INTRODUCTION TO SUPERCONDUCTIVITY Second Edition Michael Tinkham Rumford Professor of Physics and Gordon McKay Professor of Applied Physics Harvard University McGraw-Hill, Inc. New York St. Louis San Francisco Auckland Bogoté Caracas Lisbon London Madrid Mexico City Milan Montreal New Delhi San Juan Singapore Sydney Tokyo TorontoThis book was set in Times Roman by Keyword Publi The editors were Jack Shira and Eleanor Castellano; the production supervisor was Elizabeth J. Strange. The cover was designed by Amy Becker. R. R. Donnelley & Sons Company was printer and binder. INTRODUCTION TO SUPERCONDUCTIVITY Copyright ©1996 by McGraw-Hill, Inc. All rights reserved. Printed in the United States of America, Except as permitted under the United States Copyright Act of 1976, no part of this publication may be reproduced or distributed in any form or by any means, or stored in a data base or retrieval system, without the prior written permission of the publisher. This book is printed on acid-free paper. 1234567890 DOC DOC 9098765 ISBN 0-07-064878-6 Library of Congress Cataloging-in-Publication Data Tinkham, Michael Introduction to superconductivity / Michael Tinkham. — 2d ed. p. cm, — (International series in pure and applied physics) Includes index. ISBN 0-07-064878-6 1. Superconductivity. I. Title. II. Series. QC611.92.T56 1996 537.6'23—de20 95-22378ABOUT THE AUTHOR A native of Wisconsin, Michael Tinkham received an A. B. degree from Ripon College and his M. S. and Ph.D. from MIT. After a postdoctoral year at the Clarendon Laboratory in Oxford, he spent 11 years teaching at the University of California, Berkeley, before moving to Harvard in 1966, where he is now the Rumford Professor of Physics and Gordon McKay Professor of Applied Physics. Over the years, he has spent sabbatical leaves at MIT and at the University of Paris, Orsay, as a Guggenheim Fellow; at the Cavendish Laboratory in Cambridge University as an NSF Senior Postdoctoral Fellow; at the Institute for Theory of Condensed Matter in Karlsruhe, Germany, as a Humboldt Prize Fellow; at the University of California in Berkeley as a Visiting Miller Professor; and as a Visiting Professor at the Technical University of Delft, the Netherlands. He is a Member of the National Academy of Sciences and a Fellow of the American Academy of Arts and Sciences, of the Amercian Physical Society, and of the American Association for the Advancement of Science. Honors from the American Physical Society include the Richtmyer lectureship and the Buckley Solid State Physics Prize for his research on the electromagnetic properties of superconductors. In 1976 he was awarded an honorary Sc.D. from Ripon College. He has also served on the US National Committee of IUPAP and as chairman of the Fritz London Award Committee. Author of over 200 research publications, he has written three previous books: Group Theory and Quantum Mechanics, Superconductivity, and the first edition of Introduction to Superconductivity, which has been translated into Russian, Japanese, and Chinese.CONTENTS Preface xili Suggestions for Using This Book xvii Preface to the First Edition xix Historical Overview 1 1.1 The Basic Phenomena 2 1.2. The London Equations 4 1.3. The Pippard Nontocal Electrodynamies 6 1.4 The Energy Gap and the BCS Theory 8 1.5. The Ginzburg-Landau Theory 9 1.6. Type II Superconductors i 1.7. Phase, Josephson Tunneling, and Fluxoid Quantization 1B 1.8 Fluctuations and Nonequilibrium Effects 15 1.9 High-Temperature Superconductivity 16 Introduction to Electrodynamics of Superconductors 17 2.1 The London Equations 18 2.2. Screening of a Static Magnetic Field 19 2.2.1 Flat Slab in Parallel Magnetic Field 20 2.2.2 Critical Current of Wire 21 2.3. Type I Superconductors in Strong Magnetic Fields: The Intermediate State 22 2.3.1 Nonzero Demagnetizing Factor 24 2.3.2 Intermediate State in a Flat Slab 25 2.3.3 Intermediate State of a Sphere 31 Intermediate State above Critical Current of a Superconducting Wire 32 High-Frequency Electrodynamics 37 2.5.1 Complex Conductivity in Two-Fluid Approximation 37 2.5.2. High-Frequency Dissipation in Superconductors 39viii 3 CONTENTS The BCS Theory 3.1 Cooper Pairs 3.2. Origin of the Attractive Interaction 3.3. The BCS Ground State 3.4 Variational Method 3.4.1 Determination of the Coefficients 3.4.2 Evaluation of Ground-State Energy 3.4.3 Isotope Effect 3.5 Solution by Canonical Transformation 3.5.1 Excitation Energies and the Energy Gap 3.6 Finite Temperatures 3.6.1 Determination of 7, 3.6.2. Temperature Dependence of the Gap 3.6.3 Thermodynamic Quantities 3.7 State Functions and the Density of States 3.7.1 Density of States 3.8 Electron Tunneling 3.8.1 The Semiconductor Model 3.8.2. Normal-Normal Tunneling 3.8.3. Normal-Superconductor Tunneling 3.8.4 Superconductor-Superconductor Tunneling 3.8.5. Phonon Structure 3.9 Transition Probabilities and Coherence Effects 3.9.1 Ultrasonic Attenuation 3.9.2. Nuclear Relaxation 3.9.3 Electromagnetic Absorption 3.10 Electrodynamics 3.10.1 Calculation of K(O, T) or Az(T) 3.10.2 Calculation of K(q, 0) 3.10.3 Nonlocal Electrodynamics in Coordinate Space 3.10.4 Effect of Impurities 3.10.5 Complex Conductivity 3.11 The Penetration Depth 3.11.1 Preliminary Estimate of \ for Nonlocal Case 3.11.2 Solution by Fourier Analysis 3.11.3 Temperature Dependence of A 3.11.4 Penetration Depth in Thin Films: Ae and A, 3.11.5 Measurement of 3.12. Concluding Summary Ginzburg-Landau Theory 4.1. The Ginzburg-Landau Free Energy 4.2 The Ginzburg-Landau Differential Equations 4.2.1 The Ginzburg-Landau Coherence Length 4.3. Calculations of the Domain-Wall Energy Parameter 4.4 Critical Current of a Thin Wire or Film 4.5 Fluxoid Quantization and the Little-Parks Experiment 4.5.1 The Fluxoid 4.5.2. The Little-Parks Experiment 118 120 123 127 127 128contents ix 4.6 Parallel Critical Field of Thin Flims 130 4.6.1. Thicker Films 131 4.7 The Linearized GL Equation 132 4.8 Nucleation in Bulk Samples: H.2 134 4.9 Nucleation at Surfaces: H.3 135 4.10 Nucleation in Films and Foils 139 4.10.1 Angular Dependence of the Critical Field of Thin Films 139 4.10.2. Nucleation in Films of Intermediate Thickness 141 4.11 The Abrikosov Vortex State at H.2 143 Magnetic Properties of Classic Type II Superconductors 148 5.1 Behavior Near H,;: The Structure of an Isolated Vortex 149 5.1.1 The High-x Approximation 151 $1.2. Vortex-Line Energy 153 5.2 Interaction between Vortex Lines: 154 5.3. Magnetization Curves 155 5.3.1 Low Flux Density 156 5.3.2. Intermediate Flux Densities 157 5.3.3. Regime Near Ho 160 5.4 Flux Pinning, Creep, and Flow 162 5.5. Flux Flow 166 5.5.1 The Bardeen-Stephen Model 167 5.5.2 Onset of Resistance in a Wire 171 5.5.3 Experimental Verification of Flux Flow 173 5.5.4 Concluding Remarks on Flux Flow 175 5.6 The Critical-State Model 176 5.7. Thermally Activated Flux Creep 179 3.7.1 Anderson-Kim Flux-Creep Theory 180 5.7.2. Thermal Instability 186 5.8 Superconducting Magnets for Time-Varying Fields 187 5.8.1 Flux Jumps 188 5.8.2 Twisted Composite Conductors 190 Josephson Effect I: Basic Phenomena and Applications 196 6.1 Introduction 196 6.2. The Josephson Critical Current 198 6.2.1 Short One-Dimensional Metallic Weak Links 198 6.2.2. Other Weak Links 200 6.2.3. Gauge-Invariant Phase 202 6.3 The RCSJ Model 202 6.3.1 Definition of the Model 202 6.3.2. I-V Characteristics at T=0 205 6.3.3 Effects of Thermal Fluctuations 207 6.3.4 rf-Driven Junctions 211 6.4 Josephson Effect in Presence of Magnetic Flux 213 6.4.1 The Basic Principle of Quantum Interference 213 6.4.2. Extended Junctions 215 6.4.3 Time-Dependent Solutions 221X CONTENTS. 6.5 6.6 6.7 SQUID Devices 6.5.1 The de SQUID 6.5.2. The rf SQUID 6.5.3 SQUID Applications Arrays of Josephson Junctions 6.6.1 Arrays in Zero Magnetic Field 6.6.2 Arrays in Uniform Magnetic Field 6.6.3. Arrays in rf Fields: Giant Shapiro Steps S-I-S Detectors and Mixers 6.7.1 S-I-S Detectors 6.7.2 S-I-S Mixers 224 225 229 232 234 236 239 242 243 244 246 7 Josephson Effect I: Phenomena Unique to Small Junctions 248 2 72 73 14 7S 7.6 Introduction Damping Effect of Lead Impedance 7.2.1 Effect on Retrapping Current 7.2.2. The Phase Diffusion Branch Quantum Consequences of Small Capacitance 7.3.1 Particle Number Eigenstates 7.3.2. Macroscopic Quantum Tunneling Introducton to Single Electron Tunneling: The Coulomb Blockade and Staircase Energy and Charging Relations in Quasi-Equilibrium 7.5.1 Zero Bias Circuit with Normal Island 7.5.2. Even-Odd Number Parity Effect with Superconducting Island 7.5.3. Zero Bias Supercurrents with Superconducting Island and Leads Double-Junction Circuit with Finite Bias Voltage 7.6.1 Orthodox Theory and Determination of the /-V Curve 7.6.2. The Special Case Rp > Ry 7.6.3 Cotunneling or Macroscopic Quantum Tunneling of Charge 7.6.4 Superconducting Island with Finite Bias Voltage 8 Fluctuation Effects in Classic Superconductors 8.1 82 Appearance of Resistance in a Thin Superconducting Wire Appearance of Resistance in a Thin Superconducting Film: The Kosterlitz-Thouless Transition Superconductivity above T, in Zero-Dimensional Systems Spatial Variation of Fluctuations Fluctuation Diamagnetism above T,. 8.5.1 Diamagnetism in Two-Dimensional Systems Time Dependence of Fluctuations Fluctuation-Enhanced Conductivity above T, 8.7.1 Three Dimensions 8.7.2. Two Dimensions 8.7.3. One Dimension 8.7.4 Anomalous Contributions to Fluctuation Conductivity 8.7.5 High-Frequency Conductivity 248 249 250 252 256 258 259 264 266 267 269 274 278 280 281 284 284 287 288 294 296 298 302 307 308 309 311 311 312 313 314CONTENTS Xi 316 9 The High-Temperature Superconductors 10 il o1 9.2 9.3 94 95 9.6 97 98 99 Introduction The Lawrence-Doniach Model 9.2.1 The Anisotropic Ginzburg-Landau Limit 9.2.2 Crossover to Two-Dimensional Behavior 9.2.3. Discussion Magnetization of Layered Superconductors 9.3.1 The Anisotropic Ginzburg-Landau Regime 9.3.2. The Lock-In Transition Flux Motion and the Resistive Transition: An Initial Overview ‘The Melting Transition 9.5.1 A Simple Model Calculation 9.5.2. Experimental Evidence 9.5.3 Two-Dimensional vs. Three-Dimensional Melting The Effect of Pinning 9.6.1 Pinning Mechanisms in HTSC 9.6.2. Larkin-Ovchinnikov Theory of Collective Pinning 9.6.3. Giant Flux Creep in the Collective Pinning Model 9.6.4 The Vortex-Glass Model 9.6.5 Correlated Disorder and the Boson Glass Model Granular High-Temperature Superconductors 9.7.1 Effective Medium Parameters 9.7.2 Relationship between Granular and Continuum Models 9.7.3. The “Brick-Wall” Model Fluxons and High-Frequency Losses Anomalous Properties of High-Temperature and Exotic Superconductors 9.9.1 Unconventional Pairing 9.9.2 Pairing Symmetry and Flux Quantization 9.9.3. The Energy Gap 9.9.4 Heavy Fermion Superconductors Special Topics 10.1 10.2 10.3 The Bogoliubov Method: Generalized Self-Consistent Field 10.1.1 Dirty Superconductors 10.1.2. Uniform Current in Pure Superconductors 10.1.3. Excitations in Vortex Magnetic Perturbations and Gapless Superconductivity 10.2.1 Depression of T. by Magnetic Perturbations 10.2.2 Density of States Time-Dependent Ginzburg-Landau Theory 10.3.1 Electron-Phonon Relaxation Nonequilibrium Superconductivity 1d 1.2 Introduction Quasi-Particle Disequilibrium 11.2.1 Energy-Mode vs. Charge-Mode Disequilibrium 11. Relaxation Times 316 384 384 386 387 388 390 391 394 399 401 403 403 404 405 407xii CONTENTS 11.3 114 11.5 11.6 Energy-Mode Disequilibrium: Steady-State Enhancement of Superconductivity 11.3.1 Enhancement by Microwaves 11.3.2 Enhancement by Extraction of Quasi-Particles Energy-Mode Disequilibrium: Dynamic Nonequilibrium Effects 11.4.1 GL Equation for Time-Dependent Gap 11.4.2. Transient Superconductivity above J, 11.4.3. Dynamic Enhancement in Metallic Weak Links Charge-Mode Disequilibrium: Steady-State Regimes 11.5.1 Andreev Reflection 11.5.2. Subharmonic Energy Gap Structure Time-Dependent Charge-Mode Disequilibrium: Phase-Slip Centers Appendix 1: Units Appendix 2: Notation and Conventions Appendix 3: Exact Solution for Penetration Depth by Fourier Analysis Bibliography Index 408 410 412 412 4l4 AIT 421 423 425 427 433 435 437 442 445PREFACE The first edition was written after the surge of activity in the 1950s and 1960s, in which the Bardeen-Cooper-Schrieffer (BCS) and Ginzburg-Landau (GL) theories were assimilated and used to develop what was referred to in the first paragraph of that edition as a “remarkably complete and satisfactory picture” of superconduct- ing phenomena. It was written with the aim of presenting that picture to its readers, and it has performed that function well during the subsequent 20 years. Very little in the first edition has proved to be incorrect or misleading. Why, then, make the very substantial effort to prepare a second edition? The major impetus has, of course, come from the discovery of high-tempera- ture superconductivity, and the immense body of activity which this has spawned. This phenomenon is not only of great interest for its own sake, but it has also provided us with a broader perspective for understanding the properties of the classic superconductors as well. Another impetus was my desire to include an introductory survey of the many developments in classic superconductivity over the past two decades. Inevitably, I have given some extra emphasis to areas in which my own group has been active, such as Josephson junctions, single-electron tunneling effects, and nonequilibrium superconductivity, but these are also topics of broad interest. Finally, I wanted to take advantage of the 20 years of teaching experience with the first edition and to try to produce a new edition which is somewhat more “user friendly” for the beginner in the field. Rather than reviewing the whole book, let me simply highlight the changes from the first edition. First, before attacking the intimidating BCS theory, I have introduced a new Chap. 2, which, drawing on the latter part of the old Chap. 3, treats example applications of the elementary London theory of the electrody- namics of superconductors, including diamagnetic screening, the intermediate state, and high-frequency absorption. For many practical purposes, this is the most important level of understanding, and it seems counterproductive to delay discussing these examples until after BCS, as was done in the first edition. Second, although the old BCS chapter has been renumbered as Chap. 3, it is largely unchanged except for the addition of a new Sec. 3.11 on the penetration xiiiXiV PREFACE depth. This new section collects the essential material from the first part of the old Chap. 3 on the implications of the BCS nonlocal electrodynamics for the deter- mination of the effective /ocal penetration depth used throughout the rest of the book. At the same time, the less essential mathematical details of this discussion are either relegated to the appendix or omitted entirely, because the conceptual importance of nonlocality is now well established; but nonlocality is not relevant to the high-temperature superconductors or the dirty materials of technical impor- tance for magnet wire. Instead, there is some new discussion of how ) is measured experimentally and other new material introducing the perpendicular penetration depth A, of thin films. The first really major change has been to replace the old Chap. 6 on the Josephson effect by two entirely new chapters. In this expanded and modernized treatment, Chap. 6 covers the more basic principles and applications, making use of the popular “‘tilted-washboard” picture for discussing the RCSJ model, includ- ing /-V characteristics of both under- and overdamped junctions. It also discusses Shapiro steps, photon-assisted tunneling, quantum interference, and sine-Gordon solitons in long junctions. This is followed by a modernized discussion of applica- tions of Josephson junctions to SQUIDs, an entirely new discussion of arrays of Josephson junctions, and a brief discussion of superconducting tunnel junctions as high-frequency detectors. The new Chap. 7 deals with the many special properties of the small, low- capacitance junctions made possible by modern nanofabrication techniques. In such junctions, the charging energy of single electrons is important and may dominate the Josephson coupling energy, so that the particle number becomes a better quantum number than the phase y. This chapter deals with such topics as the Coulomb blockade, the single-electron tunneling transistor, and macroscopic quantum tunneling, as well as the importance of the damping of high-resistance junctions by low-impedance leads. Although these topics are currently the focus of intense research activity and can considerably deepen one’s understanding of the phase-number uncertainty relation in superconductors, this chapter can be skimmed over in a first course on superconductivity. The second major change is the introduction of an entirely new Chap. 9, dealing primarily with the high-temperature superconductors. Because the funda- mental mechanism responsible for the high 7, remains to be identified with cer- tainty, we sidestep this question and emphasize instead the many properties of these materials which can be understood in the framework of the classic Lawrence-Doniach model of layered superconductors. These include the magnetic anisotropy and the implications of the flux-line-lattice melting transition for the resistive transition. Although its applicability is not restricted to the high- temperature superconductors, we next review the Larkin-Ovchinnikov model of collective pinning, including a short discussion of flux creep in this model and also in the vortex-glass and Boson glass models. The chapter concludes with a discussion of anomalous properties of the high-temperature supercon- ductors which cannot be understood in terms of standard s-wave BCS super- conductivity but, instead, suggest d-wave pairing.PREFACE XV The third major change is the addition of the new Chap. 11, dealing with nonequilibrium superconductivity, using a simplified version of the Schmid- Schén formalism to discuss the many implications of quasi-particle disequilibrium in driven superconductors. This chapter includes discussions of the enhancement of superconductivity by microwave radiation, by quasi-particle tunneling, and by dynamic nonequilibrium effects associated with a time-dependent energy gap. Also discussed here are phase-slip centers and the interconversion of normal and supercurrent by Andreev reflection at NS interfaces. The latter is a relatively old subject which, applied to more general configurations, has recently enjoyed a resurgence of interest. The other five chapters have been lefi largely unchanged, not because they could not be improved but, rather, because doing so would have unduly, perhaps indefinitely, delayed completion of this new edition. Nontheless, some changes were made. A number of new references were added to reflect the progress made in the intervening years. Also, brief discussions of a number of new topics were inserted, such as the Kosterlitz-Thouless resistive transition in two-dimensional superconductors. While the expansion of the text which was required to include new develop- ments will make it more useful as a reference for researchers, it also makes the book too long to be covered completely in a one-semester course. On the other hand, the instructors in such a course can take advantage of this plethora of material to pick topics to their own taste, leaving other topics to be pursued in individual study, perhaps leading to the preparation of a term paper. McGraw-Hill and | would like to thank the following reviewers for their many helpful comments and suggestions in the early stages of the development of the second edition: Alex de Lozanne, Philip Duxbury, Richard S. Newrock, John Ruvalds, Mark Rzchowski, and Dale Van Harlingen. Finally, | am pleased to acknowledge the assistance of many other colleagues in encouraging my efforts, in helping to guide the focus of the revisions, and in providing generous assistance in improving the quality of the presentation. | am uniquely indebted to Rick Newrock for his careful and speedy reading of chapter after chapter and for his extensive detailed criticism of the manuscript as it neared final form, which spurred me on to make many improvements and clarifications. It is, of course, impossible to acknowledge all those who have helped in many ways over the many years (and I apologize in advance to all those inadvertently omitted), but I should at least mention (in alphabetical order) Ryogo Aoki, Mac Beasley, Chuck Black, Greg Blonder, John Clarke, Dick Ferrell, Michael Flatté, Pierre-Gilles de Gennes, Rolf Glover, Ashraf Hanna, Jack Hergenrother, Marco Tansiti, Mark Itzler, Charlie Johnson, Teun Klapwijk, Kostya Likharev, Chris Lobb, Hans Mooij, David Nelson, Miguel Octavio, Dan Prober, Dan Ralph, Mark Rzchowski, Albert Schmid, Gerd Schén, Bill Skocpol, Mark Tuominen, and Valerii Vinokur. Of course, I cheerfully accept full responsiblity for any errors or misunderstandings which may appear despite their assistance! Michael TinkhamSUGGESTIONS FOR USING THIS BOOK The first edition of this book was sufficiently slender so that most of its contents could be covered in a briskly paced one-semester course. This is no longer the case after the expansion required to bring the second edition up to date. The best choice of topics to cover will depend on the length of the course, the level of students, and the interest of the instructor. The following suggestions are offered as guidance in structuring a one-semester introductory course at the beginning graduate or advanced undergraduate level based on the material in this book. The historical overview in Chap. | should be read primarily to provide a bird’s-eye view of the subject for orientation. The discussion of electrodynamics in Sec. 1.3 can be used in an introductory lecture, with the rest of the material essentially deferred until treated carefully in later chapters. Chapter 2 presents a systematic treatment of the electrodynamics of classic superconductors, the prop- erty that gives the subject much of its interest and importance, at the simple, but very useful, level of the London equations. It should be covered carefully, except that Sec. 2.4 can be omitted to save time. Although the discussion of the BCS theory in Chap. 3 is kept as simple as possible without loss of rigor, it is still the most technically intimidating chapter. Because of its importance, it should all be examined at whatever depth seems appropriate to the class. In an elementary course, one might focus on the key results: the ground state, the energy gap, the density of states, electron tunneling, and the penetration depth, skimming lightly over the rest. Chapters 4 and 5 on the Ginzburg-Landau theory and type II superconductors are central and should be covered carefully, except for Secs. 5.7 and 5.8, which can be skimmed. Likewise, the Josephson effect, treated in Chap. 6, is fundamentally very important, but the more specialized Secs. 6.6 and 6.7 can be skimmed or omitted in a first course. The special features of very small Josephson junctions xviiXvili_ SUGGESTIONS FOR USING THIS BOOK treated in Chap. 7 are of considerable research interest at the present time, but this chapter can be omitted or skimmed in a first course, unless it is of special interest to the instructor. Similarly, much of the discussion of fluctuation effects in Chap. 8 is of somewhat specialized interest, but Secs. 8.1 and 8.2 on fluctuation-induced electrical resistance should be covered, and Secs. 8.3 through 8.5 should at least be skimmed. With the intense current interest in high-temperature superconductivity, much of Chap. 9 should be covered, at least at the qualitative level. The material from Secs. 9.6.3 through 9.8 is of more specialized interest and might be skimmed or omitted. Finally, the material in Chaps. 10 and 11 considerably enriches our under- standing of superconductivity by considering such topics as dirty superconduc- tors, gapless superconductors, and time-dependent and nonequilibrium regimes of superconductivity. Unfortunately, time limitations in a short course will probably only allow these topics to be skimmed over lightly for the general ideas, with special attention to Sec. 10.1. In summary, the more advanced and specialized material, which can be skimmed or omitted in a one-semester course, is that in Chaps. 7, 10 and II, and in some of the latter parts in Chaps. 3, 5, 6, 8, and 9.PREFACE TO THE FIRST EDITION This book has evolved from a set of lecture notes originally written for a graduate course at Harvard University during the fall term of 1969. They were subsequently rewritten during a sabbatical leave at the Cavendish laboratory in 1971-1972 and during a repeat of the course in 1973 The objective of the lectures, and of this book, is to provide an up-to-date introduction to the intriguing subject of superconductivity and some of its potential applications. The emphasis is on the rich array of phenomena and how they may be understood in the simplest possible way. Consequently, the use of thermal Green functions has been completely avoided, despite their fash- ionability and undeniable power in the hands of skilled theorists. Rather, the power of phenomenological theory in giving insight is emphasized, and micro- scopic theory is often narrowly directed to the task of computing the coefficients in phenomenological equations. It is hoped that this emphasis will make the treatment more palatable to the experimentalist, and also complement the more generous coverage of the formal theoretical aspects of the subject in most books presently available. Finally, the author was motivated by the hope that if the theoretical techniques were kept as elementary as possible, the work might have more value to undergraduates and technologists with incomplete backgrounds in theoretical physics. In a sense this book forms an updated and greatly expanded version of the Les Houches lectures of the author, written in 1961. However, so much development of the subject has occurred in the intervening years that these notes were really rewritten (twice) from start to finish. In the process, the author has drawn fre- quently on the excellent book of de Gennes, Superconductivity in Metals and Alloys, and on the two-volume treatise Superconductivity edited by Parks. There is little in the book which has not been published previously in some form, but xixXX PREFACE TO THE FIRST EDITION some topics—particularly fluctuation effects—have developed too recently to have appeared in previous books. No attempt has been made to give an exhaustive or definitive treatment. Such a treatment required the two-volume Parks treatise mentioned above. Rather, the author has chosen to introduce the reader to a selection of topics which reflect his own focus on the electrodynamic properties of superconductors, which, after all, give the subject its unique interest. The time limitation of a semester lecture course provided unrelenting discipline in limiting the number of topics and the depth of treatment. The book starts with an introductory survey which lays out the ground to be covered in the book, and gives some of the milestones in the historical develop- ment of the subject. The reader is advised to treat this as an overview only, intended to introduce concepts and language, with the detailed explanations to be developed in subsequent chapters. He definitely should not puzzle over issues which are only sketchily introduced at this point. The second chapter is devoted to “basic BCS,” the microscopic theory devel- oped by Bardeen, Cooper, and Schrieffer to explain the superconducting state. This theory is placed at the beginning because no serious discussion of super- conductivity is possible without concepts derived from the theory. Unfortunately, this chapter has by far the most forbidding formal nature of any part of the book, but this should not be allowed to discourage the reader. Little use of the math- ematical details will be made in the following chapters, and so this chapter can be skimmed for the general ideas (which are summarized in the concluding section), and referred to later if more detailed understanding of some particular point is required. With Chap. 3, we move into the phenomenological level of treatment, which characterizes the rest of the book. First, the implications of the nonlocal electro- dynamics in determining the effective penetration depth of a magnetic field into bulk and thin film superconductors are explored, the thorough discussion of the latter topic reflecting a historical interest of the author. A simplified discussion is then given of the intermediate state, in which superconducting and normal mate- rial coexist in the presence of a magnetic field. Chapter 4 develops the Ginzburg-Landau theory from the same phenomen- ological point of view used by the original authors. The theory is then applied to an extensive catalog of classic problems: domain-wall energy, critical-current density, fluxoid quantization, critical fields of films and foils, the upper critical field H.2, the Abrikosov vortex state, and the surface nucleation field H,3. The concepts treated here underlie the subjects treated in the following chapters, in addition to illustrating the power of the Ginzburg-Landau approach. In Chap. 5, the magnetic properties of type II superconductors are developed in some detail. After the equilibrium flux density has been worked out, attention is focused on the creep and flow of the flux under the influence of transport currents. In this way, insight is obtained into the considerations which limit potential applications of type II superconductors in high-field magnets. The chapter con- cludes with a discussion of the factors governing the design of superconductingPREFACE TO THE FIRST EDITION XXi magnets to cope with time-varying fields, including the use of twisted multicore composite conductors to minimize ac losses while maintaining thermal stability. Chapter 6 is devoted to the Josephson effect and macroscopic quantum phe- nomena. These subjects represent some of the purest and most fundamental aspects of superconductivity, yet also provide the basis for sensitive instruments which have revolutionized electromagnetic measurements. Both aspects are reflected in the treatment given; in particular, the detailed discussion of practical SQUID magnetometers is the first to appear in a textbook. Although for years it was thought that the effects of thermodynamic fluctua- tions were unobservably small in superconductors, the advent of the supercon- ducting dectectors just mentioned has made it possible to observe such effects both above and below T,. Chapter 7 surveys these phenomena in both electrical conductivity and diamagnetism. For example, it is shown how fluctuation effects put a limit (though an astronomical one) on the lifetime of “persistent” currents below T., and how they also give rise to “precursors” of superconductivity above T.. Because this subject has flowered since the date of the Parks treatise, this book is the first containing a thorough discussion of this interesting and informative new aspect of superconductivity. The final chapter is devoted to introductory discussions of three topics: the Bogoliubov method, gapless superconductivity, and time-dependent Ginzburg-Landau theory. These topics go beyond the elementary Ginzburg- Landau phenomenology and bring in more microscopic considerations. Yet the basic concepts and conclusions have been drawn inevitably into the dis- cussions of the topics treated earlier; morover, taken together, they lay the groundwork for work going on at the present frontiers of research. Hence, it seems fitting to close the book with a peek at these topics, where the last word is by no means in Finally, the author is pleased to thank the reviewers of the manuscript for constructive suggestions; the detailed reading of the final manuscript by Dr. Richard Harris is especially appreciated. The comments of students who have used the notes also were particularly helpful. The speedy and accurate typing of Miss Patricia McCarthy in preparing the final manuscript was an invaluable incentive to continued progress. More generally, the author wants to thank his numerous students, colleagues, and collaborators, especially in Berkeley, Orsay, Harvard, and the other Cambridge, for making his exploration of superconduc- tivity the pleasure it has been. Although it would be impossible to list them all here, I cannot close this Preface without explicitly acknowledging numerous semi- nal discussions over the years with M. R. Beasley, J. Clarke, P. G. de Gennes, R. A. Ferrell, and R. E. Glover III. If this book serves to initiate others into the fascination I have found in this subject, it will have well served its intended Purpose. Michael TinkhamINTRODUCTION TO SUPERCONDUCTIVITYCHAPTER ] HISTORICAL OVERVIEW Superconductivity was discovered in 1911 by H. Kamerlingh Onnes' in Leiden, just 3 years after he had first liquefied helium, which gave him the refrigeration technique required to reach temperatures of a few degrees Kelvin. For decades, a fundamental understanding of this phenomenon eluded the many scientists who were working in the field. Then, in the 1950s and 1960s, a remarkably complete and satisfactory theoretical picture of the classic superconductors emerged. This situation was overturned and the subject was revitalized in 1986, when a new class of high-temperature superconductors was discovered by Bednorz and Miiller.? These new superconductors seem to obey the same general phenomenology as the classic superconductors, but the basic microscopic mechanism remains an open and contentious question at the time of this writing The purpose of this book is to introduce the reader to the field of super- conductivity, which remains fascinating after more than 80 years of investigation. To retard early obsolescence, we shall emphasize the aspects which seem to be reasonably securely understood at the present time. The goal of this introductory chapter is primarily to give some historical perspective to the evolution of the subject. All detailed discussion is deferred to later chapters, where the topics are examined again in much greater depth. We start by reviewing the basic observed electrodynamic phenomena and their early 'H. Kamerlingh Onnes, Leiden Comm. 120b, 122b, 124e (1911). °G. Bednorz and K, A. Miiller, Z. Phys. B64, 189 (1986).2 INTRODUCTION TO SUPERCONDUCTIVITY phenomenological description by the Londons. We then briefly sketch the sub- sequent evolution of the concepts which are central to our present understanding. This quasi-historical review of the development of the subject is probably too terse to be fully understood on the first reading. Rather, it is intended to provide a quick overview to help orient the reader while reading subsequent chapters, in which the ideas are developed in sufficient detail to be self-contained. In fact, some readers have found this survey more useful to highlight the major points after working through the details in subsequent chapters. 1.1 THE BASIC PHENOMENA What Kamerlingh Onnes observed was that the electrical resistance of various metals such as mercury, lead, and tin disappeared completely in a small tempera- ture range at a critical temperature 7,., which is characteristic of the material. The complete disappearance of resistance is most sensitively demonstrated by experi- ments with persistent currents in superconducting rings, as shown schematically in Fig. 1.1. Once set up, such currents have been observed to flow without measurable decrease for a year, and a lower bound of some 10° years for their characteristic decay time has been established by using nuclear resonance to detect any slight decrease in the field produced by the circulating current. In fact, we shall see that under many circumstances we expect absolutely no change in field or current to occur in times less than 10!°” years! Thus, perfect conductivity is the first traditional hallmark of superconductivity. It is also the prerequisite for most potential appli- cations, such as high-current transmission lines or high-field magnets. The next hallmark to be discovered was perfect diamagnetism, found in 1933 by Meissner and Ochsenfeld.** They found that not only a magnetic field is excluded from entering a superconductor (see Fig. 1.2), as might appear to be FIGURE 1.1 Schematic diagram of persistent current experiment. 2W. Meissner and R. Ochsenfeld, Naturwissenschaften 21, 787 (1933). “Actually, the diamagnetism is perfect only for bulk samples, since the field does penetrate a finite distance , typically approximately 500 A.HISTORICAL OVERVIEW 3 Le & W = j I i i Il \ \ FIGURE 1.2 Schematic diagram of exclusion of magnetic flux from interior of massive superconductor. is the penetration depth, typically only 500 A. explained by perfect conductivity, but also that a field in an originally normal sample is expelled as it is cooled through T,. This certainly could nor be explained by perfect conductivity, which would tend to trap flux in. The existence of such a reversible Meissner effect implies that superconductivity will be destroyed by a critical magnetic field H., which is related thermodynamically to the free-energy difference between the normal and superconducting states in zero field, the so- called condensation energy of the superconducting state. More precisely, this thermodynamic critical field H, is determined by equating the energy H?/8m per unit volume, associated with holding the field out against the magnetic pressure, with the condensation energy. That is, HAT) _ go THAT) ~ LAT) (ll) where f, and f, are the Helmholtz free energies per unit volume in the respective phases in zero field. It was found empirically that H,(T)) is quite well approxi- mated by a parabolic law HAT) = H-(0)[1 ~ (1/T-)] (1.2) illustrated in Fig. 1.3. While the transition in zero field at 7, is of second order, the transition in the presence of a field is of first order since there is a discontin- uous change in the thermodynamic state of the system and an associated latent heat.4 INTRODUCTION To SUPERCONDUCTIVITY H HO) Superconducting FIGURE 1.3 0 T. T Temperature dependence of the critica! field 1.2. THE LONDON EQUATIONS These two basic electrodynamic properties, which give superconductivity its unique interest, were well described in 1935 by the brothers F. and H. London,> who proposed two equations to govern the microscopic electric and magnetic fields a = 5, (Ads) (1.3) ~c curl (AJ;) (1.4) _ fk? im where A (1.5) eC ne is a phenomenological parameter. It was expected that n,, the number density of superconducting electrons, would vary continuously from zero at T, to a limiting value of the order of n, the density of conduction electrons, at T< T,. In (1.4), we introduce our notational convention of using h to denote the value of the flux density on a microscopic scale, reserving B to denote a macroscopic average value. Although notational symmetry would suggest using e for the microscopic local value of E in the same way, to avoid constant confusion with the charge e of the electron, we shall do so only in the few cases° where it is really useful. These notational conventions are discussed further in the appendix. SE, and H. London, Proc. Roy. Soc. (London) A149, 71 (1935). The fundamental basis for our notational asymmetry in treating E and B is in the Maxwell equations curl h = 4zJ/e and curl e = —(1/c)dh/At, Superconductors in equilibrium can have nonzero J,, as described by the London equations, causing h to vary on the scale of A. But in equilibrium, or even steady state, Jh/At = 0, so that ¢ is zero, or at least constant in space, so the use of both e and E offers no advantage. The distinction is useful only in discussing time-dependent phenomena such as motion of flux-bearing vortices in type II superconductors.HISTORICAL OVERVIEW 5 The first of these equations (1.3) describes perfect conductivity since any electric field accelerates the superconducting electrons rather than simply sustain- ing their velocity against resistance as described in Ohm’s law in a normal con- ductor. The second London equation (1.4), when combined with the Maxwell equation curl h = 4nJ/c, leads to 2 h Vh=s (1.6) This implies that a magnetic field is exponentially screened from the interior of a sample with penetration depth A, i.c., the Meissner effect. Thus, the parameter A is operationally defined as a penetration depth; empirically, the temperature depen- dence of \ is found to be approximately described by MT) © MONI = (T/T)? (1.7) The implications of the London equations are illustrated much more thoroughly in Chap. 2. A simple, but unsound,“derivation” of (1.3) can be given by computing the response to a uniform electric field of a perfect normal conductor, ie., a free-electron gas with mean free path ¢ =o. In that case, d(mv)/dt = eE, and since J = nev, (1.3) follows. But this computation is not rigorous for the spatially nonuniform fields in the penetration depth, for which (1.3) and (1.4) are most useful. The fault is that the response of an electron gas to electric fields is non- local; i-e., the current at a point is determined by the electric field averaged over a region of radius ~£ about that point. Consequently, only fields that are uniform over a region of this size give a full response; in particular, the conducti becomes infinite as £— oo only for fields filling all space. Since we are dealing here with an interface between a region with field and one with no field, it is clear that even for ¢ = ov, the effective conductivity would remain finite. For the case of a high-frequency current, this corresponds to the extreme anomalous limit of the normal skin effect, in which the surface resistance remains finite even as bo. A more profound motivation for the London equations is the quantum one, emphasizing use of the vector potential A, given by F. London’ himself, Noting that the canonical momentum p is (mv + eA/c), and arguing that in the absence of an applied field we would expect the ground state to have zero net momentum (as shown in a theorem® of Bloch), we are led to the relation for the local average velocity in the presence of the field —eA Vs) = —— TF. London, Superfluids, vol. I, Wiley, New York, 1950. “This theorem is apparently unpublished, though famous. See p. 143 of the preceding reference.6 INTRODUCTION TO SUPERCONDUCTIVITY This will hold if we postulate that for some reason the wavefunction of the super- conducting electrons is “rigid” and retains its ground-state property that (p) =0. Denoting the number density of electrons participating in this rigid ground state by n,, we then have —neA A J, = ney) = = (1.8) ‘Taking the time derivative of both sides yields (1.3) and taking the curl leads to (1.4). Thus, (1.8) contains both London equations in a compact and suggestive form.’ This argument of London leaves open the actual value of 1, but a natural upper limit is provided by the total density of conduction electrons n. If this is inserted in (1.5), we obtain 12 me AL(0) = (#5) (1.9) The notation here is chosen to indicate that this is an ideal theoretical limit as T— 0. Note that 1, is expected to decrease continuously to zero as T > T., causing A(T’) to diverge at T, as described by (1.7). Careful comparisons of the rf penetration depths of samples in the normal and superconducting states have shown that the superconducting penetration depths are always larger than A,(0), even after an extrapolation of the data to T = 0. The quantitative expla- nation of this excess penetration depth required introduction of an additional concept by Pippard: the coherence length &). 1.3 THE PIPPARD NONLOCAL ELECTRODYNAMICS Pippard'® introduced the coherence length while proposing a nonlocal general- ization of the London equation (1.8). This was done in analogy to Chambers’s nonlocal generalization'' of Ohm’s law from J(r) = cE(r) to Re Ee ye“* Since (1.8) is evidently not gauge-invariant, it will only be correct for a particular gauge choice, This choice, known as the London gauge, is specified by requiring that div A = 0 (so that div J = 0), that the normal component of A over the surface be related to any supercurrent through the surface by (1.8), and that A — 0 in the interior of bulk samples. 1A. B. Pippard, Proc. Roy. Soc. (London) A246, 547 (1953). This approach of Chambers is discussed, e.g., in J. M. Ziman, Principles of the Theory of Solids, Cambridge University Press, New York (1964), p. 242.HISTORICAL OvERVIEW 7 where R = r — r’; this formula takes into account the fact that the current at a point r depends on E(r’) throughout a volume of radius ~/ about r. Pippard argued that the superconducting wavefunction should have a similar character- istic dimension & which could be estimated by an uncertainty-principle argument, as follows: Only electrons within ~kT.. of the Fermi energy can play a major role in a phenomenon which sets in at 7,, and these clectrons have a momentum range Ap = kT./vp, where vr, is the Fermi velocity. Thus, Axzh/Ap © hog/kT. leading to the definition of a characteristic length __ hor &= “TT, where a is a numerical constant of order unity, to be determined. For typical elemental superconductors such as tin and aluminum, €, > (0). If €, represents the smallest size of a wave packet that the superconducting charge carriers can form, then one would expect a weakened supercurrent response to a vector poten- tial A(r) which did not maintain its full value over a volume of radius ~£, about the point of interest. Thus, £, plays a role analogous to the mean free path ¢ in the nonlocal electrodynamics of normal metals. Of course, if the ordinary mean free path is less than €,, one might expect a further reduction in the response to an applied field. Collecting these ideas into a concrete form, Pippard proposed replacement of (1.8) by (1.10) Js 2 [RBA cy ai ~ 4n€, Ne R# where again R = r — r’ and the coherence length € in the presence of scattering was assumed to be related to that of pure material & by HE € & €& Using (1.11), Pippard found!” that he could fit the experimental data on both tin and aluminum by the choice of a single parameter a = 0.15 in (1.10). [We shall see in Chap. 3 that the microscopic theory of Bardeen, Cooper, and Schrieffer'® (BCS) confirms this form, with the numerical constant a = 0.18.) For both metals, A is considerably larger than A, (0) because A(r) decreases sharply over a distance \ <<, giving a weakened supercurrent response, and hence an increased field penetration. Moreover, the increase of A with the decreasing mean free path predicted by (1.11) and (1.12) was consistent with data on a series (1.12) "7. E, Faber and A. B. Pippard, Proc. Roy. Soc. (London) A231, 336 (1955). '3J, Bardeen, L. N. Cooper, and J. R. Schrieffer, Phys. Rev. 108, 1175 (1957).8 INTRODUCTION TO SUPERCONDUCTIVITY of tin-indium alloys with a varying mean free path. Thus, Pippard’s nonlocal electrodynamic equation (1.11) not only fitted the experimental data, but it also anticipated the form of electrodynamics found several years later from the micro- scopic theory. 1.4 THE ENERGY GAP AND THE BCS THEORY The next step in the evolution of our understanding of superconductors was the establishment of the existence of an energy gap A, of order AT,, between the ground state and the quasi-particle excitations of the system. This concept had been suggested earlier by Daunt and Mendelssohn’* to explain the observed absence of thermoelectric effects, and it had been postulated theoretically by various workers.'*'® However, the first quantitative experimental evidence arose from precise measurements of the specific heat of superconductors by Corak et al.'’ These measurements showed that the electronic specific heat well below T, was dominated by an exponential dependence so that Cos Teac PTI? (1.13) where the normal-state electronic specific heat is C,, = 77, and a and b are numerical constants. Such an exponential dependence, with b found to be ~1.5, implies a minimum excitation energy per particle of ~1.5k7;. At about the same time, measurements of electromagnetic absorption in the region of fiw ~ kT, were first carried out. Using millimeter-microwave techniques, Biondi et al.'® reached this region in aluminum, which has a low 7, ~ 1.2 K and hence a small gap, but they were not able to carry the measurements to tempera- tures much below 7,.. Working from the far-infrared side as well as from the microwave side, Glover and Tinkham'® were able to make a more complete study of thin lead films at temperatures far below T, = 7.2 K. These measure- ments and similar ones on tin films could be interpreted quite convincingly in terms of an energy gap of 3 to 4 times kT. This result was consistent with the calorimetric one if excitations always were produced in pairs, as would be expected if they obeyed Fermi statistics. The spectroscopic measurement gives '4],G. Daunt and K. Mendelssohn, Proc. Roy. Soc. (London) A185, 225 (1946). "See, e.g. V. L. Ginzburg, Fortschr. Phys. 1, 101 (1953) and references cited therein. '*J, Bardeen, “Theory of Superconductivity,” in S. Fligge (ed.), Handbuch der Physik, yol. XV, Springer Verlag, Berlin (1956), pp. 303-310. (This article showed explicitly that an energy gap would account for the Pippard nonlocal electrodynamics.) “iw. §. Corak, B. B. Goodman, C. B. Satterthwaite, and A. Wexler, Phys. Rev. 96, 1442 (1954); 102, 656 (1956) '8M_ A. Biondi, M. P, Garfunkel, and A. 0. McCoubrey, Phys. Rev. 102, 1427 (1956). "RE. Glover and M. Tinkham, Phys, Rev. 104, 844 (1956), 108, 243 (1957).HISTORICAL OVERVIEW 9 the minimum total energy £, required to create the pair of excitations; the thermal one measures the energy E,/2 per ically independent particle. At this point, Bardeen, Cooper, and Schrieffer?” (BCS) produced their epoch-making pairing theory of superconductivity, which forms the subject of Chap. 3. In the BCS theory, it was shown that even a weak attractive interaction between electrons, such as that caused in second order by the electron-phonon interaction, causes an instability of the ordinary Fermi-sea ground state of the electron gas with respect to the formation of bound pairs of electrons occupying states with equa] and opposite momentum and spin. These so-called Cooper pairs have a spatial extension of order & and, crudely speaking, comprise the super- conducting charge carriers anticipated in the phenomenological theories. One of the key predictions of this theory was that a minimum energy E, = 2A(T) should be required to break a pair, creating two quasi-particle excitations. This A(7) was predicted to increase from zero at T, to a limiting value E,(0) = 2A(0) = 3.528kT; (1.14) for T < T,. Not only did this result agree with the measured gap widths, but the BCS prediction for the shape of the absorption edge above hug = Ey wi quantitative agreement with the data of Glover and Tinkham. Th provided one of the most decisive early verifications of the microscopic theory. 1.5 THE GINZBURG-LANDAU THEORY Although a considerable body of work followed the appearance of the BCS theory, serving to substantiate its predictions for various processes such as nuclear relaxation and ultrasonic attenuation in which the energy gap and excitation spectrum play a key role, the most exciting developments of the ensuing decade came in another direction. This direction is epitomized by the Ginzburg-Landau (GL) theory of superconductivity, which concentrates entirely on the supercon- ducting electrons rather than on excitations, and was actually proposed in 1950, 7 years before BCS. Ginzburg and Landau”! introduced a complex pseudowave- function yas an order parameter within Landau’s general theory of second-order phase transitions. This y describes the superconducting electrons, and the local density of superconducting electrons (as defined in the London equations) was given by ns = |W(x)[? (1.15) . Bardeen, L. N. Cooper, and J. R. Schrieffer, Phys. Rev. 108, 1175 (1957), ?!V. L. Ginzburg and L. D. Landau, Zh. Eksperim. i Teor. Fiz. 20, 1064 (1950).10 itRopuctioN To suPERCONDUCTIVITY Then, using a variational principle and working from an assumed series expansion of the free energy in powers of w and Vz with expansion coefficients a and 3, they derived the following differential equation for y: ae (FV-S4) vt sly? ¥= -a(r yp (16) 2m Note that this is analogous to the Schrédinger equation for a free particle, but with a nonlinear term. The corresponding equation for the supercurrent hh 2 =S—W' Vy yy) -= lwPA (1.17) ~ 2m mc was also the same as the usual quantum-mechanical current expression for par- ticles of charge e* and mass m*. With this formalism they were able to treat two features which were beyond the scope of the London theory, namely: (1) non- linear effects of fields strong enough to change ny (or |2)|7) and (2) the spatial variation of n,. A major early triumph of the theory was in handling the so-called intermediate state of superconductors (discussed in Chap. 2), in which supercon- ducting and normal domains coexist in the presence of H ~ H,. The interface between two such domains is shown schematically in Fig. 1.4. When first proposed, the theory appeared rather phenomenological, and its importance was not generally appreciated, especially in the western literature. However, in 1959, Gor’kov” was able to show that the GL theory was, in fact, a limiting form of the microscopic theory of BCS (suitably generalized to deal with spatially varying situations), valid near 7, in which y is directly proportional to the gap parameter A. More physically, y can be thought of as the wavefunction of the center-of-mass motion of the Cooper pairs. The GL theory is now universally accepted as a masterstroke of physical intuition which embodies in a simple way AT) lvran, ci A(x) Superconducting Normal FIGURE 1.4 g7) Interface between superconducting and normal domains in the intermediate state. *L. P. Gor'kov, Zh. Eksperim. i Teor. Fiz. 36, 1918 (1959) [Sov. Phys.—JETP 9, 1364 (1959)]}HIsToRICAL overview IT the macroscopic quantum-mechanical nature of the superconducting state that is crucial for understanding its unique electrodynamic properties. The GL theory introduces a characteristic length, now usually called the GL coherence length, h 1) ara (1.18) which characterizes the distance over which u(r) can vary without undue energy increase. In a pure superconductor far below T,, &(T) = &, the (temperature- independent) Pippard coherence length; near T., however, &(T) diverges as (T. — Ty-'/? since @ vanishes as (T — T,). Thus, these two “coherence lengths” are related but distinct quantities. The ratio of the two characteristic lengths defines the GL parameter (1.19) Since ) also diverges as (T, — T)'/? near T,, this dimensionless ratio is approxi- mately independent of temperature. For typical classic pure superconductors, d= 500 A and € = 3,000 A, so & <1. In this case, one can show? (see Chap. 4) that there is a positive surface energy associated with a domain wall between normal and superconducting material. This positive surface energy stabilizes a domain pattern in the intermediate state, with a scale of subdivision intermediate between the microscopic length € and the macroscopic sample size. 16 TYPE If SUPERCONDUCTORS In 1957 (the same year as BCS), Abrikosov™* published a remarkably significant paper, almost overlooked at the time, in which he investigated what would hap- pen in GL theory if « were large instead of small, i.e., if €< A, rather than the reverse. Reversing the argument cited above, this should lead to a negative surface energy, so that the process of subdivision into domains would proceed until it is limited by the microscopic length €, below which the gradient energy term would become excessive. Because this behavior is so radically different from the classic intermediate-state behavior described earlier, Abrikosov called these type I super- conductors to distinguish them from the earlier ype J variety. He showed that the exact breakpoint between the two regimes was at « = 1/2. For materials with «> 1/V2, he found that instead of discontinuous breakdown of superconductiv- ity in a first-order transition at H,, there was a continuous increase in flux pene- tration starting at a lower critical field H., and reaching B = H at an upper >The physical reason is that there is an interfacial layer of thickness ~(€ — A) which pays the energetic cost of excluding the magnetic field without enjoying the full condensation energy of the superconduct- ing state. “A. A. Abrikosov, Zh. Eksperim. i Teor. 32, 1442 (1957) [Sov. Phy's—JETP 5, 1174 (1957)]12 iNtRODUCTION To SUPERCONDUCTIVITY 0 FIGURE 1.5 ‘Comparison of flux penetration behavior of type I and type I superconductors with the same thermo- dynamic critical field H,. Hig = V2eH. The ratio of B/H.z from this plot also gives the approximate variation of R/R,, where R is the electrical resistance for the case of negligible pinning, and Ry, is the normal-state resistance. critical field H,2, as shown schematically in Fig. 1.5. Because of the partial flux penetration, the diamagnetic energy cost of holding the field out is less, so Hy (which turns out to be given by V2kH,) can be much greater than the thermo- dynamic critical field H, (at which nothing special happens). This property has made possible high-field superconducting magnets. Another result of Abrikosov’s analysis was that, in the so-called mixed state, or Schubnikoy phase, between H., and H_2, the flux should not penetrate in laminar domains but, rather, in a regular array of flux tubes, each carrying a quantum of flux @y = = 2.07 x 10-7G — cm? (1.20) Within each unit cell of the array, there is a vortex of supercurrent concentrating the flux toward the vortex center. Although Abrikosoy predicted a square array, it was later shown, upon correcting a numerical error, that a triangular array should have a slightly lower free energy. This vortex array was first demonstrated experi- mentally by a magnetic decoration technique coupled with electron microscopy.”* More recently, scanning tunneling microscope measurements”® have not only *5U_ Essmann and H. Trduble, Phys. Lett. 244, 526 (1967). 26H. F. Hess et al. Phys. Rev. Lett. 62, 214 (1989); Phys. Rev. Lett. 64, 2711 (1990).HISTORICAL OvERVIEW 13 confirmed the existence of the vortex array, but they have also made possible detailed measurements of the density of electronic states in the quasi-normal core at the center of each vortex. Of course, random inhomogeneities in the underlying material lead to “pinning™ of vortices at favorable locations, so that in some cases one finds a glasslike pattern of flux tubes. We have already noted that type IT superconductors are not perfectly dia- magnetic, and since |x)|* turns out to go to zero in the centers of the vortices, we are not surprised to find that there is no energy gap in the cores. Thus, we are led to ask whether the first hallmark—perfect conductivity—is also lost. The answer is a bit equivocal, and the details are the subject of ongoing research. In the presence of a transport current. the flux tubes experience a so-called Lorentz force J x &/c per unit length (analogous to the macroscopic force density J x B/c) tending to make them move sideways, in which case a longitudinal “resistive” voltage is induced. In an ideal homogeneous material, Bardeen and Stephen?” showed that this flux motion is resisted only by a viscous drag, and that type II superconductors should show a resistance comparable to that in the normal state, only reduced by a factor ~B/H.2. In real materials, however, there is always some inhomogeneity to pin the flux, so that there is essentially no resistance until a finite current is reached, such that the Lorentz force exceeds the pinning force. In superconducting magnet wire, the pinning is deliberately made strong enough to give large critical currents. 1.7 PHASE, JOSEPHSON TUNNELIN AND FLUXOID QUANTIZATION Faced with these fallen hallmarks, one might well ask what really is the essential universal characteristic of the superconducting state. The answer is the existence of the many-particle condensate wavefunction _(r), which has amplitude and phase and which maintains phase coherence over macroscopic distances. This condensate is analogous to, but not identical to, the familiar Bose-Einstein con- densate, with Cooper pairs of electrons replacing the single bosons which con- dense in superfluid helium, for example. Since the phase and particle number are conjugate variables, reflecting com- plementary aspects of the wave-particle dualism, there is an uncertainty relation AN Ay 21 (1.21) which limits the precision with which N and y can be simultaneously known. However, since N ~ 10” in a macroscopic sample, both N and y can be known to within small fractional uncertainties, and the phase may be treated as a semi- classical variable. As we shall see in Chap. 7, however, this is not the case in very small mesoscopic structures. ?7J, Bardeen and M, J. Stephen, Phys. Rev. 140, A1197 (1965).14 rRoDUCTION To SUPERCONDUCTIVITY The physical significance of the phase degree of freedom was first empha- sized in the work of Josephson,** who predicted that pairs should be able to tunnel between two superconductors even at zero voltage difference, giving a supercurrent density J = Je sin (1 — ¢2) (1.22) where J, is a constant and y; is the phase of 2) in the ith superconductor at the tunnel junction. He also predicted that a voltage difference Vj) between the electrodes would cause the phase difference to increase with time as 2eVjt/h, so that the current would oscillate with frequency w = 2eV\2/h. Although ori- ginally received with some skepticism, these predictions have been extremely thoroughly verified. Subsequently, Josephson junctions have been utilized in ultrasensitive voltmeters and magnetometers, and in making the most accurate available measurements of the ratio of fundamental constants h/e. In fact, the standard yolt is now defined in terms of the frequency of the ac Josephson current. The most basic implication of the existence of a phase factor in 2(F) =|) |e", however, is operative in the simple case of a superconducting ring. In that case, the single-valuedness of w requires that y(r) return to itself (modulo 27) on going once around the ring on any path. Just as the correspond- ing condition in an atom leads to the quantization of orbital angular momentum in integral multiples of A, here this condition requires that the fluxoid &’ take on only integral multiples of © = hc/2e. The fluxoid is a quantity introduced by F London, which can be written o! = a+ epn-s (1.23) eS lel where ® = $A-ds is the ordinary magnetic flux through the integration loop. Since the current sustaining the flux in the ring flows only in a layer of thickness ~) on the inner surface of the ring, if the ring is thick compared to 2, the path of integration can be taken deeper inside the wall of the ring, where J, = 0. Then (1.23) implies that & = ', so that the flux itself has the quantized value n®9. This property was demonstrated experimentally”? in 1961. When J, is not small, as in the vortices in a type II superconductor, both terms in (1.23) may be equally important, and the value of the flux @ inside a given contour is itself unrestricted; only the fluxoid ®', directly related to the line integral of dy, always has precise quantum values. 8B. D. Josephson, Phys. Lett. 1, 251 (1962). B.S. Deaver and W. M. Fairbank, Phys. Rev. Lett. 7, 43 (1961); R. Doll and M. Nabauer, Phys. Rev. Lett. 7, 51 (1961).HisTORICAL overview 15 L8 FLUCTUATIONS AND NONEQUILIBRIUM EFFECTS The preceding discussion has been simplified by an implicit assumption that superconductors will always be found in the lowest-energy eigenfunction of the GL equation. While this is indeed the most probable single possibility, the presence of the thermal energy ~k7 implies that the system will fluctuate into other low- lying states with a finite probability. At temperatures below T,, the fluctuations which are most prominent are those which allow finite resistance to appear even at currents below the nominal critical current, which is defined as the maximum current before fluctuations of zero energy allow a resistive voltage to appear. The other side of the coin is that above T, fluctuations cause some vestiges of superconductivity to remain. These were first observed by Glover,” who found that the conductivity of amorphous films of superconductors diverges as (T — T,)~! as one approaches T, from above. This “Curie-Weiss” form of tem- perature dependence with an appropriate coefficient was also predicted theoreti- cally at about the same time. Somewhat later, the corresponding effect was also observed” in the diamagnetic susceptibility of pure bulk samples. In this case, the basic divergence is as (T— T, yl? These measurements and the associated theory show that in principle the effects of the superconducting interaction persist to arbitrarily high temperatures but that in practice a fairly strong cutoff sets in at about 27,. Thus, there is not only some resistance below 7, but also some superconductivity above T,, although the apparently abrupt switchover observed by Kamerlingh Onnes is still a good working approximation for most purposes. The superconducting transition in the classic superconductors is much sharper than other second- order phase transitions, such as those in magnetic materials, because the coher- ence length & is much larger than the interatomic distance, so that each electron interacts with many others. However, this is not the case in the high-temperature superconductors, where the coherence length is comparable to atomic dimensions, leading to much more prominent fluctuation effects. In addition to phenomena in which condensate states other than the ground state are explored by thermal fluctuations in the context of thermal equilibrium, nonequilibrium regimes have also been studied, as described in Chap. 11. In the simplest examples, energy is fed in from an external source to drive the quasi- particle population out of equilibrium. In this case, one can distinguish two classes of nonequilibrium, involving energy and charge, respectively, as codified by Schmid and Schén. The former category includes regimes in which 7, can actually be raised by as much as a factor of 2, but the energy gap A is not raised above its equilibrium value at T= 0. “OR, E. Glover, II, Phys. Lett. 25A, 542 (1967). 3!J, P, Gollub, M. R. Beasley, R. 8, Newbower, and M. Tinkham, Phys. Rev. Leit, 22, 1288 (1969). A. Schmid and G, Schdn, J. Low Temp. Phys. 20, 207 (1975).