download all chapters
com
https://textbookfull.com/product/electronic-devices-and-circuits-6th-
edition-bogart/
textbookfull.com
https://textbookfull.com/product/the-byzantine-
empire-1025-1204-a-political-history-second-edition-michael-angold/
textbookfull.com
https://textbookfull.com/product/diplomacy-in-a-globalizing-world-2nd-
edition-dr-pauline-kerr/
textbookfull.com
https://textbookfull.com/product/mcgraw-hill-education-social-studies-
workbook-for-the-ged-test-3rd-edition-mcgraw-hill-editors/
textbookfull.com
https://textbookfull.com/product/r-n-kao-gentleman-spymaster-1st-
edition-nitin-a-gokhale/
textbookfull.com
Handbook of Environmental Engineering 3rd Edition Myer
Kutz
https://textbookfull.com/product/handbook-of-environmental-
engineering-3rd-edition-myer-kutz/
textbookfull.com
Memristive Devices for Brain-Inspired
Computing
From Materials, Devices, and Circuits to Applications —
Computational Memory, Deep Learning, and Spiking Neural
Networks
Woodhead Publishing Series in Electronic and
Optical Materials
Edited by
Sabina Spiga
Abu Sebastian
Damien Querlioz
Bipin Rajendran
Woodhead Publishing is an imprint of Elsevier
The Officers’ Mess Business Centre, Royston Road, Duxford, CB22 4QH, United Kingdom
50 Hampshire Street, 5th Floor, Cambridge, MA 02139, United States
The Boulevard, Langford Lane, Kidlington, OX5 1GB, United Kingdom
Copyright © 2020 Elsevier Ltd. All rights reserved.
No part of this publication may be reproduced or transmitted in any form or by any means,
electronic or mechanical, including photocopying, recording, or any information storage and
retrieval system, without permission in writing from the publisher. Details on how to seek
permission, further information about the Publisher’s permissions policies and our
arrangements with organizations such as the Copyright Clearance Center and the Copyright
Licensing Agency, can be found at our website: www.elsevier.com/permissions.
This book and the individual contributions contained in it are protected under copyright
by the Publisher (other than as may be noted herein).
Notices
Knowledge and best practice in this field are constantly changing. As new research and
experience broaden our understanding, changes in research methods, professional practices, or
medical treatment may become necessary.
Practitioners and researchers must always rely on their own experience and knowledge in
evaluating and using any information, methods, compounds, or experiments described herein.
In using such information or methods they should be mindful of their own safety and the safety
of others, including parties for whom they have a professional responsibility.
To the fullest extent of the law, neither the Publisher nor the authors, contributors, or editors,
assume any liability for any injury and/or damage to persons or property as a matter of
products liability, negligence or otherwise, or from any use or operation of any methods,
products, instructions, or ideas contained in the material herein.
List of contributors xv
Preface xix
Part I
Memristive devices for brain inspired computing
1. Role of resistive memory devices in brain-inspired
computing 3
Sabina Spiga, Abu Sebastian, Damien Querlioz and Bipin Rajendran
1.1 Introduction 3
1.2 Type of resistive memory devices 4
1.3 Resistive memory devices for brain-inspired computing 8
1.4 Conclusions and perspectives 10
References 11
v
vi Contents
3. Phase-change memory 63
Manuel Le Gallo and Abu Sebastian
3.1 Introduction 63
3.1.1 Historical overview of phase-change memory 63
3.1.2 Applications of phase-change memory 64
3.2 Essentials of phase-change memory 67
3.3 A detailed description of the write operation 70
3.3.1 SET/RESET operation 70
3.3.2 Switching process 73
3.3.3 Multilevel operation 75
3.4 A detailed description of the read operation 76
3.4.1 Subthreshold electrical transport: voltage and
temperature dependence 77
3.4.2 Resistance drift 78
3.4.3 Noise 81
3.5 Key enablers for brain-inspired computing 82
3.5.1 Multilevel storage 82
3.5.2 Accumulative behavior 84
3.5.3 Inter and intradevice randomness 86
3.6 Outlook 90
References 91
Part II
Computational memory
6. Memristive devices as computational memory 167
Abu Sebastian, Damien Querlioz, Bipin Rajendran and
Sabina Spiga
6.1 Introduction 167
6.2 In-memory computing 167
6.3 Future outlook 171
References 172
Part III
Deep learning
12. Memristive devices for deep learning applications 313
Damien Querlioz, Sabina Spiga, Abu Sebastian and Bipin Rajendran
12.1 Quick introduction to deep learning 314
x Contents
Part IV
Spiking neural networks
15. Memristive devices for spiking neural networks 399
Bipin Rajendran, Damien Querlioz, Sabina Spiga and Abu Sebastian
15.1 Introduction 399
15.2 Signal encoding and processing with spikes 400
15.3 System architecture 402
15.4 Memristive devices for Spiking neural networks 402
15.5 Future outlook 403
References 404
Index 531
List of contributors
Stefano Ambrogio IBM Research Almaden, 650 Harry Road, San Jose, CA,
United States
Rotem Ben-Hur Technion Israel Institute of Technology, Haifa, Israel
Stephanie Bohaichuk Stanford University, Stanford, CA, United States
Stefano Brivio CNR—IMM, Unit of Agrate Brianza, Agrate Brianza, Italy
Geoffrey W. Burr IBM Research Almaden, 650 Harry Road, San Jose, CA,
United States
Meng-Fan Chang Electrical Engineering Department, National Tsing Hua
University, Hsinchu, Taiwan, ROC
Solomon Amsalu Chekol Center for Single Atom-based Semiconductor Device and
Department of Materials Science and Engineering, Pohang University of Science
and Technology (POSTECH), Pohang-si, South Korea
Thomas Dalgaty University of Grenoble Alpes, CEA, LETI, Grenoble, France
Chunmeng Dou Electrical Engineering Department, National Tsing Hua University,
Hsinchu, Taiwan, ROC
Ahmed Eltawil Department of Electrical Engineering and Computer Science,
University of California—Irvine, Irvine, CA, United States
Mohammed E. Fouda Department of Electrical Engineering and Computer Science,
University of California—Irvine, Irvine, CA, United States
Bin Gao Institute of Microelectronics, Tsinghua University, Beijing, P.R. China
Julie Grollier Unité Mixte de Physique, CNRS, Thales, Université Paris-Saclay,
91767 Palaiseau, France
Ameer Haj-Ali Technion Israel Institute of Technology, Haifa, Israel
Liza Herrera Diez Center for Nanosciences and Nanotechnology, CNRS, Université
Paris-Saclay, Palaiseau, France
Hyunsang Hwang Center for Single Atom-based Semiconductor Device and
Department of Materials Science and Engineering, Pohang University of Science
and Technology (POSTECH), Pohang-si, South Korea
Daniele Ielmini Department of Electronics, Information and Bioengineering,
Polytechnic University of Milan and IU.NET, Milan, Italy
Giacomo Indiveri Institute of Neuroinformatics, University of Zurich and ETH
Zurich, Zürich, Switzerland
xv
xvi List of contributors
Suhas Kumar Hewlett Packard Labs, Palo Alto, CA, United States
Fadi Kurdahi Department of Electrical Engineering and Computer Science,
University of California—Irvine, Irvine, CA, United States
Shahar Kvatinsky Technion Israel Institute of Technology, Haifa, Israel
Raphaël Laurent HawAI.tech S.A.S., Grenoble, France
Manuel Le Gallo IBM Research Zurich, Rüschlikon, Switzerland
Haitong Li Department of Electrical Engineering, Stanford University, Stanford,
CA, United States
Seokjae Lim Center for Single Atom-based Semiconductor Device and Department
of Materials Science and Engineering, Pohang University of Science and
Technology (POSTECH), Pohang-si, South Korea
Bernabé Linares-Barranco Instituto de Microelectrónica de Sevilla IMSE-CNM,
CSIC and Universidad de Sevilla, Sevilla, Spain
Nicolas Locatelli Center for Nanosciences and Nanotechnology, CNRS, Université
Paris-Saclay, Palaiseau, France
Wei D. Lu Department of Electrical Engineering and Computer Science, University
of Michigan, Ann Arbor, MI, United States
Charles Mackin IBM Research Almaden, 650 Harry Road, San Jose, CA,
United States
Stephan Menzel Peter-Grünberg-Institut 7, Forschungszentrum Jülich GmbH,
Juelich, Germany
Rivu Midya Department of Electrical and Computer Engineering, University of
Massachusetts, Amherst, MA, United States
Thomas Mikolajick NaMLab gGmbH, Dresden, Germany; Institute for
Semiconductors and Microsystems, TU Dresden, Dresden, Germany
Valerio Milo Department of Electronics, Information and Bioengineering,
Polytechnic University of Milan and IU.NET, Milan, Italy
Subhasish Mitra Department of Electrical Engineering, Stanford University,
Stanford, CA, United States
Alice Mizrahi Unité Mixte de Physique, CNRS, Thales, Université Paris-Saclay,
91767 Palaiseau, France
Pritish Narayanan IBM Research Almaden, 650 Harry Road, San Jose, CA,
United States
Emre Neftci Department of Cognitive Sciences and Department of Computer
Science, University of California—Irvine, Irvine, CA, United States
Jaehyuk Park Center for Single Atom-based Semiconductor Device and Department
of Materials Science and Engineering, Pohang University of Science and
Technology (POSTECH), Pohang-si, South Korea
Melika Payvand Institute of Neuroinformatics, University of Zurich and ETH
Zurich, Zürich, Switzerland
List of contributors xvii
xix
xx Preface
Despite our ability to train deep neural networks with brute-force optimi-
zation, the computational principles of neural networks remain poorly under-
stood. Hence significant research is aimed at unraveling the principles of
computation or information processing in large biological neuronal networks.
It is widely believed that because of the added temporal dimension, spiking
neural networks (SNNs) are computationally more powerful. The fourth part
of the book (Chapters 15 19) provides an overview of how memristive
devices could efficiently implement these novel spike-based algorithms.
This book targets a broad and interdisciplinary audience working in the
field of brain-inspired computing from materials scientists, physicists, and
electrical engineers to computer scientists. We hope it will be a valuable and
timely resource to researchers from both academia and industry and at vari-
ous levels of expertise, including masters and PhD students.
We would like to express our sincere appreciation to the authors who
provided excellent contributions and have been willing to go through multi-
ple iterations in order to improve the quality and consistency within the
book. We would also like to thank the colleagues at our respective institu-
tions, CNR-IMM-Agrate Brianza, IBM Research—Zurich, Université Paris-
Saclay/CNRS, and King’s College London, for all the support during the
preparation of this book. In particular we would like to thank Anne-Marie
Cromack from IBM Research—Zurich for editorial assistance.
1.1 Introduction
This chapter introduces various types of resistive memory devices (also
named memristive devices) of current interest for brain-inspired computing.
These memristive device technologies include a broad class of two- or three-
terminal devices whose resistance can be modified upon electrical stimuli.
The resistance changes can last for short- or long-time scales, leading to a
volatile or nonvolatile memory effect, respectively. Memristive devices are
based on a large variety of physical mechanisms, such as redox reactions and
ion migration, phase transitions, spin-polarized tunneling, and ferroelectric
polarization. The switching geometry can involve a volume, interfacial, or
confined 1D filamentary regions [1 8].
Although these technologies have been mainly developed as nonvolatile
memory devices for storage applications, recently, they have been receiving
increasing interest for brain-inspired computing, and many exciting develop-
ments are underway in this direction [1,9 23]. Today we are facing a revo-
lution driven by the increasing amount of data generated each day, which
need to be stored, classified, and processed, leading to the paradigm of data-
centric-computing. On the other hand current computing systems are inher-
ently limited in energy efficiency and data bandwidth by the physically sepa-
rated memory and processing units (von Neumann bottleneck), as well as by
the latency mismatch between the memory and processing units (memory
wall) [9,10,13]. Memristive devices have the potential to meet the consider-
able demand for new devices that enable energy-efficient and area-efficient
FIGURE 1.1 (A) Sketch of resistive random access memory (RRAM) devices featuring fila-
mentary (top) and interfacial switching (bottom). The corresponding representative current vol-
tage characteristics are reported on the right, indicating bipolar switching from a high- to low-
resistance state (set) and vice versa (reset). For filamentary systems an initial high-voltage form-
ing is required. (B, Top) Schematic drawing of phase-change memory (PCM) cell in the crystal-
line (LRS) and amorphous state (HRS) of the chalcogenide material. (Bottom) Current voltage
and resistance-programming power characteristics. (C) Ferroelectric random access memory
(FeRAM) and its polarization electric field hysteresis (left), and ferroelectric field effect transis-
tor (FeFET), with corresponding drain current versus gate voltage characteristics (right). (D)
Sketch of a magnetic tunnel junction (MTJ) (top) and evolution of the resistance versus applied
voltage (bottom) for the parallel and antiparallel configurations of the magnetization orientation
of the pinned and free layers. The MTJ is the key element for a magnetic random access mem-
ory (MRAM).
Role of resistive memory devices in brain-inspired computing Chapter | 1 5
That being said, it is worth noting that these characteristics are subject to
continuous improvement with the potential for novel outcomes thanks to
intensive research efforts worldwide. Now we introduce the novel applica-
tions beyond storage envisaged for the above-described devices.
Im Werk ist ein alter Bergmann, der alte Pönisch (Abb. 19), der
einen besonderen sechsten Sinn hat, mit dem er einen guten
Anbruch »an Silber und an Bleien reich« anscheinend zu wittern
vermag. Er riecht das Erz! In der Unterwelt ist er zu Hause. In den
Tiefen von 600 Meter und mehr unter Tag ist sein nächtliches Reich,
wo er beim schwachen Lichtlein seiner Blende die Erzadern der
Berge spürt und die Pulse der Tiefe fühlt, in denen das Leben der
Grube quillt. Mit seinem faltenreichen grauen Pergamentgesicht
voller Runen, den listigen Augen mit »Silberblick« und der
gedrungenen Gestalt mit hängenden Schultern auf den kurzen
Kurven der kräftigen Beine, gleicht er in seiner verwitterten
Bergmannstracht, mit der Blende am Strick, einem der schlauen
Zwerge der Unterwelt, welche die Schätze hüten, die Menschen bald
narren und necken, bald ihnen zu Reichtum verhelfen. –
Starker Sturm, der in der Nacht vom 9. zum 10. Februar dieses
Jahres wütete, hat einem altehrwürdigen Naturdenkmale der
Rochlitzer Gegend, der fünfhundertjährigen Streitlinde, argen
Schaden zugefügt. Einer der breit ausladenden, weit über zehn
Meter langen Äste, wurde von dem Hauptstamme losgetrennt und
liegt nun, wie unsere Abbildungen zeigen, auf dem Boden. Nur an
einer Stelle besteht noch etwas Zusammenhang zwischen Stamm
und Ast.
Die Nachricht von dieser Zerstörung fand in der Bevölkerung
allseitiges Interesse, und so lenkten viele ihre Schritte nach
Königsfeld, um den beschädigten Baum aufzusuchen. Mancher hat
auf diese Weise das erstemal von der Streitlinde gehört und sie
gesehen, obwohl sie einen Hauptbestandteil des Landschaftsbildes
ausmacht.
Wie und wann der Baum zu seinem Namen gekommen ist, läßt
sich nicht nachweisen. Da die Linde im Grenzgebiet zwischen
Königsfeld und Köttwitzsch steht, hat sie wahrscheinlich ihren
Namen von einem alten Grenzstreit. Der Sage nach sollen sich dort
zwei adlige Brüder bekämpft haben.
Willst du, geneigter Leser, den Baum aufsuchen, so wirst du ihn,
sobald die Staatsstraße Rochlitz–Geithain die Königsfelder Höhe
erreicht hat, links von dem Vorwerk »Heide« erblicken. Mit seiner
mächtigen Krone macht er aus der Ferne einen stattlichen Eindruck,
obwohl er eigentlich keinen entwickelten Baum darstellt. Der Stamm
ist höchstens zwei Meter hoch. Offenbar ist die Linde in ihrer Jugend
geköpft worden und hat nun die eigentümliche, wagerecht
ausladende Astbildung aufzuweisen. Die Hauptäste zeigen zum Teil
den Umfang von kräftigen Baumstämmen. Den Aufnahmen sieht es
wohl niemand an, daß der untere Teil des Stammes einen Umfang
von sieben Meter und siebzig Zentimeter hat. Es scheint so, als ob in
früheren Zeiten auf den Ästen ein gelegentlicher Aufenthaltsort für
Menschen gewesen sei. Wenn man den Erzählungen eines
Landwirts aus dem nahen Köttwitzsch Glauben schenken darf, so
soll unter der Streitlinde einst der Rittergutsherr seinen Frönern ein
Fest gegeben haben als Dank dafür, daß sie ihm die Ernte gut
haben einbringen helfen, ehe langandauerndes Regenwetter
einsetzte.