Modern Aspects of

Superconductivity
Theory of Superconductivity

Sergei Kruchinin
Bogolyubov Institute for Theoretical Physics, Kiev, Ukraine

Hidemi Nagao & Shigeyuki Aono

Kanazawa University, Japan

NEW JERSEY t LONDON t SINGAPORE t BEIJING t SHANGHAI t HONG KONG t TA I P E I t CHENNAI

Published by
World Scientific Publishing Co. Pte. Ltd.
5 Toh Tuck Link, Singapore 596224
USA office: 27 Warren Street, Suite 401-402, Hackensack, NJ 07601
UK office: 57 Shelton Street, Covent Garden, London WC2H 9HE

British Library Cataloguing-in-Publication Data

A catalogue record for this book is available from the British Library.

MODERN ASPECTS OF SUPERCONDUCTIVITY

Theory of Superconductivity
Copyright © 2011 by World Scientific Publishing Co. Pte. Ltd.
All rights reserved. This book, or parts thereof, may not be reproduced in any form or by any means, electronic or
mechanical, including photocopying, recording or any information storage and retrieval system now known or to
be invented, without written permission from the Publisher.

For photocopying of material in this volume, please pay a copying fee through the Copyright Clearance Center,
Inc., 222 Rosewood Drive, Danvers, MA 01923, USA. In this case permission to photocopy is not required from
the publisher.

ISBN-13 978-981-4261-60-9
ISBN-10 981-4261-60-2

Typeset by Stallion Press

Email: [email protected]

Printed in Singapore.
“The single reason for our inability to treat the problems of
superconductivity consists in the absence of a sufficient
imagination.”

— Richard P. Feynman
 Preface

Studies of superconductivity theory are among the most fruitful and promis-
ing trends in the theoretical physics of condensed matter, since supercon-
ductivity remains one of the most interesting research areas in physics.
The goal of this book is to give a representation of certain modern
aspects of superconductivity. We discuss important aspects of the theory of
superconductivity, such as the nature of high-Tc superconductivity, two-gap
superconductivity, room-temperature superconductivity, mesoscopic super-
conductivity, the pairing state and the mechanism of cuprate high-Tc super-
conductivity.
In Chap. 1, we consider the field-theoretical method of superconductivity
and discuss the basic idea of superconductivity and the elaboration of the
Ginzburg–Landau and Bardeen–Cooper–Schrieffer theories in the frame of
many-particle quantum field theory.
In Chap. 2, we consider the structures of high-Tc superconductors, phase
diagrams and the problem of pseudogaps, and analyze the mechanisms of
superconductivity. We present general arguments regarding the pairing sym-
metry in cuprate superconductors and investigate their thermodynamical
properties within the spin-fluctuation mechanism of superconductivity, by
using the method of functional integrals.
Chapter 3 concentrates on two-band and multiband superconductiv-
ity. We consider the physical properties of the superconductor MgB2 and
use our two-band model to explain the two coupled superconductor gaps
of MgB2 . To study the effect of the increasing Tc in MgB2 , we use the
renormalization-group approach and phase diagrams. In the field of super-
conductivity we meet the problem-maximum, which consists in the creation
of room-temperature superconductors. We consider this problem in our book
and make some recommendations on the search for these superconductors.

vii
viii Modern Aspects of Superconductivity: Theory of Superconductivity

Chapter 4 deals with the problem of mesoscopic superconductivity. We

consider the two-band superconductivity in ultrasmall grains, by extending
the Richardson exact solution to two-band systems, and develop the theory
of interactions between nanoscale ferromagnetic particles and superconduc-
tors. The properties of nanosize two-gap superconductors and the Kondo
effect in superconducting ultrasmall grains are investigated as well.
We also consider the ideas of quantum computing and quantum informa-
tion in mesoscopic circuits. The theory of the Josephson effect is presented,
and its applications to quantum computing are analyzed.
This book deals with a wide scope of theoretical and experimental top-
ics in superconductivity, and has been written for advanced students and
researchers in that field.

S. Kruchinin, H. Nagao and S. Aono

Kiev, Kanazawa, July 2009
 Contents

Preface vii

1. Theory of Superconductivity 1
1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Spinors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.2.1 Spinor . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.2.2 Noether theorem and Nambu–Goldstone
theorem . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.3 Propagator . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.3.1 Hamiltonian . . . . . . . . . . . . . . . . . . . . . . 11
1.4 Noninteracting . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.5 Interacting . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
1.5.1 Unrestricted Hartree–Fock (HF) . . . . . . . . . . . 19
1.5.2 Gap equation for superconductivity . . . . . . . . . 20
1.6 Illustrative Example, Critical Temperature . . . . . . . . . 21
1.6.1 Bond alternation . . . . . . . . . . . . . . . . . . . . 21
1.6.2 Deformation energy . . . . . . . . . . . . . . . . . . 23
1.6.3 Polyacene, gap equation, critical temperature . . . . 24
1.6.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . 27
1.7 Linear Response Magnetic Resonance
in Normal and Superconducting Species;
Spin–lattice Relaxation Time . . . . . . . . . . . . . . . . . 28
1.7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . 28
1.7.2 T1 in NMR . . . . . . . . . . . . . . . . . . . . . . . 31
1.7.3 Theory with Green’s function . . . . . . . . . . . . 32
1.7.4 Noninteracting . . . . . . . . . . . . . . . . . . . . . 35
1.7.5 Interacting; normal . . . . . . . . . . . . . . . . . . 36

ix
x Modern Aspects of Superconductivity: Theory of Superconductivity

1.8 Interacting; Superconductor . . . . . . . . . . . . . . . . . . 39

1.8.1 The extended Nambu spinor . . . . . . . . . . . . . 39
1.8.2 Green’s function . . . . . . . . . . . . . . . . . . . . 40
1.8.3 Spin dynamics . . . . . . . . . . . . . . . . . . . . . 41
1.8.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . 42
1.9 Ginzburg–Landau Theory from the BCS Hamiltonian . . . 43
1.9.1 BCS theory . . . . . . . . . . . . . . . . . . . . . . 44
1.9.2 Hubbard–Stratonovitch transformation . . . . . . . 46
1.9.3 Fourier transform . . . . . . . . . . . . . . . . . . . 48
1.9.4 Nambu spinor . . . . . . . . . . . . . . . . . . . . . 48
1.9.5 Critical temperature . . . . . . . . . . . . . . . . . . 51
1.9.6 Temperature dependence of φ . . . . . . . . . . . . 52
1.9.7 Dynamics of the boson field; symmetry breaking . . 53
1.9.8 Instability . . . . . . . . . . . . . . . . . . . . . . . 55
1.9.9 Supersymmetry . . . . . . . . . . . . . . . . . . . . 56
1.9.10 Towards the Ginzburg–Landau equation . . . . . . 58
1.9.11 Discussion . . . . . . . . . . . . . . . . . . . . . . . 59

2. Physics of High-Tc Superconductors 61

2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
2.2 History of the Development of Superconductivity . . . . . . 64
2.3 Structural Properties of High-Temperature
Superconductors . . . . . . . . . . . . . . . . . . . . . . . . 65
2.3.1 Phase diagram of cuprate superconductors . . . . . 69
2.4 Mechanisms of Pairing of HTSCs . . . . . . . . . . . . . . . 72
2.4.1 Specific mechanisms of pairing in superconductivity 72
2.4.2 Magnetic mechanism of pairing . . . . . . . . . . . 74
2.4.3 Exciton mechanism of pairing . . . . . . . . . . . . 75
2.4.4 Anharmonic model and superconductivity . . . . . 76
2.4.5 Van Hove singularities . . . . . . . . . . . . . . . . 77
2.4.6 Plasmon mechanism of pairing . . . . . . . . . . . . 78
2.4.7 Bipolaronic mechanism of superconductivity . . . . 78
2.5 Symmetry of Pairing in Cuprate Superconductors . . . . . . 80
2.5.1 The superconductor’s order parameter . . . . . . . 80
2.5.2 Classification of the superconductor’s order param-
eter by the representations of symmetry groups . . 82
2.6 Experimental Studies of the Symmetry
of the Superconducting Order Parameter . . . . . . . . . . . 84
2.6.1 Measurements of the Josephson tunnel current . . . 84
 Contents xi

2.6.2 Measurements of the quantization of a flow

by the technique of the three-crystal unit . . . . . . 86
2.7 Thermodynamics of the d Pairing
in Cuprate Superconductors . . . . . . . . . . . . . . . . . . 88
2.7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . 88
2.7.2 Antiferromagnetic spin fluctuations in HTSCs . . . 89
2.7.3 Continual model of antiferromagnetic spin
fluctuations . . . . . . . . . . . . . . . . . . . . . . . 92
2.7.4 Equation for the superconducting gap . . . . . . . . 97
2.7.5 Thermodynamic potential of antiferromagnetic
spin fluctuations . . . . . . . . . . . . . . . . . . . . 101
2.7.6 Heat capacity of the d pairing . . . . . . . . . . . . 103
2.7.7 Heat capacity jump in superconductors . . . . . . . 104
2.8 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
3. Multiband Superconductivity 108
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
3.2 Multiband Hamiltonian . . . . . . . . . . . . . . . . . . . . 110
3.2.1 Hamiltonian . . . . . . . . . . . . . . . . . . . . . . 110
3.2.2 Two-particle Green function . . . . . . . . . . . . . 112
3.2.3 Traditional superconductivity . . . . . . . . . . . . 114
3.2.4 Copper oxides . . . . . . . . . . . . . . . . . . . . . 115
3.2.5 Cooperative mechanism . . . . . . . . . . . . . . . . 115
3.2.6 Room-temperature superconductors . . . . . . . . . 116
3.3 Two-gap Superconductivity in MgB2 . . . . . . . . . . . . . 119
3.3.1 The physical properties of MgB2 . . . . . . . . . . . 119
3.3.2 Theoretical model . . . . . . . . . . . . . . . . . . . 122
3.3.3 Superconductivity in MgB2 . . . . . . . . . . . . . . 124
3.4 Theoretical Studies of Multiband Effects
in Superconductivity by Using
the Renormalization Group Approach . . . . . . . . . . . . 127
3.4.1 Theoretical model . . . . . . . . . . . . . . . . . . . 128
3.4.2 Renormalization group approach . . . . . . . . . . . 130
3.4.3 Vertex correction and response function
for Cooper pairs . . . . . . . . . . . . . . . . . . . . 131
3.4.4 Renormalization equation . . . . . . . . . . . . . . . 132
3.4.5 Phase diagrams . . . . . . . . . . . . . . . . . . . . 133
4. Mesoscopic Superconductivity 135
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
xii Modern Aspects of Superconductivity: Theory of Superconductivity

4.2 Nanosize Two-Gap Superconductivity . . . . . . . . . . . . 137

4.2.1 Hamiltonian for nanosize grains . . . . . . . . . . . 138
4.2.2 Path-integral approach . . . . . . . . . . . . . . . . 139
4.2.3 Condensation energy . . . . . . . . . . . . . . . . . 141
4.2.4 Critical level spacing . . . . . . . . . . . . . . . . . 142
4.2.5 Parity gap . . . . . . . . . . . . . . . . . . . . . . . 143
4.3 Exact Solution of Two-Band Superconductivity
in Ultrasmall Grains . . . . . . . . . . . . . . . . . . . . . . 145
4.3.1 Exact solution for two-band superconductivity . . . 146
4.3.2 Hamiltonian . . . . . . . . . . . . . . . . . . . . . . 146
4.3.3 Exact solution . . . . . . . . . . . . . . . . . . . . . 148
4.3.4 Preprocessing for numerical calculations . . . . . . 151
4.3.5 Results and discussion . . . . . . . . . . . . . . . . 153
4.3.6 Pair energy level . . . . . . . . . . . . . . . . . . . . 153
4.3.7 Condensation energy . . . . . . . . . . . . . . . . . 154
4.3.8 Parity gap . . . . . . . . . . . . . . . . . . . . . . . 155
4.4 Kondo Effect Coupled to Superconductivity . . . . . . . . . 157
4.4.1 Kondo regime coupled to superconductivity . . . . 158
4.4.2 Model . . . . . . . . . . . . . . . . . . . . . . . . . . 159
4.4.3 Mean-field approximation . . . . . . . . . . . . . . . 160
4.4.4 Critical level spacing in the Kondo effect . . . . . . 161
4.4.5 Kondo effect coupled to superconductivity . . . . . 162
4.4.6 Exact solution for the Kondo regime . . . . . . . . 163
4.5 Interaction of Nanoscale Ferromagnetic Granules
in London Superconductors . . . . . . . . . . . . . . . . . . 166
4.5.1 Magnetic field of ferromagnetic inclusions
in a London superconductor . . . . . . . . . . . . . 168
4.5.2 Magnetic field of ferromagnetic quantum dots
in a superconducting nanocomposite material . . . 169
4.5.3 Interaction energy of quantum dots
in a superconducting nanocomposite material . . . 170
4.5.4 Spin-orientation phase transitions in a nanocom-
posite material with arrays of ferromagnetic quan-
tum dots . . . . . . . . . . . . . . . . . . . . . . . . 171
4.6 Spin-Orientation Phase Transitions
in a Two-Dimensional Lattice of Ferromagnetic
Granules in a London-Type Superconductor . . . . . . . . . 174
4.7 Quantum Computer on Superconducting Qubits . . . . . . 181
4.7.1 Principle of quantum computers . . . . . . . . . . . 181
 Contents xiii

4.7.2 Superconducting qubits . . . . . . . . . . . . . . . . 182

4.7.3 Josephson effect . . . . . . . . . . . . . . . . . . . . 183

Summary and Conclusions 191

Appendix A: Two-particle Green Function for Multiband

Superconductors 193

Appendix B: A Solution Method for the Ferromagnetic

Granules in London Superconductors 199

Bibliography 203

Index 215
 CHAPTER 1

Theory of Superconductivity

1.1 Introduction
In the preface to the book Superconductivity, edited by Parks,1 one of the
editors says: “During the preparation of this treatise one of the authors com-
mented that it would be the last nail in the coffin [of superconductivity].”
Some specialists in superconductivity told us that we would hardly be able
to find articles useful for our future investigations, except for Anderson’s
comments at the end of that book. Further, we learned that Anderson was
pessimistic about further advance of superconductivity; for example, high-
temperature or room-temperature superconductivity was most unlikely.
However, against his expectation, the discovery of high-temperature super-
conductivity, due to Bednorz and Müller,2 is astonishing. The traditional
Bardeen–Cooper–Schrieffer (BCS) theory3 has failed to explain the mecha-
nism of such superconductivity. Anderson5 proposed a new idea called the
resonating valence bond (RVB) theory or the t−J model (the term t implies
the transfer integral, and J the electron correlation). We have never known
his theory to be successful.
A comment by Feynman, found in his book Statistical Mechanics,4 says
that it will take almost 50 years for the problem of superconductivity to be
reduced to that of explaining the gap. Following the BCS theory, we will
explain the gap, and the theory is essentially correct, but we believe that
it needs to be made obviously correct. As it stands now, there are a few
seemingly loose ends to tie up.
The theory of superconductivity seems to be founded on the London pos-
tulate.7 Associated with the gauge transformation, the conserved current is
i
j = − (φ∗ ∇φ − φ∇φ∗ ) − e|φ|2 A. (1.1)
2

1
2 Modern Aspects of Superconductivity: Theory of Superconductivity

The current due to the first term is called the paramagnetic current; and that
due to the second, the diamagnetic current. In the superconducting state,
the first term on the right-hand side changes very slightly, and sometimes
the wave function is quite rigid, that only the diamagnetic current survives.
In this respect, the superconductor is the perfect diamagnetic substance.
The current is dominated by

j = −k2 A, (1.2)

where k is a properly chosen positive constant. The Meissner effect is easily

derived from Ampére’s equation:

∇ × B = j. (1.3)

Taking rotation gives

∇2 B = k 2 B (1.4)

or

Bx = B0 e−kx . (1.5)

It is important to note that |φ|2 in Eq. (1.1) is very large, i.e. it is the classic
scale quantity, so that the magnetic field in Eq. (1.5) damps very rapidly.
This is the Meissner effect. On the book cover, we see the very spectacular
experiment demonstrating the Meissner effect — a permanent magnet hov-
ers above a superconducting plate. We know of similar phenomena — the
screening of the Coulomb interaction, or quark confinement.
The boson model of Cooper pairs is considerably successful. Equa-
tion (1.3) yields ∇2 A = k2 A, which is

∂ μ Aμ = −k2 Aμ (1.6)

in the covariant form, suggesting that a photon is massive, which is a fun-

damental aspect of superconductivity.8
An essentially similar treatment has been presented by Ginzburg and
Landau (GL).9 The superconducting state is the macroscopic state; in other
words, a thermodynamic phase. They characterized this phase by introduc-
ing the order parameter, Ψ. This looks like the Schrödinger function Ψ, and
then the primitive quantum theorists got confused, saying that behind the
Iron Curtain the quantum theory was different in features from that of the
Western countries.
 Theory of Superconductivity 3

The theory is handled as the phase transition. The Lagrangian for the
superconducting state is postulated as
 
1 1  e∗ A 2 h2
2
Fs = F0 + a|Ψ| + b|Ψ| + 4
−i∇ +  + 8π , (1.7)
2 2m∗  c
where ∗ indicates the quantities in question referred to the superconductor
(×2). Note that there are |Ψ|2 and |Ψ|4 terms in the potential parts. These
drive the system to spontaneous symmetry breaking and lead to a phase
transition for the suitable choice of constants a and b. This is now called
the Higgs mechanism.8 Certainly, the GL treatment is a few years ahead of
the Nambu–Goldstone suggestion.
The GL theory is known as the macroscopic quantum mechanism, and
in this sense, the big Ψ is sometimes amusingly called the cat’s Ψ.
A very instructive presentation of the macroscopic quantum theory is
found in The Feynman Lectures on Physics. Vol. III, Chap. 21. Various
topics there, such as the Josephson junction, are quite readable.
The microscopic theory was prepared by Bardeen, Cooper and
Schrieffer.3 However, as a preliminary discussion, we present the Bogoliubov
treatment. The superconductivity is a kind of many-electron problems. The
most general Hamiltonian should be
 
1
H = dxψ+ (x)h(x)ψ(x) + dx dx ψ+ (x)ψ(x)v(x, x )ψ+ (x )ψ(x ).
2
(1.8)
Since the algebra of electrons is the spinor, the terms other than the
above identically vanish. In other words, the three- or four-body interac-
tions are useless. First, we specify the spin indices, and next the plane-wave
representation for the spatial parts. Equation (1.8) is simply written as
∗ ∗
H = k (a+
k + ak ) + vk,−k ak a−k a−k ak . (1.9)
The simplification or the mean-field approximation for the two-body part is
twofold — say,
Δk,−k akα akα a+
−kβ a−kβ 
+
(1.10)
Δk,−k a+
kα akβ akα a−kβ  + c.c.
+

The latter looks curious, since such an expectation value, akα a−kβ , vanishes
identically. Against this common sense, Bogoliubov put the Hamiltonian
∗ ∗ ∗
k + ak ) + Δk ak a−k + Δk a−k ak .
H = k (a+ (1.11)
4 Modern Aspects of Superconductivity: Theory of Superconductivity

We understand that Bogoliubov presumed the Cooper pair, and provided

the effective Hamiltonian for pairs. His theory may be a shorthand treatment
of the BCS theory. This Hamiltonian is diagonalized by the so-called
Bogoliubov transformation which defines the quasiparticle responsible for
the superconductivity as
    
γk↑ uk −vk ck↑
= , (1.12)
γ+
−k↓ vk uk c+
−k↓

with
u2k − vk2 = 1. (1.13)
The spirit of the Bogoliubov transformation is to mix operators ck↑ and
c+
−k↓ , which are different in spin. The quasiparticle yields the new ground
state, so that the particle pair or the hole pair arises near the chemical
potential. The stabilization energy thus obtained is called the gap energy
Δk .3, 10
We now follow the BCS microscopic treatment. The Green function has
been effectively used by employing the Nambu spinor. This makes the uni-
fied treatment of normal and superconducting states possible. However,
the temperature Green function (Matsubara function) is used from the
beginning.

1.2 Spinors
We start with the general many-electron Hamiltonian not restricted to the
BCS Hamiltonian. The BCS state responsible for the superconducting state
is easily recognized from the formal description. The simplest method of the
quantum chemistry should be the Hückel theory. This consists of the single
energy matrix,

βrs = dx h(x)ρrs (x), (1.14)

where h(x) is the single-particle quantum-mechanical Hamiltonian, and the

electron density ρrs (x) is given by the product of the single-particle (atomic)
orbitals χr and χs . Note that this is the spinless theory.
We then extend the treatment into the spin space:
 ↑↑ 
βrs β↑↓
rs
βrs = . (1.15)
β↓↑ ↓↓
rs βrs
 Theory of Superconductivity 5

Any 2 × 2 matrix is expanded in the Pauli spin matrices together with the
unit matrix:
       
1 0 1 0 0 1 0 −i
σ =
0
, σ = 3
, σ =
1
, σ =
2
.
0 1 0 −1 1 0 i 0
(1.16)
However, we employ other combinations:
 
↑1 1 0
σ = (σ0 + σ3 ) = ,
2 0 0
 
↓ 1 0 0 0
σ = (σ − σ ) =
3
,
2 0 1
  (1.17)
1 1 0 1
σ = (σ + iσ ) =
+ 2
,
2 1 0
 
1 0 0
σ− = (σ1 − iσ2 ) = .
2i 1 0
In Eq. (1.24), we then have
βrs = βμrs σμ ,
(1.18)
βμrs = Tr(σμ βrs ),
in detail:
β↑rs = β↑↑
rs β↓rs = β↓↓
rs ,
↓↑
(1.19)
rs = βrs
β+ β− ↑↓
rs = βrs .

As to the Pauli matrices, the ordinary commutators are

⎫
[σi , σj ] = 2iijk σk ⎪
⎪
⎪
⎪
[σ3 , σ+ ] = 2σ+ ⎬
, (1.20)
[σ3 , σ− ] = −2σ− ⎪ ⎪
⎪
⎪
⎭
[σ+ , σ− ] = σ3
⎫
[σ↑ , σ↓ ] = 0 ⎪
⎪
⎪
⎪
↑
[σ , σ ] = σ
+ + ⎪
⎪
⎪
⎪
⎪
[σ , σ ] = −σ−⎬
↑ −
, (1.21)
[σ↓ , σ+ ] = −σ+ ⎪
⎪
⎪
⎪
[σ↓ , σ− ] = σ− ⎪⎪
⎪
⎪
−
⎪
⎭
[σ , σ ] = σ
+ 3
6 Modern Aspects of Superconductivity: Theory of Superconductivity

and the anticommutators are


[σi , σj ]+ = 2iδij
. (1.22)
[σ+ , σ− ]+ = 1
We then express the matrix in Eq. (1.15) as
βrs = βμrs σμ ,
(1.23)
βμrs = Tr(σμ βrs ),
in detail:
β↑rs = β↑↑
rs β↓rs = β↓↓
rs ,
(1.24)
↓↑
rs = βrs
β+ β− ↑↓
rs = βrs .

Here, if the quantum-mechanical Hamiltonian has the single-particle char-

acter without the external field causing a rotation in the spin space, the
off-diagonal elements are meaningless. The Hückel theory involves the spin
diagonal terms.
However, if we take the electron–electron interaction into account, even in
the mean-field approximation, the off-diagonal elements become meaningful
and responsible for the superconductivity. This is what we investigate here.

1.2.1 Spinor
The algebra representing electrons is the spinor. The Dirac relativistic (spe-
cial relativity) function describes this property well. However, the relativity
seems not so important for the present problem. We now concentrate on the
spinor character of the electron. The field operator has two components in
the spin space:
 
φ↑ (x)
φ(x) = ,
φ+
↓ (x)
 
1 0
φ̄(x) = φ+ (x)σ3 = (φ+
↑ (x) φ↓ (x))
0 −1
= (φ+
↑ (x) −φ↓ (x)). (1.25)
This is called the Nambu representation.11 The negative sign in front of φ↓
is seen in the Dirac conjugate φ+ → φ̄.
The field operators satisfy, of course, the anticommutators
 
[φα (x), φ+
β (s )]+ = δα,β δ(x − x ), (1.26)
 Theory of Superconductivity 7

where (α, β) = (↑, ↓) and x = (r, t). Then, for the spinors (1.25), the matrix
commutator holds:
  
 φ↑ (x)  +  
[φ(x), φ̄(x )]+ = , φ↑ (x ) −φ↓ (x )
φ+↓ (x) +
 + 
   
φ↑ (x), φ↑ (x ) + 
φ↓ (x ), φ↑ (x) +
=  + 
  
φ↓ (x), φ+ ↑ (x ) + φ↓ (x ), φ+
↓ (x) +
 
δ(x − x ) 0
= . (1.27)
0 δ(x − x )

1.2.2 Noether theorem and Nambu–Goldstone theorem

We seek for the meaning of the Nambu spinor.12 Consider a global trans-
formation of fields with the constant Λ:

φα (x) → φα (x)eiΛ ,
−iΛ
(1.28)
α (x) → φα (x)e
φ+ .

It is recognized that the Hamiltonian and the equation of motion are invari-
ant under this transformation. We can see that this transformation is a
rotation around the σ3 axis with Λ,
3
φ(x) → eiσ Λ φ(x), (1.29)

since
3
eiσ Λ = cos(σ3Λ) + i sin(σ3Λ)
 
(σ3Λ)2 (σ3Λ)4 (σ3Λ)3 (σ3Λ)5
=1− + + ··· + i 1 − + + ···
2! 4! 3! 5!
 
Λ2 Λ4 Λ3 Λ5
=1− + + · · · + iσ 1 −
3
+ + ···
2! 4! 3! 5!
 iΛ 
e 0
= σ cos Λ + iσ sin Λ =
0 3
.
0 e−iΛ

Here, we discuss briefly the Noether theorem and the Nambu–Goldstone

theorem. The latter makes a profound investigation possible. If the
Lagrangian of the system in question is invariant under some transformation
which is just the present case, we have the continuity relation. Notice that
8 Modern Aspects of Superconductivity: Theory of Superconductivity

the density and the current are, in terms of the Nambu spinor,
j 0 = φ+ (x)σ3 φ(x),
 (1.30)
j = −i (φ+ (x)∇φ(x) + ∇φ+ (x)φ(x)).
2m
Then the continuity relation is
∂t j0 (x) + ∇ · j = 0. (1.31)
If the system is static, the density must be conserved:
∂t j 0 = 0. (1.32)
Put 
G= dxj 0 (x), (1.33)
and if it is found that
[G, Ψ (x)] = Ψ(x), (1.34)
and the expectation value of Ψ(x) over the ground state does not vanish,
0|Ψ|0 = 0, (1.35)
i.e. if the ground state satisfying the relation (1.33) does not vanish, we can
expect the appearance of a boson Ψ(x), whose mass is zero. This boson is
called a Goldstone boson, and the symmetry breaking takes place in the
system. This is what the Goldstone theorem insists on. The details are in
the standard book on the field theory.8
Now we apply this theorem to the superconductivity. The invariant
charge is
 
Q = d3 xφ+ α (x)φ α (x) = d3 xφ+ (x)σ3 φ(x). (1.36)

In terms of the Nambu spinor, here σ3 is crucial. In the commutator (1.33),

we seek for the spin operators which do not commute with σ3 and find, say,
σ± . We then have the Goldstone commutator as
d3 x [φ+ (x )σ3 φ(x ), φ+ (x)σ± φ(x)]t=t

= d3 x(±2)0|φ+ (x)σ± φ(x )|0δ(r − r ). (1.37)
In detail,
 0  
+ 1 φ↑ (x)
φ (x)σ φ(x) = φ↑ (x) φ↓ (x)
+ +
0 0 φ+↓ (x)
= φ+ +
↑ (x)φ↓ (x), (1.38)
φ+ (x)σ− φ(x) = φ↑ (x)φ↓ (x).
 Theory of Superconductivity 9

Notice that here the same symbol φ is used for the ordinary field with spin
and the Nambu spinor. The above are nothing but the Cooper pairs, and
we now find the Goldstone bosons Ψ∗ and Ψ. In literature, it is noted that
Δ±
0|φ+ (x)σ± φ(x)|0 = ± = 0, (1.39)
g
where Δ and g are the gap and the coupling parameter, respectively. We
expect the estimate
Δ+ ∼ Δ− = Δ
to be reasonable.
The Cooper pairs are now the Goldstone bosons. A comment about
the Goldstone boson or the massless elementary excitation with k = 0 is
required. Using Eq. (1.33), we write the Goldstone commutator (1.34) as

d3 y[0|j0 (y)|nn|φ (x)|0

− 0|φ (x)|nn|j0 (y)|0]x0 =y0 = 0, (1.40)

where |n is the intermediate state, and x0 = y0 implies that this is the
equal time commutator.
Since
j0 (y) = e−ipy j0 (0)eipy , (1.41)
it is seen that

d3 y[0|J0 (0)|nn|φ (x)|0eipn y

− 0|φ (x)|nn|J0 (0)|0e−ipn y ]x0 =y0 (p|n = pn |n)

= δ(pn )[0|J0 (0)|nn|φ (0)|0eipn0 y
− 0|φ (x)|nn|J0 (0)|0e−ipn0 y ]x0 =y0
= δ(pn )[0|J0 (0)|nn|φ (0)|0eiMn y0
− 0|φ (0)|nn|J0 (0)|0e−iMn y0 ]x0 =y0 = 0. (1.42)
In order to obtain the first equality, spatial integration is carried out. Then,
considering pμ = (p, M ), we retain the fourth component. In the last equa-
tion, when Mn = 0, cancellations will arise for the summation over n. We
thus obtain, only for Mn = 0, the finite result
0|j0 (0)|nn|φ (0)|0 − 0|φ (0)|nn|j0 (0)|0
= Im 0|j0 (0)|nn|φ (x)|0, (1.43)
10 Modern Aspects of Superconductivity: Theory of Superconductivity

which is met with the requirement (1.34). The excitation with Mn = 0 needs
no excitation energy, suggesting the Goldstone boson.
We further note that the mass-zero excitation is the imaginary quantity.
This suggests the current to be the phase current, as is seen in the Josephson
effect.
Before closing this preliminary discussion, we want to make a few
remarks. At the beginning, we mentioned London’s postulate that the super-
conducting state is characterized by a statement that the wave function is
rigid, so that the current is entirely the diamagnetic current due to only the
vector potential A. “Rigid” is not really rigid, but it is understood that the
spatial derivative is vanishing, or the current flows along the entirely flat
path, which is described in a textbook as the path going around the top of
a Mexican hat. Boldly speaking, the electron in the superconducting state
is massless. Also, we have pointed out that the vector potential A, which
leads to the Meissner effect, satisfies the covariant relation (1.6),
∂μ Aμ = −k2 , (1.44)
so that a photon is massive in the superconductor.
In the following chapters, we develop a substantial microscopic explana-
tion of the above assertions.

1.3 Propagator
In the previous section, we have found that, as is seen in Eq. (1.41), the
superconducting state strongly concerns the gap function or the anomalous
Green function. We want to deal with the solid-state substances. However,
the infinite crystals described by the single band have already been fully
investigated in literature, and the recent investigations were carried out on
objects with multiband structure.13, 14 The infinite system with many bands
is constructed from the unit cell, which is really a chemical molecule. The
atoms in this molecule give the band index of the real crystal. In this respect,
we first investigate the Green function of a unit cell. The Green function is
now shown in the site representation.
Corresponding to the spinors (1.24), we define the spinor in the site
representation as
 
ar↑
ar = , ār = (a+r↑ −ar↓ ). (1.45)
a+
r↓
 Theory of Superconductivity 11

Due to this definition, it is unnecessary to insert σ3 in the matrix G, as is

seen in Schrieffer’s book.3 The commutator is
   
[ar↑ , a+
s↑ ]+ [as↓ , ar↑ ]+ δrs 0
[ar , ās ]+ = = = δrs 12×2 . (1.46)
[a+ + +
r↓ , as↑ ]+ [as↓ , ar↓ ]+ 0 δrs
The matrix propagator is defined by

Grs (τ) = −ar (τ1 ), ās (τ2 )

 
ar↑ (τ1 ), a+ s↑ (τ 2 ) −ar↑ (τ 1 ), as↓ (τ 2 )
=−
a+r↓ (τ 1 ), as↑ (τ2 ) −ar↓ (τ1 ), as↓ (τ2 )
+ +

= −θ(τ1 − τ2 )ar (τ1 )ās (τ2 ) + θ(τ2 − τ1 )ās (τ2 )ar (τ1 ), (1.47)
where τ is the imaginary time, so that the propagator is the temperature
Green function or the Matsubara function. In what follows, we put

τ = τ1 − τ2 , (1.48)

and the system depends on τ1 − τ2 . If we want to obtain the gap function,

we consider

Tr(σ+ Grs (τ)) = a+ +

r↓ (τ1 ), as↑ (τ2 ), (1.49)
and the standard procedure will be followed.

1.3.1 Hamiltonian
Various Hamiltonians can be written by using the charge density matrix,
 
ρr↑s↑ (x) ρr↑s↓ (x)
ρrs (x) = = σμ ρμrs , (1.50)
ρr↓s↑ (x) ρr↓s↓ (x)
where the basis orbitals are put to be real, so that we do not need the
conjugation procedure for field operators.
The Hückel Hamiltonian or the single-particle Hamiltonian has the struc-
ture

H 0 = hμrs ār σμ ar , (1.51)

where h includes the chemical potential and can be explicitly written as
   
ρr↑s↑ (x) ρr↑s↓ (x)
hrs = ρrs h(x) = dxh(x) = σμ hμrs . (1.52)
ρr↓s↑ (x) ρr↓s↓ (x)
12 Modern Aspects of Superconductivity: Theory of Superconductivity

Hereafter, we use the summation convention that repeated indices imply

that the summation is carried out to facilitate manipulations. Other two-
particle Hamiltonians are given in a similar way. Noting that

μυ
vrs;tu = dx dx ρμrs (x)v(x − x )ρυtu (x ), (1.53)

we have17
1
H dir = (ār σa as )vrs;tu
ab
(āt σb au ) (a, b =↑, ↓),
2
1
H ex = − (ār σa as )vrs;ut
aa
(āt σa au ) (a =↑, ↓),
2 (1.54)
1
H super = {(ār σ+ as )vrs;tu
+−
(āt σ− au )
2
−+
+ (ār σ− as )vrs;tu (āt σ+ au )}.
For the direct interaction, the quantum-mechanical Hamiltonian is

H = dx dx χ∗r↑ (r)χs↑ (r)(x)v(x − x )χ∗↓ (x )χ∗u↓ (x ).
dir

For the field-theoretical Hamiltonian, the wave functions are replaced by

the creation–annihilation operators a+ r↑ , ar↑ , and so on. These are written
in the spinor notation; for example,
 1   
 as↑
+ +
ar↑ as↑ = ar↑ ar↓ 1 0
0 a+
s↓
 
1 0 ↑
= a+ a = a+ r σ as .
r
0 0 s
Note that H ex is obtained by reversing indices (u ↔ t) in vrs;tu .

1.4 Noninteracting
“Noninteracting” implies that the Hamiltonian is bilinear with respect to
operators so that diagonalization is always possible. It should be instruc-
tive to begin with the single-particle case, since even if we manipulate the
complicated two-particle case, the procedures are almost the same when the
mean-field approximation is employed.
The energy for the Hamiltonian (1.52) is
E 0 = Tr(ρ̂H 0 ) = hrs ār as , (1.55)
where ρ̂ is the statistical operator,
0 −Ω)
ρ̂ = eβ(H , (1.56)
 Theory of Superconductivity 13

with the normalization factor Ω. We now define the temperature Green

function, in which τ is the imaginary time, τ = it,15, 16
Grs (τ) = −Trρ̂[(θ(τ1 )ar (τ1 )a+
s (τ2 ) − θ(−τ)as (τ2 )ar (τ1 )]
+

= ar (τ1 )a+

s (τ2 ), (1.57)
where
τ = τ1 − τ2 , (1.58)
and it is assumed that the system is dependent only on the relative time τ.
Then we can write E 0 in terms of the temperature Green function:
E 0 = Tr[hrs Gsr (τ = 0− )]. (1.59)
Note that hrs and Gsr are matrices.
The equation of motion (1.57) for the noninteracting Hamiltonian (1.51)
can be read as
∂τ1 as (τ1 )ār (τ2 )
= δ(τ1 − τ2 )[as (τ1 ), ār (τ2 )]+  + [as (τ1 ), ās ar hs r (τ1 )]− , ār (τ2 )
= δ(τ1 − τ2 )δsr + δss hr s (τ1 )ar (τ1 ), ār (τ2 )
= δ(τ1 − τ2 )δsr + {hsr }ar (τ1 ), ār (τ2 ). (1.60)
It can be solved using the Fourier transformation15
1  iωn τ
as (τ1 )ār (τ2 ) = e as ār ; ωn , (1.61)
β n
with
(2n + 1)π
ωn = , (1.62)
β
where the odd number indicates that particles are fermions. Then Eq. (1.50)
becomes
(iωn δsr − hsr )ar ār ; ωn  = δsr . (1.63)
In this step, the matrix structure of the above should be carefully investi-
gated. Let us assume that the single-particle Hamiltonian is spin-diagonal:
 ↑ 
hsr 0
hsr = . (1.64)
0 h↓sr
It is preferable to introduce the flame diagonalizing each element:
h↑sr = (s|ii|h|i|iir) ↑ = s|i↑ i↑ i|r↑ . (1.65)
14 Modern Aspects of Superconductivity: Theory of Superconductivity

Then we have
1 1
as ār  = ⎛ ⎞
β n (s|ii|r)↑
0
⎝ iωn −i↑ ⎠
(s|ii|r)↓
0 iωn −i↓
⎛ ⎞
(s|ii|r)↑
0
1 ⎜⎜ iωn − i↑
⎟
⎟
=
β n ⎝ (s|ii|r)↓ ⎠
0
iωn − i↓
 
(s|in(i )i|r)↑ 0
= , (1.66)
0 (s|in(i )i|r)↓
where
1
n(i↑ ) = . (1.67)
1+ ei↑ /kB T
Another sophisticated way starts from the decomposed Hamiltonian
H 0 = ār σμ hμrs as . (1.68)
The commutator is evaluated as
[as , ār σμ hμr s as ] = δsr σ0 σμ hμr s as = σμ hμss as . (1.69)
The equation of motion
(iωn δss − σμ hμss )as , ār ; ωn  = σ0 δsr . (1.70)
In the matrix notation,
(iωn − σμ hμ )a, ā; ωn  = σ0
or
σ0
a, ā; ωn  = . (1.71)
iωn − σμ hμ
In the representation where h is diagonal,
r|h|s = r|ii|h|ii|s = i r|ii|s, (1.72)
the relation (1.71) becomes
(|ii|)(iωn + σμ μi )
a, ā; ωn  =
(iωn − σμ μi )(iωn + σμ μi )
(|ii|)(iωn + σμ μi )
= . (1.73)
(iωn )2 − (μi )2
 Theory of Superconductivity 15

Now
iωn + σμ μi iωn σμ μi
μ = +
(iωn )2 − (i )2 (iωn )2 − (i )2 (iωn )2 − (μi )2
μ

   
1 1 1 σμ 1 1
= + + −
2 iωn − μi iωn + μi 2 iωn − μi iωn + μi
 
1 1 1
= + (summing ωn )
2 iωn + μi iωn + μi
1 σμ
→ (n(μi ) + n(−μi )) + (n(μi ) − n(−μi )). (1.74)
2 2
We evaluate
 
σ0 μ μ σμ μ μ
G↑rs ↑
= Trσ r|ii|s (n(i ) + n(−i )) + (n(i ) − n(−i ))
2 2
= [r|ii|s]↑ n(↑i ). (1.75)
We then have
1 ↑ − 1  r|ii|s
Grs (0 ) = = r|in(↑i )i|s. (1.76)
β n β n iωn − ↑i
This relation holds for both ↑ and ↑, and we recover the result (1.66).
Along the way, we give the energy expression for Eq. (1.51):
E 0 = Tr(hrs Gsr )
  
h↑rs 0 s|in(↑i )i|r 0
= Tr ↓ . (1.77)
0 hrs 0 s|in(↓i )i|r
It may be needless to present another illustration:
E 0 = Tr(hrs Gsr ) = Tr(σμ hμrs )(συ Gυsr ). (1.78)
The result is meaningful if σμ συ = σ0 , which leads to, in the present case,
σμ = συ = σ↑ , or σμ = συ = σ↓ .
Such manipulations will be used in the later investigation.

1.5 Interacting
In this chapter, the electron–electron interactions are taken into account,
and we will discuss how they lead to the superconducting state. The
16 Modern Aspects of Superconductivity: Theory of Superconductivity

Hamiltonians given in Eq. (1.54) are

1
H dir = (ār σa as )vrs;tu
ab
(āt σb au ) (a, b = ↑, ↓),
2
1
H ex = − (ār σa as )vrs;ut
aa
(āt σa au ) (a = ↑, ↓),
2
1 −+
+−
H sup = {(ār σ+ as )vrs;tu (āt σ− au ) + (ār σ− as )vrs;tu (āt σ+ au )}.
2

These are written in the mean-field approximation. The estimate beyond

this approximation is not the case of the present consideration. We have
H 0 = hars ār σa ar (a = ↑, ↓),
H dir = (ār σa as )vrs;tu
ab ā σb a  = (ā σa a )D a:dir
t u r s rs (a, b = ↑, ↓),
H ex = −(ār σa a s )vrs;ut āt σ au  = (ār σ as )Drs
aa b a a:ex (a = ↑, ↓),
(1.79)
−+
H sup = (ār σ+ as )vrs;ut
+−
āt σ− au  + (ār σ− as )vrs;ut āt σ+ au 
−:sup
= (ār σ+ as )Drs + ār σ− as )Drs
+:sup
,
where
a:dir
Drs ab
= vrs;tu āt σb au ,
a:ex
Drs = −vrs;ut
ab
āt σb au δab , (1.80)
+:sup
Drs +−
= vrs;tu āt σ− au ,
where · · · implies the ground-state average, which was actually obtained
with wave functions in the previous step during the self-consistent field
(SCF) calculations.
These Hamiltonians are classified into two kinds called modes — the
normal many-electron problem and that for the superconductivity:
H = H norm + H sup , (1.81)
where
H norm = (ār σa as )(hars δrs + Drs
a:dir + D a:ex ),
rs
∓:sup
(1.82)
H sup = ār σ± as Drs .
The main difference between the two is that in the former we have the
single-particle Hamiltonian, and in the latter we do not. Various Drs are
complicated, but they are merely the c-numbers in this treatment. The
propagator in question in Eq. (1.57) is presented here again:
 
Grs (τ) = −Trρ̂ (θ(τ)ar (τ1 )a+
s (τ2 ) − θ(−τ)as (τ2 )ar (τ1 )
+

= ar (τ1 )a+

s (τ2 ).
 Theory of Superconductivity 17

The equation of motion can be read as

∂τ1 as (τ1 )ār (τ2 )
= δ(τ1 − τ2 )[as (τ1 ), ār (τ2 )]+  + [as (τ1 ), ās ar Hsa r (τ1 )]− , ār (τ2 )
(a; norm, sup)
= δ(τ1 − τ2 )δsr + Dsr
a
 σ ar  (τ1 ), ār (τ2 ).
a
(1.83)
Making the Fourier transformation with respect to τ = τ2 − τ2 gives
(iωn − Drs
a
 σ )Gs r = δsr
a
(1.84)
or, in the matrix form,
1 iωn + Da σa
G(ωn ) = = . (1.85)
iωn − Da σa (iωn )2 − (Da )2
Note that D a consists of the single-electron part and the two-electron inter-
action term which involves another mate ρatu combined with the propagator
āt σa au :

Da (x) = ha (x) + dx v(x − x )ρbtu (x )σb āt at . (1.86)

However, the mean-field approximation makes this as if it were the single-

electron interaction. A few comments will be given about the matrix char-
acter of G. This is a big matrix with site indices, and each element is a 2 × 2
matrix in the spin space. The index a characterizes the mode of the mean-
field potential. All the modes are independent of each other and are individ-
ually diagonalized. We now introduce a flame, in which these are diagonal:
ia |Da |ia  = ηai . (1.87)
Look at the right-hand side of Eq. (1.85) and remember the Einstein con-
vention that repeated indices imply summation:
(D a )2 = (η↑ )2 + (η↓ )2 + (η+ )2 + (η− )2 = (ηnorm )2 + (ηsup )2 , (1.88)
where the second line is in a simple notation. Then we have
|ia (iωn + ia |Da |ia σa )ia |
G(ωn ) = norm + ηsup )} .
{iωn − (ηnorm
i + ηsup
i )}{iωn + (ηi i
 
a 1 1 1
= |i  sup + norm + ηsup ) i |
a
2 iωn − (ηnormi + ηi ) iω n + (ηi i
Diia σa
+ |ia 
2(ηnorm
i + ηsup
i )
 
1 1
× − ia |. (1.89)
iωn − (ηnorm
i + ηsup
i ) iωn − (ηnorm
i + ηsup
i )
18 Modern Aspects of Superconductivity: Theory of Superconductivity

Taking the (r, s) matrix elements and the mode c which is achieved by the
operation,
Trσc Grs = Gc rs, (1.90)
we select the terms on the right-hand side with the mode c satisfying
Trσc σa = 1. (1.91)
Otherwise, the Tr operation leads to the vanishing result.
Carrying out the summation over ωn , we get
1 c
Gcrs (0− ) = G (ωn )
β n rs
 
1 c a 1 1 1
= r |i  +
β n 2 iωn − (ηnorm
i + ηsup
i ) iωn + (ηnorm
i + ηsup
i )

1 c a Diia
× ia |sc  + r |i 
β n 2(ηnorm
i + ηsup
i )
 
1 1
× − ia |sc 
iωn − (ηnorm
i + ηsup
i ) iωn + (ηnorm
i + ηsup
i )
1
= r c |ia  (n(ηnorm
i + ηsup norm
i ) + n(−ηi − ηsup
i ))i |s 
a c
2
Diia
+ r c |ia 
2(ηnorm
i + ηsup
i )
× (n(ηnorm
i + ηsup
i ) − n(−ηi
norm
− ηsup
i )) i |s 
a c
 
1 Diia ηi
= δrs − r |i 
c a
sup tanh ia |sc , (1.92)
2 2(ηnorm
i + ηi ) 2kB T
where
1
n(ηai ) = a (1.93)
1+ eηi /kBT
and
n(η) + n(−η) = 1,
 
η
n(η) − n(−η) = −tanh .
2kB T
In the estimation of matrix elements, the chemical potential which has been
disregarded up to now is taken into account. Namely, the Hamiltonian has
an additive term −μār ar , which causes
ηai → ηai − μ < 0, while −ηai → −(ηai − μ) > 0.
 Theory of Superconductivity 19

The mean-field potentials are carefully treated. In modes with ↑ and ↓,

we have the nonvanishing single-particle parts, h↑rs δrs and h↓rs δrs , which
are usually negative. However, for superconducting modes, h± rs = 0, and
the chemical potential is lost for the same reason. The latter may be
closely related to the fact that the number of particles is not con-
served in a superconductor. These circumstances are crucial for the super-
conducting mode.

1.5.1 Unrestricted Hartree–Fock (HF)

Let us review the SCF procedure. We now discuss the ordinary many-
electron system. As an example, the propagator with the up spin,
Eq. (1.89), is
 a 
1 Diia ηi
G↑rs (0− ) = r ↑ |s↑  − r ↑ |ia  sup tanh ia |s↑ ,
2 2(ηnorm
i + ηi ) 2kB T
(1.94)
where r ↑ |s↑  is the overlap integral between the sites r and s and approxi-
mately vanishes, and
Diia = h↑ii + vii;tu
↑a ↑↑
Gatu − vii;tu G↑tu .
(a :↑, ↓) (1.95)
Note that  
ηai
tan h < 0, since ηai < 0. (1.96)
2kB T
Look at the potential of the mode ↑:

D (x) = h (x) + dx v(x − x )ρbtu (x )σb āt at .
↑ ↑
(1.97)

In this case, there is h↑ (x), whose matrix element should be negative, so

that even if the matrix elements of the second terms are positive, the (r, s)
matrix element of D ↑ (x) is probably negative. This is usually the case in
atoms, molecules and solids. In evaluating D↑ (x), the propagators (wave
functions) of all other modes are required. In this respect, Eq. (1.94) is
the self-consistent relation between propagators. Usually, the self-consistent
relation between wave functions is given in a such a way that, at the begin-
ning, the total energy is given by the potentials given in terms of tentatively
approximated wave functions, and the new approximate wave functions in
the next step are obtained by optimizing the total energy. This procedure
is lacking in the present consideration.
20 Modern Aspects of Superconductivity: Theory of Superconductivity

1.5.2 Gap equation for superconductivity

In the case of superconductivity, since σ+ (σ− ) is traceless, the first term of
Eq. (1.92) vanishes, and also h+ = 0. Now Eq. (1.92) can be read as
   + 
+−
  −)
+ −
vii;tu a¯t a u ηi
Grs (0 ) = − r |i 
+ +
sup i |s  tanh
+ +
2(ηinorm + ηi ) 2kB T
 +− −
  + 
vii;tu Gtu ηi
= − r |i 
+ +
sup i |s  tanh
+ +
. (1.98)
2(ηinorm + ηi ) 2kB T

This complicated equation gives the relation between G+ −

rs and Grs , both
referring to the superconductivity, and is called the gap equation. A few
points should be presented. While selecting the superconducting mode, we
used
    
+ − 0 1 0 0 1 0
Tr(σ σ ) = Tr = Tr = 1, (1.99)
0 0 1 0 0 0
so that
+−
Tr(σ+ v +a σa ) = vii;tu . (1.100)
The relation (1.98) yields
 +−

vii;tu
− r + |i+  i+ |s+  > 0. (1.101)
2(ηnorm
i + ηsup
i )

As has been mentioned previously, we have no chemical potential in the

superconducting state, so that η+ +−
i is positive, the same as vii;tu is. Therefore,
it is required that r + |i+ i+ |s+  be negative so that Eq. (1.101) can hold.
This is really possible, as will be mentioned below. Thus, we are not at all
concerned with the electron–phonon coupling. Let us perform the successive
approximations as

occ
r + |i+ vii;tu
+− + +
i |s  tan h(η+
i /2kB T ) ≈ r + |i+ vii;tu
+− + +
i |s  (T → 0)
i
+−
= qrs vii;tu (<0). (1.102)
+−
Here, vii;tu is the electron–electron interaction between two electron den-
sities and is certainly positive (repulsive). However, the bond order is not
necessarily so, but q14 < 0 in the following example. The last relation was
usually assumed at the beginning of the superconductivity theory.
 Theory of Superconductivity 21

We thus obtain the condition for the superconducting state to appear; it

is purely electronic and is apart from the electron–phonon coupling mecha-
nism. Actually, for the chain molecule of four carbon atoms, called butadi-
ene, the matrix qrs (r, s) = 1 − 4 is
⎛ ⎞
1.000 0.894 0.000 −0.447
⎜ 1.000 0.447 0.000⎟
{qrs } = ⎜
⎝
⎟. (1.103)
1.000 0.894⎠
1.000
We can clearly see that
+−
v11;14 q14 < 0. (1.104)

1.6 Illustrative Example, Critical Temperature

The gap equation (1.98) is, in appearance, considerably different from the
usual one. We rewrite this equation in a form similar to the usual one. To
this end, we adopt, as an example, a polyacene high polymer. Benzene,
naphthalene, anthracene, etc. are a series of polyacene, shown in Fig 1.1.
Here, the unit cell which is the butadiene molecule is in the dotted rectangle
numbered by n. The interactions t1 and t2 are given for the corresponding
bonds.

1.6.1 Bond alternation

At the beginning, we discuss the bond alternation or the Peierls instability of
these molecules. The infinite chain of acetylene, the so-called polyacetylene,
has the bond alternation, i.e., the long and short bonds do not lose their
memories in the limit where an infinite chain has been formed. This is
popular with chemists,19 but physicists call it Peierls distortion.20 The bond
alternation causes the gap between the conduction and valence bands. It has

polyacene
t1

n n+1
t2

unit cell

Fig. 1.1 Polyacene.

22 Modern Aspects of Superconductivity: Theory of Superconductivity

been said that this discontinuity prevents the superconducting phase from
arising. However, this is an old-fashioned assertion and now seems sceptical.
In the Hückel theory, the interaction matrix elements are put as t1 for
the shorter bond and t2 for the longer bond:
H 0 = −t1 (a+ 2n a1n + a1n a2n ) − t2 (a1n+1 a2n + a2n an+1 ),
+ + +
(1.105)
where we consider the neighboring unit cells numbered n and n + 1, and
each cell has two kinds of bonds. The transfer  integrals are parametrized as
δ
t1 = t − δ, t2 = t + δ δ > 0,  1 , (1.106)
t
and then the Hamiltonian is easily diagonalized as
k = ±[t2 (1 + cos k) + δ2 (1 − cos k)]1/2 , (1.107)
where + and − correspond to the conduction and valence bands, respec-
tively. We are interested in the features at the zone boundary, k = π:
cπ = 2δ, vπ = −2δ. (1.108)
Here, the superscripts c and v indicate the conduction and valence bands,
respectively. When δ = 0, certainly we have the gap, and if δ = 0, the two
bands continuously join into a single band called the half-filled band.
Next, we turn to the polyacene, whose unit cell is the butadiene molecule.
In this case, we obtain four bands:
 1
ck = [t3 + (t23 + 4|t̃k |2 )1/2 ] = −vk ,
2
(1.109)
1
k = [−t3 + (t3 + 4|t̃k | ) ] = −k ,
c 2 2 1/2 v
2
where
t̃k = t1 + t2 eik . (1.110)
The usual pairing property in alternant hydrocarbons is also seen in this
case. Employing the parametrization (1.106) gives, at k = π,
 2
1 δ
vπ = [t3 + (t23 + 16δ2 )1/2 ] ≈ 4t3 . (1.111)
2 t1
When δ = 0 (without bond alternation), we have
1
vk = {t3 − [t23 + 8t2 (1 + cos k)]1/2 }. (1.112)
2
Consider single-particle states. If δ = 0 and k = π we have, from Eq. (1.110),
t̃ = 0, so that the amplitudes at sites 2 and 5 vanish. Therefore, when δ = 0,
we get at k = π
1
|ψvk  = (a+ − a+ 4k )|0,
2 1k
(1.113)
1 +
|ψk  = (a1k + a4k )|0.
c +
2
 Theory of Superconductivity 23

It is seen that |ψvk  is antisymmetric about the C2v symmetry axis, and |ψvk 
is symmetric. Thus, v and c bands are not continuous at k = π.

1.6.2 Deformation energy

For these systems, let us study whether the bond alternation is energetically
favorable or not. We assume that the energy gain due to the bond alternation
mainly contributes to the highest valence band energy, vk .
The case of polyacetylene. The energy gain ΔE is
 2π
dk v
ΔE = ( − vk (0)), (1.114)
0 2π k
where the second term refers to the case without bond alternation
(δ = 0). In Eq. (1.107), we shift the integration origin from 0 to π, then
approximate
p2
cos k = cos(π + p) ≈ −1 + .
2
For small p, we obtain

dk 
2π !
ΔE = − (tp)2 + 4δ2 − tp
0 2π
 2  
2t δ δ
= ln (<0). (1.115)
π t πt
For polyacene, similar treatment of vk of Eq. (1.109) leads to
⎛  " ⎞
 2  2 #1/2
1 δ δ
vk = ⎝1 − 1 + 16 + 4p2 1 − ⎠, (1.116)
2 t t

and then the deformation energy becomes

 2    2
δ π 2 δ
ΔE = t − ln 4π ≈ −0.83t . (1.117)
t 4 π t
The bond alternation looks favorable for both cases. However, when the
effect of the σ bond is taken into account, this almost cancels out the sta-
bilization energy of the π system in the case of polyacene. On the other
hand, this is not the case for polyacetylene due to the singular term in the
relation (1.117).
Therefore, in what follows, by concentrating on polyacene, we are free
from the bond alternation.
24 Modern Aspects of Superconductivity: Theory of Superconductivity

1.6.3 Polyacene, gap equation, critical temperature

The unit cell of polyacene is a butadiene molecule composed of four 2pπ car-
bon atoms. The Hamiltonian in the tight-binding approximation is given as
H 0 = t(a+ + + +
n1 an2 + an2 an3 + an4 an3 + H.c.)

+ t(a+ + +
n+1,1 an2 + an+1,4 an3 + H.c.), (1.118)
where the second line connects the unit cells n and n + 1.21
The band structure of levels and the linear combination of atomic orbitals
(LCAO) coefficients U are
t
1 (k) = {1 + s(k)},
2
t
2 (k) = − {1 − s(k)},
2
(1.119)
t
3 (k) = {1 − s(k)},
2
t
4 (k) = − {1 + s(k)},
2
with
√
s(k) = 9 + 8 cos k, (1.120)
⎛ ⎞
N4 N3 N2 N1
⎜−N4 4 /t̃k −N3 3 /t̃k −N2 2 /t̃k −N1 1 /t̃k ⎟
U =⎜
⎝ N4 4 /t̃k
⎟, (1.121)
−N3 3 /t̃k N2 2 /t̃k −N1 1 /t̃k ⎠
−N4 N3 −N2 N1
where, for example,
|t̃k |2 1 eik/2
N12 = , = . (1.122)
2(|t̃k |2 + 21 ) t̃k 2t cos(k/2)
At this stage, we have completed, in principle, the usual many-electron
problem. The mean-field approximation makes the interaction problem a
one-particle problem even though the SCF treatment is required at each
step. In other words, from the viewpoint of the Hückel theory, the spin-
diagonal parts provide the answer. On the other hand, the spin-off diagonal
part, which means less in the case without electron–electron interactions, is
responsible for the superconductivity.
Up to the previous chapter, the problem had been investigated in the
site representation. That is to say, the system is considered to be composed
 Theory of Superconductivity 25

of N sites. However, the real substance is formed from unit cells, so that the
system is a repetition of the unit cell. The usual band theory of polyacene
has thus been completed at this stage.
We turn to the onset of superconductivity. In this case, the single-particle
approach is almost meaningless, but the pair state — say, the wave function
of a Cooper pair — should be investigated. For this purpose, the Green
function of a Cooper pair is most preferable. The gap equation (1.98) is
nothing but the SCF equation for a Cooper pair.
The electronic structure of the single butadiene molecule referring to
k = 0 is suggestive. Let the total number of sites of high polymer polyacene
be N . The number of sites in the unit cell is four. Then N = 4n, with
the number of unit cells n. The numbers 1–4 are the band indices; then
we have the chemical potential between the 2n level and 3n level. The
Cooper pair should be the hole pair of the 2n level indicated by mode (−)
or the particle pair of the 3n level. The discussion is confined to these levels
in solid-state physics, even if the interaction with other bands is taken into
account.
The 2n and 2n + 1 levels are called highest occupied molecular
orbital (HOMO) and lowest unoccupied molecular orbital (LUMO). These
features and behaviors are not so far or qualitatively the same as those
for k = 0.
The electronic structure of a single butadiene molecule is suggestive. The
levels and the bond orders are (the unit = t)
1 −1.618
2 −0.618
(1.123)
3 0.618
4 1.618.
Here, 1 and 2 are occupied (valence) levels, while 3 and 4 are unoccupied
(conduction) ones. Let the probability amplitude, with which the electron
on level i is found at the site r, be r|i, then the bond order qrs is defined
as (at the zero temperature)

occ
qrs = r|ii|s, (1.124)
i
where the summation includes the spin state. In the determination of
the attractive electron–electron interaction, the bond order is of crucial
importance.
26 Modern Aspects of Superconductivity: Theory of Superconductivity

The bond orders are

Site
1 1.000 0.894 0.000 −0.447
2 0.894 1.000 0.447 0.000 (1.125)
3 0.000 0.447 1.000 0.894
4 −0.447 0.000 0.894 1.000.
We indicate the negative value of q14 . In solid-state physics, the discussion
is concentrated on the highest valence band or the lowest conduction band.
In the quantum-chemical language, the partial bond orders referring only
H , which seem not so far from those with k, are listed as
to level 2, qrs
Site
1 0.362 0.224 −0.224 −0.362
2 0.224 0.138 −0.138 −0.224 (1.126)
3 −0.224 −0.138 0.138 0.224
4 −0.362 −0.224 0.224 0.362.
H.
We indicate the negative value of q14
H , it is multiplied by
If we want to look at the band structure, such as qrs
the third column of Eq. (1.121).
Based on these results, we may perform successive approximations or
simplifications:

(1) The denominator of Eq. (1.98) is the sum of the ordinary Hartree–Fock
energy and that of the superconducting state, and the latter is
+− −
ηsup
i = −vii;tu Gtu . (1.127)

The total energy is obtained by summing

ηi = ηnorm
i + ηsup
i (1.128)

with respect to the chemical potential.

(2) Each level with ηi has really a band structure and is then written as
ηi;k , by stressing the band structure by k with the band index i. Then
i is put to be the highest occupied level, and the integration over k is
carried out. The bond orders thus obtained are approximated by the
partial bond orders qrs H.

(3) The electron–electron interaction, which is effectively negative due to

the chemical structure of species, can be simply written for g < 0:
−+
g = qrs vrs;tu . (1.129)
 Theory of Superconductivity 27

(4) The energy interval establishing the superconductivity in the BCS the-
ory is related to the electron–phonon interaction, ωD (Debye fre-
quency). In the present theory, it is replaced by the band width nearly
equal to |g|.
(5) Assuming that
− −
G+ +
s↑,r↓ (0 ) = Gt↑,u↓ (0 ) (1.130)
in Eq. (1.98), we have
 
|g| 
dξ ξ + ηsup
1 = gN (0) tanh , (1.131)
0 ξ + ηsup 2kB T
where ξ = ηnorm . The critical temperature TC is determined by the
condition that ηsup vanishes at this temperature. The integration in
Eq. (1.131) is carried out as usual. Approximating
 
d3 k
= N (0) dξ
(2π)3
gives
 |g|    Z
dξ ξ dz |g|
tanh = tanh z, z=
0 ξ 2kB T 0 z 2kB T
 ∞
= [ln z tanh z]Z
0 − dz ln z sech2 z, (1.132)
0
where the upper limit in the second integration is replaced by ∞, which
makes it integrable15 :
 ∞
4eγ
dz ln z sec h2 z = − ln ; γ is the Euler constant.
0 π
A simple rearrangement of the result gives
2eγ
kB TC = |g|e−1/N (0)g ∼ 1.13|g|e−/N (0)g. (1.133)
π
The result is entirely the same as the current one. However, since it is
probable that
|g|
∼ 100, (1.134)
kB T
the critical temperature is, at most, enhanced by this value, even though it
is considerably reduced by the factor e−1/N (0)g .

1.6.4 Conclusion
As has been presented, superconductivity is not a too-complicated phe-
nomenon. If we employ the spinor representation, superconductivity is
28 Modern Aspects of Superconductivity: Theory of Superconductivity

described in parallel with the normal electronic processes. If we find, in

the copper oxide complex, the four-site unit as a butadiene molecule, it
might be the origin of the superconductivity of this material. We think that
it is not so difficult a problem for quantum chemists.

1.7 Linear Response Magnetic Resonance

in Normal and Superconducting Species;
Spin–lattice Relaxation Time
1.7.1 Introduction
The theory of linear response is one of the main topics in solid-state physics,
and its application to superconductivity is also a fundamental problem.
Perhaps the most important problem is the Meissner effect. However, we
are now interested in the magnetic resonance, whose main theme should
concern the relaxation time. Let us discuss the spin–lattice relaxation time
T1 in the nuclear magnetic resonance. An elegant theory has been provided
by Kubo and Tomita,22 and revised by us with the temperature Green
function.23
The spin–lattice relaxation time T1 increases remarkably in a supercon-
ductor just below the critical temperature. This is explained by the BCS
pairing theory and is said to be its brilliant triumph.3, 24, 25 The external
perturbation acting on the electron is written as
H  = Bkσ,k σ c+
kσ ck  σ , (1.135)
where c+ kσ , ckσ , etc. are the creation and annihilation operators of an electron
in the normal phase, and Bkσ,k σ is the matrix element of the perturbation
operator between the ordinary one-electron states in the normal phase. The
problem is as follows: If we rewrite it in terms of operators of a quasiparticle
in the superconducting phase, what will arise?
The time reversal to the above, B−k −σ ,−k−σ , has the same absolute
value, but the phase is the same or the reverse.
It is possible to classify as follows:
1± . The spin flip-flop does not arise:

kσ ck  σ ± c−k  −σ c−k−σ ).
Bkσ,k σ (c+ +

2± . The spin flip-flop does arise:

Bkσ,k −σ (c+ +
kσ ck  −σ ± c−k  σ c−k−σ ).
 Theory of Superconductivity 29

As seen above, the theory implicitly assumes that the system is a perfect
crystal and is described by the single wave vector k. The positive and nega-
tive signs of k indicate waves propagating from the vertex or off the vertex.
Before entering into the discussion about the relaxation time of the magnetic
resonance, we briefly review the Bogoliubov theory.3, 10, 15 The Bogoliubov
transformation defines a quasiparticle responsible for the superconductiv-
ity as
    
γk↑ uk −vk ck↑
= (1.136)
γ+
−k↓
vk uk c+−k↓
with
u2k − vk2 = 1.
The spirit of the Bogoliubov transformation is to mix the operators ck↑ and
c+
−k↓ , which are different in spin (as to the wave number, this mixing is not so
serious) and not mixed in the normal situation. The quasiparticle yields the
new ground state near the chemical potential. The stabilization energy thus
obtained is called the gap energy, Δk . What we have done in Sec. 1.3 is a sub-
stantial understanding of this reason. However, if we want to make the ±k
distinction meaningful, it is natural to adopt the four-component spinor —
say, the extended Nambu spinor (perhaps spurious).26 For ordinary states,
⎛ ⎞
ck↑
  ⎜ + ⎟
⎜c−k↓ ⎟
ck = ck↑ c−k↓ c−k↑ ck↓ , ck = ⎜
+ + +
⎜c+ ⎟,
⎟ (1.137)
⎝ −k↑ ⎠
ck↓
and, for the superconducting state,
⎛ ⎞
γk↑
  ⎜ + ⎟
⎜γ−k↓ ⎟
γ+
k = γ+
k↑ γ−k↓ γ−k↑ γ+
k↓ , γk = ⎜
⎜γ+ ⎟.
⎟ (1.138)
⎝ −k↑ ⎠
γk↓
These are connected with each other by the Bogoliubov transformation as
γk = Uk ck , γ+ +
k = ck Uk , (1.139)
where
   
uk 0 uk vk
Uk = , with uk = . (1.140)
0 u+
k −vk u+
k
Careful manipulation is instructive.
30 Modern Aspects of Superconductivity: Theory of Superconductivity

Case 1+ :
kσ ck  σ + c−k  −σ c−k−σ ) = ck Σ ck  = γk Uk Σ Uk  γk  ,
(c+ + + 3 + 3 +
(1.141)
where  
σ3
Σ =3
. (1.142)
−σ3
This is the 4 × 4 matrix manipulation; however, it is enough to note the
upper half of the result.
The upper half of Eq. (1.139) is
  
+ 1 0 ck ↑
(ck↑ c−k↓ )
0 −1 c+
−k  ↓
     
+ u v 1 0 u −v  γk ↑
= (γk↑ γ−k↓ )
−v u 0 −1 v u γ+−k  ↓
   
uu − vv  −uv  − vu γk ↑
= (γ+ γ−k↓ ) , (1.143)
k↑ −vu − uv  
vv − uu 
γ+−k  ↓
where u and v are the abbreviations of uk and vk , respectively, while u and
v  are those of uk and vk .
Case 1− :

(c+ + + + +
kσ ck  σ − c−k  −σ c−k−σ ) = ck 1ck  = γk Uk 1Uk  γk  . (1.144)
σ
The upper half of the above is
   uu + vv   
+ −uv  + vu γk ↑
γk↑ γ−k↓ . (1.145)
−vu + uv  vv  + uu γ+
−k  ↓
As is shown clearly, the original term is transformed in the quasiparticle
representation into a combination of the scattering term with the diagonal
element in the uk matrix and the creation or annihilation of a pair with the
off-diagonal element of u. These matrix elements are called the coherent
factors.
Let us turn to the case where the spin flip-flop is allowed.
Case 2+ :
kσ ck  −σ + c−k  σ c−k−σ = ck Σ ck  = γk Uk Σ Uk  γk 
c+ + + J + J +
(1.146)
with
⎛ ⎞
1
⎜ −1 ⎟
ΣJ = ⎜
⎝ −1
⎟.
⎠ (1.147)
1
 Theory of Superconductivity 31

This relation connects the left half of γ+

k and the lower half of γk  , by giving

   uu + vv   
+ uv  − vu γ−k ↑
Eq. (1.147) = γk↑ γ−k↓ . (1.148)
−vu + uv  −vv  − uu γ+
k ↓

Case 2− :

kσ ck  −σ − c−k  σ c−k−σ = ck Jck  = γk Uk JUk  γk  ,

c+ + + + +
(1.149)

with
⎛ ⎞
1
⎜ 1 ⎟
J=⎜
⎝ 1
⎟.
⎠ (1.150)
1

This relation also connects the left half of γ+

k and the lower half of γk  , and
we have
  
+ uu − vv  uv  + vu γ−k ↑
Eq. (1.150) = (γk↑ γ−k↓ ) . (1.151)
−vu − uv  −vv  + uu γ+
k ↓

Note that, at present, the scattering terms are off-diagonal, and the cre-
ation and annihilation terms are diagonal.
By the use of the relations3, 10, 15
   
1 k 1 k
2
uk = 1+ 2
vk = 1− ,
2 Ek 2 Ek (1.152)
Ek = k + k
2 2 2

(k is the gap energy), the coherent factors are expressed substantially as
 
1 k k k k 
1± : (uu ∓ vv  )2 = 1+ ∓ ;
2 Ek Ek Ek Ek
  (1.153)
  2 1 k k 
2± : (uv ∓ vu ) = 1∓ .
2 Ek Ek
In case 1+ , we have the ultrasonic attenuation, while the electromagnetic
interaction is in case 2+ and the magnetic resonances are in case 2− .

1.7.2 T1 in NMR
A detailed analysis of the spin–lattice relaxation time T1 in the nuclear
magnetic resonance will be presented in the next section. Here, the results
32 Modern Aspects of Superconductivity: Theory of Superconductivity

are given briefly.

  
21 k k 
T1 ∼ |Bkk | 1+ nk (1 − nk )δ(Ek − Ek − ω), (1.154)

2 Ek Ek
kk
where ω is the applied radio frequency. By converting the summation
(explicitly shown) to the integration, and further by using the relation of
state densities,
N (E) dE = N () d, then
⎧
⎪
⎨ E
N (E) d (E > ),
= (E −  )
2 2 1/2
= (1.155)
N () dE ⎪
⎩0 (E < ),
we rewrite Eq. (1.152) for ω   as
 ∞  
1 2
T1 ∼ |B| N (0)
2 2
1+

2 E(E + ω)
E(E + ω)kB T (−∂n/∂E) dE
× , (1.156)
(E 2 − 2 )1/2 [(E + ω)2 − 2 ]1/2
where the coupling constant and the state density are replaced by their suit-
able averages. This integral is divergent. Therefore, T1 of a superconductor
is strongly enhanced just below the critical temperature. This phenomenon
was observed and explained by Slichter et al.24, 25 This was said to be one of
the brilliant victories of the BCS theory. However, it has been found, in the
recent experiments on the high-temperature superconductors or the copper-
oxide superconductors, that the T1 enhancement is lost. This phenomenon is
considered deeply connected with the mechanism of the high-temperature
superconductivity of these species, and it attracted the interest of many
investigators.27, 29 However, as far as we know, the theory of magnetic reso-
nance of a superconductor has been done almost entirely under the scheme
mentioned in this introduction. Then it will be preferable to develop the
theory of magnetic resonance in accordance with the sophisticated recent
theory of superconductivity.

1.7.3 Theory with Green’s function

Our idea is as follows: the algebra of electrons is related to their field oper-
ators. In the same way, we assume the field for nuclei. For example, the
creation operator for a nucleus a+KM yields the nuclear motion with K and
M , which are spatial and spin quantum numbers, respectively. The energy
 Theory of Superconductivity 33

spectrum of the propagator GKM (τ) = aKM (τ)a+ KM  gives the line shape
of the magnetic resonance.
The nuclear propagator GKM (τ) sees the electron sea, followed by the
electron excitation in the spin space. This gives the additional line width
of the nuclear magnetic resonance. The phenomenon looks like the vacuum
polarization in quantum electrodynamics. The self-energy part that has thus
arisen in the nuclear energy is the source of the line shape of the nuclear
magnetic resonance.23
The spin–lattice relaxation time of the nuclear spin I z is given by the
imaginary part of the magnetic susceptibility χzz , which is equal to (χ+− +
χ−+ )/2 in the spatially homogeneous system. Here, ± correspond to (I x ±
iI y )/2, respectively. The ensemble average of a change, δI + (t), is given by
the linear response theory as
 t
δI + (t) = i dt Tr{ρG [H ex (t ), I + (t)]− }, (1.157)
−∞
where ρG is the grand canonical statistical operator. However, the chemical
potential is not given explicitly, unless otherwise stated. The rotating mag-
netic field causing the magnetic transition is, assuming a single mode for
simplicity,
H ex (t) = HR (I + (t)eiωt + I − (t)e−iωt ). (1.158)
As has been said, the spin–lattice relaxation arises from the interaction
between the nuclear spin and the electron spin. In other words, the electron
spins play the role of a lattice system.
H  = γgβB F (R, r)I · S, (1.159)
where F (R, r) is a function of spatial coordinates of the nucleus, R, and
that of the electron, r. The term γ is the gyromagnetic ratio of the nucleus,
and g and βB are the g factor and the Bohr magneton of the electron,
respectively.
Now the second quantization of the above is carried out. First of all, the
orthonormalized wave function describing the nuclear behavior, |ξK (R)M ,
is introduced as
(HN + HM )|ξK (R)M  = (K + M )|ξK (R), M 
= KM |ξK (R, )M , (1.160)
with
KM = K + M,
34 Modern Aspects of Superconductivity: Theory of Superconductivity

where HN is the spatial part and HM is the Zeeman part. Then we have
I α → ξK (R)M |I α |ξK  (R)M  a+
KM aK  M 

= M |I α |M  a+
KM aK  M  δKK  . (1.161)
A similar equation is given for electrons,
(HS + Hm )φk (r)|m = (k + m)|φk (r), m = km |φk (r), m, (1.162)
where
km = k + m,
so that
H  → hγgβB M |I β |M  m|S α |m ξK (R)φk (r)|F (R, r)|ξK  (R)φk (r)
× (a+ +
KM aK  M  )(ckm ck  m ). (1.163)
If the nuclear motion is assumed to be that of a harmonic oscillator, a+KM
and aKM are the creation and annihilation operators of vibrational excita-
tions. When the nuclei carry noninteger spins, these are considered to obey
the Fermi statistics or satisfy the anticommutation relation
[a+
KM , aK  M  ]+ = δKK  δM M  . (1.164)
However, as will be seen in the following, this selection of the statistics is
not fatal for the theory. Needless to say, the operators for electrons satisfy
the anticommutation relations.
The change of I + in Eq. (1.157) can be written, in the interaction rep-
resentation, as (we retain the I − term in Eq. (1.158))
 t

δI (t) = iγHR
+
dt e−iωt
−∞

× Tr{ρG M − 1|I − |M M |I + |M − 1

 
K,M −1 (t )aK,M (t ), aK,M (t)aK,M −1 (t)]− }
× [a+ +

 t

= −γHR dt e−iωt DK,M −1,M (t − t)
−∞
 ∞ 
1
= − γHR e−iωt dse−iωs DK,M −1,M (s)
2 −∞ KM
1
= − γHR e−iωt DK,M −1,M (ω). (1.165)
2
From this result, the magnetic susceptibility of the present system is
γ
χ+− (ω) = − DK,M −1,M (ω). (1.166)
2
 Theory of Superconductivity 35

In the course of derivation, the matrix elements of spin operators are put to
be equal to 1, and then the Tr operation is carried out. Here, DK,M −1,M (s)
is a retarded Green function,
DK,M −1,M (s) = −iθ(s)Tr{ρG [a+ +
KM −1 (s)aKM (s), aKM aKM −1 ]− }, (1.167)

and DK,M −1,M (ω) is its Fourier transform. Now our problem is to estimate
this retarded function.
The retarded Green function is easily obtained by analytical continua-
tion from the Matsubara function (or the temperature Green function with
imaginary time τ) which is causal in τ,
 ' (!
DK,M −1,M (τ) = −Tr ρG T τ a+ K,M −1 (τ)aK,M (τ)a+
a
K,M K,M −1 , (1.168)
for which the Feynman diagram analysis is available.15

1.7.4 Noninteracting
Here, we deal with the case without the spin–lattice interaction. It might
be trivial; however, it seems instructive for the later investigation. By the
use of the simplified notation, · · · = Tr(ρG · · · ), the Green function in this
case is written as
+ +
DK,M
0
−1,M (τ) = T τ[aKM −1 (τ)aK,M (τ)aKM aKM −1 ]

= GKM
0 0
(τ)GKM −1 (−τ), (1.169)
where
GKM
0
(τ) = −T τ[aKM (τ)a+
KM ], (1.170)
and the corresponding Fourier transform of Eq. (1.169) is
1 0
DK,M
0
−1,M (ωn ) = GKM (υn )GKM
0
−1 (υn − ωn ), (1.171)
β υ
n

where
1
GKM
0
(υn ) = . (1.172)
iυn − KM
Therefore,
1 1 1
DK,M
0
−1,M (ωn ) = ·
β υ iυn − iωn − KM −1 iυn − KM
n
 
1 1 1 1
= −
β υ iυn − iωn − KM −1 iυn − KM iωn + KM −1 − KM
n

1
= [n(KM −1 ) − n(KM )] , (1.173)
iωn − KI
36 Modern Aspects of Superconductivity: Theory of Superconductivity

where
KI = KM − KM −1 ,
and it is noted that ωn is even, and that it does not matter in
obtaining the particle number. The retarded Green function is obtained
simply by replacing iωn with ω + iη (η is a positive infinitesimal). Thus, we
obtain
1
δI + (t) = e−iωt HR (n(K− ) − n(K+ )) , (1.174)
ω − KI + iη
where the radio frequency stimulating the resonance is rewritten by ω0 . The
magnetic susceptibility χ+− thus becomes
1
χ+− (ω) = γ(n(K− ) − n(K+ )) ,
ω − KI + iη
γ = e−iωt HR , (1.175)
whose imaginary part is
χ+− (ω) = −πγ(n(K− ) − n(K+ ))δ(ω − KI ). (1.176)
This gives the sharp δ function-type energy spectrum. We have no line width
or the relaxation time, and states are stationary.

1.7.5 Interacting; normal

In the interacting system, G 0 in Eq. (1.169) has to be replaced by G, includ-
ing the interaction, which in the present case is the spin–spin interaction
between nuclei and electrons, as has been given in Eq. (1.157),
1
DKM −1,M (ωn ) = GKM (υn )GKM −1 (υn − ωn ), (1.177)
β υ
n

where, for instance,

GKM (υn ) = [(GKM
0
(υn ))−1 − ΣKM (ω)]−1 , (1.178)
ω being the interaction energy.
Our procedure is as follows. The two Green functions with the self-energy
part are evaluated. Combining them gives the retarded Green function D
for estimating the magnetic susceptibility.
The most important (divergent) self-energy part of this self-energy is due
to the ring diagram shown in Fig. 1.2. The problem is to examine how the
 Theory of Superconductivity 37

ω
m+1
λ ρ+ω
M-1 ρ
m

M ω

Fig. 1.2 Self-energy part G(υn ).

energy of the nucleus propagator (unequal-dashed line) is changed by the

ring diagram of the electron (full line). That is to say,
G(υn )−1 = G(υn )0 + Σ(ω),
(1.179)
Σ(ω) = −|K|2 S(ρn )S(ρn + ω),
where S is the electron propagator, and the minus sign is due to the fermion
loop. The self-energy part includes the coupling terms, where
|K|2 = M |I + |M − 1m − 1|S − |m|2
× |χK (R)φk (r)|F (R, r)|φk (r)χK  (R)|2 , (1.180)
and the minus sign is due to a Fermion loop. The ring diagram is calculated
as follows:
Skm (ρn )Skm−1 (ρn + ω)
 1 1
= ·
ρn
iρn − km iρn + ω − km+1
 1 1

1
= −
ρ
iρn − km iρn + ω − km+1 ω − km+1 + km
n

1
= (n(km ) − n(km+1 )) . (1.181)
ω − km+1 + km
The lattice (electron) gets ω from a nucleus to lift the electron spin from m
to m + 1, and at the other vertex, the inverse process occurs. This looks like
the radiation process in the photochemistry. We thus have the self-energy
part of GK;M −1,M ,
1
ΣK;M −1,M (υn ) = |K|2 (n(km ) − n(km+1 )) , (1.182)
ω − ks
where
ks = km+1 − km .
38 Modern Aspects of Superconductivity: Theory of Superconductivity

Another propagator GK;M , has the self-energy part ΣK;M,M −1 (υn ). This
is built by replacing ρn + ω with ρn − ω in Eq. (1.181).
1
ΣK;M,M −1 (υn ) = |K|2 (n(km ) − n(km−1 ))
ω − km−1 + km
1
= |K|2 (n(km ) − n(km−1 ))
,
ω − ks
1
≈ |K|2 (n(km+1 ) − n(km )) . (1.183)
ω − ks
We now turn to the evaluation of the propagator DK;M −1,M (τ):
DK;M −1,M (ωn )
1
= GKM (υn )GKM −1 (υn − ωn )
β υ
n

1
= (iυn − KM − ΣKM − )−1 (iυn − iωn − KM −1 − ΣK;M,M −1 )−1
β υn
 
1 1 1
= −
β υn
iυn − iωn − KM −1 − ΣKM −1 iυn − KM − ΣKM
 
1
×
iωn − KM −1 − ΣKM −1 + KM + ΣKM
 
1
= (N (KM ) − N (KM −1 )) . (1.184)
iωn − KI + ΣKM − ΣKM −1
Here,
KI = KM − KM −1 ,
and the self-energy parts are disregarded in obtaining the particle density.
We can see that the additional terms in the denominator modify the line
shape.
If we put iωn → ω + iη, we can obtain the retarded Green function,
whose imaginary part gives the line shape:
 
1
D(ω) ∼
ω + iη − KI + ΣKM − ΣKM −1
   
1
= P − iπδ(ω − KI + ΣKM − ΣKM −1 )
ω − −ΣKM −1 KI
i(Im)
= (at the resonance point). (1.185)
(Re)2 + (Im)2
The line shape is now changed from the δ function type to the Lorentz type,
as expected.
 Theory of Superconductivity 39

1.8 Interacting; Superconductor

1.8.1 The extended Nambu spinor
Now, we investigate how the line shape obtained above is further modified
in a superconductor. The electron propagators in the previous section are
replaced by those in a superconductor. They have already been studied in
Sec. 1.3 and are presented here in a new fashion adequate for the following
investigation. As has been done by BCS, let us consider the attractive two-
body potential g, which is assumed constant for simplicity, and further keep
in mind the Cooper pair. The Hamiltonian
 
1 + +
Hel = kα ckα ckα + gckα c−kβ c−kβ ckα ,
+
(1.186)
2
where kα is the orbital energy, includes the Zeeman energy in the present
case.
Now we use the extended Nambu representation of Eq. (1.137),
⎛ ⎞
ckα
⎜ + ⎟  
⎜c−kβ ⎟
ck = ⎜⎜c+ ⎟
⎟, c+ = c+ c−kβ c−kα c+
k kα kβ (1.187)
⎝ −kα ⎠
ckβ
with the equal-time commutator:
[ck , c+
k  ]+ = 1δkk  . (1.188)
In these terms, the Hamiltonian is rewritten as
1 + −
Hel = k c+
k Σ ck + 2 g(ck Σ ck )(ck  Σ ck  ),
3 + +
(1.189)

where k is the diagonal matrix of γ :

⎛ ⎞
kα
⎜ −kβ ⎟
k = ⎜⎝
⎟.
⎠ (1.190)
−kα
kβ
Let us define
   + 
σ3 0 σ 0
Σ3 = , Σ+
= ,
0 −σ3 0 −σ−
 
σ− 0
Σ− = , (1.191)
0 −σ+
40 Modern Aspects of Superconductivity: Theory of Superconductivity

with
     
1 0 0 1 0 −i
σ =
3
, σ =
1
, σ =
2
,
0 −1 1 0 i 0
1 1
σ+ = (σ1 + iσ2 ), σ− = (σ1 − iσ2 ). (1.192)
2 2
As for the electron–electron interaction, only those for the Cooper pairs are
selected; namely, Σ+ selects a Cooper pair in the particle state, and Σ− in
the hole state.

1.8.2 Green’s function

Here, the discussions done in Sec. 1.4 are repeated briefly. The above
Hamiltonian is invariant under the scale transformation. Then we have a
current conservation, especially the charge conservation in the static state
(Noether’s theorem). If we have any quantity which does not commute
with this invariant charge, we can expect a phase transition (the Goldstone
theorem).
The charge proportional to

kγ ckγ  =
c+ c+
k Σ ck 
3
(1.193)
k>0

is invariant under the rotation about the Σ3 axis in the space spanned by
Σ3 , Σ+ and Σ− . Observing that
[Σ3 , Σ± ] = ±2Σ± (1.194)
suggests the phase transitions along the Σ± directions.
If we define
c̄ = c+ Σ3 ,
as has been done in Sec. 1.43, the discussions parallel to those there will be
possible in the following. However, this is not employed in this case.
The phase transition cannot be achieved by the perturbational approach,
but the effective Hamiltonian giving the phase transition should be included
at the beginning. For example, the modified Hamiltonian
−
k (k Σ + ρΣ + ηΣ )ck ,
H 0 = c+ 3 +

1  + + − −
(1.195)
H int = g {(ck Σ ck ) · (c+
k  Σ ck  ) − ck (ρΣ + ηΣ )ck },
+ +
2  kk
 Theory of Superconductivity 41

where ρ and η are the so-called gap energies which are assumed to be
independent of k for simplicity. Note that the Hamiltonian H 0 is already
symmetry-broken. Here, we adopt a conventional method. In what follows,
H int is neglected, i.e. we start from the Green function due to the effec-
tive Hamiltonian and then take the interaction (1.158) into account. At this
stage, we may expect the result that will be obtained in such a manner that
the normal and superconducting states contribute additively.
The temperature Green function with the imaginary time τ is defined as
G0kk (τ) = −Tτ [ck (τ)c+
k  ], (1.196)
where · · · = Tr{ρG · · · }. The equation of motion of G0kk is
∂τ G0kk (τ) = −∂τ [θ(τ)ck (τ)c+ +
k  (0) − θ(−τ)ck  (0)ck (τ)]

= δ(τ)[ck (τ), c+ +
k  (0)]+  + T τ[ck (τ), H ]− , ck (0)
0

= δkk + (k Σ3 + ρΣ+ + ηΣ− )G0kk (1.197)

In the course of the above derivation, the commutator in Eq. (6.65) was
used.
We make the Fourier transformation,
1  −iωn τ 0
G0kk (τ) = e Gkk (ωn ), (1.198)
β ω
n

where β = kB T and kB is the Boltzmann constant. Then the equation of

motion becomes
(iωn + k Σ3 + ρΣ+ + ηΣ− )G0kk (ωn ) = δkk . (1.199)
Namely,
1
G0kk (ωn ) =
iωn − k Σ3 − ρΣ+ − ηΣ−
 
iωn + k Σ3 + ρΣ+ + ηΣ−
= δkk , (1.200)
(iωn )2 − Ek2
where
Ek2 = 2k + ρη. (1.201)

1.8.3 Spin dynamics

The propagator in Eq. (1.200) is the 4 × 4 matrix. If we ignore ρ and η,
this is reduced to the normal propagator. We have already tried this
simple case.
42 Modern Aspects of Superconductivity: Theory of Superconductivity

Now we assume the terms responsible for the superconductivity as a

perturbation. Namely (S is the electron propagator),
iυn + k Σ3 + ρΣ+ + ηΣ−
S(υn ) =
(iυn )2 − Ek2
iυn + k Σ3 ρΣ+ + ηΣ−
= + + ··· . (1.202)
(iυn )2 + 2k (iυn )2 + 2k
It is helpful to separate this relation into components and to manipulate
each one individually. Noticing that
σ0 + σ3 = σ↑ + σ↓ ,
we have, for example by operating σ↑ followed by Tr,
iυn + k Σ3 + ρΣ+ + ηΣ− 1 ρ+η
S ↑ (υn ) = = + + ···
(iυn ) − Ek
2 2 iυn − k↑ (iυn )2 + 2k
 
1 ρ+η 1 1
= + − + ··· (1.203)
iυn − k↑ 2k↑ iυn − k↑ iυn + k↑
In order to get the self-energy part of the nuclear propagator, we have
to evaluate
1
Σk;M −1,M (ω) = K 2 S(υn )S(υn + ω). (1.204)
β υ
n

The first-order term has already been evaluated in Eq. (1.181). Then we
have to carry out the complicated manipulation due to the second term of
the last line in Eq. (1.203). However, the combinations other than those
satisfying the resonance condition ω ∼ ks [see Eq. (1.182)] should give a
small effect which is to be neglected.
We then consider that the procedures of the previous section need not
be repeated, and we are allowed to multiply the results there by the factor
(ρ + η)/2k↑ as the perturbing correction or the superconducting effect.

1.8.4 Conclusion
If we review the present investigations from the viewpoints of the line shape
problem in the magnetic resonance, three cases are clearly distinguished.
In the noninteracting case, the line shape is written in terms of the delta
function. In the interacting case of the normal phase, it is presented by
the Lorentz-like function, and in the superconducting phase it is further
multiplied by a coherent factor; whereas the statistical factors referring to
the nuclear states never change throughout.
 Theory of Superconductivity 43

For a superconductor, the present theory has almost nothing to give

more than the current one has done. However, the theory is not merely to
reproduce the experimental result, but to predict the mechanism hidden in
observations, in such a way that the manipulations reveal step by step the
working mechanism inside the matter. We might be satisfied with a slightly
deeper understanding of the superconductivity.
Among various opinions we give, as an example, Scalpino’s.29 He pointed
out three possibilities regarding the loss of the T1 enhancement in the copper
oxide superconductor:

(1) The d-wave single-particle density of states has logarithmic singularities,

rather than the square-root singularity for the s wave gap.
(2) The coherent factor for the quasiparticle scattering vanishes for k ∼
(π, π) for a dx2 −y2 gap.
(3) The inelastic scattering acts to suppress the peak just as for an s wave.

Scalpino had the opinion that the theory of superconductivity is already

so well furnished that other fundamental ideas beyond the original BCS one
are almost needless, except for some smart equipment. As the phenomena
observed in the copper oxide superconductor are rather qualitative and fairly
clear-cut, the explanation for them must be simple. It is expected that a
quantum-chemical speculation could make this possible.
Let us address scalpino’s opinion. The divergent character of Eq. (1.154)
seems a merely mathematical problem. The difference between Eq. (1.185)
and that (which will be obtained) for the superconducting case is the coher-
ent factor, (1.203). In the case of a BCS superconductor, it holds that
Ek  ρ so that the enhancement of the spin–lattice relaxation time T1 is
observed due to this factor, which leads to the superconductivity. However,
in the case of a high-temperature superconductor, ρ, η ∼ Ek , as has been
seen in the previous chapter, we cannot observe the sharp onset of super-
conductivity. Then we miss the coherent effect.
There is presumably a simpler reason why the T1 enhancement is not
observed. In copper oxide, electrons responsible for superconductivity are
probably the d electrons; then we have the vanishing interaction term if
it is the Fermi contact term between a nucleus and an electron — say,
F (R, r) ∼ δ(R − r) in Eq. (1.159).

1.9 Ginzburg–Landau Theory from the BCS Hamiltonian

The macroscopic quantum theory of superconductivity has been given by
Ginzburg and Landau.9 This looks rather phenomenological; however, since
44 Modern Aspects of Superconductivity: Theory of Superconductivity

a microscopic justification has been provided by Gorkov and others,16, 30

it has a substantial foundation. We also reviewed the GL theory, in the
introduction of Sec. 1.4, from the viewpoint of Landau’s general theory of
phase transitions. It is crucial that the Lagrangian of the system is written
as the fourth-order function of the order parameter Ψ, which is the electron
field. If the coefficients of the second- and fourth-order terms are suitably
chosen, the new ground state shapes a champagne bottle, or a Mexican hat
is built. If electrons moves on this flat route around the top, the derivative
of the orbital vanishes or the kinetic energy vanishes, which implies that
the wave function is rigid. We may say that the electron mass is effectively
zero. We thus have the current only due to the vector potential, i.e. the
diamagnetic current. This causes the Meissner effect.
The GL theory is quite useful for applications, since the microscopic the-
ory by itself is too complicated for manipulating large-scale problems. If we
can solve the GL equation for a real problem under an appropriate bound-
ary condition, various information on this system can be obtained.10 The
macroscopic wave function or GL order parameter Ψ is related to the gap
function and is understood as the field of Cooper pairs. The parameters in
the GL equation are also written in the microscopic terms or by the exper-
imental values. In Ref. 28, the GL function Ψ is derived directly from the
BCS Hamiltonian, not via the gap function or the anomalous Green function
related to the gap function. The electron–electron interaction composed of
the four fermion operators is changed by the Hubbard–Stratonovitch trans-
formation to an auxiliary complex boson field φ, in which electrons behave
as if they were free. Using the path-integral method, we can carry out the
integration up to the quadratic terms of the electron operators. If we care-
fully analyze the resulting effective Lagrangian for the boson field, we will
find that this boson field, which is described by a complex function, suggests
a phase transition; the condensation arises in particles described by the real
part of the boson field, while particles in the imaginary (or phase) part turn
out to be massless Goldstone bosons. It is then clear that the boson field
is to be the GL order parameter. In the course of analysis, the concept of
supersymmetry is effectively used.

1.9.1 BCS theory

We assume the classical field of an electron is described by the Grassmann
algebra or the anticommuting c-number; namely, the creation and annihi-
lation operators a∗k and ak are treated as anticommuting c-numbers. The
 Theory of Superconductivity 45

BCS Hamiltonian is written as

H = H0 + Hint ,

H0 = kσ a∗kσ akσ , (1.205)
kσ

Hint = −gk−k,−kk (a∗k↑ a∗−k↓ )(a−k ↓ ak ↑ ), gk−k,k−k > 0,
kk 
and 
−gk−k,−kk = dr1 dr2 χ∗k↑ (r1 )χ∗−k↓ (r2 )v(r1 , r2 )χ−k ↓ (r2 )χk ↑ (r1 ),
(1.206)
where v(r1 , r2 ) is the effective coupling giving an attractive character for the
electron–electron interaction. Here, the summation convention that repeated
indices imply summation is used. This is helpful in facilitating the manipu-
lation. We would consider that the BCS Hamiltonian is an attractive inter-
action between Cooper pairs rather than an attractive interaction between
electrons, (Figs. 1.3 and 1.4). In the following, it is assumed that kσ is
independent of spin and gk−k,k−k is independent not only of spin but also,
finally, of k and k .

Fig. 1.3 In the BCS model, superconductivity arises due to the attractive electron–
electron interaction which appears owing to the scattering by phonons.

Fig. 1.4 Excitation of the ground state of a Cooper pair demonstrates the presence of
the energy gap Δ, which is (very approximately) isotropic in the momentum space.
46 Modern Aspects of Superconductivity: Theory of Superconductivity

Let us investigate the partition function

Z = Tre−β(H0 +Hint ) . (1.207)
This is expressed by the use of the path-integral method.31, 32 The spirit of
the Feynman path-integral is sometimes written as follows: the (imaginary)
time interval 0 → β is sliced into numerous pieces, each being labeled by τp .
Since each slice is made arbitrarily small, the quantum effect arising from the
commutation relation can be neglected, so that the operators are regarded as
c-numbers in each slice. Instead, this c-number function can take any value
even in the small time slice. Connecting these precise values from 0 → β, we
can draw all of the paths in this interval. If we count the effects from all of
these paths, the quantum-mechanical result of the subject can be obtained.
However, this statement seems rather misleading.
Feynman’s path-integral is the third method of quantization.33 We begin
with the classical treatment, and then if we apply the path-integral proce-
dure to it, the quantum effect is certainly taken into account. This cor-
responds to the conceptual development from the geometric optics to the
physical optics. The Bose system (in the quantum-mechanical sense) is writ-
ten by an ordinary c-number, but the Fermi system should be described by
a Grassmann number. Thus, we do not worry about the commutation rela-
tions of field operators and obtain
Z = Tr e−βĤ
 ) N  β
= lim da∗ (τp ) da(τp )exp [(ȧ∗ (τp )a(τp ) − H(τp )], =
N →∞
p
N
p=1
  β
= Da∗ (τ)Da(τ) exp dτ{a∗kσ (τ)(−∂τ − kσ )akσ (τ)
0
∗ ∗
+ gk−k,−k k ak↑ (τ)a−k↓ (τ)a−k ↓ (τ)ak ↑ (τ)}, (1.208)
where
∞
)
∗
Da (τ)Da(τ) = da∗kσ (τp ) dakσ (τp ). (1.209)
p=1

1.9.2 Hubbard–Stratonovitch transformation

Difficulty lies in the quartic term of the electron–electron interaction. Let
us define
Bα∗ = a∗k↑ a∗−k↓ , Bα = a−k ↓ ak ↑ (1.210)
 Theory of Superconductivity 47

so as to write, in each time slice,

gk−k,−kk a∗k↑ a∗−k↓ a−k ↓ ak ↑ = gαα Bα∗ Bα . (1.211)
This is simplified by using the identity called the Hubbard–Stratonovitch
transformation.34 We introduce a complex boson field φ, since (Bα )∗ = Bα :
 ∞  ∞ √ √
∗ −πφ∗ φ ∗ ∗
1= dφ dφe = dφ∗ dφe−π(φ −i g/πB )(φ+i g/πB)
−∞
∞ √
−∞
∗ ∗ ∗ ∗
= dφ∗ dφe−πφ φ+i gπ(B φ−Bφ ) e−gB B ,
−∞
namely
 ∞
∗ φ+i√gπ(B ∗ φ−Bφ∗ ) ∗B
dφ∗ dφe−πφ = egB , (1.212)
−∞
where we have the definition
dφ∗ dφ = d(Re φ) d(Im φ).a (1.213)
We cannot apply this identity for the quantum-mechanical partition func-
tion because of the noncommutativity of the operators involved, but now
we have no trouble since, in the present path-integral treatment, operators
turn out to be c-numbers. Then Eq. (1.208) becomes

Z= Da∗ (τ)Da(τ) dφ∗ (τ) dφ(τ)
 β
× exp dτ{−a∗kσ (τ)−1 ∗
kσ (τ)akσ (τ) − πφα (τ)φα (τ)
0
√ √
+ i πgk−k,α a∗k↑ (τ)a∗−k↓ (τ)φα (τ) − i πgα,−k k a−k ↓ (τ)ak ↑ (τ)φ∗α (τ)},
(1.214)
with
Δ−1
kσ (τ) = −∂τ − kσ .

The essential feature of this expression is that the operator appearing

in the exponent in Eq. (1.214) is quadratic only in ak and a∗k , so that the
evaluation with respect to the fermion variables is similar to the evaluation
for the noninteracting system, in which particles are moving in an effective
boson field φk . Equation (1.214) shows that, inside the exponent, Z is a
weighted average of the second and third terms over the field φα .

a
Actually, dφR dφI = 1
2i
dφ∗ dφ. The constants are adsorbed in the normalization factor.
48 Modern Aspects of Superconductivity: Theory of Superconductivity

1.9.3 Fourier transform

The Fourier transformation with respect to τ is performed as
 
ak (τ) = eiωn τ ak (ωn ), φα (τ) = eiυn τ φα (υn ),
ωn υn

(2n + 1)π 2nπ

ωn = , υn = . (1.215)
β β
Note that ωn refers to a fermion, and υn to a boson.
Using the above and the relation
 β
dτei(ωn −ωm )τ = βδnm , (1.216)
0
we have

Z= Da∗ (ωn )Da(ωn )Dφ∗ (υn )Dφ(υn )

× exp {−βa∗kσ (ωn )kσ (iωn )−1 akσ (ωn ) − βπφ∗α (υn )φα (υn )
ω n υn
√
+ i πgk−k,α a∗k↑ (ωn + υn )a∗−k↓ (ω)φα (υn )
√
− i πgα,−k k ak ↓ (ωn )a−k ↑ (ωn + υn )φ∗α (υn )}, (1.217)
where kσ (iωn ) is Green’s function, given as
1
kσ (iωn ) = . (1.218)
iωn − kσ
We can observe the following in Eq. (1.217): the third line tells us that,
at the vertex indicated by the coupling gk−k,α , the boson field φα sinks,
and then the particle pair a∗k↑ a∗−k↓ arises. The last line displays the reverse
phenomenon. Note the energy conservations at vertices.

1.9.4 Nambu spinor

At this stage, we would rather use a spinor notation due to Nambu11 that
is very useful for investigating the superconductivity:
 
ak↑
ak = , a∗k = (a∗k↑ a−k↓ ),
a∗−k↓
 
a−k↓
a−k = , a∗−k = (a∗−k↓ ak↑ ). (1.219)
a∗k↑
 Theory of Superconductivity 49

It is instructive to manipulate the complicated last line in (1.217) in the

present language: it has a structure such that
   
∗ ∗ ∗ ∗ 1 ak↑
φα ak↑ a−k↓ − φα a−k ↓ ak ↑ = φα (ak↑ a−k↓ ) (0 1) ∗
0 a−k↓
   
∗ ∗ 0 a−k ↓
− φα (a−k ↓ ak ↑ )
 (1 0)
1 a∗k ↑
= φα a∗k σ+ ak + φ∗α a∗−k σ− a−k .
The positive sign in the second term is due to the Grassmann character of
ak↑ and a−k↓ . Also,
a∗kσ k akσ = a∗ σk3 k ak , (1.220)
since k = −k .
In the above, the Pauli matrices and related ones are rewritten as
     
0 1 0 −i 1 0
σ =
1
, σ =
2
, σ =
3
,
1 0 i 0 0 −1
   
1 1 0 1 − 1 1 0 0
σ = (σ + iσ ) =
+ 2
, σ = (σ − iσ ) = 2
. (1.221)
2 0 0 2 1 0
Equation (1.217) is now written as
  
Z= Da∗ DaDφ∗ Dφ exp {−βπφ∗α (υ)φα (υ)
ω n υn

+ βa∗k (ωn )(iωn− k σ )ak (ωn )

3

√
+ iβ πgα,k −k φα (υn )a∗k (ωn + υn )σ+ ak (ωn )
√
+ iβ πgα ,k−k φ∗α (υn )a∗−k (ωn )σ− a−k (ωn + υn )}. (1.222)
We can integrate with respect to fermion variables: note that31

∗
dz ∗ dze(z Az) = det A for a fermion,

∗
dz ∗ dze(z Az) = (det A)−1 for a boson. (1.223)

Using the upper one, we can immediately obtain

 
 *
*
Z = Dφ∗ Dφ exp −βπφ∗α (υ)φα (υ) + det* β(iωn − k σ3 )
υn ωn

*
√ *
+ iβ πgα,k −k (φα (υn )σ+ + φ∗α (υn )σ− )* , (1.224)
50 Modern Aspects of Superconductivity: Theory of Superconductivity

where we have made use of a symmetry property, gα,k −k = gα ,k−k , and
 · · ·  stands for a matrix. This can be also rewritten as

Z = Dφ∗ Dφ exp−βW (φ), (1.225)

with the effective action for the boson field

 1
W (φ) = π φ∗α (υn )φα (υn ) − Tr log β
υn
β ω
n
 
√
× (iωn − k σ3 ) + i uα,k (φα (υn )σ+ + φ∗α (υn )σ− ) , (1.226)
υn
where the relation log · det = Tr · log and the simplified notation
uα,k = πgα,k −k = πgα ,k−k (1.227)
have been used.
By employing the steepest descent method, we can obtain a particu-
lar φ, by optimizing W (φ), which will be denoted as φ̄. For example, by
differentiating W (φ) with respect to φ∗δ (υn ), we have
√
∂W (φ) 1 i uδ,k σ−
= πφα − Tr √
∂φ∗δ β ω (iωn − k σ3 ) + i uα,k (φα (υn )σ+ + φ∗α (υn )σ− )
n

= 0. (1.228)
We thus obtain
1 (uδ,k uα,k ) · φ̄α
πφ̄δ (υn ) = , (1.229)
β (iωn )2 − Ek2
iωn
where
+
Ek2 (υ) = 2k + (uα,k uα ,k )φ̄∗α (υn )φ̄α (υn ). (1.230)
Note that the field strength φ∗α (υn )φα (υn ) has the dimension of energy.
In obtaining Eq. (1.229), we first rationalize the denominator, and then
the Tr operation on σ’s is carried out. The numerator that remains is the
coefficient of Tr σ+ σ− = 1. A similar equation is obtained for φ∗ . Equation
(1.229) corresponds to the gap equation.
Now we employ an approximation in the rest of this study where, as was
mentioned at the beginning, uα,k = u (a positive constant). Thus, to the
first approximation, φ is nearly constant:
π 1 1
= . (1.231)
u β ω (iωn )2 − Ek2
n
 Theory of Superconductivity 51

Doing the frequency sum, we get

1 1
= ±n(x), (1.232)
β ω iωn − x
n

where n(x) is the Fermi or Bose function according to whether ωn is odd or

even, respectively, and we obtain (for fermions)
π n(Ek ) − n(−Ek ) tan h(βEk /2)
= = . (1.233)
u 2Ek 2Ek
From the above relation combined with Eq. (1.229), we can estimate φ̄,
which corresponds to the gap energy.

1.9.5 Critical temperature

It is straightforward to obtain the critical temperature, Tc , for the supercon-
ductivity from Eq. (1.233).15 Here, it is repeated for completeness. In the
limit T → Tc , the gap, or φα in the present case, disappears. The summation
over k is approximately replaced by the integration
 ωD
2π d 
= tanh , (1.234)
uN (0) 0  2kB Tc
where N (0) is the state density at the Fermi surface, and the cutoff param-
eter ωD is, in the case of the BCS theory, the Debye frequency of phonons
associated with the attractive interaction between electrons.
The integration in Eq. (1.241) proceeds as follows:
 ωD  βC ωD /2  
d  dz βc
tanh = tanh z z=
0  2kB Tc 0 z 2
 βc ωD /2
β ω /2
= log z tanh z]0c D − dz log z sech2 z
0
 ∞  
βc ωD βc ωD
≈ log − dz log z sech2 z tanh ≈1
2 0 2
βc ωD 4eγ
= log + log . (1.235)
2 π
Combined with (1.233), this yields
βc ωD 2π 4eγ
log = − log , (1.236)
2 uN (0) π
where γ = 0.5772 is Euler’s constant. This relation will be used in the next
section. Simple rearrangements yield
Tc ≈ 1.13ωD e−2π/N (0)u . (1.237)
52 Modern Aspects of Superconductivity: Theory of Superconductivity

1.9.6 Temperature dependence of φ

For the later investigation, we derive a temperature dependence of φα near
the critical temperature.16 Coming back to (1.231) and restoring the hidden
summation with respect to k, we can rewrite this as

2π 1  ωD 1
= d . (1.238)
uN (0) β ω 0 (iωn ) − (2 + u|φα |2 )
2
n

Considering that |φα |2  1 near the critical temperature, we expand the

above as
 
2π 1  ωD 1 u|φα |2
= d +
uN (0) β ω 0 (iωn )2 − 2 ((iωn )2 − 2 )2
n

(u|φα |2 ))2
+ + ··· . (1.239)
((iωn )2 − 2 )3

For the convergent integrals which are the second and third terms, the
integration limit is extended to infinity, and then we obtain (putting  =
ω tan θ)
 ωD  1
2π d β 1 π
= tanh + u|φα |2
uN (0) 0  2 β4 ω3n
1 3π  1
+ (u|φα |2 )2 + ··· . (1.240)
β 16 ω5n
The integration of the first term on the right-hand side, which is similar
to (1.235), has β instead of βc . Then, using (1.236), we have, in the lowest
approximation,
βc 1π  1
log = u|φα |2 + ··· . (1.241)
β β4 ω3n
Using
∞
 1 2p − 1
= ζ(p) (1.242)
(2n + 1)p 2p
0

in (1.241), we have the temperature dependence of φα ,

, -
16 T
u|φα |2 = πkB Tc 1 − , ζ(3) = 1.202, (1.243)
7ζ(3) Tc

which is the same as that of the energy gap.15

 Theory of Superconductivity 53

1.9.7 Dynamics of the boson field; symmetry breaking

The boson field φα (υn ), which satisfies (1.227), is written as φ̄α , so that
φα (υn ) = φ̄α + (φα − φ̄α ) = φ̄α + Φα (υn ). (1.244)
Now Φα (υn ) becomes a physical or fluctuation component in solid-state
physics. Our aim is to find the effective Lagrangian or Hamiltonian for it.
In order to find the terms proportional to Φα (υn ) in (1.244), we first
observe that
φα (υn )τ+ + φ∗α (υn )τ− ≡ φα (υn ) · τ = φα (υn )τ(a) ,
(a)
a = 1, 2
= φ̄α (υn ) · τ + Φα (υn ) · τ. (1.245)
Expanding W (φ) in (1.226) in Φα (υn ) and noticing that, upon the Tr oper-
ation, the odd terms with respect to τ will vanish. We have
 1 √
W (φ∗ , φ) = π φ∗α φα − Tr log β(iωn − τ3 − i u(φ̄α · τ)
υ
β ω
n n

1  1
even
√
− Tr [(ωn )i u(Φα (υn ) · τ)]n , (1.246)
βω υ n n
n n

where
√
iωn + k τ3 + i u(φ̄α · τ)
k (ωn ) = . (1.247)
(iωn )2 − Ek2
Note that now the propagator includes the mean-field effect. What we are
interested in is the second line of Eq. (1.246).
The term with n = 2 is precisely written as
−1     
(a) ∗(a ) 
Tr uk (ωn ) Φα (υn )τ(a) k+α (ωn + υn ) Φα (υn )τ(a )
2β ω υ
n n

= u|Φα (υn )|2 Pα (υn ), (1.248)
υn

where careful manipulations about τ’s are required. Except for the explicit
ones combined with Φ’s, we have τ’s inside ’s. Let us call the terms with
a = a = 1 and a = a = 2 the direct interactions, and the terms with
a = 1, a = 2 and a = 2, a = 1 the cross interactions. We then have
−2  iωn (iωn + iυn ) − k k+α + uφ̄∗ · φ̄
Pα (υn ) = 2 ], (1.249)
β ω [(iωn )2 − Ek2 ][(iωn + iυn )2 − Ek+α
n

where the first two terms in the numerator arise from the direct interaction
and the third from the cross interaction. Here, use has been made of the
54 Modern Aspects of Superconductivity: Theory of Superconductivity

fact that Tr τ3 τ+ τ3 τ− = −1. Employing (1.231), we can carry out a lengthy

but not difficult calculation of Pα (υn ). Note that an even frequency υn has
nothing to do with obtaining the Fermi functions. The result is
Ek+α − Ek
Pα (υn ) = −n(Ek )(1 − n(Ek+α ))
(iυ) − (Ek+α − Ek )2
2

Ek+α Ek − k+α k + u(φ̄∗ · φ̄)
× . (1.250)
Ek+α Ek
Here, we again note that, in the square brackets, the first two terms are
obtained from the direct interaction and the third term from the cross inter-
action.
It might be convenient, for the later investigation, to do the frequency
sum for Pα (υn ):
 Ek+α − Ek
υn
(iυn ) − (Ek+α − Ek )2
2


11 1 1
=β −
β υ 2 iυn − (Ek+α − Ek ) iυn + (Ek+α − Ek )
n

β
= [−nB (Ek+α − Ek ) + nB (−Ek+α + Ek )],
2

−β β(Ek+α − Ek )
= cot h ,
2 2
where nB is the Bose function, and use of Eq. (1.232) has been made for
the even frequency υn . Then Eq. (1.250) becomes
 
β β(Eα+γ − Eα )
Pγ (υn ) = n(Eα )(1 − n(Eα+γ )) cot h
υn
2 2

Eα+γ Eα − α+γ α + u(φ̄∗ · φ)
× . (1.251)
Eα+γ Eα
We then obtain the effective action for φ up to the second order in Φ, as
follows:
 1 
Wα (φ∗ , φ) = π φ∗α (υn )φα (υn ) − Tr
υn
βω υ
n n
√
× log{−β(iωn − k τ − uφ̄α (υn ) · τ)}
3

 Ek+α − Ek
− n(Eα )(1 − n(Eα+γ ))
υn
(iυ) − (Ek+α − Ek )2
2


Ek+α Ek − k+α k + u(φ̄∗ · φ̄)
× . (1.252)
Ek+α Ek
 Theory of Superconductivity 55

1.9.8 Instability
In order to discuss the dynamics of Φα (υn ), we must obtain an expression
for the action of Φα (υn ) similar to that for aα (ωn ) given in the exponent of
Eq. (1.208). Since we are interested in the energy region much lower than
that for the electronic excitation (Eα+γ − Eα ), the denominator of Pα (υn )
in Eq. (1.250) is expanded as

1 −1 (iυn )2
= 1+ + ··· .
(iυ)2 − (Ek+α − Ek )2 (Ek+α − Ek )2 (Ek+α − Ek )2
(1.253)
Substituting this into Pα (υn ) in Eq. (1.250) and taking up the terms of the
order of (iυ)2 in the above, we have
(Ek+α − Eγ )2
(iυn )2 + u(φ̄∗ φ̄) + ··· . (1.254)
Ek+α Ek
The second term is a small and complicated, but positive quantity. We thus
obtain the effective Lagrangian for Φ up to the second order of Φ∗ Φ:
LB ∼ Φ∗ (iυn )2 Φ + ηΦ∗ Φ + · · · , η > 0. (1.255)
In the above, iυn is replaced by ∂/∂τ in the future. At this stage, the details
of η are immaterial except that this is positive.
To put forward the problem, we need the term proportional to Φ∗ Φ∗ ΦΦ.
This is obtainable from the term with n = 4 in Eq. (1.246). Without any
detailed calculation, we may presume a positive quantity for this, noted
by Λ2 . Then the total Lagrangian of the boson field becomes
LB ∼ Φ∗ (iυn )2 Φ + ηΦ∗ Φ − Λ2 Φ∗ Φ∗ ΦΦ. (1.256)
This looks like the Lagrangian that Ginzburg and Landau have used for
investigating the superconductivity. We thus expect a kind of phase transi-
tion. Let us optimize, with respect to Φ∗ , the potential part of Eq. (1.256):
V = −ηΦ∗ Φ + Λ2 Φ∗ Φ∗ ΦΦ, (1.257)
∂V
= −ηΦ + 2Λ2 Φ∗ ΦΦ = (−η + 2Λ2 Φ∗ Φ)Φ = 0. (1.258)
∂Φ∗
If Φ = 0, nothing happens; while, in the case of Φ = 0, it is possible that
η
Φ∗ Φ = ≡ ā2 . (1.259)
2Λ2
We now put
Φ = (ρ + ā)eiθ . (1.260)
56 Modern Aspects of Superconductivity: Theory of Superconductivity

Namely, only the radial part is affected. This choice is allowed in the spirit
of the unitary gauge. If we insert this into (1.257), it follows that

V = Λ2 [(ρ + ā)2 − ā2 ]2 − Λ2 ā4 = Λ2 [Φ∗ Φ − ā2 ]2 − Λ2 ā4 . (1.261)

Observing the final result, we realize that the potential becomes mini-
mum at |Φ| = ā; namely, at this position apart from the origin, |Φ| = 0,
which corresponds to the HF state. The mass of the field Φ is given by
the coefficient of Φ∗ Φ, which is 4Λ2 ā2 for a particle of the ρ field, while
the term involving θ is completely lost. That is to say, the particle in the
θ field is massless, and it is called the Goldstone boson. This is due to the
occurrence of the infinite degeneracies along the direction perpendicular to
the ρ coordinate.

1.9.9 Supersymmetry
The total Lagrangian of the system is

Ltot = Le + LB + Lint , (1.262)

where

Le = a∗i (iωn − i τ3 )ai ,

LB = −Φ̇∗ Φ̇ − Λ2 [Φ∗ Φ − ā2 ]2 , (1.263)
√
Lint = −i u(Φa∗ τ+ a + Φ∗ a∗ τ− a).

In the above, the second line is obtained from Eqs. (1.256) and (1.261),
and then the third line from Eq. (1.245) supplied by necessary terms, and a
compact notation is used, and the immaterial constant terms are omitted.
The L in the above is, strictly speaking, the Lagrangian density, and the
action which we are interested in is the space–time integral of the Lagrangian
density, so that the partial integration with respect to τ (after the replace-
ment iυn → ∂τ ) gives the first term with the minus sign. If we define the
momentum conjugate to Φ,
∂Ltot ∂Ltot
Π= = −Φ̇∗ , Π∗ = = −Φ̇, (1.264)
∂ Φ̇ ∂ Φ̇∗
the total Hamiltonian for the boson field is obtained as

Htot = Π∗ Φ̇ + ΠΦ̇∗ − L
√
= Π∗ Π + i u(a∗ Φτ+ a + a∗ Φ∗ τ− a) + Λ2 [Φ∗ Φ − ā2 ]2 . (1.265)
 Theory of Superconductivity 57

The momenta, Π∗ and Π conjugate to Φ∗ and Φ respectively, have the

properties
∂ ∂
[Π, f (Φ∗ , Φ)]− = f (Φ∗ , Φ), [Π∗ , f (Φ∗ , Φ)]− = f (Φ∗ , Φ),
∂Φ ∂Φ∗
(1.266)
since we are now dealing with the classical operators in the framework of
the path-integral formalism.
Let us turn to a supersymmetric treatment.35–37 The fermionic composite
charge operators are defined as
√
Q∗ = Π∗ a∗ − i u(Φ∗ Φ − ā2 )a∗ τ− ,
√ (1.267)
Q = Πa + i u(Φ∗ Φ − ā2 )τ+ a.
First of all, we have to manipulate the commutator
√ √
[Q∗ , Q]+ = [Π∗ a∗ − i u(Φ∗ Φ − ā2 )a∗ τ− , Πa + i u(Φ∗ Φ − ā2 )τ+ a]+ .
(1.268)
Straightforward calculations give
[Π∗ a∗ , Πa]+ = [Π∗ a∗ , a]+ Π − a[Π, Π∗ a∗ ]+
= Π∗ [a∗ , a]+ Π − a[Π, Π∗ ]+ a∗
= Π∗ Π.
Here, for the Grassmann numbers, use has been made of the relation
∂ ∂ ∂a ∂ ∂
[a∗ , a]+ = a−a = +a −a = 1.
∂a ∂a ∂a ∂a ∂a
Further,
√
[Π∗ a∗ , i u(Φ∗ Φ − ā2 )τ+ a]+
√ √
= [Π∗ a∗ , i u(Φ∗ Φ − ā2 )]− τ+ a − i u(Φ∗ Φ − ā2 )[τ+ a, Π∗ a∗ ]+
√ √
= [Π∗ , i u(Φ∗ Φ − ā2 )]− a∗ τ+ a − i u(Φ∗ Φ − ā2 )τ− Π∗ [a, a∗ ]+
√
= i uΦa∗ τ+ a
and
√
[i u(Φ∗ Φ − ā2 )a∗ τ− , Πa]+
√ √
= [Πa, i u(Φ∗ Φ − ā2 )]− a∗ τ− − i u(Φ∗ Φ − ā2 )[a∗ τ− , Πa]+
√ √
= [Π, i u(Φ∗ Φ − ā2 )]− aa∗ τ− − i u(Φ∗ Φ − ā2 )τ− Π[a∗ , a]+
√
= i uΦ∗ aa∗ τ−
√
= −i uΦ∗ a∗ τ− a.
58 Modern Aspects of Superconductivity: Theory of Superconductivity

The final term must be symmetrized:

1
[a∗ τ+ , τ− a]+ → {[a∗ τ+ , τ− a]+ + [a∗ τ− , τ+ a]+ } (1.269)
2
1
= {[(0, a∗ ), (0, a)]+ + [(b, 0), (b∗ , 0)]+ }
2
 
1 [b, b∗ ]+ 0
= (1.270)
2 0 [a∗ , a]+
1
= 1. (1.271)
2
In the second line, b and b∗ , which are used in literature, are defined.
We thus obtain the fundamental relation in the supersymmetric quantum
mechanics,
H = [Q, Q∗ ]+ , (1.272)
if we endow, to the constant Λ in (1.264), a specified value,
2

Λ2 = 2u, (1.273)
37
which is reasonable. However, as mentioned before, this had to be obtained
analytically from the term with n = 4 in Eq. (1.252). If it is assumed that
the present system is the case of supersymmetry, we can avoid this tedious
calculation.
The nilpotency of the charge operators,
Q2 = 0, (Q∗ )2 = 0, (1.274)
is clearly preserved from a2 = 0 and (a∗ )2 = 0. We have thus completed a
supersymmetric analysis.

1.9.10 Towards the Ginzburg–Landau equation

Now we can think about the GL equation. Since it concerns only the static
part of the condensed boson field, we keep the first and third terms of
Eq. (1.265). The first term is the kinetic part, while the last term is the
potential part:
HGL = Π∗ Π − ηΦ∗ Φ + Λ2 (Φ∗ Φ)2 , (1.275)
with
(Ek+α − Eγ )2
η = u(φ̄∗ φ̄) .
Ek+α Ek
As usual, the coefficients of Φ∗ Φ and (Φ∗ Φ)2 in the GL equation are a
and b, respectively. It is crucial that a is negative, which is given by −η
in the present consideration. The temperature dependence of η is found
through u(φ̄∗ φ̄), as in Eq. (1.265). This is exactly the same as that obtained
by Gorkov.30
 Theory of Superconductivity 59

1.9.11 Discussion
Some people seemed to be puzzled by the curious look of the GL equa-
tion when it was published: the superconductivity being a macroscopic,
thermodynamically stable phase — why is it predicted by a Schrödinger-
like wave function, to which the microscopic phenomena are subjected?
Several years later, the microscopic theory was established by BCS, and
soon Abrikosov had successfully correlated these two treatments. He made
clear that the GL function is deeply concerned with the gap function char-
acteristic of the superconductivity and is a wave function describing the
condensed Cooper pairs.
The superconducting state is certainly a stable thermodynamical state,
and a phase transition from the normal state to the superconducting state is
interpreted as a long-distance correlation between the Cooper pairs. Yang38
developed a unified treatment of the phase transition in terms of the density
matrix. According to his treatment, the onset of the superconductivity is
understood in such a way that the off-diagonal long-range order (ODLRO)
of the second-order density matrix has a nonvanishing value. This concept
is clearly related to London’s rigid function, the quasiboson condensation
being widely seen as a powerful model of superconductivity and the vari-
ational wave function tried by BCS.3 Recently, Dunne et al.39 applied the
concept of ODLRO to argue the high-temperature superconductivity in cop-
per oxide, where the attractive interaction between electrons is assumed to
have originated from Friedel oscillations in the screened potential.
However, the previous presentation of Abrikosov looks to be a detour;
a complicated and tedious procedure in which the condensed pair is char-
acterized by an anomalous temperature Green function which is difficult to
manipulate for beginners. Therefore, the direct way to reach the GL theory
from the BCS Hamiltonian should be preferable. What we have done in the
present investigation is the following: the auxiliary boson field driven by
the Hubbard–Stratonovitch transformation to eliminate the quartic term of
electron operators is just the GL function.
In conclusion, from the treatments used so far, the conditions which are
necessary for the occurrence of superconductivity are:

(1) The wave function which means the ground-state average of operators
of a Cooper pair must be complex or two-dimensional. If one of these
two degrees of freedom gets a new stable structure, the other degree of
freedom, whose direction is perpendicular to the former, offers the infi-
nite degeneracy. In the Nambu theory, the first refers to the σ1 direction,
and the second to σ2 in the fictitious spin space.
60 Modern Aspects of Superconductivity: Theory of Superconductivity

(2) The electron–electron interaction should be attractive, otherwise the

negative coefficient of Φ∗ Φ in Eq. (1.275) cannot be obtained. In the
present consideration, we have observed that this condition is estab-
lished in the effective electron–electron interaction of the system involv-
ing the multiband structure.
However, we will discuss a little more the normal state under the usual
condition. In the normal species, the coupling constant between electrons
is intrinsically positive. However, if we are interested in the exchange inter-
action, i.e. the so-called Fock term with negative coupling, this is met by
condition 2. This is a short range interaction, while the direct coupling is
so strong as to overwhelm the exchange interaction.
Let us turn to the behavior of the quartic electron operators. It is unex-
pected that these are grouped into the pair operators of particle–particle
and hole–hole, which do not conserve the particle number. Thus, it is nat-
ural to group them into a couple of particle–hole pairs. This choice makes
the auxiliary boson function real.
The Hamiltonian which satisfies the above two conditions, looking like
the BCS Hamiltonian, is the one with the dipole–dipole interaction. The
simplest case is that of the intermolecular interaction due to the induced
dipole–dipole interaction with a coupling constant −d:
Hd = −da∗r b∗r bs as , (1.276)
where, for example, a∗r and b∗r are the creation operators for a particle and a
hole, respectively. If the electron–electron interaction is screened sufficiently,
this may be another possibility for the superconductivity of a molecular
complex.
 CHAPTER 2

Physics of High-Tc Superconductors

2.1 Introduction
In 1986, K. A. Müller and J. G. Bednorz discovered the phenomenon of
superconductivity in Cu-oxide compounds of lanthanum and barium at
the temperature Tc = 35 K,2 and they were awarded the Nobel Prize
in 1987. This discovery gave an additional push to the intensification of
the scientific activity in the field of superconductivity. Over the last 23
years, the superconducting transition temperature has been increased to
140 K. Moreover, one may expect the discovery of new superconductors with
higher critical temperatures. Recently, the high critical temperatures of new,
Mg-containing superconductors (MgBr2 ) (Tc = 39 K) and superconducting
oxypnictides LnO(1−x) Fx FeAs (Tc = 50 K) were registered.
The main purpose of solid-state physics is the development and fabri-
cation of substances which possess superconductivity at room temperature.
At present, there is an intensive search for such high-temperature supercon-
ductors (HTSCs). Especially promising is the method of producing super-
conducting materials with the help of laser spraying of layers.
After the discovery of HTSCs, the following problems became urgent:
(1) Conceptual–theoretic: the problem of clarifying mechanisms of high-
temperature superconductivity.
(2) Engineering and technical: the problem of practical applications of
HTSCs.
(3) Research: the problem of the search for materials with higher Tc .
The great interest in the phenomenon of superconductivity is caused by the
basic possibility of using it in the future for electric power transfer with-
out losses and for the construction of quantum high-power generators. This

61
62 Modern Aspects of Superconductivity: Theory of Superconductivity

phenomenon could be applied to superconductive electronics and computer

techniques (superconducting elements of memory). At present, a new trend
in technology, namely the construction of quantum computers on the basis
of HTSCs with d pairing, can be seen.
HTSCs have unique physical properties in both the normal state and
the superconducting one. To comprehend the physics of these complex com-
pounds is one of the main tasks of the theory of superconductivity, whose
solution will allow one to explain the mechanism ensuring high-temperature
superconductivity.
The disappearance of resistance on the cooling of superconductors down
to a certain critical temperature is one of the most characteristic effects
in superconductors. But, in order to understand the reason for the occur-
rence of superconductivity, it is necessary to study the other effects which
accompany this phenomenon.
Consider briefly the main properties of HTSCs:
(1) New HTSCs have a great anisotropy across the axis c and possess a
multilayer structure. The main block defining the metallic and super-
conductive properties is the plane with CuO2 , which form the square
lattices of Cu ions.
(2) HTSCs are superconductors of the second kind (l/  1, where l is the
coherence length and  the penetration depth of a magnetic field).
(3) They have high critical temperatures Tc .
(4) They have antiferromagnetic ordering of spins of Cu in the CuO2 planes
and powerful spin fluctuations with a wide spectrum of excitations.
At present, there exists no theoretical approach which would explain the
totality of the thermodynamical, magnetic and superconductive properties
of HTSCs from a single viewpoint.
The electron–phonon mechanism of pairing, being the principal one in
standard superconductors, makes a considerable contribution to the estab-
lishment of the superconducting state in HTSCs. But, in order to obtain a
proper description, it is necessary to consider the other mechanisms inherent
in HTSCs.
Among such mechanisms is the spin-fluctuation one proposed by D. Pines
(USA). The reason for the pairing of electrons can be the scattering of
electrons on spin fluctuations. This model of pairing is one of the most real
models of high-temperature superconductivity.
The thermodynamics of superconductors at low temperatures is
determined by the excitation of two quasiparticles. In the traditional
 Physics of High-Tc Superconductors 63

superconductors with pairing of the BCS type, the energy gap is isotropic
(s pairing), and the temperature dependence of the heat capacity has an
exponential form ∼ exp−Δ/kB T , where Δ is the superconductor’s gap. In
superconductors with anisotropic pairing, the temperature dependence of
the heat capacity has a power character, namely T n . The appearance of
such temperature dependences is related to the fact that the superconduc-
tor’s gap is zero on the Fermi surface. The development of the thermody-
namics of this model is the actual problem of HTSCs.
In the following section, we consider the history of the development of
studies on the phenomenon of superconductivity. The structure of HTSCs
and their physical properties are analyzed. We make a survey of the mech-
anisms of superconductivity and discuss the problem on the symmetry of
the order parameter.
We draw the conclusion that, on the whole, the most probable is the “syn-
ergetic” mechanism of pairing which includes, as components, the electron–
phonon and spin-fluctuation interactions in cuprate planes.
Only if the interaction of all the degrees of freedom (lattice, electronic
and spin ones) is taken into account can a number of contradictory prop-
erties observed in the superconducting and normal phases be explained.
Moreover, one needs to consider the complicated structure of HTSCs. It
will be emphasized that new experiments should be executed in order to
explain the available experimental data.
In this section, we also consider the thermodynamics of the spin-
fluctuation mechanism of pairing, present the method of functional inte-
gration for the calculation of thermodynamical properties on the basis of
the Pines spin-fluctuation Hamiltonian, and deduce the Schwinger–Dyson
equations for Green functions and equations for the thermodynamic poten-
tial. Based on the Schwinger–Dyson equation, we obtain equations for the
superconductor gap, which are used in numerical calculations of the thermo-
dynamics of HTSCs. We present analytic formulas for the thermodynamic
potential and its jump. The numerical calculations of the temperature
dependence of the electron heat capacity indicate that the temperature
dependence of the heat capacity is proportional to the square of the tem-
perature. We emphasize that such temperature dependence is related to the
d pairing. It is shown that the measurement of the temperature dependence
of the heat capacity can be a supplementary test for the establishment of
a type of the symmetry of pairing in HTSCs. The jump of the heat capac-
ity of cuprate superconductors near the critical temperature is evaluated
as well.
64 Modern Aspects of Superconductivity: Theory of Superconductivity

2.2 History of the Development of Superconductivity

The phenomenon of superconductivity was discovered by the physicist H.
Kamerlingh-Onnes at the Leiden Laboratory (the Netherlands) in 1911, in
the same year when Rutherford discovered the atom. Kamerlingh-Onnes reg-
istered the disappearance of resistance of Hg at the temperature T = 4.5 K.
This state was called superconducting. Being cooled down to a temperature
less than the above-mentioned critical temperature, many conductors can
be transferred in the superconducting state, in which electrical resistance
is absent. The disappearance of resistance is the most dramatic effect in
superconductors. But, in order to comprehend the reason for the origin of
superconductivity, we need to study the other effects accompanying this
phenomenon. The dissipationless current states in superconductors were a
puzzle for a long period.
The phenomenon of superconductivity is an excellent example of the
manifestation of quantum effects on the macroscopic scale. At present, it
occupies the place of the most enigmatic phenomenon in condensed-state
physics, namely the physics of metals.
The main purpose of solid-state physics is the creation of superconductors
which have the superconductive property at room temperature. At present,
researchers are continuing the wide-scale search for such HTSCs by testing
a variety of substances.
A number of phenomenological models were proposed to clarify the phe-
nomenon of superconductivity, namely the models advanced by London and
Ginzburg–Landau.9, 40 The essence of these theories which have played an
important role in the development of ideas of superconductivity was pre-
sented in Chap. 1.
The success of the GL theory was related to the circumstance that it is
placed in the mainstream of the general theory of phase transitions.
Among the first microscopic theories devoted to the consideration of
the electron–phonon interaction, the work by H. Fröhlich41 is of the great-
est importance. The following stage in the development of superconduc-
tivity started from the universal theory of BCS published in 1957,104 which
strongly promoted the subsequent study of superconductivity.42 The authors
of this theory were awarded by the Nobel Prize.
The BCS theory gave the possibility of elucidating a lot of experiments
on the superconductivity of metals and alloys. However, during the develop-
ment of the theory, some grounded assumptions and approximations were
 Physics of High-Tc Superconductors 65

accepted. Therefore, the BCS theory was needed in substantiation with the
help of stricter arguments. This was made by academician N. N. Bogoliubov
in Ref. 42. He developed a microscopic theory of the phenomena of super-
conductivity and superfluidity. By using the Fröhlich effective Hamiltonian,
Bogoliubov calculated the spectrum of excitations of a superconductor
within the method of canonical transformations proposed by him in 1947.
In 1962, B. D. Josephson advanced the theory of tunnel effects in super-
conductors (the Josephson effect), which was marked by the Nobel Prize in
1973.43 This discovery strongly intensified the experimental studies of super-
conductivity and was applied to electronics and computer technologies.
In 1986, K. A. Müller and J. G. Bednorz discovered HTSCs,2 which ini-
tiated the huge “boom” manifested in numerous publications. The principal
thought of Müller was the following: by selecting the suitable chemical com-
position, one can enhance the electron–phonon interaction and thus increase
the critical temperature Tc . Müller and Bednorz found a new class of HTSCs,
the so-called cuprate superconductors.
HTSCs have been studied for 23 years by making significant efforts, but
the pattern is still not completely clear. This is related to the complicated
structure of cuprates, difficulties in the production of perfect single crystals,
and difficult control over the degree of doping. The comprehension of HTSCs
will be attained if our knowledge about them approaches a critical level
sufficient for the understanding of a great amount of experimental data from
a single viewpoint. In Fig. 2.1, we present a plot of the critical temperatures
of superconductivity over years.44, 45

2.3 Structural Properties of High-Temperature

Superconductors
The properties of new HTSCs differ essentially from those of traditional
superconductors, which are described by the BCS theory. Let us briefly
consider the main properties of HTSCs. As is known, the atomic structure
defines the character of chemical bonds in solids and a number of relevant
physical properties. Even small changes of the structure frequently lead to
significant changes of their electron properties — for example, at the metal–
dielectric phase transitions. Therefore, the study of a crystal structure with
long-range atomic order and its dependence on the temperature, pressure
and composition is of great importance for HTSCs. Such investigations are
66 Modern Aspects of Superconductivity: Theory of Superconductivity

Fig. 2.1 The evolution of critical temperatures since the discovery of superconductivity.

significant for the comprehension of mechanisms of high-temperature super-

conductivity.
At present, we have several families of HTSCs: thermodynamically sta-
ble copper oxides containing lanthanum, yttrium, bismuth, thallium and
mercury.45
We note that the structure of all copper-oxide superconductors has a
block character. The main block defining the metallic and superconduc-
tive properties of a compound is the plane with CuO2 which forms the
square lattice of copper ions coupled with one another through oxygen ions.
Depending on the composition, the elementary cell of a high-temperature
compound can have one, two or more cuprate layers. In this case, the critical
temperature of the superconducting transition increases with the number of
cuprate layers. In Fig. 2.2, we present an elementary cell of the orthorhombic
structure of yttrium ceramics YBa2 Cu3 O7−δ , δ ≈ 0.1 − 0.3. The maximum
Tc ≈ 90 K. The size of the cell is characterized by the following parameters:
a = 3.81 Å, b = 3.89 Å, c = 11.7 Å.
Thus, HTSCs are characterized both by large volumes of elementary cells
and by a clearly manifested anisotropy of layers.
 Physics of High-Tc Superconductors 67

Fig. 2.2 Elementary cell of yttrium ceramics YBa2 CuO7 : 1–yttrium; 2–barium; 3–
oxygen; 4–copper.

Of significant interest is the discovery of the compounds Bi2 CaSr2 Cu2 O8

(Bi/2-1-2-2) and Tl2 CaBa2 Cu2 O8 (Tl/2-1-2-2) with the temperature of the
superconducting transition Tc > 100 K. These compounds can have differ-
ent numbers of cuprate layers and are described by the general formula
A2 Can Y2 Cun O2n4 , where A = Bi(Tl) and Y = Sr(Ba). Their temperature
Tc depends on the number of cuprate layers and takes values of 10, 85 and
110 K for the compounds with Bi, and 85, 105 and 125 K for the compounds
with Tl for n = 1, 2, 3, respectively. For the compounds with Tl with the
single layer Tl-O and the general formula TlCan−1 Ba2 Cun O2n+3 (Tl/1 . . .),
the number of cuprate layers reaches n = 5.
We present the structures of the family of Tl-based superconductors in
Figs. 2.3 and 2.4. In Table 2.1, we show the dependences of Tc and the
interplane distance (Å) on the number of layers in the unit cell N . The
explanation for the dependence of the critical temperature on the number
of cuprate layers is an urgent problem.46, 47 All HTSCs have a complicated
multiband structure,98 shown in Fig. 2.5.
68 Modern Aspects of Superconductivity: Theory of Superconductivity

Fig. 2.3 Schematic distribution of ions in the unit cell of superconductors

Tl1 Ba2 CaN−1 CuN O2N+4 or, in the brief notation, Tl(1 : 2 : N − 1 : N ) at N = 1, 2, . . . , 5.

Fig. 2.4 Schematic distribution of ions in the unit cell of superconductors

Tl1 Ba2 CaN−1 CuN O2N+4 or, in the brief notation, Tl(1 : 2 : N − 1 : N ) at N = 1, 2, . . . , 5.

Table 2.1 Tl-based systems Tl1 Ba2 CaN−1 CuN O2N+3 .

N 1 2 3 4 5 6

a (Å) — 3.8500 3.8493 3.8153 3.8469 —

Tc (K) 13–15 78–91 116–120 122 106 102
 Physics of High-Tc Superconductors 69

Fig. 2.5 Structure of energy bands near the Fermi energy for (a) high-temperature super-
conductor Bi2 Sr2 CaCu2 O8 and (b) low-temperature superconductor Bi2 Sr2 CaCuO6 .

We note the importance of the structure of zones near the Fermi level.
For example, the structures of the bands of two superconductors (high-
temperature Bi2 Sr2 CaCu2 O8 and low-temperature Bi2 Sr2 CaCuO6 ) are sig-
nificantly different at point K. The energy interval from the band bottom
to the Fermi surface is equal to 0.7 eV at point K for the high-temperature
superconductor and 0.1 eV for the low-temperature one. The explanation
for the influence of the multiband structure on Tc is given in Chap. 3.
It is worth noting that all copper-oxide superconductors have a block
character in their structure. The main block defining the metallic and super-
conductive properties of a compound is the plane with CuO2 which con-
tains the square lattice of copper ions coupled with one another through
oxygen ions. Depending on the composition, the elementary cell of a high-
temperature compound can have one, two, three or more cuprate layers.
Moreover, the critical temperature of the superconducting transition has a
nonmonotonous dependence on the number of cuprate layers.46

2.3.1 Phase diagram of cuprate superconductors

All electron properties of HTSCs depend strongly on the doping. HTSCs
without doping are dielectrics and antiferromagnetics. As the concentration
70 Modern Aspects of Superconductivity: Theory of Superconductivity

Fig. 2.6 Phase diagram of cuprate superconductors in the temperature-doping variables.

TN — the Néel temperature; Tc — the critical temperature of the superconducting tran-
sition; T ∗ — the characteristic temperature of a pseudogap; T 0 — the upper crossover
temperature.

x increases, these materials become metals. Superconductivity arises at large

x, behind the limits of the magnetically ordered phase. The experiments
showed that the charge carriers have the hole character for all classes of
HTSCs.
It has become clear recently that the high-temperature superconductivity
is related to peculiarities of the behavior of these compounds in the normal
phase. As seen from the phase diagram (Fig. 2.6), the superconducting states
arise near the antiferromagnetic phase. In yttrium-containing systems, the
antiferromagnetic and superconducting regions adjoin one another.
The experiments on the inelastic magnetic scattering of neutrons indi-
cate the existence of strong magnetic fluctuations in the doped region, even
beyond the limits of the antiferromagnetic phase. This points out the impor-
tant role of antiferromagnetic fluctuations in the compounds with high-
temperature superconductivity.
In HTSCs, the gap is present in the absence of the phase coherence,
i.e. in nonsuperconducting specimens. This gap is called a pseudogap. A
pseudogap is shown in Fig. 2.6. It appears at temperatures less than some
characteristic temperature T ∗ which depends on the doping. Its nature has
still not been completely explained. The study of a pseudogap in the electron
spectrum of HTSCs was carried out in many works.48
 Physics of High-Tc Superconductors 71

Metals become superconductors if their free electrons are bound in

Cooper’s pairs. Moreover, the pairs are formed in such a way that their
wave functions have the same phase. The phase coherence is responsible for
the change of the resistance on the cooling below the critical temperature Tc .
The presence of coupled pairs in a superconductor causes the appearance
of a gap in the spectrum of excitations. In standard superconductors, the
phase coherence of pairs appears simultaneously with the appearance of
pairs. From one viewpoint, a pseudogap is related to the appearance of
coupled pairs, which is not related to the phase coherence.
Another viewpoint consists the following. The pseudogap arises in
HTSCs in connection with the formation of magnetic states which com-
pete with superconducting states. The efforts of experimenters aimed at the
resolution of this dilemma are complicated by a strong anisotropy of the
superconductor gap. Some physicists believe that the most probable situ-
ation is related to the creation of the superconducting state with paired
electrons at a certain doping which coexists with antiferromagnetism. It is
possible that this is just the “new state of matter” which has been widely
discussed in the last few years in connection with HTSCs.

2.3.1.1 Antiferromagnetism of HTSCs

An interesting peculiarity of copper-oxide compounds which has a universal

character is the presence of the antiferromagnetic ordering of spins of Cu
in the CuO2 planes. The sufficiently strong indirect exchange interaction
of spins of Cu induces the 3D long-range antiferromagnetic order with the
relatively high Néel temperatures TN = 300–500 K.49 Though the long-
range order disappears in the metallic (and superconducting) phase, strong
fluctuations with a wide spectrum of excitations are conserved. This allows
one to advance a number of hypotheses on the possibility of electron pairing
in copper-oxide compounds through the magnetic degrees of freedom.
Therefore, the study of the antiferromagnetic properties of HTSCs is of
importance for the verification of the hypotheses on the magnetic mechanism
of superconductivity. The interaction of spins of Cu in a plane has a 2D
character, and their small values, S = 1/2, lead to significant quantum
fluctuations.
The first indications of the existence of antiferromagnetism in the copper-
oxide compounds were obtained on the basis of macroscopic measurements
of susceptibility. The detailed investigation of both the magnetic structure
72 Modern Aspects of Superconductivity: Theory of Superconductivity

and spin correlations in the metallic phase became possible only with the
help of the neutron scattering.

2.4 Mechanisms of Pairing of HTSCs

To explain high-temperature superconductivity, a lot of models and mecha-
nisms of this unique phenomenon have been proposed. The key question is
the nature of the mechanism of pairing of carriers. There are many different
models of superconductivity available, among which we mention the follow-
ing: the magnon model, exciton model, model of resonant valence bonds,
bipolaronic model, bisoliton model, anharmonic model, model of local pairs
and plasmon model. We give a classification of the mechanisms of pairing
of HTSCs, which is shown in Fig. 2.7, according to Ref. 50.
This classification demonstrates the diverse physical patterns of HTSCs.
Apart from the ordinary BCS mechanism based on the electron–phonon
interaction, there exist many other mechanisms, as was mentioned above.
All these models use the concept of pairing with a subsequent formation
of a Bose condensate at temperatures Tc irrespective of the nature of the
resulting attraction.

2.4.1 Specific mechanisms of pairing in superconductivity

Consider the models most popular at present. Along with the ordinary BCS
mechanism based on the electron–phonon interaction, we turn to the mag-
netic, exciton, plasmon and bipolaronic mechanisms of pairing. All these
models apply the concept of pairing with the subsequent formation of a
Bose condensate (at certain temperatures Tc ) regardless of the reasons for
the attraction.
The BCS theory presents the formula for the critical temperature Tc in
the case of the weak electron–phonon interaction:
 
−1
Tc = 1.14Θ exp , (2.1)
N0 V
where Θ = ΩD /kB , ΩD is the Debye energy, N0 is the density of states
of the Fermi level and V is the attractive pairing potential acting between
electrons.
The maximum value of the critical temperature given by the BCS the-
ory is 40 K. Thus, there arises the question about the other mechanisms of
pairing. The interaction of electrons is repulsive, i.e. one needs to seek a
 Physics of High-Tc Superconductors 73

Fig. 2.7 Classification of the mechanisms of pairing of HTSCs.

“transfer system” in metals which is distinct from the phonon system. The
general scheme for the interaction of electrons via a transfer system X can
be presented as 
e1 + X → e11 + X ∗ ,
(2.2)
e2 + X ∗ → e12 + X,
where ei corresponds to an electron with momentum pi , X is the ground
state and X ∗ is the excited state of the transfer system.
As a result of this reaction, the system returns to the initial state, and
the electrons make an exchange by momenta.
74 Modern Aspects of Superconductivity: Theory of Superconductivity

It can be shown that such an interaction leads to the attraction, and the
critical temperature is given by the formula
 
−1
Tc ∼ ΔE exp ,
λ
where ΔE is the difference in energies of the states X and X ∗ , and λ depends
on the interaction of electrons with the system X.51

2.4.2 Magnetic mechanism of pairing

The magnetic mechanism of pairing of HTSCs was examined by many
researchers. These studies used the assumption that the pairing is realized
due to the exchange by spin excitations — magnons, i.e. the transfer sys-
tem is the system of spins in a magnetic metal. Magnon is a quasiparticle
which is the quantum synonym of a spin wave of excitation in a magneti-
cally ordered system. Great attention was attracted to the pioneering work
by A. I. Akhiezer and I. Ya. Pomeranchuk.52, 53 They considered the inter-
action between conduction electrons caused by both the exchange by acous-
tic phonons and an additional interaction related to the exchange by spin
waves (magnons). It was shown that superconductivity and ferromagnetism
can coexist in the same spatial regions. At sufficiently low concentrations of
the ferromagnetic component, an increase in its concentration leads to an
increase of Tc in the case of the triplet pairing. Within the phonon mecha-
nism, the pairing occurs in a singlet state; then an increase in the concentra-
tion of the ferromagnetic component induces a decrease of Tc . It is known
that ferromagnetism competes with superconductivity. The different situa-
tion is characteristic of antiferromagnetism. HTSCs are antiferromagnetic
dielectrics. As was mentioned above, they reveal strong magnetic fluctua-
tions in the region of doping, which can be responsible for the pairing. In
this chapter, we will consider the spin-fluctuation model of pairing.
After the discovery of HTSCs, a lot of relevant works dealt with the
problem of the evolution of a system of Cu ions under the transition from
the dielectric antiferromagnetic state to the metallic one.
It is also worth noting the model of resonance valence bonds (RVBs),
which was advanced by Nobel Prize winner P. Anderson in 1987,5 who
made an attempt to explain the Cooper pairing in HTSCs by the par-
ticipation of magnetic excitations. The Anderson model is based on the
concept of magnetic ordering. The RVB mechanism ensures the joining
of carriers in pairs with compensated spin — the so-called spinons. On
 Physics of High-Tc Superconductors 75

the doping of HTSC compounds, there arise holes which can form the
complexes with spinons — holons. Superconductivity is explained by the
pairing of holons, i.e. by the creation of spinless bosons with a double
charge. That is to say, the pairing of carriers in the RVB model is realized
due to the exchange by magnons. At low temperatures, the paired holons
form a superconducting condensate. The RVB model has played a positive
role, by attracting the attention of researchers to the study of antiferromag-
netism in HTSCs, though spinons and holons have not been experimentally
identified.

2.4.3 Exciton mechanism of pairing

According to the general principle concerning the “transfer system” in super-
conductors, the electron–phonon interaction should not be obligatorily real-
ized. Some other interaction ensuring the pairing of electrons can be suitable.
In principle, the mechanism of superconductivity can be switched on by
bound electrons which interact with conduction electrons. The first exciton
model, in which the pairing is realized due to electron excitations, was pro-
posed by W. A. Little54 for organic superconductors, and by V. L. Ginzburg
and D. A. Kirzhnitz40 for layered systems. In the construction of this model,
it was necessary to assume the existence of two groups of electrons: one of
them is related to the conduction band, where the superconducting pairing
occurs due to the exchange by excitons which are excitations in the second
group of almost-localized electrons. In view of the many-band character of
the electron spectrum, layered structure and other peculiarities of the elec-
tron subsystem in HTSCs, such a distribution of electron states is quite
possible. This underlies the development of a lot of exciton models. The
search for superconductivity in organic materials was stimulated to a signif-
icant degree by the idea of Little about the possibility of high-temperature
superconductivity due to the excitonic mechanism of the Cooper pairing of
electrons in long conducting polymeric chains containing lateral molecular
branches–polarizers. Since the mass M of such excitonic excitations is small,
one would expect to observe a high value of the temperature, Tc ∼ M −1/2 .
But this model was not practically realized, since high-energy intramolecular
excitonic excitations cannot ensure the binding of electrons in pairs.
At present, a number of quasi-one-dimensional organic superconductors
with metallic conductance have been synthesized. They pass to the super-
conducting state at T = 10 K. Crystals of similar organic superconductors
76 Modern Aspects of Superconductivity: Theory of Superconductivity

consist of, as a rule, planar molecules packed in zigzag-like stacks which

form chains. The good overlapping of electron wave functions of neighbor-
ing molecules in a stack ensures the metallic conductance along a chain. The
overlapping of electron wave functions of neighboring chains is small, which
leads to the quasi-one-dimensional character of the electron spectrum and
to a strong anisotropy of electronic properties of a crystal. Up to now, no
experimental proofs of the manifestation of the excitonic mechanism is such
systems are available.
As an example of a laminar system, we mention a quasi-two-dimensional
structure of the “sandwich” type (dielectric–metal–dielectric). In such a
structure, the Cooper pairing of electrons in a metal film occurs due to the
attraction caused by their interaction with excitons in the dielectric plates.

2.4.4 Anharmonic model and superconductivity

It is known that the appearance of superconductivity is often preceded by
structural transformations in a crystal. These are usually explained within
the anharmonic model. In the opinion of some researchers,45 such structural
transformations arising before the start of superconductivity decrease signif-
icantly the frequencies of phonons and subsequently increase the parameter
of electron–phonon interaction. The softening of the phonon spectrum is
caused by the great amplitudes of displacements of ions in the two-well
potential which models the structural transformations. In some works,49
the effect of a structural transformation on superconductivity in the limit-
ing case of a weak pairing interaction and the isotropic gap was studied. The
properties of high-temperature superconductivity were also studied within
the model, where the superconductivity is enhanced due to the singularity
of the density of electron states which appears at structural or antiferromag-
netic phase transitions. The weak point of the models relating the supercon-
ductivity to structural phase transitions is due to a significant temperature
interval between the known structural transitions and the temperature Tc .
The works, where nonphonon pairing mechanisms are introduced, include
the studies49 which use the Hubbard Hamiltonian in systems, where only the
interaction of the repulsive type is present. It is considered that the effect
of pairing is caused by the kinematic interaction at a noncomplete occu-
pation of Hubbard subbands. Unfortunately, no clear comprehension of the
nature of the arising attraction has been attained in this case. It is possible
that the bound state of quasiparticles is virtual, i.e. it decays. The search
for new mechanisms of superconductivity caused by a strong correlation of
 Physics of High-Tc Superconductors 77

electrons in cuprate superconductors with quasi-two-dimensional electron

structure has been reduced to a search for new, nonstandard ground states.
We mention the polaron mechanism of pairing, which is related to the Jahn–
Teller effect, well known in the quantum chemistry of complex compounds.
The essence of the Jahn–Teller effect is that a nonlinear system in the pres-
ence of the electron degeneration deforms spontaneously its structure so
that this degeneration disappears or decreases. The Jahn–Teller effect leads
to rearrangement of the atomic orbitals of copper under conditions of the
octahedral oxygen surroundings. A displacement of oxygen ions inside of an
elementary cell which is caused by the appearance of a quasiparticle induces
the displacement of equilibrium positions in the neighboring cells. In such a
way, there arises the strong electron–phonon interaction of a quasiparticle
with a local deformation.

2.4.5 Van Hove singularities

A van Hove singularity (vHs) in the density of states (DOS) N (E) was
proposed as a Tc -enhancement mechanism for intermetallic superconduc-
tors two decades ago. All cuprate superconductors possess two-dimensional
elements of their structures. In the construction of the microscopic theory
of high-temperature superconductivity, it is important to clarify the specific
features of the dispersion E(k) and the behavior of the DOS N (E). For a
two-dimensional problem (n = 2; 2D), the DOS is independent of the energy,
N (E) = const, and the band is dispersionless. The photoemission experi-
ments indicate the existence of an almost-flat band near the Fermi surface
for cuprate superconductors. The presence of a flat band and an isoenergetic
surface in the form of an elongated saddle leads to the existence of vHs’s in
the DOS near the Fermi surface. In this model, in the calculations of Tc by
the BCS formula, N (E) is replaced by N (EF ). The formula for Tc looks like
 
−1
Tc = 1.14Θ exp , (2.3)
λ
where λ = V N (E).
If the function N (E) has the corresponding singularity at E = EF , it
will be related to the vHs. In the two-dimensional case, the presence of
a logarithmic singularity of the DOS on the Brillouin zone boundary is
possible:
 
 D 

N (E) ∼ ln , (2.4)
E − EF 
78 Modern Aspects of Superconductivity: Theory of Superconductivity

where D is the characteristic energy cutoff. Then Tc has the form

 
−1
Tc ∼ D exp √ . (2.5)
λ
We note that, in connection with the quasi-two-dimensionality of lattices of
HTSCs, the hypothesis of anyonic superconductivity is of certain interest
(Fig. 2.7). Anyons are quasiparticles with intermediate statistics (between
the Bose and Fermi statistics) which can exist just in two-dimensional struc-
tures. The term “anyon” was introduced by F. Wilczek in the framework of
the concept of supersymmetry.

2.4.6 Plasmon mechanism of pairing

Many works have been devoted to attempts at explaining high-temperature
superconductivity on the basis of the idea of the pairing as a result of the
exchange by quanta of longitudinal plasma waves — plasmons.
Longitudinal plasma waves are formed in solids in the region of frequen-
cies, at which the dielectric permeability of the medium becomes zero. The
characteristic frequency of plasma waves in 3D crystals is defined by the
formula
4πe2 N
ω̃p = , (2.6)
m
where N is the concentration of electrons, and e and m are their charge
and mass, respectively. At the electron density N ∼ 1–3 × 1022 cm−3 , the
plasma frequency ω̃p ∼ 1015 –1016 . We might assume that the exchange by
plasmons, rather than by phonons, would induce an increase of the pre-
exponential factor in the formula deduced in the BCS theory,
 
1
Tc = Θ exp , (2.7)
λ − μ∗
by 2–3 orders, if Θ = ω̃p /kB . However, such an increase does not cause a
significant growth of Tc , because the plasmons at the frequency ω̃p , which
is close to the frequency of electrons, EF /, cannot cause the superconduct-
ing pairing and their role is important only for the dielectric properties of
crystals.44

2.4.7 Bipolaronic mechanism of superconductivity

One of the attempts to explain the phenomenon of high-temperature super-
conductivity is called the bipolaronic theory. Bipolarons are Bose particles
like the ordinary Cooper pairs.
 Physics of High-Tc Superconductors 79

In the theory of bipolarons, the superconductivity is caused by the super-

fluidity of the Bose condensate of bipolarons.
The ideas of polarons and bipolarons were used by A. S. Aleksandrov
and J. Ranninger55 to clarify high-temperature superconductivity.
The idea of a polaron is based on the assumption about the autolocal-
ization of an electron in the ion crystal due to its interaction with longi-
tudinal optical vibrations under the local polarization, which is caused by
the electron itself. The electron is confined to the local-polarization-induced
potential well and conserves it by its own field. The idea of autolocalization
of electrons in ion crystals was intensively developed by S. I. Pekar.56
The efficiency of the interaction of an electron with mass m and charge
e with long-wave longitudinal optical vibrations with frequency Ω in the
medium is characterized by the dimensionless parameter
-
e2 m
g= , (2.8)
ε̃ 2Ω2
introduced by H. Fröhlich. Here, ε̃ is the dielectric permeability of the inertial
polarization. The interaction is assumed to be small if g < 1. Due to a high
frequency, Ω, the deformation field is a faster subsystem. Therefore, it has
time to follow the movement of an electron. This field accompanies the
movement of the electron in the form of a weak cloud of phonons. The
energy of interaction of the field and the electron is proportional to the first
degree of g.
In the BCS theory, the pairing of conduction electrons is realized due to
the interaction with acoustic phonons and is characterized by the dimen-
sionless constant of interaction
λ = V N (EF ), (2.9)
where N (EF ) is the density of energy states of electrons on the Fermi sur-
face, and the quantity V is inversely proportional to the coefficient of elas-
ticity of a crystal.
In the bipolaronic model, we have the parameter
2λ2 Ωz − Vc
λ∗ = , (2.10)
D
where λ is defined by the relation (2.9), z is the number of nearest neighbors
and D is the width of the conduction band of free quasiparticles.
A bipolaron, like a Cooper pair, has charge 2e, and its effective mass is
determined by the formula

m̃ ≈ m exp(λ2 ). (2.11)
D
80 Modern Aspects of Superconductivity: Theory of Superconductivity

The effective mass m̃ can be very high and can exceed the mass of a free
quasiparticle in the conduction band by several orders.
To calculate the superconducting transition temperature Tc which cor-
responds to the Bose condensation, the authors applied the formula for the
ideal Bose gas:

3.312 N 2/3
kB Tc = . (2.12)
m̃
At m̃ = 100 and N = 1021 cm−3 , this formula yields Tc ∼ 28 K.
Analogously to the bipolaronic model, the bisoliton mechanism of high-
temperature superconductivity was proposed in Ref. 45 and 46.

2.5 Symmetry of Pairing in Cuprate Superconductors

Of great importance for HTSCs is the pairing symmetry or the symmetry
of the order parameter. This question was considered at many conferences
and seminars all over the world. Several NATO seminars and conferences
on this trend were held in Ukraine in the town of Yalta, organized by one
of the authors of this book.57–60
The development of the microscopic theory of superconductivity was
followed by interest in the question about the nontrivial superconductivity
corresponding to the Cooper pairing with nonzero orbital moment.
The system, in which the nontrivial pairing was first discovered, is He3 .
To explain this phenomenon, it was necessary to introduce a supplementary
mechanism of pairing due to spin fluctuations.

2.5.1 The superconductor’s order parameter

Most physical properties depend on the properties of the symmetry of a
superconductor’s order parameter, which is defined by the formula

Δαβ (k) = aαk aβ−kα . (2.13)

The problem of pairing symmetry is a problem of the pairing of charged

fermions into states with the final orbital moment.
As usual, both the standard pairing called the s pairing and the non-
standard d pairing are considered. They differ by the orbital moment of
the pair: in the first and second cases, the moments are L = 0 and L = 2,
respectively.
 Physics of High-Tc Superconductors 81

We also note that the continuous symmetry group in crystals is broken,

and it is necessary to speak not about the orbital moment, but about the
irreducible representations, by which the order parameter is classified. We
will consider this question in the following subsection.
Usually, the standard pairing and the nonstandard pairing are distin-
guished. For the nonstandard pairing, the symmetry of the order parameter
is lower than that of a crystal.
For a two-dimensional tetragonal crystal (square lattices), the possible
symmetries of the superconductor’s order parameter were enumerated by
M. Sigrist and T. M. Rice61 on the basis of the theory of group representa-
tions. The basis functions of relevant irreducible representations define the
possible dependence of the order parameter on the wave vector.
It is worth noting that the anisotropic pairing with the orbital moment
L = 2, i.e. dx2 −y2 , has the following functional form in the k space:

Δ(k) = Δo [cos(kx a) − cos(ky a)], (2.14)

where Δo is the maximum value of the gap, and a is the lattice constant.
The gap is strongly anisotropic along the direction (110) in the k space. In
this case, the order parameter sign is changed in the directions along kx
and ky .
Apart from the d symmetry, it is worth considering also the s symmetry,
for which we can choose two collections of basis functions:

Δ(k) = Δo . (2.15)

The anisotropic s pairing is considered as well. This form of pairing is

analyzed in works by P. W. Anderson5 and coworkers who have studied the
mechanism of pairing on the basis of the tunneling of electrons between lay-
ers. In these states, the order parameter sign is invariable, and its amplitude
is varied along the direction (110):

[anisotropic s]Δ(k) = Δo [cos(kx a) − cos(ky a)]4 + Δ1 , (2.16)

where Δ1 corresponds to the minimum along the direction (110).

It follows from the symmetry-based reasoning that the mixed states
with various symmetries can be realized. We mention the states which
are mostly in use. The “extended” s-coupled states were considered in
works by D. J. Scalapino.63 A possible functional form of these states is as
follows:

[extended s wave] Δ(k) = Δo {(1 + γ2 )[cos(kx a) − cos(ky a)]2 − γ2 }. (2.17)

82 Modern Aspects of Superconductivity: Theory of Superconductivity

They have eight parts with alternating signs, and eight nodes which are split
by ±γπ/2 along the direction (110). G. Kotliar and coworkers used mixed
s + id states63 :
[s + id] Δ(k) = Δo [ε + i(1 − ε)[cos(kx a) − cos(ky a)]]. (2.18)
R. B. Laughlin62 analyzed the mixed states dx2 +y2 + idxy on the basis of
the anyonic mechanism of pairing:
[d + id] Δ(k) = Δo [(1 − ε)[cos(kx a) − cos(ky a)]
+ iε[2 sin(kx a) sin(ky a)]], (2.19)
where ε is the share of s or dxy states mixed with the dx2 +y2 states, and εΔo
is the minimum value of the energy gap. These mixed states are of interest
because they are not invariant with respect to the inversion in the time.
The value and phase of the superconductor’s order parameter, as functions
of the direction in cuprate planes CuO2 , are given in Fig. 2.8 for various
kinds of pairing symmetry.

2.5.2 Classification of the superconductor’s order parameter

by the representations of symmetry groups
It should be noted that there is no classification of states by the orbital
moment for crystals. The general theory of nonstandard pairing has been
developed on the basis of the analysis of point symmetry groups.
J. Annett was one of the first to classify superconducting states by the
irreducible representations of groups for HTSCs.64 As for superconductors
with heavy fermions, the group analysis was carried out in Ref. 65.
Of importance is the question whether the symmetry of the order param-
eter in HTSCs is lower than the symmetry of the crystal lattice.
The order parameter can be represented as a linear combination:

Δk = ηΓi fΓi (k), (2.20)
where ηΓi is the irreducible representation of groups, by which the order
parameter is transformed, fΓi (k) is the basis function of the irreducible
representation; A1g corresponds to the anisotropic s symmetry; and B1g
corresponds to the d pairing.
An analogous decomposition of the superconducting order parameter was
performed for a superconductor with heavy fermions:
U P t3 .
 Physics of High-Tc Superconductors 83

Fig. 2.8 Types of pairing symmetries which are considered for HTSCs.

The symmetric states of the system can be presented by the indication of all
possible subgroups of the full group, relative to which the order parameter is
invariant.66 The full symmetry group of a crystal includes a point symmetry
group G, operations of inversion of the time R, and the group of calibration
84 Modern Aspects of Superconductivity: Theory of Superconductivity

transformations U (1). That is to say, we have the following partition into

subgroups:
G × R × U (1). (2.21)
The La- and Y-based superconductors which have the tetragonal symmetry
of crystal lattices, i.e. G = D4h = D4 ×I,58 are most-studied. In the Y-based
compounds, one observes small orthorhombic distortions of a crystal lattice.
Group D4h includes the operations of rotations Cn , around the z axis by
angles of πn/2, and rotations Un by angles of π:
 πn   πn 
x cos + y cos , (2.22)
2 2
where n = 0, 1, 2, 3 have five reducible representations: four one-dimensional
(A1g , A2g , B1g , B2g ) and one two-dimensional (E).

2.6 Experimental Studies of the Symmetry

of the Superconducting Order Parameter
At present, the results of three groups of experiments, in which the symme-
try of the order parameter is revealed, are available.
The first group joins different low-temperature characteristics of super-
conductors, such as the Knight shift, the rate of relaxation in NMR, the
temperature dependence of the heat capacity, and the penetration depth.
If the superconductor’s order parameter has zeros on different areas of
the Fermi surface (as in the case of the dx2 −y2 symmetry), the mentioned
quantities will have the power temperature dependence, rather than the
exponential one.
The second group of experiments is based on the direct measurement of
the phase of the order parameter with the help of interference phenomena
on Josephson junctions in a magnetic field.
The third group deals with the direct measurements of a value of the gap
by means of spectroscopic experiments. Here, the most interesting results
are presented by photoemission spectroscopy with angle resolution, as well
as Raman and neutron spectroscopies.

2.6.1 Measurements of the Josephson tunnel current

The most definite information about a symmetry of the superconducting
order parameter can be obtained from the studies of the phase of the order
 Physics of High-Tc Superconductors 85

parameter, in which the critical current in Josephson junctions positioned in

a magnetic field is measured. The critical current for a rectangular Josephson
junction oscillates with the field by the Fraunhofer diffraction law:
sin(πΦ/Φ0 )
Ic (Φ) = J0 A , (2.23)
πΦ/Φ0
where Φ is the magnetic flux through the junction, Φ0 is the quantum
flux, J0 is the density of the critical current in the zero field, and A is the
junction area. A diffraction pattern is shown in Fig. 2.9(a). Let us consider
the superconductor YBCO, which possesses the tetragonal symmetry. Let
its axis be oriented normally to the plane of the figure, and its edges to
be perpendicular to the axes A and B of the base plane. In the tunnel
junction with corner geometry, another superconductor is applied to both
edges which are perpendicular to A and B and, are joined with each other
[Figs. 2.9(b) and 2.9(c)].
This experiment can be compared to that involving a two-junction
SQUID, in which there occurs a superposition of tunnel currents produced
by electrons with wave vectors kx and ky , so that the resulting diffraction
pattern depends on the symmetry of the order parameter of a superconduc-
tor under study. For the s symmetry, the order parameter on both edges of
the corner junction is the same, and the resulting diffraction pattern will be
similar to the case of the standard junction. But, in the case of d symmetry,
the order parameter on the corner junction edges has different signs, and
this changes the diffraction pattern.
The total current is illustrated in Fig. 2.10(c). In the zero field, the
critical current turns out to be zero due to the mutual compensation of its
two components.
In the symmetric contact, the dependence on the field is determined by
the formula67
sin2 (πΦ/2Φ0 )
Ic (Φ) = J0 A , (2.24)
πΦ/2Φ0
corresponding to the pattern in Fig. 2.10(c).
Thus, by the difference of a diffraction pattern from both the standard
one and that corresponding to the corner Josephson junction, we can predict
the symmetry of the order parameter. In Ref. 67, a similar experiment was
carried out on the tunnel junction YBCO–Au–Pb. The results presented in
Fig. 2.10 testify to the d symmetry of the order parameter in superconduct-
ing YBCO.
86 Modern Aspects of Superconductivity: Theory of Superconductivity

Fig. 2.9 Critical current in a Josephson junction vs the applied magnetic field: (a) stan-
dard tunnel contact, (b) corner tunnel junction for a superconductor with the s symmetry
of the order parameter, (c) corner tunnel junction for a superconductor with the d sym-
metry of the order parameter.67

All the above-mentioned experiments with Josephson junctions were per-

formed with a single crystal YBa2 Cu3 O6 .

2.6.2 Measurements of the quantization of a flow

by the technique of the three-crystal unit
Another type of experiment on the determination of a symmetry of the
order parameter is based on the measurement of a flow quantum in a
 Physics of High-Tc Superconductors 87

Fig. 2.10 Critical current as a function of the magnetic field in a Josephson junction
YBCO-Au-Pb in two geometries: (a) standard and (b) corner.67

superconducting ring fabricated from three superconducting single crystals

of yttrium with different orientations.
The idea of such an experiment is based on the theoretical result obtained
by M. Sigrist and T. M. Rice61 : for superconductors with the d symmetry,
the tunnel current between two superconducting crystals separated by a thin
boundary depends on the orientation of the order parameter with respect
to the interface. The current between the superconductors with numbers ij
is given by the formula

Isij = (Aij cos 2θi cos 2θj ) sin ΔΦij . (2.25)

Here, Aij is the constant which characterizes the junction of ij; θi and
θj are the angles of the crystallographic axes with the boundary plane;
and Φij is the difference of phases of the order parameters on both sides
88 Modern Aspects of Superconductivity: Theory of Superconductivity

Fig. 2.11 Scheme of the experiment on the measurement of a half-integer number of

quanta of the flow captured by a superconducting ring with three Josephson junctions
with the d symmetry of the order parameter.34

of the boundary. It was shown that a spontaneous magnetization, which

corresponds to the flow equal to a half of Φ0 , appears in a superconducting
ring with a single Josephson junction with the phase difference π. If the
ring has an odd number of π junctions, the result is the same. The direct
measurement of a half-quantum of the flow through such a ring would testify
to the d symmetry of the order parameter. The direct measurement of a half-
quantum of the flow was realized in Ref. 68. A scheme of the experiment is
shown in Fig. 2.11.

2.7 Thermodynamics of the d Pairing

in Cuprate Superconductors
2.7.1 Introduction
The basic question of the theory of superconductivity concerns the mech-
anism ensuring the pairing of electrons. In the BCS theory, it is the
electron–phonon interaction. Some recent theoretical models postulate the
mechanism of antiferromagnetic spin fluctuations,69–75 so that the electron
scattering on them can be the reason for the pairing of electrons. Spin fluctu-
ations play an important role in superconductors with heavy fermions.72, 73
The authors of Refs. 64 and 65 performed calculations of a value of the
 Physics of High-Tc Superconductors 89

superconductor gap, the critical temperature, the temperature dependence

of the resistance, and many other quantities, but the thermodynamics was
not considered. For this reason, it is of interest to calculate the thermo-
dynamics of antiferromagnetic spin fluctuations, namely the temperature
dependence of the heat capacity, its jump near Tc , and the parameter
R = ΔC/γTc equal to 1.42 by the BCS theory.
The purpose of this section is the calculation of the thermodynamics of
antiferromagnetic spin fluctuations, namely the electron heat capacity and
the jump of the heat capacity.
The thermodynamics of superconductors at low temperatures is deter-
mined by the excitation of two quasiparticles. In the traditional supercon-
ductors with pairing of the BCS type, the energy gap is isotropic (s pairing),
and the temperature dependence of the heat capacity has the exponential
form ∼ exp−Δ/kB T , where Δ is the superconductor’s gap. In the supercon-
ductors with anisotropic pairing, the temperature dependence of the heat
capacity has a power character, namely T n . The appearance of such tem-
perature dependences is related to the fact that the superconductor’s gap
has zeros on the Fermi surface.
As was noted above, the anisotropic pairing with the orbital moment
L = 2, i.e. with the dx2 −y2 symmetry, has the following functional form in
the k space:

Δ(k) = Δo [cos(kx a) − cos(ky a)], (2.26)

where Δo is the maximum value of the gap and a is the lattice constant.
The gap is strongly anisotropic in the direction (110) in the k space, and
the sign of the order parameter is changed along the directions kx and ky .
The main results of this section have been published in Refs. 78–81.

2.7.2 Antiferromagnetic spin fluctuations in HTSCs

For the first time, the idea of the possibility of electron pairing through spin
fluctuations was advanced by A. I. Akhiezer and I. Ya. Pomeranchuk.81
They showed that the indirect interaction of electrons through spin waves
in a ferromagnetic metal has the character of attraction in the triplet state
and hence, can lead to triplet pairing. Consider some experiments and facts
on antiferromagnetic spin fluctuations.
The basis for the hypothesis on the spin-fluctuation mechanism of pair-
ing consists in the fact that the stoichiometric compounds La2 CuO4 and
90 Modern Aspects of Superconductivity: Theory of Superconductivity

YBa2 Cu3 O6 are antiferromagnetic dielectrics. The doping of superconduc-

tors leads to the appearance of the metallic state and superconductivity.
The closeness of HTSCs to the antiferromagnetic transition with the wave
vector Q = (π/a, π/a) defines the important role of spin fluctuations, inter-
action with which forms the quasiparticle spectrum of electrons and can
simultaneously result in the Cooper pairing.
HTSCs are referred to the class of strongly correlated systems which
are theoretically studied in the frame of the Hubbard model. This model
describes the hops of electrons in the lattice with the matrix element t for the
nearest neighbors with regard to the Coulomb repulsion U , when the elec-
trons are positioned at the same site. The model is set by the Hamiltonian
 † 
H = −t Ciσ Cjσ + U ni↑ ni↓ , (2.27)
i,j,σ i,
†
where Ciσ (Cjσ )is the operator of creation (annihilation) of an electron at
†
the site i with spin σ, and ni↑ = Ciσ Cjσ is the number of electrons at
the site. In the given region of the parameters t, U and n (the electron
concentration), the appearance of magnetically ordered phases is possible.
Near the boundary of the existence of such a phase from the side of the
paramagnetic region, strong fluctuations of the magnetic order parameter,
i.e. paramagnons, must be manifested.
In the two-dimensional system of CuO layers in cuprate superconductors,
the electron spectrum is presented by the formula
ε(t) = −2t cos kx a + cos ky a), (2.28)
and the chemical potential μ is determined by the given electron concentra-
tion n. On the half-filling (n = 1), this spectrum has the nesting at the wave
vector q = Q, which induces a sharp peak in the spin susceptibility near this
point. This means an instability of the system relative to the formation of
the antiferromagnetic state with the wave vector Q and the intensification
of spin fluctuations near the point of the magnetic phase transition.
Near the half-filling, when the system is really antiferromagnetically
unstable, the numerical calculations indicate that the superconducting order
parameter has d symmetry, i.e. the gap depends on the wave vector by the
relation (2.26).
The gap (2.26) is an alternating function of the wave vector (Fig. 2.12)
and has zero values on the diagonals.
Figure 2.12 shows that the wave function of a Cooper pair is equal to
zero just on the diagonals of the square. Therefore, the repulsive interaction
on these diagonals does not act on the pair, and a Cooper pair with the
 Physics of High-Tc Superconductors 91

Fig. 2.12 Distribution of signs of the gap function in the limits of the Brillouin first band.

d symmetry survives even at large values of U . The superconductors with

the d pairing should have a number of particular properties which can be
observed in experiments. Many of these peculiarities are related to the zeros
of the order parameter. The quasiparticle spectrum at low temperatures
must give the power contribution to the thermodynamic properties, such
as the heat capacity, parameters of NMR, and the penetration depth of
a magnetic field, rather than the exponential one as in ordinary isotropic
superconductors.
The observation of such power contributions will indicate the presence of
a nontrivial order parameter with zeros on the Fermi surface. The totality of
experimental data for various HTSCs indicates the certain realization of the
anisotropic order parameter in them and, with a high probability, with the d
symmetry. The last circumstances present the important argument in favor
of the spin-fluctuation mechanisms of HTSCs. One of them was intensively
developed by Pines and his coworkers,69, 71 who used a phenomenological
form of the magnetic susceptibility with the parameters determined from
the experiments on cuprates.
Let us consider this model in more detail. We introduce a Hamiltonian
which involves antiferromagnetic spin fluctuations, as was done in works by
Pines69, 70 :

H = H0 + Hint , (2.29)

where H0 is the Hamiltonian of free electrons. The interaction is described

by the Hamiltonian
1 
Hint = g(q)s(q)S(−q), (2.30)
Ω q
92 Modern Aspects of Superconductivity: Theory of Superconductivity

where Ω is the cell volume and g(q) the interaction constant;

1 +
s(q) = Ψk+q σαβ Ψkβ (2.31)
2
α,β,k

where s(q) is the operator of spin density, σαβ is the Pauli matrix, Ψ+ k+q,α
is the operator of creation of an electron with the momentum k + q and
the spin projection α, Ψkβ is the operator of creation of a hole with the
momentum k and the spin projection β, and S(−q) is the operator of spin
fluctuations, whose properties are set by the correlator χ(q, ω)69, 70 ;
χij (n, m) the spin susceptibility which is modeled by

χQ
χ(q, ω) = , (2.32)
1+ ξ2 (q − Q)2 − iω/ωSF
qx > 0, qy > 0,

where χQ is the static spin susceptibility with the wave vector Q =

(π/a, π/a), ξ is the temperature-dependent antiferromagnetic correlation
length, and ωSF is the characteristic frequency of spin fluctuations of the
paramagnon energy. All parameters are taken from experiments, including
the data from NMR studies.69 In this case, the interaction constant is a free
parameter of the theory. It can be determined by calculating some quan-
tity with the help of the Hamiltonian and by comparing the result with
experiments.
We now define the quantities χQ and ωSF as
 2
ξ
χQ = χ0 β1/2 , (2.33)
a
Γ
ωSF = , (2.34)
(π(ξ/a)2 β1/2 )

where χ0 is the experimentally measured long-wave limit of the spin sus-

ceptibility, β = π2 , and Γ is the energy constant. The NMR data for the
compounds yield ξ(Tc ) = 2.3a, ωSF = 8 meV, Γ = 0.4 meV.
It is foreseen that the phenomenological Hamiltonian will give a self-
consistent description of the spin dynamics of the system in the sense that
the spin susceptibility calculated with its help (through the characteristics
of a quasiparticle spectrum, which themselves depend on the susceptibility)
will agree with values determined by the formula (2.32).
 Physics of High-Tc Superconductors 93

2.7.3 Continual model of antiferromagnetic spin fluctuations

The authors of Refs. 78–81 proposed a continual model for the spin-
fluctuation mechanism of pairing which allows one to efficiently solve the
problem of thermodynamics for this mechanism.
Let us consider this model in more detail. We calculate the thermody-
namics which is set by the Pines spin-fluctuation Hamiltonian. We write the
Hamiltonian in the lattice representation
H = H0 + Hint , (2.35)
 1 
H = −t ψ+
α (n)ψα (n + p) + Si (n)χ−1
ij (n, m)Sj (m)
n,p
2 n,m
 σ 
i
+g ψ+α (n) ψβ (n)Si (n), (2.36)
n
2 αβ

 1
H0 = −t ψ+
α (n)ψα (n + p) + Si (n)χ−1
ij (n, m)Sj (m), (2.37)
n,p
2 n,m
 σ 
i
Hint = g ψ+
α (n) ψβ (n)Si (n), (2.38)
n
2 αβ

where the sum is taken over all the sites of the infinite lattice (the lattice
constant is equal to a), ρ is the unit vector joining the neighboring sites, Si
.
is the spin operator, N = n ψ+ α (n)ψα (n) is the operator of the number of
particles, t is the half-width of the conduction band69, 70 and χij (n, m) is
the spin correlation function.
It is necessary to calculate the grand partition function:
exp{−βΩ(μ, β, g)} ≡ Tr exp{−β(H − μN )}, (2.39)
1
β= ,
kT
where μ is the chemical potential, g is the coupling constant and Ω(μ, β, q)
is the thermodynamic potential.
It is convenient to use the formalism of continual integration for a sys-
tem of Fermi particles. The method of functional integration, i.e. the inte-
gration in the space of functions, was proposed by N. Wiener in 1925, but
the physicists paid no attention to this method. Continual integrals were
introduced into physics by R. Feynman4 in the 1940s and were used for the
reformulation of quantum mechanics. Continual integration is one of the
most powerful methods of contemporary theoretical physics, allowing one
to simplify, accelerate and clarify the process of analytic calculations. The
94 Modern Aspects of Superconductivity: Theory of Superconductivity

application of this method to a system with an infinite number of degrees

of freedom enables one to develop, in such a way, the diagram theory of
perturbations.
The grand partition function can be written in the form of a continual
integral83 :
 )   β 
exp −βΩ = N dSi (n)dψα (n, τ)dψα (n, τ) exp −
+
dτL(τ) , (2.40)
n 0

with
   
∂
L(τ) = ψ+
α (n, τ)− μ ψα (n, τ) − t ψ+
α (n, τ)ψα (n + p, τ)
n
∂τ n,p
 σ 
i
+g ψ+
α (n, τ) ψβ (n, τ)Si (n, τ)
n
2 α,β

1
+ Si (n, τ)χ−1
ij (n, m, τ)Sj (m, τ), (2.41)
2 n,m
where L(τ) is the Lagrangian of the system, and N is the normalizing factor:
 )   β 
−1
N = dSi (n) dψα (n, τ) dψα (n, τ) exp −
+
dτL(τ, μ = g = 0) .
n 0
(2.42)
We will use the matrix formalism to construct the theory of perturbations
for the Green functions. It is convenient to introduce a four-component
bispinor (Majorana):
 
ψ
Ψ= , (2.43)
−σ2 ψ∗
 
ψ1
ψ= ,
ψ2
where σ1 are the Pauli spin matrices. The Majorana spinor is a Weyl spinor
written in the four-component form.
Now, we can write L in the form
  
1 0 ∂
L= Ψ(n, τ) Γ − Γ Γ5 μ δ̂ − t(τ − δ̂)Γ Γ5
0 0
(2.44)
2 n,p ∂τ
g
× ψ(n + p, τ) + Ψ(n, τ)Γi Γ5 Ψ(n, τ)S i (n, τ)
4 n
1
+ Si (n, τ)χ−1
ij (n, m, τ)Sj (m, τ), (2.45)
2 n,m
 Physics of High-Tc Superconductors 95

with
 
0 I
Γ =0
, (2.46)
I 0
 
0 −σi
Γ =
i
,
σi 0
 
1 0
I= ,
0 1

where Γi are the Dirac gamma matrices, Ψ̄ ≡ Ψ+ Γ0 , /δ = δn,n+p , and Γi are

given in the chiral representation.
There exists a connection between Ψ and Ψ̄ : Ψ = C Ψ̄T , where C is the
matrix of charge conjugation, C = Γ0 Γ2 , C T = −C. In terms of Ψ, the
partition function takes the form
 ) Rβ
−βΩ
e =N n dSi (n, τ) dΨ(n, τ)e− 0 dτL(τ) . (2.47)

We now use the method of bilocal operators84 to calculate the grand

partition function. In this way, we obtain the Schwinger–Dyson equation
and the equation of free energy. The details of calculations can be found in
Refs. 78–81.
In order to calculate Ω, we introduce a source of the bilocal operator84 :
  β "
e−βΩ(J) = N dS dΨe − dτ L(τ) (2.48)
0
#
1  β
+ Ψ̄(n, τ)J(m, n, τ, τ )Ψ(m, τ ) .
2 n,m 0

The full Green fermion function is determined by the formula


Gnm (J) = eβΩ(J) dS dΨ Ψ(n, τ)Ψ̄(m, τ ) exp{· · · } (2.49)

= 0|T Ψ(n, τ)Ψ̄(m, τ )|0.

Let us write the Schwinger–Dyson equation for the Green function G, by
using the method of the bilocal operator81 :
δF
= 0, (2.50)
δG
F (G) = βΩ(J = 0). (2.51)
96 Modern Aspects of Superconductivity: Theory of Superconductivity

First, we consider the case of a free system. By integrating over the fermion
fields in the functional integral, we obtain
1 1
F0 = βΩ(J) = − Tr Ln(G−1 0 − Γ Γ5 μ + J) + Tr Ln G0 ,
0 −1
(2.52)
2 2
−1
0 ∂
G0 = Γ − t(1 − δ̂)Γ Γ5
0
. (2.53)
∂τ
For the interacting system, the free energy takes the form
F = F0 + Fint . (2.54)
.
We represent Fint as a series in g: Fint = n=1 (g2 )Fn . In the lowest order
in g, we get the relation
g2
F1 = − Tr{Γi G T rΓj G − 2GΓi GΓj }χij , (2.55)
32
where the Green function for fermions and spins is as follows:
Ψ(x, τ)Ψ̄(y, τ ) = G(x, τ; y, τ ),
Si (x, τ)Sj (y, τ ) = χij (x, τ; y, τ ). (2.56)
Taking the condition δF/δG = 0 into account, we obtain the Schwinger–
Dyson equation:
g2 i
G−1 = G−1 0 − Γ 0
Γ5 μ + {Γ T rΓj Gχij − 2Γi GΓj χij }, (2.57)
12
where Γ0 and Γi are Dirac matrices (Fig. 2.13).
Here, the straight line corresponds to the Green function for fermions G,
and the wavy line to the Green function for spins χi,j (= δij χ). The equation
for free energy
g2
F1 = − Tr{Γi G TrΓj G − 2GΓi GΓj }χij , (2.58)
32

Fig. 2.13 Graphical form of Eq. (2.57).

Fig. 2.14 Graphical form of the right-hand side of Eq. (2.58).

 Physics of High-Tc Superconductors 97

where G and χij are, respectively, the Green fermion and spin functions,
corresponds to the contribution of two vacuum diagrams (Fig. 2.14).

2.7.4 Equation for the superconducting gap

From Eq. (2.57), it is easy to deduce the equation for a gap, which can be
found in Ref. 70. By executing the Fourier transformation
∞  π/a

 (2n+1)π
d2 k iωn τ−ik x ωn = β for fermions,
G(x, τ) = G(k, ωn )e
n=−∞ −π/a
(2π) 2
ωn = β
2nπ
for bosons,

we rewrite Eq. (2.57) in the momentum space as

∞  π/a 2
0−1 g2  d p
G−1 (k, iωn ) = G0 (k, iωn ) +
4β m=−∞ −π/a (2π)2

× [Γi Γ5 T rΓi Γ5 G(p, iωm ) − 2Γi Γ5 G(p, iωm ) Γi Γ5 ]

× χ(k − p, iωn − iωm ), (2.59)

where we set χij = δij χ,

 
0 {iωn − (ε(k) − μ)}I
G−1 (k, iωn ) =
0 {iωn + ε(k) − μ}I 0
≡ Γ0 iωn − Γ0 Γ5 (ε(k) − μ), (2.60)
ε(k) = −2t[cos kx a + cos ky a]. (2.61)

In accordance with the standard relation of the diagram technique, we

denote the free energy by Σ(k, iωn ). Then we have

0
G−1 (k, iωn ) = G−1
0 (k, iωn ) − Σ(k, iωn ), (2.62)

and Eq. (2.57) is the equation for Σ. We write the solution for Σ in the form

Σ = Γ0 A + Γ0 Γ5 B + Δ + Γ0 Γ5 Γi Δi , (2.63)

where A, B, Δ and Δi are functions of k and iωn .

By determining the matrix which is inverse to (2.62), we get

Γ0 (ωn − A) − Γ0 Γ5 (ε(k) − μ + B) + Δ + Γ0 Γ5
G(k, iωn ) = . (2.64)
(ωn − A)2 − (ε(k) − μ + B)2 − Δ2 − Δ2i
98 Modern Aspects of Superconductivity: Theory of Superconductivity

Substituting (2.64) and (2.63) in (2.63), we obtain the system of equations

for the functions A, B, Δ and Δi :

3g2  d2 p iωn − A(p, iωm )
A(k, iωn ) =
4β m (2π)2 D(iωm , p)

× χ(k − p, iωn − iωm ), (2.65)


3g2  d2 p ε(p) − μ + B(p, iωm )
B(k, iωn ) =
4β m (2π)2 D(iωm , p)

× χ(k − p, iωn − iωm ), (2.66)


3g2  d2 p Δ(p, iωm )
Δ(k, iωn ) =
4β m (2π)2 D(iωm , p)

× χ(k − p, iωn − iωm ), (2.67)


g2  d2 p Δi (p, iωm )
Δi (k, iωn ) = −
4β m (2π)2 D(iωm , p)

× χ(k − p, iωn − iωm ), (2.68)

where

D(iωm , p) = (ωn − A)2 − (ε(k) − μ + B)2 . (2.69)

Equations (2.67) and (2.68) correspond, respectively, to the singlet

and triplet pairings. Moreover, the singlet channel is characterized by the
repulsion, and the triplet one by the attraction. Therefore, it is necessary to
take the trivial solution, Δ = 0, of Eq. (2.67). In the following calculations,
we will neglect the contributions from the functions A and B, which leads
only to the renormalization of the wave function and the chemical poten-
tial, and we will study only Eq. (2.67). By neglecting Δ2 in the deno-
minator, we get the linearized equations for Δ0 which determine the critical
temperature Tc .
For the correlation function χ(q, iωn ), we will use the dispersion relations
 ∞
1 dω Im χ(q, ω )
χ(q, iωn ) = −
π −∞ iωn − ω
 ∞  ∞ 
1 dω Im χ(q, ω ) dω Im χ(q, −ω )
=− + . (2.70)
π 0 iωn − ω 0 iωn + ω
 Physics of High-Tc Superconductors 99

Fig. 2.15 Integration contour C = C1 + C2 .

In order to sum over m in (2.70), we consider the contour C = C1 + C2

(Fig. 2.15).
The following formula for ωn = (2n+1)π
β is valid:
∞
  
β F (ω) dω β F (ω) dω
F (iωn ) = − =− , (2.71)
n=−∞
2πi C exp βω + 1 2πi C exp −βω + 1
or
∞
 
β 1 βω
F (iωn ) = − F (ω)th dω. (2.72)
n=−∞
2πi 2 C 2

Taking into account that Im χ(q, −ω) = −Im χ(q, ω), Eq. (2.70) can be
reduced to
 
1 ∞   1 1
χ(q, iωn ) = − dω Im χ(q, ω ) − . (2.73)
π 0 iωn − ω iωn + ω
With the help of the formula (2.72), we can rewrite (2.67) in the form
   ∞
g2 d2 p
Δ(k, iωn ) = dωF (p, ω) dω Im χ(k − p, ω )
8π(2πi) (2π)2 C 0

1 1 βω
× 
− 
th , (2.74)
iωn − ω − ω iωn − ω + ω 2
where
Δ(k, iωn)
F (p, ω) = .
ω2 − (ε(p) − μ)2
100 Modern Aspects of Superconductivity: Theory of Superconductivity

After some calculations and transformations, we get the general formula

for the superconducting parameter79–81 :
  β(ε(p)−μ)
g2 π/a d2 p th
Re(k, ω) = 2
8 −π/a (2π) 2 2(ε(p) − μ)
× [Re χ(k − p, ω − (ε(p) − μ))
+ Re χ(k − p, ω + (ε(p) − μ))]
 ∞
dυ βυ
× ReΔ(p, υ) − 2 cth Im χ(k − p, υ)
0 π 2

(ω − υ)2 − (ε(p) − μ)2 − δ2
×
((ω − υ)2 − (ε(p) − μ)2 − δ2 )2 + 4(ω − υ)2 δ2
× ReΔ(p, ω − υ)
(ω + υ)2 − (ε(p) − μ)2 − δ2
+
((ω + υ)2 − (ε(p) − μ)2 − δ2 )2 + 4(ω − υ)2 δ2
#
× ReΔ(p, ω + υ) (2.75)

ω2 − (ε(p) − μ)2 − δ2
Re F (p, ω + iδ) = ReΔ(p, ω) . (2.76)
[ω2 − (ε(p) − μ)2 − δ2 ]2 + ω2 δ2
If the superconducting gap depends weakly on the frequency, then
Δ(k, ω)  Δ(k, 0). In this case, we set ω = 0 in Eq. (2.75). Considering
the relation Reχ(q, ω) = Re χ(q, −ω) and making the simple transforma-
tions, we obtain an equation similar to that in Ref. 70:
 
g2 π/a d2 p th β(ε(p)−μ)
Δ(k) = Re χ(k − p, ε(p) − μ) 2
8 −π/a (2π)2 ε(p) − μ
 ∞
dυ βυ
+ 2 cth Im χ(k − p, υ)
0 π 2

(ε(p) − μ)2 − υ2 + δ2
× Δ(p). (2.77)
[(ε(p) − μ)2 − υ2 + δ2 ]2 + 4υ2 δ2
It is important that our method of calculations yields a more general
equation for the superconducting gap than that obtained by D. Pines. Our
equation coincides with the Pines equations after some simplification, which
is the test for our calculations.
 Physics of High-Tc Superconductors 101

2.7.5 Thermodynamic potential of antiferromagnetic

spin fluctuations
For the free energy, we have the equation
1
F (G) = βΩ = − Tr[lnGo G−1 + (G−1
o + Γ Γ5 μ)G ]
o −1
2
g2
− Tr{Γi GT rΓj Gχij − 2GΓi GΓj χij }, (2.78)
32
where G satisfies Eq. (2.57). We now multiply Eq. (2.58) by G and take the
trace, Tr. This allows us to obtain
g2
Tr{Γi GT rΓj Gχij −2GΓi GΓj χij } = −Tr{(G−1
o +Γ Γ5 μ)G−1}.
o
(2.79)
8
Using (2.57), we can rewrite the relation (2.79) for the functional of free
energy calculated with the use of solutions to the Schwinger–Dyson equation
in the form

1 −1 1 −1 1
βΩ = − Tr lnGo Γ + (Go + Γ Γ5 μ)G − .o
(2.80)
2 2 2
By implementing the Fourier transformation, we get
∞  π/a
1 d2 k
Ω= V Tr ln Go (k, iwn )G−1 (k, iwn )
2β n=−∞ −π/a (2π)2

1 −1 1
+ (Go (k, iwn ) + Γ Γ5 μ)G(k, iwn ) − ,
o
(2.81)
2 2
where V is the two-dimensional volume (the area of the cuprate plane) and
Tr stands for the trace of a matrix.
For the free energy functional Ω(Δ), we have the formula
∞ 
V  d2 k
Ω=− Tr ln Go (k, iωn )−1 (k, iωn )
2β n=−∞ (2 ∗ π)2

1 −1 1
+ (Go (k, iωn ) + r r5 μ)G(k, iωn ) − ,
o
(2.82)
2 2

where G−1
o (k, iωn ) = r iωn − r r5 ε(k), and G(k, iωn ) is given by the for-
o o

mula (2.57). We note that the functional Ω(Δ, μ) is normalized so that

Ω(Δ = 0, μ = 0) = 0.
102 Modern Aspects of Superconductivity: Theory of Superconductivity

By calculating the trace of the r matrix, we obtain the expression

∞ 
V  d2 k ω2n + (ε − μ)2 + Δ2
Ω(Δ, μ) = − ln
β n=−∞ (2 ∗ π)2 ω2n + ε2

Δ2
− 2 . (2.83)
ωn + (ε − μ)2 + Δ2
By using the formula
∞  " √
 ω2 + b2 ∞
β β a2 + x
ln 2n = dx √ th
n=−∞
ωn + a2 0 2 a2 + x 2
√ #
β β b2 + x ch βb
− √ th = 2 ln βa2
, (2.84)
2
2 b +x 2 ch 2
we get the equation for the thermodynamic potential:

 
β
d2 k 2 ch 2 (ε − μ)2 + δ2
Ω(Δ) = V − ln
(2π)2 β ch βε
2

Δ2 β (ε − μ)2 + Δ2
+ th . (2.85)
2 (ε − μ)2 + Δ2 2
By making some transformations and calculating the traces of Γ matrices,
we arrive at the equation
 
V dw d2 k 4
Ω(Δ) − Ω(0) = Δ (k, w)
2 C 2πi (2π)2 i
1 1
× . (2.86)
eBw + 1 [w − (ε(k − μ)2 ]2
2

By expanding the contour C and calculating the contribution to the

integral at poles w = ±(ε(k) − μ), we get

V d2 k Δ4i (k)
Ω(Δ) − Ω(0) =
8 (2π)2 [ε(k) − μ]2
  
β 2 β(ε(k) − μ)
× 1 − th
2 2
1 ε(k) − μ
− th , (2.87)
ε(k) − μ 2
where ε(k) describes the spectrum of two-dimensional electrons; the free
energy F is connected with the thermodynamic potential Ω by the relation
 Physics of High-Tc Superconductors 103

F = βΩ; Ω(Δ) and Ω(0) are the thermodynamic potentials at T < Tc and
T > Tc , respectively, and V is the two-dimensional volume (the area of a
cuprate layer).
It is easy to verify that despite the presence of the factors [ε(k) − μ] in
the denominator of the formula (2.87), no singularity on the Fermi surface
ε(k) − μ is present.
Equations (2.85) and (2.87) (at T ∼ Tc ) can be a basis for calculations of
various thermodynamic quantities, including a jump of the heat capacity.
A similar method of calculations was developed by S. Weinberg.85

2.7.6 Heat capacity of the d pairing

Equation (2.85) yields the following formula for the thermodynamic poten-
tial: Ω(Δ)79

 
β
d2 k 2 ch 2 ( (ε(k) − μ)2 + Δ(k)2 )
Ω(Δ) = V − ln
(2π)2 β ch βε
2

Δ(k)2 (ε(k) − μ)2 + Δ(k)2
+ th . (2.88)
2 (ε(k) − μ)2 + Δ(k)2 2
Here, V is the two-dimensional volume (the area of a cuprate layer), Δ(k)
is the superconductor gap, k is the momentum of an electron, and ε(k) =
−2t[cos(kx a) + cos(ky a)] gives the spectrum of two-dimensional electrons.
The heat capacity is calculated by the formula
∂2Ω
C = −T . (2.89)
∂T 2
The results of calculations of the heat capacity C are given in Ref. 45.
We have carried out computer-based calculations of the heat capacity of
the compound YBa2 Cu3 O6.63 . The equation for the superconductor gap Δ
was obtained in Sec. 2.7.5. The task of solving the integral equation was
reduced to that of an algebraic equation which was solved by the method
of iterations.
The equation for the superconductor gap depends on the spin correla-
tion function. For it, the necessary data were taken from Refs. 69 and 70.
In calculations, we used the following values of parameters of the correlator:
ωSF (TC ) ≈ 7.7 meV, χS (Q) = 44 eV, ξ/a ≈ 2.5, t = 2 eV, μ = 0.25 and
Tc = 95 K. We chose a value of the constant g which satisfies the relation
2Δ/kTc = 3.4, similar to that in Refs. 66 and 67. The results of numerical
104 Modern Aspects of Superconductivity: Theory of Superconductivity

calculations are presented in Fig. 2.15, where we can see the temperature
dependence of the electron heat capacity: 1 — a curve which is an approxi-
mation of the results of computer-based calculations (crosses); 2 — a curve
which describes the exponential BCS dependence. These results are in good
agreement with experiments made by A. Kapitulnik and co-workers86 at low
temperatures.
The calculations indicated a square dependence of the heat capacity on
the temperature in the temperature range from zero to the superconducting
temperature. It was shown45,46,79,80 that the dependence has the form
C = AT 2 , (2.90)
where A = 0.126 J/K3 mol.

2.7.7 Heat capacity jump in superconductors

In what follows, we present the results of calculations of a jump of the heat
capacity near the critical temperature,
∂ 2 ΔΩ 1
ΔC = −T 2
, β= , (2.91)
∂T T
and evaluate the parameter R = ΔC/γTc . Omitting the awkward and quite
complicated details, we will give the final results.
We obtained R = 1.6. This value of the parameter is larger than that in
the BCS theory, where R = 1.43. It is worth noting that the heat capacity
jump is very sensitive to the doping. Figure 2.16 shows the results of recent
calculations by one of the authors81 for R as a function of the doping. It
is seen that this parameter depends strongly on the doping. By using these
results, it is possible to evaluate the condensation energy for HTSCs which
is proportional to R2 .82
The calculations imply that the heat capacity depends on the temper-
ature as T 2 . Analogous results were obtained for the physics of heavy
fermions.91 It was shown in this work that d symmetry leads to a linear
dependence of C/T on the temperature, whereas s symmetry is related to
the exponential dependence of this quantity on the temperature. The linear
dependence of C/T on the temperature was observed in the experimen-
tal works (Refs. 87–90) for YBaCuO superconductors. The experimental
results are presented in Fig. 2.17. As was mentioned above, the problem of
the determination of the symmetry of a gap in cuprate superconductors is
 Physics of High-Tc Superconductors 105

Fig. 2.16 Temperature dependence of the electron heat capacity: 1 — a curve which
is an approximation of the results of computer-based calculations (crosses); 2 — a curve
which describes the exponential BCS dependence, with the points giving the experimental
data.86

Fig. 2.17 The parameter R(Tc ), which characterizes the heat capacity jump as a function
of the doping for Y0.8 Ca0.2 Ba2 Cu3 O7−x . The solid curve presents the results of theoretical
calculations; the points mark the experimental data.89
106 Modern Aspects of Superconductivity: Theory of Superconductivity

urgent at the present time. Many experiments have confirmed the d sym-
metry of the pairing.67, 68 In particular, we mention new experiments,68 in
which the researchers have studied the quantization of magnetic flows in a
ring which includes three Josephson junctions. These works supporting the
existence of d pairing in HTSCs were marked by the Barkley Prize. It was
also shown that the sign of the order parameter in YBa2 Cu3 O7−δ depends
on the direction. This dependence corresponds to the dx2 −y2 symmetry.
The temperature dependence of the electron heat capacity obtained in
our calculations79, 80 is related to the d pairing. These thermodynamic cal-
culations can be a supplementary test in the determination of the pairing
symmetry in HTSCs.
It should be noted that thermodynamic calculations clarifying the behav-
ior of HTSCs have also been performed in Refs. 92–95 in the frame of the
other mechanisms of pairing.

2.8 Summary
High-temperature superconductivity is a dynamical field of solid-state
physics which has been intensively developed by theorists and experi-
menters. By summarizing the physical properties and the mechanisms of
superconductivity in new HTSCs , we wish to separate the main properties
and the theoretical problems arising in the studies of HTSCs.
In order to comprehend the nature of the superconducting state, it is nec-
essary to construct a consistent microscopic theory which would be able to
describe superconductive and normal properties of HTSCs. It is seen from
the above-presented survey that many mechanisms of pairing in HTSCs,
proposed as the explanations for this phenomenon, have been advanced. On
the whole, the most probable seems to be the “synergetic” mechanism of
pairing, whose constituents are the electron–phonon interaction, spin fluc-
tuation, and other types of interaction in cuprate planes.
We believe that the known challenging properties — which are contra-
dictory to a certain extent — of many chemical compounds in the super-
conducting and normal phases can be explained only by considering the
interaction of all the degrees of freedom, such as lattice-, electron- and spin-
related ones. In this case, it is also necessary to take into account the com-
plicated structure of HTSCs. Further development of the theory will require
not only the execution of bulky numerical calculations, but also the solu-
tion of a number of fundamental problems concerning the strong electron
 Physics of High-Tc Superconductors 107

correlations. It seems to us that one of the key problems in the field of

superconductivity is the mechanism and the symmetry of pairing.
We have shown above that some thermodynamical problems of high-
temperature superconductivity, in particular the problem of antiferromag-
netic spin fluctuations, can be efficiently solved within the method of
continual integrals. We conclude with the following remarks:
(1) The thermodynamic theory of antiferromagnetic spin fluctuations in
HTSCs has been constructed. The method of functional integration was
applied to the calculations of the superconductor gap and the thermo-
dynamic potential. The Schwinger–Dyson equation and the equation for
the free energy were deduced.
(2) From the Schwinger–Dyson equation, the equations for the supercon-
ductor gap, which are used in the numerical calculations within the
thermodynamics of HTSCs, were constructed. The analytic formulas
for the thermodynamic potential and its jump were obtained.
(3) Numerical calculations of the temperature dependence of the electron
heat capacity were carried out, and it was shown that the electron heat
capacity is proportional to the square of the temperature. It was empha-
sized that such a temperature dependence is related to the d pairing.
It was shown that the measurement of the temperature dependence of
the heat capacity can be a supplementary test in the determination of
the type of pairing symmetry in HTSCs. The jump of the heat capacity
of cuprate superconductors near the critical temperature was evaluated
as well.
 CHAPTER 3

Multiband Superconductivity

3.1 Introduction
In this chapter, we present the theory of superconductivity with regard
to a complicated multiband structure of superconductors. Calculations of
the band structure of cuprate superconductors indicate that several energy
bands intersect one another on the Fermi surface in these compounds,96 and
the Fermi surface passes through high-symmetry points which correspond
to the Lifshitz electronic topological transition. In addition, the discovery of
two-gap superconductivity in two-band superconductors MgB2 allows one
to consider the possibility of using a multiband theory of superconductivity.
In Sec. 3.2, we analyze the problems of multiband superconductivity and
superconductivity at room temperature. In Sec. 3.3, we study the physical
properties of superconductors MgB2 and problems of two-gap supercon-
ductivity, as well as the phase diagram for superconductors MgB2 on the
basis of the renormalization group approach. In the field of superconduc-
tivity, we meet the problem-maximum — the creation of room-temperature
superconductors. We consider this problem in our book and give some rec-
ommendations on the search for these superconductors. The results of this
chapter were obtained by the authors and published in Refs. 96–103.
The theory of superconductivity arising from the electron–phonon inter-
action mechanism by Bardeen, Cooper and Schrieffer (BCS)104 has been
well established, and it is now the standard theory for superconductiv-
ity.2, 3, 6, 15, 16, 104 In Chap. 2, we have considered non-BCS mechanisms via
spin fluctuations, charge fluctuations (plasmons) and electron excitations
(excitons), which have attracted great interest, for example in relation to the
possibility of high-Tc superconductivity. These mechanisms have the com-
mon characteristic that the electron–electron (e–e) interaction is an origin

108
 Multiband Superconductivity 109

of the superconductivity. After the discovery of the high-Tc copper oxides,2

Anderson6 emphasized an important role of the e–e interaction. Over the
past decade, many non-BCS theories106–111 have been proposed, but they
do not converge as a unified and well-accepted theory yet. On the other
hand, experimental studies on copper oxides have revealed the following
characteristics:

(1) These species are antiferromagnets before doping, in accordance with

the importance of the e–e interaction;
(2) The high-Tc superconductivity appears in the intermediate region of the
metal–insulator transition and disappears in the metallic or overdoped
region.112–114

Accumulated experimental results on the species and related materials

suggest a guiding principle that the doping in magnetic systems, more gen-
erally charge-transfer (CT) insulators, may provide several exotic phases
which are

(a) Ferromagnetic metal or insulator,

(b) Spin glass,
(c) Paramagnetic metals,
(d) Antiferromagnetic metals,
(e) Ferrimagnetic metal or insulator,
(f) Charge- or spin-mediated superconductor.

Relative stabilities of these phases should be dependent on several factors.

The theoretical description of such phases and phase transitions in a sys-
tematic fashion is quite hard. Recently, the importance of multiband effects
in high-Tc superconductivity has been pointed out.3, 16, 104 In the framework
of the two-particle Green function techniques,98, 101 it is shown that a class
of new so-called coupled states arises in the electron–phonon system. The
model numerical calculations have shown that the superconducting (SC)
gap depends on the number of bands crossing the Fermi level, and the tem-
perature dependence of the SC gap for high-Tc superconductors is more
complicated than that predicted in the BCS approach. We have also inves-
tigated anomalous phases in a two-band model by using the Green function
techniques.101, 103 The expressions for the transition temperature for sev-
eral phases have been derived, and the approach has been applied to the
superconductivity in molecular crystals by charge injection and field-induced
superconductivity.
110 Modern Aspects of Superconductivity: Theory of Superconductivity

In this chapter, we investigate the superconductivity by using the two-

band model and the two-particle Green function technique. In the framework
of the two-band model, the coupled states in the electron system and the
conditions under which the coupled states can appear are investigated. We
apply the model to the electron–phonon mechanism within the traditional
BSC method, the electron–electron interaction mechanism for high-Tc super-
conductivity, and the cooperative mechanism in relation to the multiband
superconductivity.

3.2 Multiband Hamiltonian

In this section, we summarize the two-band model for superconductivity,
introduce a two-particle Green function and investigate the spectral prop-
erties of the model.

3.2.1 Hamiltonian
We start from the Hamiltonian for two bands i and j :
H = H0 + Hint , (3.1)
with
 
H0 = [εi − μ]a+
ikσ aikσ + [εj − μ]ajkσ ajkσ ,
+
(3.2)
k,σ

1  
Hint = αβγδ aip1 α aip2 β aip3 γ aip4 δ + (i → j)
Γiiii + +
4
δ(p1 +p2 ,p3 +p4 )αβγδ

+ Γiijj
αβγδ aip1 α aip2 β ajp3 γ ajp4 δ + (i → j)
+ +


+ Γijij + +
αβγδ aip1 α ajp2 β aip3 γ ajp4 δ + (i → j) , (3.3)

where Γ is the bare vertex part,

Γijkl
αβγδ = ip1 α jp2 β | kp3 γ lp4 δ δαδ δβγ − ip1 α jp2 β | lp4 δ kp3 γ δαγ δβδ ,
(3.4)

with

ip1 αjp2 β|kp3 γlp4  = dr1 dr2 φ∗ip1 α (r1 )φ∗jp2 β (r2 )

× V (r1 , r2 )φkp3 γ (r2 )φlp4 δ (r1 ), (3.5)

 Multiband Superconductivity 111

and a +ipσ (a ipσ ) is the creation (annihilation) operator corresponding to the

excitation of electrons (or holes) in the ith band with spin σ and momentum
p, μ is the chemical potential and φ∗ipα is a single-particle wave function.
Here, we suppose that the vertex function in Eq. (3.3) consists of the effec-
tive interactions between the carriers caused by the linear vibronic coupling
in several bands and the screened Coulombic interband interaction of car-
riers. When we use the two-band Hamiltonian (3.1) and define the order
parameters for the singlet exciton, triplet exciton and singlet Cooper pair,
the mean field Hamiltonian is easily derived.96–100, 115, 116 Here, we focus on
four-electron scattering processes:

g1 = ii | ii = jj | jj, (3.6)

g2 = ii | jj = jj | ii, (3.7)
g3 = ij | ij = ji | ji, (3.8)
g4 = ij | ji = ji | ij. (3.9)

g 1 and g 2 represent the intraband two-particle normal and umklapp scat-

terings respectively, g 3 is the interband two-particle umklapp process and
g 4 indicates the interband two-particle interaction on different bands (see
Fig. 3.1). Note that Γ’s are given by
jjjj
αβγδ = Γαβγδ = g1 (δαδ δβγ − δαγ δβδ ),
Γiiii

Γiijj jjii
αβγδ = Γαβγδ = g2 (δαδ δβγ − δαγ δβδ ),

Γijij jiji
αβγδ = Γαβγδ = g3 δαδ δβγ − g4 δαγ δβδ , (3.10)

where an antisymmetrized vertex function Γ is considered to be a constant

independent of the momenta.

Fig. 3.1 Electron–electron interactions. Dependence of g on the direction in the momen-

tum space is ignored in this model, gx (k) ≈ gx (x = i, j). We assume that g x is constant.
112 Modern Aspects of Superconductivity: Theory of Superconductivity

The spectrum is elucidated by the Green function method. Using Green

functions which characterize the CDW (charge-density-wave), SDW (spin-
density-wave) and SSC (singlet superconducting) phases, we obtain a
self-consistent equation, according to the traditional procedure.13, 96, 115, 116
Then, we can obtain expressions for the transition temperature for some
cases.
In the framework of the one-band model, the electronic phases are char-
acterized by
−g2 − 2g3 + g4 > 0, for CDW,
g2 + g4 > 0, for SDW,
−g1 > 0, for SSC.
In the framework of the two-band model, we have already derived
expressions of the transition temperature for CDW, SDW and SSC. In
the previous papers,13, 96, 115, 116 we have investigated the dependence of Tc
on the hole or electron concentration for the superconductivity of copper
oxides by using the two-band model and have obtained a phase diagram of
Bi2 Sr2 Ca1−x Yx Cu2 O8 (Bi-2212) by means of the above expressions for the
transition temperature. The dependence of Tc on Δp can be reproduced in
agreement with the experiment.13, 96, 115, 116 Recently, we have also obtained
phase diagrams of copper oxides, anthracene, oligothiophene and C60 crys-
tals by using the analytic solutions.13, 96, 115, 116

3.2.2 Two-particle Green function

In this subsection, we introduce a two-particle Green function96–98 (see also
Appendix A) to investigate the physical properties of superconductivity in
the two-band model. In statistical mechanics, Green functions are a conve-
nient generalization of the notion of correlation functions. Like the latter,
the former are closely related to the calculations of observables and give the
well-known advantages in the construction and the solution of equations.
First, let us define one-particle Green functions:
1  2
Gυσ (k, t − t) = −iT aυkσ (t)a+ 
υkσ (t ) , (3.11)
where σ and υ represent labels for a spin and a band, respectively. The
equation for a Green function derived by using the two-band model (3.1) is
written as
3 4
∂1  
2  ∂aυkσ (t) +   
i −iT aυkσ (t)aυkσ (t ) = δ(t − t ) + −iT i
+
aυkσ t , (3.12)
∂t ∂t
 Multiband Superconductivity 113

where
∂aυkσ (t)
i = [aυkσ , H]. (3.13)
∂t
The equation for the Green function (3.12) is rewritten after inserting
Eq. (3.13) as
∂1  2
i −iT aυkσ (t)a+ υkσ (t )
∂t
1  2
= δ(t − t ) + [ευ − μ] −iT aυkσ (t)a+ 
υkσ (t )
1 
+ g1 (δσδ δβγ − δσγ δβδ )Gγδβσ 
2υυυυ (p3 , p4 , p2 , k; t, t )
2
βγδ δ(k+p2 ,p3 +p4 )

1  
+ g2 (δσδ δβγ − δσγ δβδ )Gγδβσ 
2υ υ υυ (p3 , p4 , p2 , k; t, t )
2 
υ βγδ δ(k+p2 ,p3 +p4 )

1  
+ (g3 δσδ δβγ − g4 δσγ δβδ )Gγδβσ 
2υυ υ υ (p3 , p4 , p2 , k; t, t ),
2 
υ βγδ δ(k+p2 ,p3 +p4 )
(3.14)

where
1  2
Gγδβσ  
2υυυυ (p3 , p4 , p2 , k; t, t ) = −iT aυp3 γ (t)aυp4 δ (t)aυp2 β (t − 0)aυkσ (t ) ,
+ +

(3.15)
1  2
Gγδβσ  
2υ υ υυ (p3 , p4 , p2 , k; t, t ) = −iT aυ p3 γ (t)aυ p4 δ (t)aυp2 β (t − 0)aυkσ (t ) ,
+ +

(3.16)
1  2
Gγδβσ  
2υυ υ υ (p3 , p4 , p2 , k; t, t ) = −iT aυp3 γ (t)aυ p4 δ (t)aυ p2 β (t − 0)aυkσ (t ) .
+ +

(3.17)

υ indicates a band different from υ. To calculate the density of electron

states, we have to focus on the case where t → t−0. The two-particle Green
functions in Eq. (3.14) is rewritten as G2 (p3 , p4 , p2 , k; t − t ) (t → t − 0).
In this study, we investigate only the spectral properties of two-particle
Green functions for the superconductivity. Therefore, we focus on the fol-
lowing two-particle Green function:
1  2
Gγδβσ  +  + 
2υυυυ (p3 , p4 , p2 , k; t, t ) = −iT aυp3 γ (t)aυp4 δ (t)aυp2 β (t )aυkσ (t ) . (3.18)

For simplicity, we consider only three cases: (i) g1 = 0 and others = 0;

(ii) g2 = 0 and others = 0; (iii) g1 = 0, g2 = 0 and others = 0.
114 Modern Aspects of Superconductivity: Theory of Superconductivity

3.2.3 Traditional superconductivity

In general, in the framework of the BCS theory, the Hamiltonian is
described by a single-band model. In the effective electron–electron inter-
action (Eq. (3.1)), we consider that g1 = 0 and others = 0 and focus only
on the single-band model. According to the approach101 used for phonon
systems which is based on the method of Bogoliubov and Tyablikov,148 we
can derive the equation for a two-particle electron Green function. The spec-
tral features of the electron system in the mentioned region of energy are
described by the Fourier transform of this function. For the simplest case of
a one-electron zone crossing the Fermi energy level, it can be given as
.
γδβσ 
f (k, k , ω) σ,σ φ(σ, σ )
G2υυυυ (p3 , p4 , p2 , k; t − t ) = . , (3.19)
1 − g1 q K(ω, k, k , q)

where
2 − nυk+q − nυk −q
K(ω, k, k , q) = . (3.20)
2ω − ευk+q − ευk −q

n k indicates the filling number of electrons and g 1 is the effective Fourier

transform of the e–e interaction. If the e–e interaction constant renormal-
ized by the electron–phonon interaction becomes negative, coupled states
will appear in the electron system. In the previous papers,96–98 we have pre-
sented analysis of the spectral properties of the two-particle Green function.
According to the same procedure,96–98 we obtain the equation for coupled
states in the electron system:
 
 Δ 

1 − g1 N (εf ) −ln1 −  = 0, (3.21)
a

where
√  
N (εf ) = 2πm∗υ m∗υ ε 2− nυk+q − nυk −q  , (3.22)
ε=εf

1
nυk =    . (3.23)
exp ευk − εf /T + 1

a = 2(ω − εf − Δυ − E), E = k2 /2m, ε = q 2 /2m and m∗υ = mυ /m. m∗υ means

the reduced effective mass of the electron in the crystal energy zone. m is
the mass of the free electron εf — the Fermi level. If g1 < 0, we can find
solutions to Eq. (3.21) for superconductivity.
 Multiband Superconductivity 115

3.2.4 Copper oxides

In copper oxides, the effective e–e interaction g2 is important for realizing
the high-Tc superconductivity.135, 136 Therefore, we consider that g2 = 0 and
others = 0. The two-particle Green function (3.18) is rewritten as
.
γδβσ 
f (k, k , ω) σ,σ φ(σ, σ )
G2υυυυ (p3 , p4 , p2 , k; t − t ) = . , (3.24)
1 − g22 q,q K2 (ω, k, k , q, q )
where
 υ υ
 υ υ

  2 − n k+q−q  − nk −q+q 2 − n k+q − n k  −q
K2 ω, k, k , q, q =     .
2ω − ευk+q−q − ευk −q+q 2ω − ευk+q − ευk −q
(3.25)
According to a similar procedure in previous papers, 96, 98 we study the sit-
uation near the extremum (minimum or maximum) of the electron zone.
Then we suppose that k = k = k0 + k , and ευk0 = ευ corresponds to the
extremum of the zone. We expand the energy in the momentum k ± q in a
series up to terms of the second order and suppose that the energy extremum
is located near the Fermi-level energy. Then, the sum in the denominator of
Eq. (3.24) is approximately reduced to the following expression:
  
  υ   υ   2Δ 2Δ 
K2 (ω, k, k , q, q ) ≈ N εf N εf ln 1 −
 
1− . (3.26)
a  υ aυ 
q,q

Therefore, we obtain the equation for coupled states in the electron system:
  
 υ   υ   2Δ 2Δ 
2 
1 − g2 N εf N εf ln 1 − 1− = 0. (3.27)
aυ aυ 
From Eq. (3.27), we find the possibility of the existence of a solution for
the coupled states, if g2 = 0. Thus, the effective e–e interaction g 2 with a
positive value contributes to the superconductivity.

3.2.5 Cooperative mechanism

Here, we consider that g1 = 0, g2 = 0 and others = 0. In a similar way, the
two-particle Green function (3.18) is approximately derived as
Gγδβσ 
2υυυυ (p3 , p4 , p2 , k; t − t )
.
f (k, k , ω) σ,σ φ(σ, σ )
= . 
. 
,
[1 − (g1 + g2 ) q K(ω, k, k , q)][1 − (g1 − g2 ) q K(ω, k, k , q)]
(3.28)
116 Modern Aspects of Superconductivity: Theory of Superconductivity

where K(ω, k, k , q) is as given by Eq. (3.20). The summation in the denom-

inator of Eq. (3.28) is performed in a similar way, and the equation for
coupled states in the electron system is approximately derived as

   
 Δ   Δ 
1 − (g1 + g2 )N (εf ) ln1 −  1 − (g1 − g2 )N (εf ) ln1 −  = 0.
a a
(3.29)
When g1 + g2 < 0 or g1 − g2 < 0, we can find solutions to this equation.

3.2.6 Room-temperature superconductors

In the previous subsections, we have approximately calculated two-particle
Green functions for three cases — traditional superconductivity, copper
oxides and the cooperative mechanism — in the framework of a two-band
model. From these Green functions, we have derived the equation for cou-
pled states for each case. In the case of a single-band model, which indi-
cates traditional superconductivity such as the BCS theory, it is necessary
that the effective e–e interaction be negative (g1 < 0) in order to realize
the superconductivity. The maximal transition temperature for the super-
conductivity predicted by the theory is about 40 K. On the other hand,
we can expect, in a two-band model for negative g 1 , that the transition
temperature becomes higher than that derived within a single-band model,
because of the tunneling of Cooper pairs between two bands. The tunneling
of Cooper pairs causes stabilization of the order parameter of the super-
conductivity.119, 120, 129 In the framework of a two-band model, we consider
that the Fermi energy level crosses with two bands. The results derived
from the two-particle Green function in the previous section suggest that
the superconductivity appears for g2 < 0 or g2 > 0. Note that g2 con-
tributes to SDW. From the results derived from calculations involving g 1
and g 2 (cooperative mechanism), we expect a higher Tc than that of copper
oxides.
On the basis of the results obtained in this section, we present a schematic
diagram for superconductivity, shown in Fig. 3.2. The mechanism of high-Tc
superconductivity of materials such as copper oxides might be around the
cooperative mechanism in the figure. In what follows, we calculate a two-
particle Green function in the two-band model and derive an equation for
coupled states. In the framework of the two-band model, the results predict
that superconductivity appears even if the e–e interaction is positive. We can
 Multiband Superconductivity 117

Fig. 3.2 Schematic diagram for superconductivity. MB means multiband effects.

expect that the transition temperature is higher (300 K) than that for copper
oxides by the cooperative mechanism.
Let us discuss the problem of superconductors operating at room tem-
perature (RTSCs). It is obvious that the main task in the field of super-
conductivity is the fabrication of materials with superconducting properties
at room temperature. The study of HTSCs is only a stage on the way to
the main purpose, namely to the development of RTSCs. At the present
time, the highest known Tc = 135 K (at the atmospheric pressure). Now
there occurs a wide-scale search for such superconductors. Since none of the
known physical laws allows one to exclude the possibility for RTSCs to exist,
the future discovery of RTSCs seems to be without doubt. New materials
are created by means of physical and chemical modifications of the known
compounds, including the application of nanotechnological methods. The
creation of new materials with preassigned physical properties is one of the
actual problems of modern science. At present, various approaches are being
developed in order to avoid the labor-consuming sorting of different chem-
ical compounds and conditions of synthesis, i.e. to optimize the solution of
this problem. One of the most efficient and promising solution is the method
of structural design. In Ref. 121, the structural design was used in solving
the problems related to the search for new HTSCs on the basis of complex
copper oxides.
We now present the recommendations concerning the search for new
HTSCs with higher critical temperatures which were advanced by the Nobel
118 Modern Aspects of Superconductivity: Theory of Superconductivity

Prize winner K. A. Müller.122 He divided the well-known HTSCs into three

classes:
(1) Layered cuprates,
(2) MgB2 ,
(3) Doped fullerenes of the type K3 C60 .
Müller emphasizes that the discovery of cuprate superconductors was pro-
moted by the concept of Jahn–Teller polarons. Two singlet polarons form a
bipolaron. Bipolarons are able, in turn, to form metallic clusters (stripes).
Müller indicated the following factors which should be taken into account
in the search for new HTSCs:
(1) Superconductivity is favored by a layered (quasi-two-dimensional)
crystalline structure;
(2) Oxygen ions are proposed as anions;
(3) Fluorine, chlorine and nitrogen can be considered as well.
It is worth noting that the recent discovery of FeAs superconductors elimi-
nated the monopoly of cuprates in the physics of HTSCs. It is possible that
new superconductors should be sought in the other rows and columns of
the periodic table. It is necessary to concentrate efforts on the purposeful
search for and the creation of new HTSCs.

3.3 Two-gap Superconductivity in MgB2

3.3.1 The physical properties of MgB2
Recently, the superconductivity of MgB2 with Tc = 39 K, which is the high-
est temperature among two-component systems, was discovered.124 The
great interest in the study of magnesium diboride is related to the fact
that MgB2 has occupied the “intermediate” place between low- and high-
temperature superconductors by the value of Tc . Therefore, modern litera-
ture calls MgB2 a “medium-Tc superconductor” (MTSC). The low cost of
this superconductor also makes it economical for use. We recall that wires
made of cuprate superconductors include 70% of silver, which is expensive.
An important peculiarity of MgB2 is its quasi-two-dimensional structure of
the AlB2 type. Interestingly, AlB2 is not superconducting. Note that MgB2
is another example of the crucial role played by the lattice structure regard-
ing superconductivity.123 It is an “old” material which has been known since
the early 1950s, but only recently discovered to be superconducting. MgB2
 Multiband Superconductivity 119

has a hexagonal structure118 ; see Fig. 3.3.23 The results of calculations of the
temperature dependence of the specific heat of MgB2 given in Fig. 3.5 follow
the corresponding experimental data. The band structure of MgB2 has been
calculated in several works since the discovery of superconductivity.128 The
electronic properties of MgB2 are plotted in Fig. 3.4. The band structure of
MgB2 is similar to that of graphite and is formed by the σ and π zones.

Fig. 3.3 The structure of MgB2 containing graphite-type B layers separated by hexagonal
close-packed layers of Mg.

Fig. 3.4 Band structure of MgB2 .

120 Modern Aspects of Superconductivity: Theory of Superconductivity

T (K)

Fig. 3.5 The MgB2 specific heat vs. temperature. The line indicates theory, and the
circles represent the experimental data.

The important problem for superconductivity in MgB2 is the mecha-

nism of superconductivity. It can be the conventional electron–phonon (e–p)
mechanism or a more exotic mechanism. The presence of an isotope effect
is a strong indicator of the phonon mediation of superconductivity. The
large isotope effect αB = 0.26130 shows that phonons associated with B
vibrations play a significant role in the MgB2 superconductivity, whereas
the magnesium isotope effect is very small: αM = 0.02.130 The total isotope
effect is αB + αM = 0.3, which supports the electron–phonon mechanism of
superconductivity.
MgB2 is a II-type superconductor.123 The discovery of the supercon-
ductivity of MgB2 has aroused great interest in multigap superconductiv-
ity.125 MgB2 has two superconducting gaps, 4 meV and 7.5 meV, due to
the π and σ electron bands. The two-gap structure was established in a
number of experiments.130 The plot of the specific heat of MgB2 vs. tem-
perature (Fig. 3.5) demonstrates good agreement between the theoretical
and experimental data. Both gaps have s symmetry and result from the
highly anisotropic layer structure of MgB2 . Although multigap supercon-
ductivity was discussed theoretically119, 120, 129 as early as 1958, it was only
observed experimentally126 in the 1980s. MgB2 is the first material in which
the effects of multigap superconductivity are so dominant, and its implica-
tions so thoroughly explored. Recent band calculations127, 128 of MgB2 with
the McMillan formula for the transition temperature have supported the
 Multiband Superconductivity 121

e–p interaction mechanism for the superconductivity. In this case, the pos-
sibility of two-band superconductivity has also been discussed in relation
to two-gap functions experimentally and theoretically. Very recently, two-
band or multiband superconductivity has been theoretically investigated
in relation to the superconductivity arising from Coulomb repulsive inter-
actions. The two-band model was introduced by Kondo.129 We have also
investigated anomalous phases in the two-band model by using the Green
function techniques.96–99, 101, 103 Recently, we have pointed out the impor-
tance of multiband effects in high-Tc superconductivity.96–98 The expres-
sions for the transition temperature for several phases have been derived,
and the approach has been applied to the superconductivity in molecular
crystals by charge injection and the field-induced superconductivity.13 In
previous papers,96–99, 101, 103 we have investigated the superconductivity by
using the two-band model and the two-particle Green function techniques.
We have applied the model to the e–p mechanism for the traditional BCS
method, the e–e interaction mechanism for high-Tc superconductivity,2 and
the cooperative mechanism. In the framework of the two-particle Green
function techniques,101 it has been shown that the temperature dependence
of the superconductivity gap for high-Tc superconductors is more compli-
cated than that predicted in the BCS approach. In Ref. 99, phase diagrams
for the two-band model superconductivity have been investigated, by using
the renormalization group approach. Below, we will discuss the possibil-
ity of the cooperative mechanism of two-band superconductivity in relation
to high-Tc superconductivity and study the effect of the increase of Tc in
MgB2 due to the enhanced interband pairing scattering. In this section,
we will investigate our two-band model for the explanation of the multi-
gap superconductivity of MgB2 . We apply the model to the e–p mechanism
for the traditional BCS method, the e–e interaction mechanism for high-Tc
superconductivity, and the cooperative mechanism in relation to multiband
superconductivity.

3.3.1.1 Mg1−x Alx B2

Critical temperature and other superconducting properties of two-band

superconductors depend on the doping level as well as the interband and
intraband scattering, which can be modified by chemical substitutions. The
influence of doping on the transition temperature in Mg1−x Alx B2 is illus-
trated in Fig. 3.6. It is seen that the doping destroys the superconductivity
in MgB2 . This can be understood as a result of the competition of two
122 Modern Aspects of Superconductivity: Theory of Superconductivity

Tc (K)

Fig. 3.6 The influence of doping on Tc in Mg1−x Alx B2 . The line indicates theory, and
the circles represent the experimental data.

effects: the first one is a coupling effect related to the changes of the car-
rier concentration, and the second one depends on the introduction of new
scattering centers leading to a modification of the interband and intraband
scattering.

3.3.2 Theoretical model

In this subsection, we will use the two-band model for superconductivity.
We start from the Hamiltonian for two bands i and j:

H = H0 + Hint , (3.30)

with
 
H0 = [i (k) − μ]a†ikσ aikσ + [j (k) − μ]a†jkσ ajkσ , (3.31)
k,σ

1    † †
Hint = αβγδ aip1 α aip2 β aip3 γ aip4 δ + (i ↔ j)
Γiiii
4
δ(p1 +p2 ,p3 +p4 ) α,β,γ,δ
† †
+ Γiijj
αβγδ aip1 α aip2 β ajp3 γ ajp4 δ + (i ↔ j)
† † 
+ Γijij
αβγδ aip1 α ajp2 β aip3 γ ajp4 δ + (i ↔ j) , (3.32)

where Γ is the bare vertex part,

Γijkl
αβγδ = ip1 αjp2 β|kp3 γlp4 δδαδ δβγ − ip1 αjp2 β|lp4 δkp3 γδαγ δβδ , (3.33)
 Multiband Superconductivity 123

with

ip1 αjp2 β|kp3 βlp4 α


= dr1 dr2 φ∗ip1 α (r1 )φ∗jp2 β (r2 )V (r1 , r2 )φkp3 β (r2 )φlp4 α (r1 ), (3.34)

where aipσ+ (aipσ ) is the creation (annihilation) operator corresponding to

the excitation of electrons (or holes) in the ith band with spin σ and momen-
tum p, μ is the chemical potential and φ is a single-particle wave function.
Here, we suppose that the vertex function in Eq. (3.32) involves the effective
interactions between the carriers caused by the linear vibronic coupling in
several bands and the screened Coulombic interband interaction of carriers.
When we use the two-band Hamiltonian (3.1) and define the order
parameters for a singlet exciton, triplet exciton and singlet Cooper pair,
the mean field Hamiltonian is easily derived.96–99, 101, 103 Here, we focus on
three electron scattering processes contributing to the singlet superconduct-
ing phase in the Hamiltonian (3.1):

gi1 = ii|ii, gj1 = jj|jj, (3.35)

g2 = ii|jj = jj|ii, (3.36)
g3 = ij|ij = ji|ji, (3.37)
g4 = ij|ji = ji|ij, (3.38)

where gi1 and gj1 represent the ith and jth intraband two-particle nor-
mal scattering processes, respectively, and g2 indicates the intraband two-
particle umklapp scattering (see Fig. 3.7). For simplicity, we consider the
three cases in Refs. 99 and 100: (i) g1 = 0 and others = 0; (ii) g2 = 0; and
others = 0; (iii) gi1 and gj1 , and others = 0, using the two-particle Green
function techniques (see Fig. 3.7).
It was shown that, possibly, the two-gap superconductivity arises in
case (iii). The superconductivity arising from the e–p mechanism (g1 < 0
and g1 < g2 ) such as for MgB2 is in the two-gap region. On the other hand,
the superconductivity of copper oxides (g1 > g2 ) is outside the two-gap
region. These results predict that we may observe two-gap functions for
MgB2 and only a single-gap function for copper oxides.
124 Modern Aspects of Superconductivity: Theory of Superconductivity

Fig. 3.7 Electron–electron interactions. Solid and dashed lines indicate π and σ bands,
respectively; gi1 , gj1 , g2 , g3 and g4 contribute to superconductivity.

3.3.3 Superconductivity in MgB2

Using case (iii) for gi1 and gj1 , g2 = 0 and others = 0 to describe the
superconductivity in MgB2 . We now have a reduced Hamiltonian:

H = H0 + Hint , (3.39)

where
 
H0 = [i − μ]a†ikσ aikσ + [j − μ]a†jkσ ajkσ , (3.40)
k,σ
  
Hint = g1i a†ik a†i−k ai−k aik + i→j+ g2 a†ik a†i−k aj−k ajk . (3.41)

We now define the order parameters which are helpful in constructing the
mean-field Hamiltonian:
1 † † 2
Δi = aip↑ ai−p↓ , (3.42)
p
1 2
Δj = a†jp↑ a†j−p↓ . (3.43)
p

The relation between two superconducting gaps of the system is as follows:

1 − gi1 ρi fi
Δj = Δi , (3.44)
g2 ρj fj
 Multiband Superconductivity 125

where
 μ−Ec
dξ (ξ2 + Δ2i )1/2
fi = tan h ,
μ (ξ2 + Δ2i )1/2 2T
 (3.45)
μ−Ej (ξ2 + Δ2j )1/2
dξ
fj = tan h ,
μ−Ec (ξ2 + Δ2j )1/2 2T

with the coupled gap equation

(1 − gi1 ρi fi )(1 − gj1 ρj fj ) = g22 fi fj . (3.46)

We have tried to estimate the coupling constant of the pair electron scat-
tering process between the π and σ bands of MgB2 and have calculated
the parameters by using a rough numerical approximation. We focus on one
π band and σ band of MgB2 and consider electrons near Fermi surfaces.
We found that the parameter g1 = −0.4 eV, by using the transfer integral
between the π and σ bands. We estimate the coupling parameter g2 of the
pair electron scattering process by the expression
 1,2
g2 = Vk1,k2 , (3.47)
k1,k2

1,2
Vk1,k2 = u∗1,r (k1)u∗1,s (k1)vrs u2,t (k2)u2,u (k2), (3.48)
r,s,t,u

where the labels 1 and 2 are the π band and the σ band, respectively, ui,r (ξ)
is the LCAO coefficient for the ith band and ξ is the moment.115, 116 The
indices k1 and k2 are summed over each Fermi surface. However, it is diffi-
cult to perform the sum exactly. In this case, we used a few points near the
Fermi surface. The coupling constant of the pair electron scattering between
the π band and the σ band is g2 = 0.025 eV. From numerical calculations
of Eqs. (3.44)–(3.46), we can also obtain the temperature dependence of
the two-gap parameters (see Fig. 3.8). We have used the density of states
of the π and σ bands (ρi = 0.2 eV −1 ; ρj = 0.14 eV−1 ), chemical potential
μ = −2.0, the top energy of the σ band Ej = −1.0 and the fitting param-
eters (gi1 = −0.4 eV; gj1 = −0.6 eV; g2 = 0.02 eV). These calculations have
qualitative agreement with experiments.118, 130, 131 The expression for the
transition temperature of superconductivity derived in a simple approxima-
tion is:
 
−1
Tc+ = 1.13(ζ − Ej ) · exp , (3.49)
g+ ρ
126 Modern Aspects of Superconductivity: Theory of Superconductivity

Fig. 3.8 Temperature dependence of two superconducting gaps.

where
1
g+ = (B + B 2 − 4A) (3.50)
24
and
A = g1i g1j − g22 , (3.51)
B = g1i + g1j , (3.52)
ζ = −μ. (3.53)
From the expressions for Tc+ , we can see the effect of increase of Tc+ due to
the enhanced interband pairing scattering (g2 ).
Figure 3.9 shows a schematic diagram of the mechanism of pairing for
two gaps. The scenario is as follows. Electrons from the π and σ bands
make up the subsystems. For g2 , we have two independent subsystems with
the different transition temperatures of superconductivity Tcπ and Tcσ and
two independent superconducting gaps. In our model, we have two coupled
superconducting gaps with the relation (3.44) and one transition tempera-
ture of superconductivity Tc+ , which is in agreement with experiments. In
this model, we have two channels of superconductivity: conventional chan-
nel (intraband g1 ) and unconventional channel (interband g2 ). Two gaps
appear simultaneously in different bands which are like BCS gaps. The gap
 Multiband Superconductivity 127

Fig. 3.9 Schematic diagram of the mechanism of pairing for two gaps.

in the π band is bigger than that for the σ band, because the density of
state is 0.25 eV in the π band and 0.14 eV in the σ band. The current of
Cooper pairs flows from the π band into the σ band, because the density
of Cooper pairs in the π band is much higher. The tunneling of Cooper
pairs also stabilizes the order parameter in the σ band. In this way, we can
predict the physical properties of the multigap superconductivity if we have
the superconductors with a multiband structure, as shown in Fig. 3.9.
Thus, we have presented our two-band model with the intraband two-
particle scattering and interband pairing scattering processes to describe
two-gap superconductivity in MgB2 . We defined the parameters of our model
and made numerical calculations of the temperature dependence of two
gaps in qualitative agreement with experiments. We have proposed a two-
channel scenario of superconductivity: the conventional channel (intraband
g1 ), which is connected with the BCS mechanism in different bands; and
the unconventional channel (interband g2 ), which describes the tunneling of
Cooper pairs between two bands. The tunneling of Cooper pairs also sta-
bilizes the order parameters of superconductivity and increases the critical
temperature of superconductivity.

3.4 Theoretical Studies of Multiband Effects in

Superconductivity by Using the Renormalization
Group Approach
We present the renormalization equations using the two-band model and
construct phase diagrams for the two-band superconductivity. In the frame-
work of the two-band model, the given results predict that superconductivity
appears, even if the e–e interaction is positive. We discuss the possibility
of a cooperative mechanism in the two-band superconductivity in relation
to high-Tc superconductivity. The recent discovery of superconductivity in
MgB2 124 has also attracted great interest to the elucidation of its mech-
anism from both the experimental and theoretical viewpoints. A crucial
128 Modern Aspects of Superconductivity: Theory of Superconductivity

role of the e–p interaction has been pointed out for the superconductiv-
ity of MgB2 . Recent band calculations of the transition temperature for
MgB2 127, 128 with the McMillan formula132 have supported the e–p interac-
tion mechanism for the superconductivity. In this case, the possibility of the
two-band superconductivity in relation to two-gap functions has also been
considered experimentally and theoretically. Very recently, the two-band or
multiband superconductivity has been theoretically investigated in relation
to the superconductivity arising from Coulomb repulsive interactions.129
The two-band model was introduced by Kondo.129 Recently, we have pointed
out the importance of multiband effects in high-Tc superconductivity.99, 138
We have also investigated anomalous phases in the two-band model by using
the Green function techniques.101, 136 The expressions of the transition tem-
perature for several phases have been derived, and the approach has been
applied to the superconductivity in several crystals by charge injection and
the field-induced superconductivity.136, 139 In the previous section, we have
investigated superconductivity by using the two-band model and the two-
particle Green function techniques.99, 101 We have applied the model to the
e–p mechanism for the traditional BSC method, the e–e interaction mecha-
nism for high-Tc superconductivity and the cooperative mechanism. In the
framework of the two-particle Green function techniques,103 it has been
shown that, in the e–p system, a class of new so-called coupled states arises.
In this section, we investigate two- or multiband effects in superconductivity
by using the two-band model within the renormalization group approach.
Renormalization equations for the two-band superconductivity are derived
from the response function and the vertex correction of the model. Phase
diagrams numerically obtained from the renormalization equations are pre-
sented. We also discuss the superconductivity arising from the e–e repulsive
interaction in relation to the two-band superconductivity.

3.4.1 Theoretical model

In this subsection, we summarize the two-band model for the superconduc-
tivity and introduce the renormalization group approach.142, 144
We consider a Hamiltonian for two bands i and j, written as
H = H0 + Hint , (3.54)
with
 
H0 = [i (k) − μ]a† aikσ + [j (k) − μ]a† aj kσ , (3.55)
ikσ j kσ
k,σ
 Multiband Superconductivity 129

1    † †
Hint = αβγδ aip1 α aip2 β aip3 γ aip4 δ + (i ↔ j)
Γiiii
4
δ(p1 +p2 ,p3 +p4 ) α,β,γ,δ

† †
+ Γiijj
αβγδ aip1 α aip2 β aj p3 γ aj p4 δ + (i ↔ j)

† † 
+ Γijij
αβγδ aip1 α aj p2 β aip3 γ aj p4 δ + (i ↔ j) , (3.56)

where Γ is the bare vertex part,

Γijkl
αβγδ = ip1 αjp2 β|kp3 γlp4 δ δαδ δβγ − ip1 αjp2 β|lp4 δkp3 γ δαγ δβδ , (3.57)

with

ip1 αjp2 β|kp3 βlp4 α


= dr1 dr2 φ∗ip1 α (r1 )φ∗j p2 β (r2 )V (r1 , r2 )φkp3 β (r2 )φlp4 α (r1 ), (3.58)

where a†ipσ (aipσ ) is the creation (annihilation) operator corresponding to the

excitation of electrons (or holes) in the ith band with spin σ and momentum
p, μ is the chemical potential and φ∗iαp is a single-particle wave function.
1
Here, we suppose that the vertex function in Eq. (3.56) involves the effec-
tive interactions between carriers caused by the linear vibronic coupling in
several bands and the screened Coulombic interband interaction of carriers.
We focus on three electron scattering processes contributing to the singlet
superconducting phase in the Hamiltonian (3.56), as shown in Fig. 3.10:

gi1 = ii|ii, (3.59)

gj1 = jj|jj, (3.60)
g2 = ii|jj = jj|ii, (3.61)

where gi1 and gj1 represent the ith and jth intraband two-particle nor-
mal scattering processes, respectively, and g2 indicates the intraband two-
particle umklapp scattering. Note that the Γ’s are given by

αβγδ = gi1 (δαδ δβγ − δαγ δβδ ),

Γiiii

Γjjjj
αβγδ = gj1 (δαδ δβγ − δαγ δβδ ),

Γiijj jjii
αβγδ = Γαβγδ = g2 (δαδ δβγ − δαγ δβδ ), (3.62)

where we assume that an antisymmetrized vertex function Γ is a constant

independent of the momenta.
130 Modern Aspects of Superconductivity: Theory of Superconductivity

i i j j

=gi1 =gj1

i i j j

i j j i

=g2 =g2

i j j i

Fig. 3.10 Electron–electron interactions. Dependence of g on the direction in the momen-

tum space is ignored in this model [gx (k) ≈ gx (x = i, j)]. We assume that gx is constant.

The spectrum is elucidated by the Green function method. Using

Green functions which characterize the CDW, SDW and SSC phases,
we obtain a self-consistent equation according to the traditional proce-
dure.115, 136, 138, 145–147 Then we can obtain expressions for the transition
temperature in some cases. Electronic phases of a one-dimensional system
have been investigated by using a similar approximation in the framework
of the one-band model.144–147 In the framework of the mean-field approxi-
mation within the two-band model, we have already derived expressions for
the transition temperature for CDW, SDW and SSC. In Refs. 136 and 138,
we have investigated the dependence of Tc on the hole or electron concentra-
tion for the superconductivity of copper oxides by using the two-band model
and have obtained a phase diagram of Bi2 Sr2 Ca1−x Yx Cu2 O8 (Bi-2212) by
means of the above expressions for the transition temperature.

3.4.2 Renormalization-group approach

The Dyson equation is invariant under a multiple renormalization of Green
function and coupling parameters g. From this invariance for a scaling pro-
cedure, we obtain differential equations for the coupling parameters and the
external vertex of a Cooper pair:
 
∂ ∂ u
y g̃i (y, u, g) = g̃i ξ, , g̃(t, u, g) |ξ=1 , (3.63)
∂y ∂ξ y
 
∂ ∂ u
y ln Λ(y, u, g) = ln Λ ξ, , g̃(t, u, g) |ξ=1 , (3.64)
∂y ∂ξ y
where y and u are parameters with the dimension of energy, g is the set of
original couplings and Λ is the external vertex.
 Multiband Superconductivity 131

3.4.3 Vertex correction and response function

for Cooper pairs
To solve Eqs. (3.63) and (3.64), we estimate the right-hand side of Eq. (3.63)
by using the perturbation theory. We consider the lowest-order correction
to the vertex for a Cooper pair, as shown in Fig. 3.11. From these diagrams,
we obtain
     2  
g̃i1 gi1 −gi1 −g22 Li
= + , (3.65)
g̃j1 gj1 −gj1
2 −g22 Lj

      
g̃2 g2 −gi1 g2 −g2 gj1 Li
= + , (3.66)
g̃2 g2 −gj1 g2 −g2 gi1 Lj
with
T 
Li = Πi (k, ω) = Gi (q, ω )Gi (k − q, ω − ω ),
N
q,ω
T 
Lj = Πj (k, ω) = Gj (q, ω )Gj (k − q, ω − ω ), (3.67)
N 
q,ω

Fig. 3.11 Diagrams of the first-order vertex correction. (a) and (b) contribute to g̃i1 , (c)
and (d) are diagrams for g̃j1 , and (e)–(h) for g̃2 .
132 Modern Aspects of Superconductivity: Theory of Superconductivity

where G and T are the temperature Green function and temperature, respec-
tively. For the special case ω = 0, k = 0, the functions Li and Lj become
    
ui /y u /y
Li = − tan h + tan h i ln ξ − 2A, (3.68)
2ξ 2ξ
    
uj /y uj /y
Lj = − tan h + tan h ln ξ − 2A, (3.69)
2ξ 2ξ
where 
A= dx ln x sech2 x, (3.70)
ui (uj ) and ui (uj ) are dimensionless functions expressed by the chemical
potential, the cutoff energy, the top energy of the jth band, and the density
of state for the ith (jth) band.
Next, we consider a first-order response function for a singlet Cooper
pair, as shown in Fig. 3.12. Then the first-order vertex function Λ for the
ith and jth bands can be written as
      
Λi + Λj 2 −g1 −g2 Li + Lj
= + . (3.71)
Λi − Λj 0 −g1 g2 Li − Lj

3.4.4 Renormalization equation

For simplicity, we hereafter assume that gi1 = gj1 . From Eqs. (3.63), (3.65)
and (3.66), we obtain the differential equations written as
∂  
g̃1 = − g̃12 + g̃22 , (3.72)
∂x
∂
g̃2 = −2g̃1 g̃2 . (3.73)
∂x

Fig. 3.12 Diagrams of the first-order response function. (a) and (b) contribute to Λi ,
and (c) and (d) show diagrams for Λj .
 Multiband Superconductivity 133

In a similar way, using Eqs. (3.64) and (3.71), we obtain the differential
equations written as
∂
ln Λ+ = −g̃1 − g̃2 , (3.74)
∂x
∂
ln Λ− = −g̃1 + g̃2 , (3.75)
∂x
where Λ+ = Λi + Λj and Λ− = Λi − Λj .

3.4.5 Phase diagrams

In the previous subsection, we have derived the basic equations (3.72)–(3.75)
to find the low-temperature phases. For the special case of g2 = 0, we obtain
an analytic solution:
1
g̃1 = , (3.76)
x + g1−1
g̃2 = 0, (3.77)
1
Λi = Λj = . (3.78)
g1 x + 1
From these solutions, we find that the superconducting phase appears only
when the intraband interaction g1 is negative. In the case of the traditional
superconductivity described by the BCS theory, it is necessary that the
effective e–e interaction be negative (g1 < 0) in order to realize the super-
conductivity. The present result agrees with that of the traditional theory for
superconductivity expressed by the one-band model. For the case of g2 = 0,
the phase diagrams numerically obtained from the above renormalization
equations are shown in Fig. 3.13. Figure 3.13(a) shows the phase diagram
for the sum of contributions to the superconductivity from the ith and jth
bands. Thus, this phase implies that the signs of the superconducting state
for the ith and jth bands are the same. From this diagram, we find the
superconductivity only for −g1 − g2 > 0, with negative g2 . On the other
hand, the phase diagram for the difference between the superconducting
states for the ith and jth bands is shown in Fig. 3.13(b). In this case, this
phase means that the sign of the superconductivity for the ith band is dif-
ferent from that of the jth band. We can see that the superconductivity
appears only for −g1 + g2 > 0 with positive g2 from Fig. 3.13(b). Thus,
the present results suggest that the two-band superconductivity appears
when the intraband umklapp repulsive scattering g2 is larger than the nor-
mal repulsive scattering g1 . In the region of g2 > g1 with g2 < 0 and
134 Modern Aspects of Superconductivity: Theory of Superconductivity

(a) (b)
4 4

2 2

2
0
2

g
g

-2 -2

-4 -4

-5 0 5 -5 0 5
g g
1 1

Fig. 3.13 Phase diagrams for superconductivity: (a) Λi + Λj , (b) Λi − Λj .

g2 < −g1 with g2 > 0 (two-gap region), we can expect that two-gap func-
tions are observed. In the former region, those superconducting gaps may
be expressed by |Λ+ | > |Λ− |, and the latter may be |Λ+ | < |Λ− |. On
the other hand, we expect only a single-gap function in the other region.
These results agree with the previous solutions96, 99 derived by using the
two-particle Green function techniques. The superconductivity arising from
the e–p mechanism (g1 < 0 and |g2 | < |g1 |) such as that in MgB2 is in the
two-gap region. On the other hand, the superconductivity such as in copper
oxides (|g2 | > |g1 |) is outside the two-gap region. These results predict that
we may observe two-gap functions for MgB2 and only a single-gap function
for copper oxides. In the two-band model for negative g1 with transferring
or tunneling of Cooper pairs between two bands, we can expect that the
transition temperature becomes higher than that derived from the single-
band model. The tunneling of Cooper pairs causes stabilization of the order
parameter of superconductivity.119, 120, 129 We can also expect higher Tc of
the superconductivity than that for copper oxides in two regions (g1 < 0,
g2 < 0 and g1 < 0, g2 > 0) by the cooperative mechanism. Phase diagrams
for CDW, SDW, and singlet superconductivity derived from a more general
Hamiltonian will be presented elsewhere.
Thus, we have derived the renormalization equations and presented the
phase diagrams for the two-band superconductivity. In the framework of the
two-band model, the present results predict that superconductivity appears
even if the e–e interaction is positive. We can expect that the transition
temperature becomes higher than that of copper oxides by the cooperative
mechanism.
 CHAPTER 4

Mesoscopic Superconductivity

4.1 Introduction
Recent advances in nanoscience have demonstrated that fundamentally new
physical phenomena are found when systems are reduced in size to dimen-
sions which become comparable to the fundamental microscopic lengths of
the investigated material. Superconductivity is a macroscopic quantum phe-
nomenon, and it is therefore especially interesting to see how this quantum
state is influenced when the samples are reduced to nanometer sizes. Recent
developments in nanotechnologies and measurement techniques allow exper-
imental investigation of the magnetic and thermodynamic superconducting
properties of mesoscopic samples in this regime.
In this book, we develop some theoretical models for describing such
nanoscale superconducting systems and explore possible new experimental
phenomena which we can predict based upon these models. In bulk sam-
ples, the standard BCS theory gives a good description of the phenomenon
of superconductivity.104 However, it was noticed by Anderson in 1959 that,
as the size of a superconductor becomes smaller, and the quantum energy
level spacing of the electrons in the sample approaches the superconduct-
ing gap, the BCS theory will fail.6 The exact solution to the reduced BCS
Hamiltonian for finite-size systems was developed by Richardson in the con-
text of nuclear physics a long time ago.149 This shows that, while the grand
canonical BCS wave function gives a very accurate solution to the BCS
Hamiltonian in the limit where the number of electrons is very large N  1,
for small values of N one has to use exact analytical methods to obtain
reliable results. The recent experimental advances in fabricating and mea-
suring superconductivity in ultrasmall mesoscopic and nanoscale grains has
renewed theoretical interest in the Richardson solution.150–153 In this book,

135
136 Modern Aspects of Superconductivity: Theory of Superconductivity

we propose to develop theoretical models for nanoscale superconducting sys-

tems and to apply these models to a variety of systems of current experimen-
tal interest. The Richardson solution depends on the electron energy level
spacings near to the Fermi level, and so these different geometric shapes will
lead to different size dependences of the thermodynamic and electronic prop-
erties. Experiments have recently demonstrated superconductivity in one-
dimensional nanowires of lead154 and carbon.155 Our theoretical predictions
will include the even–odd parity effects in tunneling spectra, which have
already been observed on the nanometer scale in Al grains, (Fig. 4.1),150
but which could also be observed in nanotube superconductors. In Figs. 4.1
and 4.2, we can see the ensemble of small metallic grains which can be
in the normal or superconducting state. For such systems, the important
parameters are the l-granular size, the electron coherence length, and the
penetration depth.123 In grains, there appears the quantum size effect (dis-
cretization of the electron energy spectrum). In Fig. 4.2, we show the size
effect in grains. We note that the level spacing depends on many parameters,
such as the electron coherence length and the penetration depth. Carbon
nanotubes were first observed in 1991 by Iijima in Japan, (Fig. 4.3).

Fig. 4.1 Illustration of superconductor grains.

Fig. 4.2 Size effect in superconductor grains.

 Mesoscopic Superconductivity 137

Fig. 4.3 Carbon nanotube revealing superconductivity.

In this chapter, we investigate properties of nanosize two-gap supercon-

ductivity by using a two-sublevel model in the framework of the mean-field
approximation. A model corresponding to nanosize two-gap superconduc-
tivity is presented, and the partition function of the nanosize system is
analytically derived by using the path-integral approach. A definition of the
critical level spacing of the two-gap superconductivity is also presented, and
we discuss the condensation energy and the parity gap of the two-gap super-
conductivity in relation to the size dependence of those properties with two
bulk gaps and the effective pair scattering process between two sublevels. We
present the theory of interactions between two nanoscale ferromagnetic par-
ticles embedded in a superconductor and spin orientation phase transitions
in such a system. We also consider the ideas of quantum computing and
quantum information in mesoscopic circuits. The theory of the Josephson
effect is presented, and its applications in quantum computing are analyzed.
The results of this chapter were obtained by the authors and published in
Refs. 123, 156–158.

4.2 Nanosize Two-Gap Superconductivity

Multiband superconductivity has been theoretically investigated in relation
to MgB2 superconductivity in Chap. 3.
Recent experiments160, 164 by Black et al. have also generated high inter-
est in the size dependence of the superconductivity. Properties of ultra-
small superconducting grains have been theoretically investigated by many
groups.161, 162, 165–169 In such ultrasmall grains, the old but fundamental the-
oretical question was noticed by Anderson.6 The standard BCS theory gives
a good description of the phenomenon of superconductivity in large sam-
ples. However, as the size of a superconductor becomes small, the BCS
theory fails. In ultrasmall Al grains, the bulk gap has been discussed in
relation to physical properties in ultrasmall grains such as the parity gap,169
condensation energy167 and electron correlation,162 with the size dependence
of the level spacing161 of samples.
138 Modern Aspects of Superconductivity: Theory of Superconductivity

In nanosize grains of a superconductor, the quantum level spacing

approaches the superconducting gap. In the case of a two-gap supercon-
ductor, we can consider a model with two sublevels corresponding to two
independent bands. In this section, we present a model for nanosize two-
gap superconductivity and an expression for the partition function of the
system.

4.2.1 Hamiltonian for nanosize grains

We consider a pairing Hamiltonian with two sublevels corresponding to two
bands 1 and 2 written as
H = H0 + Hint , (4.1)
where
 
H0 = [ε1j − μ]a†jσ ajσ + [ε2k − μ]b†kσ bkσ , (4.2)
j,σ k,σ
 
Hint = −g1 a†j↑ a†j↓ aj  ↓ aj  ↑ − g2 b†k↑ b†k↓ bk ↓ bk ↑
j,j  ∈I k,k  ∈J
 
+ g12 a†j↑ a†j↓ bk↓ bk↑ + g12 b†k↑ b†k↓ aj↓ aj↑ . (4.3)
j∈I,k∈J j∈I,k∈J

Here, a†jσ (ajσ ) and b†jσ (bj,σ ) are the creation (annihilation) operator in sub-
levels 1 and 2 with spin σ and energies ε1j and ε2j , respectively, the operators
for each sublevel satisfy the anticommutation relations, and the operators
between sublevels are independent, μ is the chemical potential, the second
term of Eq. (4.1) is the interaction Hamiltonian, g1 and g2 are the effective
interaction constant for sublevels 1 and 2, and g12 is an effective interac-
tion constant which corresponds to the pair scattering process between two
bands. The sums of j and k in Eq. (4.3) are over the set I of N1I states
which correspond to the half-filled band 1 with fixed width 2ω1D and the
set J of N2J states for band 2, respectively.
In this study, we assume that the Debye energies for two sublevels are the
same: ω1D = ω2D = ωD . Within this assumption, N1I and N2J are relatively
estimated by the density of state (DOS) for two bands as N1I /N2J = ρ1 /ρ2 ,
where ρ1 and ρ2 are the DOS for two bands. The interaction constants
g1 and g2 can be written as d1 λ1 and d2 λ2 , respectively. d1 = 2ωD /N1I
and d2 = 2ωD /N2J are the mean energy level spacings, and λ1 and λ2 are
the dimensionless parameters for two sublevels. We take the intersublevel
 Mesoscopic Superconductivity 139

√
interaction constant g12 = d1 d2 λ12 . We then obtain the relation ρ1 /ρ2 =
N1I /N2J = d2 /d1 .

4.2.2 Path-integral approach

It is convenient to introduce a path-integral approach for the treatment of
fluctuations of the order parameters. This approach gives an exact expres-
sion for the grand partition function of a superconductor.

H − μN
Z(μ, T ) = Tr exp − , (4.4)
T
where T is the temperature and N is the number operator in the grain. The
idea of the path-integral approach is to replace the description of a system
under study in terms of electronic operators by an equivalent description in
terms of the superconducting order parameter.
By the path-integral approach, we obtain an expression for the grand
partition function for the Hamiltonian (4.1):

Z(μ, T ) = DΔ1 DΔ∗1 DΔ2 DΔ∗2 e−S[Δ1 ,Δ2 ] . (4.5)

Here, the action S[Δ1 , Δ2 ] is defined as

 ξ1j
 
ξ2k

−1 −1
S[Δ1 , Δ2 ] = − Tr ln G1j − − Tr ln G2k −
T T
j k
 1/T
1
2 [g2 |Δ1 (τ)| + g1 |Δ2 (τ)|
2 2
+ dτ
0 g1 g2 − g12
+ g12 (Δ1 (τ) Δ2 (τ)∗ + Δ1 (τ)∗ Δ2 (τ))]. (4.6)
Δ1 and Δ2 are bulk gaps for sublevels 1 and 2, respectively, ξ1j = ε1j − μ
and ξ2k = ε2k − μ, and the inverse Green functions

−1 d
G1j (τ, τ ) = − − ξ1j σ − Δ1 (τ)σ − Δ1 (τ)σ δ(τ − τ ),
 z + ∗ −
(4.7)
dτ

d
G2k (τ, τ ) = − − ξ2k σ − Δ2 (τ)σ − Δ2 (τ)σ δ(τ − τ ),
−1  z + ∗ −
(4.8)
dτ
where σ± = σx ± iσy , and σx,y,z are the Pauli matrices. G−1 −1
1 and G2 satisfy
antiperiodic boundary conditions.
In the case of a stronger interaction, Δ1  d1 and Δ2  d2 , we con-
sider the mean-field approximation for the order parameters in the path-
integral approach. Substituting the time-independent order parameters into
140 Modern Aspects of Superconductivity: Theory of Superconductivity

the action (4.6), we have

 
Ω(μ) = (ξ1j − 1j ) + (ξ2k − 2k )
j k

1 ∗ ∗
2 [g2 Δ1 + g1 Δ2 + g12 (Δ1 Δ2 + Δ1 Δ2 )],
2 2
+ (4.9)
g1 g2 − g12
where 1j = (ξ21j + Δ21 )1/2 and 2k = (ξ22k + Δ22 )1/2 . In Eq. (4.9), the values
of Δ1 and Δ2 must be chosen in such a way as to minimize Ω. From the
minimization of Ω, we obtain a coupled gap equation at zero temperature
for the two-gap system:
   g . 1 . 
−g12 k 212k Δ 
Δ1 1 j 21j 1
= . 1 . 1 . (4.10)
Δ2 −g12 j g2 k Δ2
21j 22k

From the coupled gap equation (4.10), we formally obtain an expression for
the bulk gap for two-gap superconductivity at zero temperature:
 
−1 1
Δ̃1 = ω sinh , (4.11)
η1
 
−1 1
Δ̃2 = ω sinh , (4.12)
η2
where
1 λ2 + α± [η1 , η2 ]λ12
= , (4.13)
η1 λ1 λ2 − λ212
1 λ1 + α−1
± [η1 , η2 ]λ12
= , (4.14)
η2 λ1 λ2 − λ212
with
 
sinh η11
α± [η1 , η2 ] = ±  . (4.15)
sinh η12

For the two-band superconductivity, we can consider two cases for the phase
of the gaps: sgn(Δ̃1 ) = sgn(Δ̃2 ) and sgn(Δ̃1 ) = −sgn(Δ̃2 ). For the same
phase, α+ is used in Eqs. (4.13) and (4.14), and we use α− for the opposite
phase. Note that Δ̃1 = −Δ̃2 in the limit of strong intersublevel coupling
λ12 , i.e. the opposite phase. For λ12 = 0, we find the same results for two
bulk gaps derived from the conventional BCS theory for two independent
sublevels.
 Mesoscopic Superconductivity 141

4.2.3 Condensation energy

In this subsection, we discuss properties such as condensation energy, critical
level spacing and the parity gap of nanosize two-gap superconductivity by
using the partition function derived in the previous subsection.
In nanosize superconductivity, the condensation energy can be defined
C (λ) = E G (0) − E G (λ) − nλd, where E G is the ground state
as EN,b N,b N,b N,b
energy of the N -electron system in the interaction band, b is the number
of electrons on single occupied levels, and λ and n are the dimensionless
coupling parameter and the number of pair-occupied levels, respectively. In
the case of a nanosize two-band system, the condensation energy can be
written as

C
EN 1 ,b1 ;N2 ,b2
(λ1 , λ2 , λ12 ) = EN
G
1 ,b1 ;N2 ,b2
(0, 0, 0) − EN
G
1 ,b1 ;N2 ,b2
(λ1 , λ2 , λ12 )
− n1 λ1 d1 − n2 λ2 d2 , (4.16)

where EN G
1 ,b1 ;N2 ,b2
(λ1 , λ2 , λ12 ) is the ground state energy of the (N1 + N2 )-
electron system. From Eqs. (4.4) and (4.9), the condensation energy of the
two-sublevel system can be expressed by the condensation energy of inde-
pendent single-level systems:

λ212
C
EN 1 ,b1 ;N2 ,b2
(λ1 , λ2 , λ12 ) = EN
C
1 ,b1
C
(λ1 ) + EN 2 ,b2
(λ2 ) −
λ1 λ2 − λ212
 2 
Δ1 Δ22 2(Δ∗1 Δ2 + Δ1 Δ∗2 )
× + + √ , (4.17)
d1 λ1 d2 λ2 d1 d2 λ12

where EN C C
(λ1 ) and EN (λ2 ) correspond to the condensation energy for
1 ,b1 2 ,b2
the single-band case. In the same phases of Δ1 and Δ2 , the condensation
energy (4.17) decreases, i.e. there appears the instability by the coupling
constant λ12 . On the other hand, in the opposite phases, the condensation
energy becomes larger, because Δ∗1 Δ2 + Δ1 Δ∗2 < 0. We can expect that
the condensation energy for two-gap superconductivity results in a higher
stability than that of two independent systems, due to the intersublevel
coupling λ12 and the opposite phases.

4.2.4 Critical level spacing

To discuss the critical level spacing for a two-gap system, we start from the
coupled gap equation (4.10). For the case of the critical level spacing of the
142 Modern Aspects of Superconductivity: Theory of Superconductivity

two-gap system, we have

 1  1  1  1
1 = λ1 + λ2 − (λ1 λ2 − λ212 ) ,
j
2|ξ̃1j | k
2|ξ̃2k | j
2|ξ̃1j | k 2|ξ̃2k |
(4.18)
where ξ̃i = ξi /di for sublevel i = 1, 2. For the odd or even cases, Eq. (4.18)
can be approximately solved by using the digamma function: for the odd
case, the critical level spacing becomes

1 d2
do1c = ωD eγ exp − , do2c = do1c , (4.19)
λ d1
and, for the even case,

γ 1 d2
d1c = 4ωD e exp − , de2c = de1c .
e
(4.20)
λ d1
Here, we use
+ 
1 1
= λ1 + λ2 − ax + (λ1 − λ2 − ax) + 4λ12 ,
2 2 (4.21)
λ 2x
with
x = λ1 λ2 − λ212 , (4.22)
d1
a = log . (4.23)
d2
From these expressions, we find some relations:
de1c = 4do1c , de2c = 4do2c , (4.24)

eγ 1 1
do1/2c ≈ exp − Δ̃1/2 . (4.25)
2 η1/2 λ
In the case of |λ1 − λ2 |  λ12 , Eq. (4.25) can be approximately rewritten
as 
eγ λ2 − λ1 + 2αλ12
d1/2c ≈
o
exp Δ̃1/2 . (4.26)
2 λ1 λ2 − λ212
On the other hand, in the limit of |λ1 − λ2 |  λ12 , we have

eγ (1 + α)λ12
d1/2c ≈
o
exp Δ̃1/2 . (4.27)
2 λ1 λ2 − λ212
For the case of λ12 = 0, Eq. (4.25) can be rewritten as do1/2c ≈
exp[γ]/2 exp[1/λ1 − 1/λ2 ] Δ̃1/2 . Therefore, when the coupling constants λ1
and λ2 take the same value, we have a relation similar to that for a single-
level system: do1/2c ≈ 0.89 Δ̃1/2 . These results suggest that the critical level
spacing strongly depends upon λ12 and the difference between the effective
interaction constants for sublevels. The relation in Eq. (4.24) is the same
relation in the conventional nanosize BCS theory.
 Mesoscopic Superconductivity 143

4.2.5 Parity gap

In this subsection, we consider a parity gap in the case of two-gap super-
conductivity in ultrasmall grains. In the case of two sublevel spacings, the
chemical potential lies halfway between the highest occupied and the low-
est unoccupied levels of a smaller level spacing in the half-filled case, as
shown in Fig. 4.4(a). We assume that d1 < d2 and that the numbers of
occupied levels corresponding to each sublevel are n1 and n2 , respectively.
Then, the total number of electrons becomes N = 2n1 + 2n2 . When we
consider N = 2n1 + 2n2 + 1, the chemical potential lies on the level ε1n1 +1 ,
as shown in Fig. 4.4(b). Figure 4.4(c) indicates the position of the chemical
potential in the case of N = 2n1 + 2n2 + 2. The parity gap of nanosize
two-gap superconductivity is written as
1 G 
Δ1p = E2n
G
1 +1+2n2 ,1
− E2n 1 +2n2 ,0
G
+ E2(n 1 +1)+2n2 ,0
. (4.28)
2
G = Ω
From Eq. (4.9) and for the ground state energy EN,b μN + μN N , we
obtain  
d1 ρ1
Δp = Δ1 −
1
−1 . (4.29)
4 ρ2

(a) d1 (b) (c)

d2

n2+1 n2+1μ n2+1 μ

n1+1 n1+1 n1+1
μ n1
n1 n2 n1 n2 n2
n1-1 n1-1 n1-1
n2-1 n2-1 n2-1

(d) (e)

n1+2 n1+2 μ
n1+1 n2+1 μ n1+1
n2+1

n1 n1 n2
n2
n1-1 n1-1
n2-1 n2-1

Fig. 4.4 Position of the chemical potential relative the electronic energy levels in a two-
gap superconducting grain. Solid and dotted lines mean two sublevels. (a) Half-filled
system with 2n1 + 2n2 electrons; (b) (2n1 + 1 + 2n2 )-electron system; (c) [2(n1 + 1) + 2n2 ]-
electron system; (d) [2(n1 +1)+2n2 +1]-electron system; (e) [2(n1 +1)+2(n2 +1)]-electron
system.
144 Modern Aspects of Superconductivity: Theory of Superconductivity

From Figs. 4.4(c)–4.4(e), we can define another parity gap:

1 G 
Δ2p = E2(n
G
1 +1)+2n 2 +1,1 − E2(n 1 +1)+2n 2 ,0 + E G
2(n 1 +1)+2(n 2 +1),0 . (4.30)
2
From the latter definition of Eq. (4.30), we have
 
d2 3ρ2
Δp = Δ2 −
2
−1 . (4.31)
4 ρ1
The present results suggest two kinds of the dependence of the parity gap on
the level spacing. The parity gap does not depend on the effective interac-
tion λ12 . The structure around the Fermi level plays an important role in the
contribution to the size dependence of the parity gap. We have investigated
the properties of nanosize two-gap superconductivity by using a two-sublevel
model in the framework of the mean-field approximation. From the discus-
sion of the condensation energy in nanosize two-gap superconductivity, the
phases of the gaps are very important for stabilizing the superconductivity.
In the same phases, the two-gap superconductivity is unstable by the cou-
pling constant λ12 . On the other hand, in the opposite phases, the super-
conductivity becomes stable. We can expect that, due to the condensation
energy, the two-gap superconductivity becomes more stable than that for
two independent systems due to the intersublevel coupling λ12 and the oppo-
site phases.
We have also discussed the critical level spacing for two-gap supercon-
ductivity in ultrasmall grains. These results suggest that the critical level
spacing strongly depends on λ12 and the difference between the effective
interaction constants for sublevels. Moreover, the relation between the crit-
ical level spacing and the bulk gaps is modified as compared with the result
obtained for ultrasmall superconducting Al grains.
For the parity gap in two-gap superconductivity, the present results sug-
gest two kinds of the dependence of the parity gap on the level spacing and
that the structure around the Fermi level plays an important role by con-
tributing to the size dependence on the parity gap. The parity gap does not
depend on the effective interaction λ12 .
In the case of a cluster system, we have to apply a more accurate approach
beyond the mean-field approximation presented in this study by investigat-
ing the physical properties, and we also have to consider the contribution
of the surface of samples to the level structure around the Fermi level.
We will present these problems in the next section. On the basis of the
presented results, we might expect the possibility of a new multigap super-
conductivity arising in the nanosize region with a higher critical transition
temperature.
 Mesoscopic Superconductivity 145

In summary, a model corresponding to nanosize two-gap superconduc-

tivity has been presented, and an expression for the partition function of
the nanosize system has been analytically derived by using the path-integral
approach. A definition of the critical level spacing of the two-gap supercon-
ductivity has also been presented, and we discussed the condensation energy
and the parity gap of the two-gap superconductivity in relation to the size
dependence of those properties with two bulk gaps, as well as the effective
pair scattering process between two sublevels.

4.3 Exact Solution of Two-Band Superconductivity

in Ultrasmall Grains
Many groups have theoretically investigated the physical properties such
as critical level spacing, condensation energy and parity gap in ultrasmall
grains with the conventional superconductivity.161, 162, 165–169 The ques-
tion concerning such nanosize superconducting grains has been discussed
by Anderson.6 The standard BCS theory becomes false when the level
spacing approaches the superconducting gap. To investigate the proper-
ties in such nanosize systems, it is necessary to perform a more accurate
treatment. Braun and von Delft165, 166, 173 have reintroduced the exact
solution for the reduced BCS Hamiltonian developed by Richardson.170–172
It is noteworthy that Richardson’s solution is applicable at distributions of
single-electron energy levels. V. N. Gladilin et al.167 have investigated the
pairing characteristics such as the condensation energy, spectroscopic gap
and parity gap, by using Richardson’s exact solution for the reduced BCS
Hamiltonian.
The recent discovery of the superconductivity of MgB2 124 with Tc = 39 K
has also attracted great interest, aimed at the elucidation of its mechanism
from both the experimental and theoretical viewpoints. Since this discovery,
the possibility of two-band superconductivity has also been discussed in
relation to the two-gap functions experimentally and theoretically.
In this section, we investigate the two-band superconductivity in ultra-
small grains. Richardson’s exact solution is extended to two-band sys-
tems, and a new coupled equation is derived according to the procedure
of Richardson’s works. The parity gap and the condensation energy of an
ultrasmall two-band superconducting grain are numerically given by solv-
ing the coupled equation. We discuss these properties of ultrasmall grains
in relation to the correlation, interband interaction and size dependence.
146 Modern Aspects of Superconductivity: Theory of Superconductivity

4.3.1 Exact solution for two-band superconductivity

In this subsection, we derive an exact solution of the two-band supercon-
ductivity for a reduced BCS Hamiltonian.

4.3.2 Hamiltonian
We consider a Hamiltonian for two bands 1 and 2 written as
H = H1 + H2 + Hint , (4.32)
where
 
H1 = ε1j a†jσ ajσ − g1 a†j↑ a†j↓ ak↓ ak↑ , (4.33)
jσ jk
 
H2 = ε2j b†jσ bjσ − g2 b†j↑ b†j↓ bk↓ bk↑ , (4.34)
jσ jk
 
Hint = g12 a†j↑ a†j↓ bk↓ bk↑ + g12 b†j↑ b†j↓ ak↓ ak↑ . (4.35)
jk jk

The first and second terms of Eq. (4.32) correspond to the reduced BCS
Hamiltonian for bands 1 and 2, respectively. The third term means a cou-
pling between them and corresponds to the pair scattering process between
these two bands (see Fig. 4.5). a†jσ (ajσ ) and b†jσ (bjσ ) are the creation
(annihilation) operator in bands 1 and 2 with spin σ and the single-particle
levels ε1j and ε2j , respectively. The sums of j and k are taken over a set of

Fig. 4.5 Two-band system. The dotted line means the chemical potential; εnj is the
single-particle energy for band n and level j; −g1 and −g2 are the intraband pair inter-
action coupling constants and g12 are the interband pair interaction coupling constant.
 Mesoscopic Superconductivity 147

N1 states for band 1 with fixed width 2ω1D and a set of N2 states for band
2 with fixed width 2ω2D , respectively.
In this study, we assume that the Debye energies for two bands coin-
cide, i.e.
ω1D = ω2D = ωD . (4.36)
Within this assumption, N1 and N2 are relatively estimated by the DOS for
two bands as
N1 ρ1
= , (4.37)
N2 ρ2
where ρ1 and ρ2 are the DOS for two bands, respectively. The interaction
constants g1 and g2 can be written as
g1 = d1 λ1 , g2 = d2 λ2 , (4.38)
where d1 and d2 are the mean single-particle level spacings,
2ωD 2ωD
d1 = , d2 = , (4.39)
N1 − 1 N2 − 1
and λ1 and λ2 are the dimensionality interaction parameters for two bands.
We define the interband interaction constant as
g12 = d1 d2 λ12 . (4.40)
In summary, we obtain the relation
ρ1 N1 − 1 d2
≈ = . (4.41)
ρ2 N2 − 1 d1
The system we are considering consists of two half-filled bands, each of
which has equally spaced Nn single-particle levels and Mn (= Nn /2) doubly
occupied pair levels (n = 1, 2). We take the single-particle level spacing as
our energy unity. Thus, the single-particle spectrum is given by
εnj = dn j − ωD , j = 1, 2, . . . , Nn (n = 1, 2). (4.42)
Richardson has obtained his solution within the single-band model for
an arbitrary set of single-particle levels. For simplicity, we assume that there
are no singly occupied single-particle levels. As can be seen from Eqs. (4.33)–
(4.35), these levels are decoupled from the rest of the system. They are said
to be blocked and contribute to the total energy with their single-particle
energies. The above simplification implies that every single-particle level j
is either empty (i.e. |vac) or occupied by a pair of electrons (i.e. a†j↑ a†j↓ |vac
and b†j↑ b†j↓ |vac). These are called the unblocked level.
148 Modern Aspects of Superconductivity: Theory of Superconductivity

4.3.3 Exact solution

In order to extend Richardson’s solution to a two-band system, we define
two kinds of hard-core boson operators as
cj = aj↓ aj↑ , c†j = a†j↑ a†j↓ , (4.43)

dj = bj↓ bj↑ , d†j = b†j↑ b†j↓ , (4.44)

which satisfy the commutation relations
c†2
j = 0, [cj , c†k ] = δjk (1 − 2c†j cj ), [c†j cj , c†k ] = δjk c†j , (4.45)

d†2
j = 0, [dj , d†k ] = δjk (1 − 2d†j dj ), [d†j dj , d†k ] = δjk d†j , (4.46)
which reflects the Pauli principle for the fermions they were constructed
from.
The Hamiltonian (4.32) for the unblocked levels can then be written as

N1 
N1 
N2 
N2
HU = 2 ε1j c†j cj − g1 c†j ck + 2 ε2j d†j dj − g2 d†j dk
j jk j jk


N1 
N2 
N2 
N1
+ g12 c†j dk + g12 d†j ck . (4.47)
j k j k

We find the eigenstates |M1 ; M2  of this Hamiltonian with M1 + M2 pairs

in the form
HU |M1 ; M2 U = E(M1 ; M2 )|M1 ; M2 U
M 
 1 M2
= E1J + E2K |M1 ; M2 U , (4.48)
J=1 K=1

where E(M1 ; M2 ) is the eigenvalue and

)
M1 )
M2
|M1 ; M2 U = CJ† †
DK |vac, (4.49)
J=1 K=1
with

N1
c†j 
N2
d†j
CJ† = , DJ† = . (4.50)
2ε1j − E1J 2ε2j − E2J
j j

Now, we define C0† and D0† as


N1 
N2
C0† = c†j , D0† = d†j . (4.51)
j j
 Mesoscopic Superconductivity 149

Then we can rewrite Eq. (4.47) as


N1 
N2
HU = 2 ε1j c†j cj − g1 C0† C0 +2 ε2j d†j dj − g2 D0† D0
j j

+ g12 C0† D0 + g12 D0† C0 . (4.52)

The commutation relations for new operators are given as

c†j d†j
[c†j cj , CJ† ] = , [d†j dj , DJ† ] = , (4.53)
2ε1j − E1J 2ε2j − E2J

N1
1 − c†j cj 
N2
1 − d†j dj
[C0 , CJ† ] = , [D0 , DJ† ] = , (4.54)
2ε1j − E1J 2ε2j − E2J
j j


N1
1 − c†j cj
[HU , CJ† ] = E1J CJ† + C0† + g1 C0†
2ε1j − E1J
j


N1
1 − c†j cj
+ g12 D0† , (4.55)
2ε1j − E1J
j


N2
1 − d†j dj
[HU , DJ† ] = E2J DJ† + D0† + g1 D0†
2ε2j − E2J
j


N2
1 − d†j dj
+ g12 C0† . (4.56)
2ε2j − E2J
j

Using the above-presented commutation relations, we find that

M 
 1 
M2
HU |M1 ; M2 U = E1J + E2K |M1 ; M2 U
J=1 K=1
⎛ ⎞
†

M1 
N1
g1 
M1
2g1
+ C0 ⎝1 − + ⎠|M1 (J); M2 U
2ε1j − E1J E − E1J
1J 
J=1 j J  =J
⎛ ⎞

M1 N1
g12 
M1
2g12
+ D† ⎝ − ⎠|M1 (J); M2 U
0
2ε1j − E1J 
E1J  − E1J
J=1 j J =J
150 Modern Aspects of Superconductivity: Theory of Superconductivity

⎛ ⎞
†

M2 
N2
g12 
M2
2g12
+ C0 ⎝ − ⎠|M1 ; M2 (K)U
2ε2j − E2K E2K  − E2K
K=1 j K  =K
⎛ ⎞

M2 
N2
g2 
M2
2g2
+ D0† ⎝1 − + ⎠|M1 ; M2 (K)U ,
2ε2j − E2K 
E2K  − E2K
K=1 j K =K
(4.57)

where

)
L−1 )
M1 )
M2
|M1 (L); M2 U = CJ† CJ†  †
DK |vac, (4.58)
J=1 J  =L+1 K=1

)
M1 )
L−1 )
M2
|M1 ; M2 (L)U = CJ† †
DK †
DK  |vac. (4.59)
J=1 K=1 K  =L+1

Comparing Eq. (4.57) with Eq. (4.48), we obtain, for arbitrary J and K,
  
  1 + g1 A1J −g12 A2K |M1 (J); M2 U
C0† †
D0 = 0, (4.60)
−g12 A1J 1 + g2 A2K |M1 ; M2 (K)U

where


Nn
1 
Mn
2
AnL = − + . (4.61)
2εnj − EnL 
EnL − EnL
j L =L

A nontrivial solution to Eq. (4.60) is derived from the determinantal

equation

FJK = (1 + g1 A1J ) (1 + g2 A2K ) − g12

2
A1J A2K = 0. (4.62)

This constitutes a set of M1 + M2 coupled equations for M1 + M2 param-

eters E1J and E2K (J = 1, 2, . . . , M1 ; K = 1, 2, . . . , M2 ), which may be
thought of as self-consistently determined pair energies. Equation (4.62)
is the exact eigenvalue equation for a two-band superconducting system
and can be regarded as a generalization of Richardson’s original eigenvalue
equation.
 Mesoscopic Superconductivity 151

4.3.4 Preprocessing for numerical calculations

To remove the divergences from the second term of AnL in Eq. (4.61), we
make changes of the energy variables:

En2λ = ξnλ + iηnλ ,

En2λ−1 = ξnλ − iηnλ , (4.63)
λ = 1, 2, . . . , Mn /2,

where we assume that the number of pairs is even. Since the complex
pair energies appear in complex conjugate pairs, the total energy will be
expressed as the real component.
A further transformation is necessary in order to remove the divergences
from the first term of AnL . We define new variables xnλ and ynλ as

ξnλ = εn2λ + εn2λ−1 + dn xnλ (xnλ ≤ 0), (4.64)

η2nλ = −(Δε2n2λ − d2n x2nλ )ynλ (ynλ ≥ 0), (4.65)

where

Δεn2λ = εn2λ − εn2λ−1 . (4.66)

Considering the sign of ynλ , we can express ηnλ as

ηnλ = |ηnλ |e−iφnλ ,


0 for Δε2n2λ − d2n x2nλ ≤ 0,
φnλ = π (4.67)
2 for Δε2n2λ − d2n x2nλ > 0.

Then we can rewrite FJK , by using the new variables, and define the result
as Fαβ . We extract the real and imaginary parts of Fαβ as
1 ∗
+
Fαβ = (Fαβ + Fαβ ) = 1 + g1 R1α + g2 R2β + (g1 g2 − g12
2
)R1α R2β
2
− (g1 g2 − g12
2
)I1α I2β cos(φ1α + φ2β )
− {g1 + (g1 g2 − g12
2
)R2β } I1α sin φ1α
− {g2 + (g1 g2 − g12
2
)R1α } I2β sin φ2β , (4.68)
− 1 ∗
Fαβ = (Fαβ − Fαβ ) = −(g1 g2 − g12
2
)I1α I2β sin(φ1α + φ2β )
2i
+ {g1 + (g1 g2 − g12
2
)R2β }I1α cos φ1α
+ {g2 + (g1 g2 − g12
2
)R1α }I2β cos φ2β , (4.69)
152 Modern Aspects of Superconductivity: Theory of Superconductivity

where

2dn xnλ (1 + ynλ )

Rnλ = −
(1 − ynλ ) Δε2n2λ − (1 + ynλ )2 d2n x2nλ
2


Mn
ξnμλ (ξ2nμλ + η2nμ + η2nλ )
+4
(ξ2nμλ + η2nμ + η2nλ )2 − 4η2nμ η2nλ
μ=λ


Nn
2εnλ − ξnλ
− , (4.70)
(2εnλ − ξnλ )2 + η2nλ
j=2λ−1,2λ

1 − ynλ
2
Inλ =
(1 − ynλ )2 Δε2n2λ − (1 + ynλ )2 d2n x2nλ

Mn
ξ2nμλ − η2nμ + η2nλ
− 4ynλ
(ξ2nμλ + η2nμ + η2nλ )2 − 4η2nμ η2nλ
μ=λ
⎫ ,
⎬  2 

Nn
1  Δεn2λ − d2n x2nλ 
+ ynλ ×  , (4.71)

(2εnλ − ξnλ )2 + η2nλ ⎭ ynλ
n=2λ−1,2λ

with

ξnμλ = ξnμ − ξnλ . (4.72)

Thus, for an arbitrary combination of α and β, we must solve the following

equations:

+
Fαβ = 0,
−
Fαβ = 0 (α = 1, 2, . . . , M1 /2; β = 1, 2, . . . , M2 /2). (4.73)

4.3.5 Results and discussion

We now apply the exact solution for a two-band system to discuss the prop-
erties of the two-band superconductivity in ultrasmall grains. The single-
particle level patterns of the (2M1 + m) + 2M2 electron system (m = 0, 1, 2)
under consideration are represented in Figs. 4.6(a), 4.6(b) and 4.6(c), respec-
tively. The dotted lines represent the chemical potential, and d1 and d2
(d1 < d2 ) are the mean level spacings. As seen in these figures, the addi-
tional electrons first occupy band 1 and then band 2.
 Mesoscopic Superconductivity 153

Fig. 4.6 Single-particle levels near the Fermi level in the case of two-band superconduc-
tivity. The dotted lines represent the chemical potential. The left and right bands are
band 1 and band 2, respectively. d1 and d2 are the mean level spacings. (a) 2M1 + 2M2
electron system, where Mn is the number of pair levels; (b) (2M1 + 1) + 2M2 electron
system; (c) (2M1 + 2) + 2M2 electron system.

Numerical calculations are carried out under the condition that N1 :N2 =
3:2, ωD = 50 and λ1 = λ2 = λ.

4.3.6 Pair energy level

By minimizing the sum of squares of Eqs. (4.69) and (4.70),
M1 /2 M2 /2 
 
+2 −2
F = Fαβ + Fαβ , (4.74)
α=1 β=1

for various interaction parameters, we obtain a behavior of pair energy levels

EnJ of two bands, as shown in Fig. 4.7, in which the solid and broken lines
correspond to the pair energy levels of band 1 and band 2, respectively. The
parameters used in this calculation are N1 = 12, N2 = 8, M1 = 6, M2 = 4,
0 ≤ λ ≤ 1.0 and 0 ≤ λ12 ≤ 0.2.
As seen in the figures, band 2 condenses into degenerate levels, but band
1 does not. In general, we can expect that the single-particle levels in a band,
whose mean level spacing d is larger than that of the other band, degenerate
faster. The behavior of the condensing band is qualitatively the same as
that in the case of calculations for the single band.166 The coexistence of
the normal band and the condensed one may be reflected in the opposite
phase of the gaps of these bands.156
154 Modern Aspects of Superconductivity: Theory of Superconductivity

Fig. 4.7 Typical behavior of pair energy levels of two bands for the ground state. The
parameters used in calculation are N1 = 12, N2 = 8, M1 = 6, M2 = 4, ωD = 50,
0 ≤ λ1 = λ2 ≤ 1.5 and 0 ≤ λ12 ≤ 0.3. The solid and broken lines correspond to the pair
energy levels of band 1 and band 2, respectively.

4.3.7 Condensation energy

The condensation energy of band n for the (2M1 +m)+2M2 electron system
can be defined as
 m
EnC (2M1 + m, 2M2 ) = En (2M1 + m, 2M2 ) + Mn + gn
2
− En0 (2M1 + m, 2M2 ), (4.75)
where En (2M1 + m, 2M2 ) and En0 (2M1 + m, 2M2 ) are the ground-state ener-
gies and the sum of the single-particle energies, respectively.
We calculate the condensation energies and show them in Figs. 4.8(a)
and 4.8(b). The parameters used in this calculation are λ = 0.5, and
λ12 = 0.01 for (a) and λ12 = 0.1  for (b). Values are normalized by the
−1 λ
bulk gap, Δ = ωD sinh λ2 −λ212
. The solid and broken lines correspond
to the condensation energy for band 1 and band 2, respectively. The lines
plotted with squares, triangles and circles are for the 2M1 + 2M2 electron
system, (2M1 + 1) + 2M2 electron system and (2M1 + 2) + 2M2 electron
system, respectively.
As seen in the figures, we can understand that band 2 condenses, but
band 1 does not because of the sign of values. This difference of signs may
also be reflected in the opposite phases of the gaps of these bands. The
behavior of the results for the condensed band (band 2) is qualitatively the
same as in the case of the single-band calculations. The condensation energy
 Mesoscopic Superconductivity 155

(a) (b)

Fig. 4.8 Condensation energy. The parameters used in calculation are ω “ D = 50 ”and
λ1 = λ2 = 0.5. Values are normalized by the bulk gap, Δ1 = ωD sinh−1 λ λ λ−λ 2
2 or
“ ” 1 2 12
−1 λ1
Δ2 = ωD sinh λ1 λ2 −λ2
. The solid and broken lines correspond to the condensation
12
energy for band 1 and band 2, respectively. The lines plotted with squares, triangles
and circles are for the 2M1 + 2M2 electron system, (2M1 + 1) + 2M2 electron system
and (2M1 + 2) + 2M2 electron system, respectively. (a) The condensation energy for the
interband coupling parameter λ12 = 0.01; (b) the condensation energy for λ12 = 0.1.

of band 2 for the (2M1 +2)+2M2 electron system is, however, different from
the others. We can also see that the condensation energy is affected by the
interband interaction λ12 . This was mentioned in our previous work.156

4.3.8 Parity gap

The parity gap of band n is defined as

Δpn = En (2M1 + 1, 2M2 )

1
− {En (2M1 , 2M2 ) + En (2M1 + 2, 2M2 )}, (4.76)
2
which was introduced by Matveev and Larkin and characterizes the differ-
ence of even–odd ground-state energies.169
We have also calculated the parity gaps shown in Fig. 4.9. The solid and
broken lines correspond to the parity gap for band 1 and band 2, respectively.
The lines plotted with triangles and squares are for the interband coupling
parameter λ12 = 0.01 and λ12 = 0.1, respectively. Other parameters used
in this calculation are the same as for the calculation of the condensation
energy. Values are normalized by the bulk gap.
For the condensed band, we obtain qualitatively the same result as in
the case of the single-band calculations, i.e. there is a minimal point and a
tendency towards 1 for d → 0. The mean level spacing giving the minimal
156 Modern Aspects of Superconductivity: Theory of Superconductivity

Fig. 4.9 Parity gap. The parameters used in calculation are ωD = 50 and λ1 = λ2 = 0.5.
The solid and broken lines correspond to the parity gap for band 1 and band 2, respectively.
The lines plotted with triangles and squares are for the interband coupling parameter
λ12 = 0.01 and λ12 = 0.1, respectively. Values are normalized by the bulk gap.

point is, however, much less than that for the case of calculations for the sin-
gle band. The parity gap is almost independent of the interband interaction
λ12 . This was also mentioned in our previous work.156
We have extended Richardson’s exact solution to the two-band system,
and have derived a new coupled equation. To investigate the properties of
the two-band superconductivity, we have solved the equation numerically
and determined the behavior of pair energy levels, the condensation energy
and the parity gap.
The band whose mean level spacing is larger than that of the other band
degenerates and condenses faster. The behavior of the condensing band is
qualitatively the same as that in the case of calculations for the single band.
The coexistence of the normal band and the condensed one may be reflected
in the opposite phases of the gaps of these bands. This phase character
appears in all the results of numerical calculations. Therefore, the phase of
a gap is important for stabilizing the two-band superconductivity.
We have also calculated the condensation energy and the parity gap for
two-band superconductivity. The results suggest that the interband inter-
action λ12 affects the condensation energy, but not the parity gap.
In summary, the expression of Richardson’s exact solution for two-band
superconductivity has been presented, by solving numerically a new cou-
pled equation. Then, the behaviors of pair energy levels, the condensation
 Mesoscopic Superconductivity 157

energy and the parity gap have been determined. The results for the con-
densed band are almost qualitatively the same as those for the single-band
calculation, and the coexistence of the normal band and the condensed one
may have originated from the opposite phases of the gaps of these bands.

4.4 Kondo Effect Coupled to Superconductivity

The Kondo effect has attracted great interest, while considering the proper-
ties of semiconductor quantum dots. It can be understood as the magnetic
exchange interaction between a localized impurity spin and free conduction
electrons.174 To minimize the exchange energy, the conduction electrons tend
to screen the spin of the magnetic impurity, and the ensemble forms a spin
singlet. In a quantum dot, some exotic properties of the Kondo effect have
been observed.175, 176 Recently, Sasaki et al. have found a significant Kondo
effect in quantum dots with an even number of electrons.177 The spacing of
discrete levels in such quantum dots is comparable with the strength of the
e–e Coulomb interaction. The Kondo effect in multilevel quantum dots has
been investigated theoretically by several groups.178–180 They have shown
that the contribution from many levels enhances the Kondo effect in nor-
mal metals. There are some investigations into the Kondo effect in quantum
dots revealing ferromagnetism,181 noncollinear magnetism,182 superconduc-
tivity183 and so on.184, 185
Properties of ultrasmall superconducting grains have also been theoreti-
cally investigated by many groups.160–162, 164, 165, 167–169, 186 Black et al. have
revealed the presence of a parity-dependent spectroscopic gap in the tunnel-
ing spectra of nanosize Al grains.160, 164 For such ultrasmall superconducting
grains, the bulk gap has been discussed in relation to physical properties
such as the parity gap,169 condensation energy165 and electron correlation166
with the size dependence of the level spacing of samples.167 In the previous
works,156 we have also discussed physical properties such as condensation
energy, the parity gap, and electron correlation of two-gap superconduc-
tivity in relation to the size dependence and the effective pair scattering
process. In addition, the possibility of new two-gap superconductivity has
been discussed by many groups.14, 97, 119, 129, 133, 135, 137, 140, 187, 188
In a standard s-wave superconductor, the electrons form pairs with
antialigned spins and are in a singlet state as well. When the supercon-
ductivity and the Kondo effect are present simultaneously, they are usu-
ally expected to be competing physical phenomena. The local magnetic
158 Modern Aspects of Superconductivity: Theory of Superconductivity

moments from the impurities tend to align the spins of the electron pairs
in the superconductor, which often results in a strongly reduced transition
temperature. Buitelaar et al. have experimentally investigated the Kondo
effect in a carbon nanotube quantum dot coupled to superconducting Au/Al
leads.183 They found that the superconductivity of the leads does not destroy
the Kondo correlations in the quantum dot at the Kondo temperature. A
more subtle interplay has been proposed for exotic and not well-understood
materials such as heavy-fermion superconductors, in which the two effects
might actually coexist.189
In this paper, we investigate the Kondo effect and the superconductivity
in ultrasmall grains by using a model which consists of the sd and reduced
BCS Hamiltonians with the introduction of a pseudofermion. The mean-
field approximation for the model is introduced, and we calculate physical
properties of the critical level spacing and the condensation energy. These
physical properties are discussed in relation to the coexistence of the super-
conductivity and the Kondo regime. Finally, we derive the exact equation
for the Kondo regime in a nanosystem and discuss the condensation energy
from the viewpoint of the correlation energy.

4.4.1 Kondo regime coupled to superconductivity

In nanosize superconducting grains, the quantum level spacing approaches
the superconducting gap. It is necessary to treat the discretized energy lev-
els of a small-size system. For ultrasmall superconducting grains, we can
consider the pairing-force Hamiltonian to describe the electronic structure
of the system172 and can determine the critical level spacing in the case
where the superconducting gap function vanishes at a quantum level spac-
ing.167 In this section, we present a model for a system in the Kondo regime
coupled to the superconductivity and discuss the physical properties such
as critical level spacing and condensation energy by using the mean-field
approximation in relation to the gap function, spin singlet order as the
Kondo effect, coexistence and so on.

4.4.2 Model
We consider a model coupled to the superconductivity for quantum dots
to investigate the Kondo effect in normal metals, which can be expressed
by the effective low-energy Hamiltonian obtained by the Schrieffer–Wolff
 Mesoscopic Superconductivity 159

transformation190 :

H = H0 + H1 + H2 , (4.77)

where
 
H0 = εk a†kσ akσ + Eσ d†σ dσ , (4.78)
k,σ σ
'  (
H1 = J S+ a†k ↓ ak↑ + S− a†k ↑ ak↓ + Sz a†k ↑ ak↑ − a†k ↓ ak↓ , (4.79)
k,k 

H2 = −g a†k↑ a†k↓ ak ↓ ak ↑ . (4.80)
k,k 

a†kσ (akσ ) and d†σ (dσ ) are the creation (annihilation) operator corresponding
to conduction electrons and the effective magnetic particle as an impurity,
respectively. In this study, we assume that the magnetic particle is a fermion
with S = 1/2, for simplicity. E represents an extraction energy given by
E↑,↓ = −E0 ±Ez , including the Zeeman effect. The second term of Eq. (4.77)
is the interaction between conduction electrons and the spin in a quantum
dot. S is the spin operator as S+ = d†↑ d↓ , S− = d†↓ d↑ and Sz = (d†↑ d↑ −
d†↓ d↓ )/2. The third term corresponds to the interaction between conduction
electrons included in the pairing-force Hamiltonian.
Here, we introduce a pseudofermion for the magnetic particle opera-
tor 191 as

d†↑ = f↓ , d↑ = f↓† ,

d†↓ = −f↑ , d↓ = −f↑† . (4.81)

For this transformation, we have the condition

f↑† f↑ + f↓† f↓ = 1, (4.82)

and we have |σ = fσ† |0. The spin operator S can be presented as S+ = f↑† f↓ ,
S− = f↓† f↑ and Sz = (f↑† f↑ − f↓† f↓ )/2. The Hamiltonian can be rewritten as
 † 
H0 = ε̃k ckσ ckσ + Efσ† fσ , (4.83)
k,σ σ

H1 = J fσ† fσ c†k σ ckσ , (4.84)
k,k  ,σ,σ
160 Modern Aspects of Superconductivity: Theory of Superconductivity


H2 = −g c†k↑ c†k↓ ck ↓ ck ↑ , (4.85)
k,k 
. . †
where ckσ = i Uik aiσ and ε̃k = i,j Uki [εi δij − J/2]Ujk . For the sake of
simplicity, we focus only on Ez = 0 without an external magnetic field:
E = E0 .

4.4.3 Mean-field approximation

In this subsection, we introduce the mean-field approximation for the present
Hamiltonian (4.77). Eto et al. have presented the mean-field approximation
for the Kondo effect in quantum dots.192
In the mean-field approximation, we can introduce the spin singlet order
parameter
1  †
Ξ= √ fσ ckσ . (4.86)
2 k,σ
This order parameter describes the spin couplings between the dot states
and conduction electrons. The superconducting gap function can be
expressed as

Δ= ck↓ ck↑ . (4.87)
k

Using these order parameters in Eqs. (4.84) and (4.85), we obtain the mean-
field Hamiltonian
 †  √ 
HMF = ε̃k ckσ ckσ + Ẽfσ†fσ + 2J [Ξfσ c†kσ + Ξ∗ ckσ fσ† ]
k,σ σ k,σ

−g [Δ∗ ck↓ ck↑ + Δc†k↑ c†k↓ ]. (4.88)
k

The constraint (4.82) is taken into account by the second term with a
Lagrange multiplier λ. In this study, we assume a constant DOS with
the energy region of the Debye energy, and the coupling constants can be
expressed as J = dJ˜ and g = dλ.

4.4.4 Critical level spacing in the Kondo effect

By minimizing the expectation value of HMF in Eq. (4.88), the order param-
eters can be determined self-consistently. First, we show the Kondo effect
without the pairing-force part (g = 0) in the framework of the mean-field
 Mesoscopic Superconductivity 161

approximation. Next, the Kondo effect in the presence of the superconduc-

tivity is discussed in relation to the critical level spacing and the conden-
sation energy. Finally, we derive the exact equation for the Kondo effect in
ultrasmall grains coupled to normal metals and discuss properties such as
the condensation energy in relation to Richardson’s exact equation for the
superconductivity.
For ultrasmall superconducting grains, the critical level spacing dBCS
c can
be expressed as dBCS = 4ω eγ exp(−1/λ) for an even number of electrons,
c D
where ωD means the Debye energy. This result suggests that the gap func-
tion of a nanosize system with the level spacing d vanishes when the coupling
parameter λc is less than the value (ln 4ωD /d + γ)−1 . The bulk gap function
Δc with λc can be expressed as Δc = ωD sh−1 (1/λc ).
Figure 4.10(a) shows the gap function of a nanosize system in the frame-
work of the standard BCS theory. We can find the region where the gap
function vanishes when the coupling becomes less than λc . This means that
the level spacing is larger than the gap function in this region.
Here, we drive the critical level spacing for only the Kondo regime
(λ = 0). The equation determining the singlet order parameter can be

Fig. 4.10 Gap function and spin-singlet order. (a) The gap function. This vanishes in the
region of λ values less than λc . (b) Spin singlet order parameter. In the case of J˜ < J˜c ,
the singlet order vanishes. The system consists of eight energy levels and eight electrons
with the level spacing d = 1.0 and ωD = 1.0.
162 Modern Aspects of Superconductivity: Theory of Superconductivity

expressed as
 Ξ(ξk − x)
Ξ= , (4.89)
(ξk − x)2 + Ξ2
k
' + (
where ξk = ε̃k −μ, x = ε̃k + Ẽ ± (ε̃k − Ẽ)2 + 4Ξ2 /2 and μ is the chemical
potential. For the case of the critical level spacing, the solution shows that
the spin singlet order parameter vanishes. From Eq. (4.89), we can find the
critical level spacing dKondo
c for the Kondo regime.

γ 1
dcKondo
= 4ωD e exp − √ . (4.90)
2 2J˜
√
When the coupling parameter J˜ is smaller than J˜c = [2 2(ln(4ωD /d) +
γ)]−1 , the spin singlet order parameter vanishes.
Figure 4.10(b) presents the spin singlet order parameter given by
Eq. (4.86) in the case g = 0. In the region of J˜ < J˜c , the order param-
eter vanishes. This result suggests the critical level spacing in the Kondo
effect.

4.4.5 Kondo effect coupled to superconductivity

Here, we consider a simple system which consists of eight energy levels
and eight electrons, and investigate the critical level spacing and the con-
densation energy of the coupled system between the superconductivity and
the Kondo regime in the framework of the mean-field approximation of
Eq. (4.88).
Figure 4.11(a) shows the spin singlet order parameter and the gap func-
tion for several cases. We can find the critical level spacings for the gap
function and for the spin singlet order parameter. When λ < λc and J˜ > J˜c ,
we can find only the spin singlet order parameters. In the region of λ/λc
from 1.4 to 1.7 with J/˜ J˜c = 0.189, we can find the coexistence of the gap
function and the spin singlet order parameter. For λ/λc larger than 1.7, only
the gap function still exists, and the spin singlet order parameter vanishes.
˜ J˜c = 0.284, we can find the coexistence in the region λ/λc = 1.7 − 2.3.
At J/
These results suggest that strong local magnetic moments from the impuri-
ties reduce the transition temperature for superconductivity. However, the
weak couplings λ of the superconductivity do not destroy the spin singlet
order parameter at all. These results are in good agreement with the exper-
imental results.183 We can find that there is a coexistence region for the
superconductivity and the Kondo regime.
 Mesoscopic Superconductivity 163

Fig. 4.11 Physical properties in a coupled system. (a) Gap function and spin singlet order
parameter. (b) Condensation energy. J/˜ J˜c = 0, 0.94741, 1.8948, 2.8422. Other parameters
are the same.

Figure 4.11(b) shows the condensation energy for several λ and J˜ values.
We have found that the condensation energy of the coupled system between
the superconductivity and the Kondo regime becomes lower than that for the
pure superconductivity. In the coexistence region, the highest value of the
condensation energy appears in all cases.

4.4.6 Exact solution for the Kondo regime

The standard BCS theory gives a good description of the phenomenon of
superconductivity in large samples. However, when the size of a supercon-
ductor becomes small, the theory fails. To investigate the physical properties
such as the condensation energy, and the parity gap, it is necessary to per-
form a more accurate treatment. For the superconductivity in ultrasmall
grains, the exact solution to the reduced BCS Hamiltonian presented by
Richardson172 has been applied to investigate the above-mentioned physical
properties.168
By using the wave function describing all pair electron excitations, we
can derive the exact solution for the pairing-force (reduced) Hamiltonian:


N
λ 
n
2λ
2− + = 0, (4.91)
ε̃k − Ei El − Ei
k=1 l=1, l=i
164 Modern Aspects of Superconductivity: Theory of Superconductivity

Fig. 4.12 Exact solution for the superconductivity. (a) Condensation energy of the exact
solution and that obtained in the mean-field approximation. (b) Pairing energy level with
energy level obtained in the mean-field approximation. Eight energy levels, eight electrons,
d = 1.0, ωD = 4.0.

where N and n are the number of orbitals and the number of the occupied
orbitals, respectively, and Ei corresponds to the exact orbital. Figure 4.12
shows the condensation energy and the pairing energy level for the nanosize
superconductivity. Note that the physical properties obtained in the mean-
field approximation give a good description of the high DOS (d → ∞).
We have found the different behavior of the condensation energy from
that obtained in the mean-field approximation, as shown in Fig. 4.12(a).
Figure 4.12(b) presents the qualitative behavior of the pairing energy level in
the ground state. At λ about 1.6, more than two energy levels in Fig. 4.12(b)
are completely paired. The pairing behavior has already been reported by
many groups.138, 172
Let us derive the exact equation for the Kondo regime in ultrasmall
grains. We can consider the Hamiltonian H = H0 + H1 in Eq. (4.77). We
introduce a creation operator describing all excited states at the spin singlet
coupling between a conduction electron and a pseudofermion:
†
 c† fσ
kσ
Bj = , (4.92)
ε̃k − Ej
k,σ
where Ej represents the exact eigenenergies in the Kondo regime. The exact
eigenstate |Ψn  for the Kondo regime can be written as |Ψn  = Πnυ=1 Bυ† |0.
Other electrons, which are not related to the spin singlet order, contribute
.
Esingle = nk=1 ε̃k to the eigenenergy. The ground-state energy EGS can be
.
written as EGS = nk=1 [Ek + ε̃k ].
 Mesoscopic Superconductivity 165

By operating the Hamiltonian to the exact eigenstate, we obtain the

condition
N
J˜
1+ = 0. (4.93)
ε̃k − Ej
k=1
This equation gives the exact solution for the Kondo regime. Note that the
creation operator (4.92) might be a true boson one compared with the case
of the reduced BCS model.
Figure 4.13 shows the condensation energy of the exact solution in the
Kondo regime with that obtained in the mean-field approximation. We can
find the different behavior of the condensation energy from that obtained
in the mean-field approximation. However, the behavior is similar to that
in the case of the superconductivity in nanosize systems.
We have investigated properties of the Kondo regime coupled to the
superconductivity in ultrasmall grains by using the mean-field approxima-
tion. In the framework of the mean-field approximation, we have found the
critical level spacing for the Kondo regime. The result suggests that the
Kondo effect vanishes when the level spacing becomes larger than the criti-
cal level spacing.
We have calculated physical properties of the critical level spacing and
the condensation energy of the coupled system by using the mean-field

Fig. 4.13 Condensation energy for the Kondo regime: all parameters used in the system
are as follows: eight energy levels, eight electrons, d = 1.0 and ωD = 4.0.
166 Modern Aspects of Superconductivity: Theory of Superconductivity

approximation. From the results, we have found that strong local magnetic
moments from the impurities reduce the transition temperature for super-
conductivity. However, weak couplings λ of the superconductivity do not
destroy the spin singlet order parameter at all. These results are in good
agreement with the experimental results.183 We have found that there is a
coexistence region for the superconductivity and the Kondo regime.
Finally, we have derived the exact equation for the Kondo regime in
a nanosystem, which was not an easy task, and have discussed the con-
densation energy from the viewpoint of energy levels. Further study of the
properties in the Kondo regime with the use of the exact equation will be
presented elsewhere.
In summary, we have investigated the Kondo effect and the supercon-
ductivity in ultrasmall grains by using a model which involves the sd and
reduced BCS Hamiltonians with the introduction of a pseudofermion. The
mean-field approximation for the model has been introduced, and we have
calculated physical properties of the critical level spacing and the conden-
sation energy. These physical properties have been discussed in relation to
the coexistence of the superconductivity and the Kondo regime. Finally, we
have derived the exact equation for the Kondo regime in a nanosystem and
discussed the condensation energy from the viewpoint of energy levels.

4.5 Interaction of Nanoscale Ferromagnetic Granules

in London Superconductors
Recent experiments have fabricated structured arrays of ferromagnetic
nanoparticles in proximity to a bulk superconductor. We consider the the-
ory of interactions between two nanoscale ferromagnetic particles embedded
in a superconductor. In the London limit approximation, we will show that
the interactions between ferromagnetic particles can lead to either parallel
or antiparallel spin alignment. The crossover between these is dependent
on the ratio of the interparticle spacing and the London penetration depth.
We show that a phase transition between spin orientations can occur under
variation of the temperature. Finally, we comment on the extension of these
results to arrays of nanoparticles in different geometries.
Magnetism and superconductivity are two competing collective ordered
states in metals. In the case of ferromagnetism, the exchange interactions
lead to parallel alignment of electronic spins, while e–p interactions in the
BCS superconductivity lead to spin singlet pairing of electrons. Clearly,
 Mesoscopic Superconductivity 167

these two types of order are in general mutually incompatible. In bulk

systems, the frustration between the electron singlet pairing and the spin
ordering is resolved by the FFLO (Fulde–Ferrell,193 Larkin–Ovchinnikov194 )
state. However, this has proven to be elusive experimentally, and few exam-
ples are known. In particular, the FFLO state appears to be highly sensitive
to disorder.
In recent years, however, there has been a great increase of interest in the
interactions between ferromagnetism and superconductivity in artificially
structured systems. Advances in nanotechnology and microfabrication have
made it possible to build hybrid structures containing both ferromagnetic
and superconducting components which interact magnetically or via the
proximity effect.195, 204 Superconductor–ferromagnet–superconductor pla-
nar structures have been found to show the π-junction Josephson behav-
ior.196–198 Ferromagnet–superconductor–ferromagnet spin valve structures
have also been fabricated,199 with potential applications to spintronics. In
addition, more complex types of structures have been produced; for exam-
ple, Moschalkov et al. fabricated arrays of ferromagnetic nanoscale dots on
superconducting substrates.200, 201 A description of such structures based
upon the G–L theory was developed by Peeters.202
In this section, we consider the interaction between magnetic nanopar-
ticles embedded in a superconductor, as shown in Fig. 4.14. Theoretical
studies of such systems can be carried out exactly in two limiting cases,
depending on the relative magnitudes of the London penetration depth, λ,
and the coherence length, ξ. If we consider the nanoparticles to be essen-
tially pointlike on the scale of both these characteristic lengths, then they
correspond to effective pointlike magnetic moments of the form

M(r) = mi δ(r − ri ), (4.94)
i

Fig. 4.14 Schematic diagram of a nanocomposite superconductor containing ferromag-

netic granules.
168 Modern Aspects of Superconductivity: Theory of Superconductivity

where the particle at ri has magnetic moment mi . Interactions between

these isolated moments arise directly from magnetic dipole–dipole forces
modified by the screening of the bulk supercurrents. A second source of the
interaction between the moments is the RKKY interaction modified by the
presence of the BCS energy gap Δ. It is clear that the range of the dipolar
forces is determined by the penetration depth λ, while the usual oscillatory
power-law RKKY interaction is truncated exponentially on a length scale
of the order of ξm . Therefore, the dipolar forces dominate for superconduc-
tors in the London limit r0  λ, while the RKKY interactions are more
important in the Pippard case r0  l, where r0−1 = ξ−1 −1
m + l , and l is the
mean free path.15 Here, we consider the London limit and neglect RKKY
interactions. Magnetic impurities interacting via RKKY interactions were
considered by Larkin.203 In the London limit, we first derive general expres-
sions for the configuration of magnetic fields and the interaction energy of
an ensemble of ferromagnetic granules. Then we consider the case of two
interacting nanoparticles. It is found that, as a function of the temperature,
an orientational phase transition can take place. The conditions for such a
phase transition to occur are derived for a chain of ferromagnetic granules.
Finally, we comment on the application of these results to the determination
of the equilibrium configurations of more general lattices of ferromagnetic
particles.

4.5.1 Magnetic field of ferromagnetic inclusions

in a London superconductor
We determine the magnetic configurations in superconducting nanocompos-
ite systems by means of Maxwell’s equations. The total current, j, includes
both normal and superconducting parts:
j = js + jn . (4.95)
The role of the normal currents in a superconductor is negligible, since
the superconductor exhibits weak magnetic characteristics in the normal
state. However, a normal current will be present within the ferromagnetic
inclusions. This normal-state current can be written in the traditional form
jn = ∇ × M, (4.96)
where M is the magnetization of a material. The supercurrent obeys the
usual London equation
ns e2
∇ × js = − B (4.97)
m
 Mesoscopic Superconductivity 169

in SI units,64 where B is the magnetic field, ns is the superfluid density, and

m and e are the electron mass and charge, respectively.
Combining the relations (4.95)–(4.97) with the Maxwell equation
∇ × B = μ0 j, the magnetic field can be found in the form
∇ × (∇ × B) + λ−2 B = μ0 ∇ × (∇ × M), (4.98)
where λ is the London penetration depth of the field in the superconductor.
Since div B = 0, Eq. (4.98) can be rewritten as
−∇2 B + λ−2 B = μ0 ∇ × (∇ × M). (4.99)
For the boundary conditions, we assume that the magnetic field in the
superconductor vanishes far from the region of the ferromagnetic inclu-
sions. According to these boundary conditions, the solution Eq. (4.99) can
be expressed as follows:

μ0
B(r) = d3 r  G(|r − r |) ∇ × (∇ × M(r )), (4.100)
4π
where the Green function is given by
exp(−|r − r |/λ)
G(|r − r |) = . (4.101)
|r − r |
After the double integration by parts from manipulation (see App. B), this
expression can be rewritten in the general form
    
μ0 3  R 3R(R · M(r )) M(r )
B(r) = d r exp − · −
4π λ R5 R3
  
R R2 2R(R · M(r ))
× 1+ + 2 − , (4.102)
λ λ λ2 · R 3
where R = r − r and R = |r − r |.
The expression (4.102) determines the magnetic field outside of the ferro-
magnetic inclusions. Since the magnetization of the system M(r) is defined
by (4.96), we find in (4.101) that M(r) = 0 outside of the volume of ferro-
magnetic granules.

4.5.2 Magnetic field of ferromagnetic quantum dots

in a superconducting nanocomposite material
Assuming that the sizes of the ferromagnetic inclusions are on the nanome-
ter length scale, they will appear essentially pointlike on the scale of the
penetration depth λ. In this case, we can approximate the magnetization
170 Modern Aspects of Superconductivity: Theory of Superconductivity

as a sum of point magnetic moments, as shown in Eq. (4.94). Using this

approximation in the general expression (4.102), we find the result (see
App. B)    
μ0  Ri 3Ri (Ri · mi ) mi
B(r) = exp − · − 3
4π λ Ri5 Ri
i
  
Ri Ri2 2Ri (Ri · mi )
× 1+ + 2 − , (4.103)
λ λ λ2 · Ri3
where Ri = r − ri and Ri = |r − ri |.
It is apparent from Eq. (4.103) that if the temperature of a superconduc-
tor approaches Tc , and the penetration depth λ → ∞, then the expression
(4.103) tends to the limit  
μ0  3Ri (Ri · mi ) mi
B(r) = − 3 , (4.104)
4π
i
Ri5 Ri
which describes the usual magnetic field of isolated dipoles in the normal-
state medium.
In the other limiting case, if the distance between granules is greater
than the depth penetration, we have  
μ0  exp(−Ri /λ) Ri (Ri · mi )
B(r) = · − mi . (4.105)
4π Ri · λ2 R 2
i
i

4.5.3 Interaction energy of quantum dots in a

superconducting nanocomposite material
To determine the collective states of the magnetic moments in supercon-
ducting nanocomposite materials in the London limit, we make use of the
expression for the free energyF of the system205 :
1 5 6
F = d3 r B2 + λ2 (∇ × B)2 . (4.106)
2μ0
Integrating by parts and using the Gauss theorem, we transform Eq. (4.106)
into  
1
F = d r B{B + λ ∇ × (∇ × B)} = −2πλ
3 2 2
d3 r B · ∇ × (∇ × M).
2μ0
(4.107)
Integrating by parts and using the Gauss theorem again, we transform
Eq. (4.107) into 
F = −2πλ2 d3 r M · ∇ × (∇ × B). (4.108)
Then, using the London equation (4.98) and omitting the magnetostatic
self-energy from consideration, we find the interaction energy of magnetic
 Mesoscopic Superconductivity 171

moments. The obtained expression can be used to determine the collective

state of magnetization of an ensemble of granules:
   
μ0   Rij 3(Rij · mj )(Rij · mi ) mi · mj
U= exp − · −
8π λ R 5
ij Rij3
i j
  
2
Rji Rij 2(Rij · mj )(Rij · mi )
× 1+ + 2 − , (4.109)
λ λ λ2 · Rij
3

where Rij = rj − ri , Rij = |rj − ri |, i = j.

4.5.4 Spin-orientation phase transitions in a nanocomposite

material with arrays of ferromagnetic quantum dots
We begin by studying the magnetic configuration of an isolated pair of
magnetic moments. The interaction energy of such a pair can be written as
   
μ0 R12 3(R12 · m1 )(R12 · m2 ) m2 · m1
U =− exp − · −
4π λ R125 3
R12
 2  
R12 R12 2(R12 · m1 )(R12 · m2 )
× 1+ + 2 − . (4.110)
λ λ λ2 · R12
3

Let us introduce a coordinate system with the origin at the first magnetic
moment m1 and the polar axis along the line connecting magnetic moments.
In this coordinate system, the magnetic moments have the components mi =
mi (cos ϕi sin θi , sin ϕi sin θi , cos θi ), and their interaction energy (4.110) is
written in the form

μ0 exp(−R12 /λ)
U= m1 m2 · 3 · f (θi , ϕi , R12 /λ), (4.111)
4π R12

where

f (θi , ϕi , x) = −2(1 + x) cos θ1 cos θ2 + (1 + x + x2 ) sin θ1 sin θ2 cos ϕ

(4.112)

and ϕ = ϕ2 − ϕ1 .
Differentiating the function f (θi , ϕi , x) with respect to angular variables
and equating the results to zero, we find that there are four possible stable
172 Modern Aspects of Superconductivity: Theory of Superconductivity

energy configurations:
θ1 = θ2 = 0,
θ1 = θ2 = π, (4.113)
0 ≤ ϕ < 2π,
π
θ1 = θ2 = ,
2 (4.114)
ϕ = π,
π
θ1 = θ2 = ,
2 (4.115)
ϕ = 0,
θ1 = 0, θ2 = π,
(4.116)
0 ≤ ϕ < 2π,
which are illustrated in Fig. 4.15.

Fig. 4.15 Four energy saddle points of a pair of ferromagnetic quantum dots, as defined
in Eq. (4.113). Of the four, 1 and 2 correspond to the ground state, depending on the
condition (4.118). States 3 and 4 are never stable.
 Mesoscopic Superconductivity 173

Further analysis shows that configurations (4.115) and (4.116) are saddle
points, not energy minima. Evaluating the second derivatives of (4.112), we
obtain the stability condition of the configuration (4.113):
 2
R R
− − 1 ≤ 0. (4.117)
λ λ
This implies that the ferromagnetic ordering of the pair of magnetic
moments is possible if
R 1 √
≤ (1 + 5). (4.118)
λ 2
It turns out that if the condition (4.118) is violated, then the alternative
configuration (4.114) is a stable energy minimum. We can conclude that if
the temperature changes, and the penetration depth parameter λ(T ) varies
in such a way that the condition (4.118) is not satisfied, then the ground-
state orientation will change from (4.113) to (4.114).
This result is readily generalized to ordered arrays of ferromagnetic gran-
ules, and so we conclude that orientational phase transitions are possible in
systems of quantum dots in a superconducting matrix. For example, it is
clear that the condition (4.118) can be applied to linear chains of quantum
dots. On the other hand, the results for square or cubic lattices remain to
be determined.
We have considered the interactions between nanoscale magnetic dots
embedded in a bulk superconducting material. Our approach is valid for
materials which are well described in the London limit r0  λ, since RKKY
interactions are negligible. We have shown that, depending on the dimen-
sionless parameter R/λ, different stable ground states occur. So, as the tem-
perature varies, orientational phase transitions will take place for periodic
arrays of such quantum dots. Of course, our calculation does not include all
types of interactions which define the orientation of magnetic moments in
space. In particular, we neglect the energy of a magnetic anisotropy of gran-
ules which is determined by the shape of granules or the type of their crystal
lattice. However, when the shape of granules is close to the spherical one
and the lattice of a ferromagnet has the cubic symmetry, then Eq. (4.109)
will essentially be exact.
In the experimental systems studied by V. V. Moshchalkov,201 a square
of Pt/Co magnetic nanodots was deposited on the surface on the Pb super-
conductor, which is of type I(k = 0.48). The dots were about 0.26 μm in
diameter, and they were deposited on the grid with a spacing of 0.6 μm.
For Pb, the penetration depth is 39 nm at low temperatures. So, this array
174 Modern Aspects of Superconductivity: Theory of Superconductivity

was in the limit R > λ and the dot–dot interaction would be expected to
correspond to the antiferromagnetic alignment shown in the second state
in Fig. 4.15. With increase in the temperature, the transition to the ferro-
magnetic alignment would occur according to Eq. (4.118) at λ = 0.36R, i.e.
219 nm.
−1/2
According to the Casimir formula λ(T ) = λ(0)(1 − t4 ) , with t =
T /Tc , this would occur at T = 7.14 K, compared with Tc = 7.2 K. Therefore,
the experimental conditions for the transition to be observed are certainly
feasible. Of course, for an exact comparison with theory in this case, our
theory should be generalized to deal with magnetic particles near the surface
rather than with those embedded in the bulk of a superconductor.
Of course, it would be interesting in the future to generalize our results
to superconductors in the Pippard limit, where the RKKY interactions
between quantum dots will dominate over dipolar forces.203
The expression (4.110) can be used to study, by means of numerical
methods, the magnet configurations and the orientational phase transitions
in an ensemble of nanogranules. It is possible to determine the conditions
of orientational transformations in the analytic form for the ordered struc-
tures (a chain of granules, plane and volume lattices). Inasmuch as the state
of the magnetic subsystem of a specimen at phase transitions is changed,
this phenomenon can be experimentally observed under the change in the
magnetic susceptibility in the region of low fields.

4.6 Spin-Orientation Phase Transitions

in a Two-Dimensional Lattice of Ferromagnetic
Granules in a London-Type Superconductor
In order to determine the collective state of magnetic moments in nanocom-
posite materials with the matrix made of a London-type superconductor,
we use the formula (4.103) for the energy of magnetic interaction.
First of all, we note that the realization of one or other magnetic con-
figuration is defined by both the competition of diamagnetic effects from
the side of the superconducting matrix and the magnetostatic interaction
in the system of ferromagnetic granules. At lower temperatures eliminat-
ing the thermal disordering of the system, the magnetostatic interaction
leads to a correlation of magnetic moments. As the main conditions for the
formation of magnetic configurations, we take the equivalence of all sites
and the zero value of the net magnetic moment of the lattice. The planar
 Mesoscopic Superconductivity 175

Fig. 4.16 Two-dimensional lattice of magnetic points. Magnetic moments (black) are
directed up or down (white); a is the lattice constant.

lattice of granules by itself sets a preferred direction in space. Therefore,

we will separate two configurations from the whole manifold of spatial
orientations of magnetic moments. The first configuration is presented in
Fig. 4.16. It is characterized by the orientation of the magnetization of
granules which is orthogonal to the base plane. The alternation of the
magnetization of neighboring granules decreases, to a certain extent, the
energy of magnetic interaction. In addition, such a distribution of magnetic
moments favors a decrease in the amplitude of a magnetic field in the super-
conducting matrix, which is also advantageous from the energy viewpoint.
Thus, the given configuration can be considered as a version of the mag-
netic order.
A configuration of the second type is shown in Fig. 4.17. It is charac-
terized by the distribution of magnetic moments in the base plane of the
lattice such that the magnetic moments are aligned as magnetic chains
with alternating directions of the magnetization. Here, like the config-
uration presented in Fig. 4.16, the main requirement, i.e. the equiva-
lence of the states of magnetic points, is satisfied. A similar distribution
also decreases the energy of magnetostatic interaction, and favors a
decrease in the amplitude of a magnetic field in a superconductor. At
the same time, the planar orientation of magnetic moments has the
basic distinction from the orthogonal one. For example, by means of
176 Modern Aspects of Superconductivity: Theory of Superconductivity

Fig. 4.17 Two-dimensional lattice of magnetic points. Magnetic moments are aligned
in the plane in the form of chains; a is the lattice constant.

a continuous deformation of the magnetization in the base plane, the

configuration in Fig. 4.16 can be transferred into the structure shown in
Fig. 4.17.
Such a system is characterized by a coherent rotation of magnetic
moments by an angle ±ϕ relative to the principal direction of the lat-
tice. In this case, there occur both the modulation of the direction of
moments at sites of the lattice and some increase in the energy of mag-
netic chains, but these processes are accompanied by the formation of mag-
netic vortices in cells, which promotes a decrease in the energy of magnetic
interaction. The states of separate magnetic points in the lattice remain
equivalent at the zero total magnetization. Thus, the following questions
arise: How does the energy of the array of magnetic moments in the base
plane shown in Fig. 4.18 depend on the angle ±ϕ, and to which value is
it equal in the equilibrium state? To answer these questions, we consider
the relation (4.110) for the interaction energy and reduce it to a single sum
by virtue of the fact that the states of magnetic points are equivalent. In
this case, in order to calculate the energy of the lattice, it is sufficient to
determine the energy of a single magnetic point, e.g. m0,0 , located at the
origin of the coordinate system, and then multiply the result by the total
number of magnetic points N . The relation (4.110) becomes significantly
 Mesoscopic Superconductivity 177

Fig. 4.18 Part of the lattice with a modulated planar distribution of the magnetization; a
is the lattice constant. Circles represent magnetic points, and the arrows on them indicate
the directions of magnetic moments mn k in the base plane (n and k are the spatial indices
of magnetic points). The angle ϕ defines the deviation of the moments of magnetic points
from the principal direction of the lattice. The states of all points in the given configuration
are equivalent. The net magnetic moment is zero. The separated circles schematically
denote magnetic vortices. The dotted lines are tangents to the directions of magnetic
moments at sites of the lattice.

simpler:
    N  α · r  (r · m )(r · m )
μ0 N ∂ ∂2  i i i i 0,0
U =− 3 1− + 2 exp −
4π 2 ∂α ∂α δ 5
ri
i
   α · r m · m
∂2 
N
N ∂ i i 0,0
+ 1− + 2 exp − . (4.119)
2 ∂α ∂α δ ri3
i
By writing the formula (4.119), we used the method of differentiation with
respect to the parameter α, which should be set equal to 1 after the calcu-
lations. The index i stands for the summation over all sites of the lattice,
δ — the penetration depth for 2D case. Performing the summation in the
relation (4.119), it is convenient to introduce the pair of indices (n, k) defin-
ing the position of a site in the lattice (Fig. 4.18) instead of the running
index of magnetic points i. It is easy to see that the system represented in
Fig. 4.18 possesses the translational invariance with a period of 2a so that
mn,k = mn+2l, k+2p ,
l, p = ±1, ±2, . . . . (4.120)
178 Modern Aspects of Superconductivity: Theory of Superconductivity

The lattice has only four types of magnetic points differ which from
one another by a spatial orientation of magnetic moments. Their vector
components depend on the angle ϕ in the following manner:
⎛ ⎞
cos ϕ
mn,k = mn+2l,k+2p , m2l,2p = m0,0 = m⎝sin ϕ ⎠,
0
⎛ ⎞
cos ϕ
m2l+1,2p = m1,0 = m⎝− sin ϕ⎠,
0
⎛ ⎞
− cos ϕ
m2l,2p+1 = m0,1 = m⎝ sin ϕ ⎠,
0
⎛ ⎞
− cos ϕ
m2l+1,2p+1 = m1,1 = m ⎝ − sin ϕ ⎠,
0
l, p = ±1, ±2, . . . ,
l, p = ±1, ±2, . . . , (4.121)
where m is the modulus of the magnetic moment of a site.
After the substitution of Eq. (4.121) in Eq. (4.119) and the summation
over sites of the unbounded lattice, we get the following interesting result.
It turned out that the interaction energy of the system of magnetic points
(see Fig. 4.18) does not depend on the angle ϕ and is determined by the
relation
U μ0 m2
= · F (a/δ), (4.122)
N 4π a3
where N is the number of sites of the lattice, m is the magnetic moment of
a granule and F (a/δ) is the energy characteristic of a magnetic state which
is a universal function of a single parameter and determines the dependence
of the energy on both the period and the field penetration depth in the case
where the magnetic moments are distributed in the base plane of the lattice.
It can be represented in the form of a sum:
  ∞ ∞
μ0 ∂ ∂2 1   exp(−2α(a/δ) l2 + p2 )
F(a/δ) = −1 + + ·
4π ∂α ∂α2 8 (l2 + p2 )3/2
l=1 p=0
  ∞ ∞
∂ ∂2 1  
− −1 + + 2 ·
∂α ∂α 4
l=−∞ p=−∞
 Mesoscopic Superconductivity 179

exp(−α(a/δ) (2l + 1)2 + (2p + 1)2 )

×
((2l + 1)2 + (2p + 1)2 )3/2
  ∞ ∞
∂ ∂2 1  
− 3−3 + ·
∂α ∂α2 2 p=−∞ l=−∞

((2l + 1)2 − (2p)2 ) ·exp(−α(a/δ) (2l + 1)2 + (2p)2 )

× .
((2l + 1)2 + (2p)2 )5/2
(4.123)
Thus, there occurs the degeneration of the state in the parameter ϕ in the
presence of a tough correlation of the mutual orientations of moments of
the ensemble of magnetic points. The energies of the configuration shown
in Fig. 4.17(1b) and the ensemble with a modulated distribution of the
magnetization (see Fig. 4.18) coincide. In turn, the determination of the
energy of magnetic interaction for the configuration possessing the orthogo-
nal orientation of magnetic moments (Fig. 4.16) requires a smaller number
of calculations, because the first sum in the formula (4.119) vanishes. The
result of calculations can be represented in the form
U⊥ μ0 m2
= · Φ (a/δ), (4.124)
N 4π a3
where Φ(a/δ) is the energy characteristic of the magnetic state which is a
universal function of the single parameter and determines the dependence
of the energy on both the period and the field penetration depth under the
distribution of magnetic moments normally to the base plane of the lattice.
This function can be represented in the form
 
μ0 ∂ ∂2 1
Φ(a/δ) = 1− + ·
4π ∂α ∂α2 2
⎧  
⎪
⎨ ∞  ∞ 4 · exp −α · (a/δ) · 2 · l 2 + 2 · p2 ∞ ∞
×  3/2 −
⎪
⎩ l=1 p=0 2 · l 2 + 2 · p2 l=−∞ p=−∞
 + ⎫
exp −α · (a/δ) · 2 · (l + 1/2)2 + 2 · (p + 1/2)2 ⎪ ⎪
⎬
×  3/2 .
⎪
⎪
2
2 · (l + 1/2) + 2 · (p + 1/2) 2
⎭

(4.125)
The calculation of the functions F (a/δ) and Φ(a/δ) on the basis of the
relations (4.123) and (4.125) is not a difficult task and can be realized with
180 Modern Aspects of Superconductivity: Theory of Superconductivity

any mathematical software. The results are presented in graphical form in

Fig. 4.19.

Fig. 4.19 Plots of the energy characteristics F (a/δ) and Φ(a/δ) of states of the lattice
with normal and planar orientations of magnetic moments, respectively. Values of F (0)
and Φ(0) correspond to the transition of the superconducting matrix to the normal state.

In Fig. 4.19, we represent the plots of the energy characteristics of two

different states of a magnetic lattice versus the ratio of the parameter of a
cell and the penetration depth of the magnetic field, a/δ. The limit a/δ → 0
corresponds to the transition of the matrix to the normal state. It is obvious
that a lattice with planar orientation of magnetic moments (Figs. 4.17 and
4.18) possesses the lower energy in the normal state at a/δ = 0. Therefore,
the state with the perpendicular direction of moments (Fig. 4.16) cannot
be realized at all in the absence of a superconductor. As the temperature
decreases and the penetration depth diminishes gradually, the parameter a/δ
begins to grow. When this parameter attains the value a0 /δ ≈ 3.3, the con-
figuration with the orthogonal orientation of magnetic moments (Fig. 4.16)
becomes more advantageous in energy, and the orientation phase transition
occurs in the system. Upon at a decrease in the temperature, a similar sce-
nario of events completely corresponds to a reorientation of the magnetic
moments of an isolated pair of magnetic points, which was considered in the
previous work.205
In conclusion, we note that an analogous phase transition can be expected
to occur in a planar lattice with a rectangular cell. The only difference
 Mesoscopic Superconductivity 181

will consist in the elimination of the degeneration relative to the direc-

tions of magnetic moments in the base plane. Of course, the direct observa-
tion of a phase transition will be hampered, because the problem involves
the magnetic lattice surrounded by a superconductor. However, a similar
orientation transformation must happen in a lattice applied on the surface
of a massive superconductor, though values of the parameter a0 /δ will be
different in this case.

4.7 Quantum Computer on Superconducting Qubits

4.7.1 Principle of quantum computers
Silicon microprocessors, being the main element of modern computers, have
reached the limit of development. The miniaturization, i.e. the aspiration
to place as many components as possible on a smaller and smaller area of
a chip, has approached the boundary of physical possibilities. Further, it
will be impossible to conserve the stability of the operation of computers.
Many researchers believe that silicon processors will begin to go into the
past in at most five years, and the production of chips will be based on
the other material — carbon nanotubes. It is worth noting that the com-
putational processes are accompanied by the release of heat. Feynman said:
“Any classical computation is a physical process running with the release
of heat.” As is known, the calculations lead to an increase in the entropy
and, hence, to the release of heat. The idea of the creation of quantum
computers arose several decades ago, when it was proposed to reject the
application of electric circuits in the processing of information and to pass
to the use of quantum mechanics. Classical computers are processing the
information only on the basis of ideas of one bit of information (it corre-
sponds to the transition from state 0 to state 1 or conversely). Quantum
computers can process the information by basing on ideas of a quantum bit
(qubit), which allows one to realize simultaneously four logical operations
(0 + 0 = 0; 0 + 1 = 1; 1 + 0 = 1; 1 + 1 = 2).
The qubit is an abridged notation for the quantum bit and represents
the unity of information coded in a quantum system which can be in the
states |0 and |1, and in any superposition of these states.
Let the state of a qubit be described by the state |f , which can be
represented as a superposition of the states |0 and |1:

|f  = a|0 + b|1,
182 Modern Aspects of Superconductivity: Theory of Superconductivity

where
a2 + b2 = 1. (4.126)
Since the devices can register only classical quantities, the measurement of
a quantum bit will give the states |0 or |1 with probabilities determined
by the squares of the coefficients a and b.
The use of the principle of superposition allows one to increase the infor-
mational space exponentially with linear growth of the size of a physical
system, because the register including n qubits can be in a superposition of
at once 2n states. In addition, quantum mechanics admits the existence of
the so-called entangled states possessing no analogs in classical physics. An
ensemble of qubits is a collection of qubits in different but given states.
It is worth noting that, even on the level of mathematics, a quantum
computer operates in a basically different way than a classical computer.
Input data are coded in “quantum cells of memory.” In this case, the col-
lection of qubits becomes a single quantum system. This system undergoes
the sequence of elementary quantum operations. Quantum computations
are a realization of the most astonishing idea of applying the principles of
quantum mechanics to the world of computers. Ordinary computers, despite
their complexity, use classical laws of mechanics. In recent years, the the-
ory of classical computations has been developed on the basis of works
by A. Turing. With the appearance of quantum computations, new possi-
bilities have arisen, and the situation has been radically changed. Quantum
methods can be successfully used in the solution of mathematical prob-
lems, though the time consumed for the solution of mathematical problems
increases exponentially with the complexity of the problem.
For the theory of quantum calculations, the physical nature of qubits is
not of crucial importance; the basically important point is that the system
in the course of calculations obeys the laws of quantum mechanics.

4.7.2 Superconducting qubits

The realization of a quantum computer requires the availability of systems
with a doubly degenerate ground state. For this reason, great attention is
paid to systems with two-level wells which can be fabricated by facilities of
solid-state electronics. The modern technology allows one to produce circuits
containing millions of transistors and Josephson junctions.
By observing the operation of electrical circuits at temperatures close
to absolute zero, researchers found a new proof of the fact that the laws
of quantum mechanics are suitable not only for the microscopic objects
 Mesoscopic Superconductivity 183

(atoms and electrons) but also for large electronic schemes, which include
superconducting bits (qubits).
Many years ago, attempts to construct a Josephson computer on the
basis of the Josephson tunnel logic failed hopelessly. The principal rea-
son for the failure was the huge technological dispersion of parameters of
tunnel junctions, which did not allow one to produce large microcircuits.
A Josephson computer can be created only on the way, having nothing in
common with that based on semiconductors; namely, it can be just a quan-
tum computer.
The hope is related to two circumstances. First, the fabrication of super-
conducting qubits is quite possible in the framework of up-to-date tech-
nology. Second, the presence of a gap in the spectrum of excitations of
a superconductor allows one to expect the suppression of generation in a
system.
Let us consider the first steps on the way toward construction of a super-
conducting quantum computer. First, we consider, in brief, the Josephson
effect.

4.7.3 Josephson effect

The Josephson effect is certainly one of the most interesting phenomena in
superconductivity. It is the passing of a superconducting current through a
thin dielectric layer separating two superconductors (the so-called Josephson
junction).
It was predicted by an English physicist, B. Josephson, on the basis
of superconductivity theory (in 1962; Nobel awarded the Prize in 1973),
and discovered experimentally in 1963. Conduction electrons pass through
a dielectric (a film of copper oxide ∼10 Å in thickness) due to the tunneling
effect.209, 210
In Fig. 4.20, we present a scheme of the Josephson tunneling between two
superconductors. The Josephson stationary effect considers the fact that the

Fig. 4.20 Scheme of the Josephson tunneling between two superconductors.

184 Modern Aspects of Superconductivity: Theory of Superconductivity

superconducting current
Jc = J0 sin ϕ, (4.127)
dϕ 2eV
= , (4.128)
dt 
where ϕ is the phase difference on the interface of superconductors, V the
applied voltage, and J0 the critical current through the junction.
The Josephson effect indicates the existence of the electron ordering in
superconductors, namely the phase coherence: in the ground state, all elec-
tron pairs have the same phase ϕ which characterizes their wave function
√
Ψ1 = ns1 eiϕ1 . According to quantum mechanics, the presence of a phase
difference must cause a current through the junction. The discovery of such
a current in experiment proves the existence of macroscopic phenomena in
nature which are directly determined by the phase of a wave function:
√
Ψ = neiψ . (4.129)

4.7.3.1 Current passing through two series-connected Josephson junctions

The Josephson current in the scheme drawn in Fig. 4.21 can be easily deter-
mined with the help of the elementary Feynman approach to a Josephson
junction as a two-level quantum-mechanical system.4 By introducing an
intermediate object, some island, with the wave function Ψ0 into the two-
level system characterized by the wave functions Ψ1 = ns1 /2eiϕ1 and
Ψ2 = ns2 /2eiϕ2 , we write the Schrödinger equation in the form
dΨ1 eV
i = Ψ1 + KΨ0 , (4.130)
dt 2
dΨ0
i = KΨ1 + KΨ2 + E0 Ψ0 , (4.131)
dt
dΨ2 eV
i = KΨ0 − Ψ2 . (4.132)
dt 2
The formula for a constant current running through two series-connected
Josephson junctions at the zero external potential difference takes the form
√
∂ nS1 K2 √
J = =− nS2 sin(ϕ2 − ϕ1 ). (4.133)
∂t E0

Fig. 4.21 Current passing through two series-connected Josephson junctions.

 Mesoscopic Superconductivity 185

This relation can be rewritten as follows:

Jc = J0 sin(ϕ), (4.134)
where ϕ = ϕ2 − ϕ1 is the phase difference. The relation obtained is called
the Josephson formula (the Josephson stationary effect) and determines the
current of superconductive electron pairs due to the tunneling transition.
The Josephson coupling energy is an important parameter of the
Josephson junction. From (4.127), we have
 π
hJ0 hJ0
E= JV dt = sin ϕ dϕ = − cos ϕ. (4.135)
2 2e
0

The Josephson effect is still one of the phenomena that make superconduc-
tors such a fascinating area of study. Despite more than 40 years of intensive
studies and numerous applications, it remains an important field of research
in connection with the use of small superconducting grains.

4.7.3.2 SQUIDs

The Josephson effect allowed one to construct superconducting interferome-

ters, called SQUIDs (superconducting quantum interference devices), which
contain parallel weak connections between superconductors.206 In Fig. 4.22,
we present a scheme with Josephson junctions. The total current running
from 1 to 2 is equal to
I = I0 d[sin(Δϕ1 ) + sin(Δϕ2 )], (4.136)

Fig. 4.22 Basic diagram of a quantum interferometer.

186 Modern Aspects of Superconductivity: Theory of Superconductivity

where
Δϕ1 = ϕ2A − ϕ1A ,
Δϕ2 = ϕ2B − ϕ1B
are the phase differences on the first and second Josephson junctions. There
occurs a distinctive interference of the superconducting currents running
through these connections.
Inside the superconductor, the current is zero: j ≡ 0. We will use the
following formula for the current j:
2e
j ∼ h∇ϕ − A. (4.137)
c
We can write 
2e
ϕ1B − ϕ1A = A · dl, (4.138)
hc C1

2e
ϕ2B − ϕ2A = A · dl. (4.139)
hc C2
Summing up these two equations, we get 7
2e Φ
ϕ1B − ϕ2B + ϕ2A − ϕ1A = A · dl = 2π . (4.140)
hc Φ0
Thus, we have
Φ
Δϕ1 − Δϕ2 = 2π , (4.141)
Φ0
where Φ is the total quantum flux.
The flux quantum is defined as
h
Φ0 = . (4.142)
2e
For a balance SQUID ring system, we can write
Φ
Δϕ1 = ϕ0 + π ,
Φ0
Φ
Δϕ2 = ϕ0 − π . (4.143)
Φ0
The total current in the SQUID is
I = I0 sin(Δϕ1 ) + I0 sin(Δϕ2 )
   
Φ Φ
= I0 sin ϕ0 + π + I0 sin ϕ0 − π
Φ0 Φ0
    
Φ  Φ 
= 2I0 sin(ϕ0 ) cos π 
= Imax cos π , (4.144)
Φ0 Φ0 
with Imax = 2I0 sin(ϕ0 ).
In this case, the critical current turns out to be periodically dependent
on the flow of an external magnetic field, which allows one to use such a
unit for exact measurement of the magnetic field.
 Mesoscopic Superconductivity 187

4.7.3.3 Flux qubit

Let us consider the first steps on the way towards creation of a super-
conducting computer. The simplest superconducting system demonstrating
the coherence is the SQUID, which is a superconducting ring including a
Josephson junction at one point. The energy of this system contains two
terms, the Josephson transition energy (cos Φ) and the energy related to
the ring L:
 
Φ (Φ − Φx )2
H = −EJ cos 2π + . (4.145)
Φ0 2L
Here, Φ is the difference of superconductive phases at the junction. The
superconductive phase in the ring is proportional to a magnetic flow applied
to the ring (quantization of the magnetic flow). If Φx is equal exactly to a
half of the magnetic flow, the potential of a SQUID becomes doubly degen-
erate.
Two minima of the well correspond to the currents in the ring passing
in the clockwise and counterclockwise directions, respectively.208
A superposition of these states in SQUIDs, was observed experimentally
in Refs. 208, 211. These experiments demonstrated clearly the possibility of
creating a superposition of states in a system with a macroscopic number of
particles. In the given case, the circular current including 1013 electrons was
registered in a loop. The states participating in a superposition were macro-
scopically distinguishable, by differing from one another by the currents,
whose difference was several microamperes. Recently, an important notion
has been introduced in the course of studies of structures with Josephson
junctions. It is the notion of macroscopic quantum coherence. In such sys-
tems, the Josephson energy can have two almost-degenerate minima at val-
ues of the phase which are separated by a potential barrier (see Fig. 4.23).
It is possible that the phase passes from one minimum to another due to
the quantum-mechanical tunneling, and the eigenstates of the system are
superpositions of the states localized in the first and second minima.

Fig. 4.23 Scheme of the Josephson tunneling between two superconductors.

188 Modern Aspects of Superconductivity: Theory of Superconductivity

The operation of a superconducting computer requires low tempera-

tures, which are needed, in particular, to suppress heat-induced excitations
destroying the quantum-mechanical state of a system.
It is worth noting that the best condition for the observation of these
current states is defined by the size of the superconducting ring. If the
ring size is taken to be 1 cm, there appear the effects of decoherence which
will destroy the current states. But if the ring is taken to be much smaller
(say, 5 μm), then it is possible to observe these states. These states were
discovered in experiments with Rabi oscillations.211

4.7.3.4 Charge qubit

The second type of superconducting bit can be realized in the “Cooper-

pair box” system, which is characterized by two charge states: without an
excess Cooper pair |0 and with a single Cooper pair |1. A Cooper-pair
box is a nanotransistor that has Coulomb blockage with controlling voltage
Vg , as shown in Fig. 4.24. It is represented by an aluminum superconduct-
ing film of the order of 1 mm in size, with a working temperature of about
several millikelvins, which is significantly lower than the superconducting
temperature Tc . Its quantum state can be characterized by the number of
Cooper pairs. According to the BCS theory, the ground state of a super-
conductor is a superposition of states with different numbers of Cooper
pairs. The excited — i.e. unpaired — states in an ordinary superconductor
are separated from the ground state by the energy gap. Therefore, as the
number of electrons varies by 1, i.e. N → N ± 1, the ground-state energy
must be changed by ±δ. The sign (plus or minus) depends on wherever the
initial number of electrons N is even or odd. The effects of parity of the
number of electrons which are considered in this chapter in connection with

Fig. 4.24 Cooper-pair box [× denotes a Josephson junction (JJ)]. Cg and Vg are the
capacitor and the controlling voltage.
 Mesoscopic Superconductivity 189

mesoscopic superconductivity were successfully measured on nanotransistors

with Coulomb blockage.
The Hamiltonian of such a system can be written as
H = Ec (n − ng )2 + Ej cos ϕ, (4.146)
where Ec and Ej are the charging and Josephson energies, and ϕ is the phase
change. In the charging mode where Ec  Ej , only two lower-charged states
are of importance. The controlling voltage Vg induces a charge in the box:
Vg
ng = Cg , (4.147)
2e
where 2e is the charge of each Cooper pair and Cg is the gate capacitance. At
ng = 1/2, such a system operates as a two-level atomic system, in which the
states |0 and |1 can be realized. The control is realized by a voltage Vg .
On the basis of such a qubit, a system of two qubits was realized,208, 209
and the formation of entangled states was demonstrated. The effects of
parity of the number of electrons which are considered in this chapter in
connection with mesoscopic superconductivity were successfully measured
on nanotransistors with Coulomb blockage.

4.7.3.5 Phase qubit

There exists one more possibility of realizing SQUIDs with the use of HTSCs
possessing lattice, d pairing (π-loop ones).210
The physics of d pairing was considered in Chap. 2. Quantum processors
on the base of these SQUIDs are being developed at the Canadian company
D-wave.
We note also that, in addition to superconducting qubits, quantum com-
puters use qubits possessing other physical properties. Scientists at Yale
University used a very fine aluminum plate in the fabrication of a quantum
chip. A single qubit consists of one billion aluminum atoms, which never-
theless behaves as a single unit that can be in two energy states, denoted as
0 and 1. Such quantum-mechanical states of a qubit cannot be long-term —
their lifetime is about one microsecond. But it is sufficient for a chip to
solve the so-called algorithm. We have considered the technologies of super-
conducting computers which represent a new type of quantum computers.
These computers are based on the other mechanism obeying the laws of
quantum mechanics. We recall that, till recently, the principle of devices
was invariable, and the archetype of such devices is mechanical clocks. In
such devices, all stages of their relative motion can be observed; therefore, it
is quite simple to understand their structure. On the contrary, the principle
190 Modern Aspects of Superconductivity: Theory of Superconductivity

of operation of quantum computers involves the specific features of quantum

mechanics which are difficult to understand. Nevertheless, by possessing the
quantum resources, we can solve the very difficult complicated problems. In
particular, the most important potential field of application of quantum
computers is the problem of the exact calculation of properties of quantum
systems.
 Summary and Conclusions

In this book we have concentrated our attention on the most urgent prob-
lems of superconductivity, such as the nature of high-temperature supercon-
ductivity, mechanisms and symmetry of pairing, two-gap superconductivity
in magnesium diborades, mesoscopic superconductivity and the problems of
room-temperature superconductivity.
We considered some questions concerning the application of quantum
field theory to the problems of superconductivity. Quantum field theory pro-
vides the original and powerful means for the solution of certain problems of
superconductivity. In the field of superconductivity, we meet the problem-
maximum — it consists the creation of room-temperature superconductors.
We considered this problem in our book, gave some recommendations on the
search for these superconductors and analyzed the possibility of the fabri-
cation of artificial materials possessing the property of superconductivity at
room temperature.
We also touched on the questions of the application of superconducting
qubits to quantum computers. It was shown that the description of super-
conducting qubits is based on the laws of quantum mechanics.
It is worth noting that HTSCs have been investigated for more than
two decades with great efforts but the whole pattern of the phenomenon
is not yet available. We are sure that the comprehension of HTSCs will be
attained when our knowledge of HTSCs reaches the critical level which will
be sufficient for understanding the huge amount of experimental data from
a single viewpoint. The study is in progress. In particular, we mention the

191
192 Modern Aspects of Superconductivity: Theory of Superconductivity

recent discovery of Cu-less superconductors with a high critical temperature

which contain layers of FeAs.
The theoretical and experimental investigations into superconductivity
and its practical applications at present constitute this wide, uncommon
and interesting branch of science.
 APPENDIX A

Two-particle Green Function

for Multiband Superconductors

In this appendix, we will solve the problem of determining a two-particle

Green function for multiband superconductors. In Chap. 3 we considered
multiband superconductivity using a two-particle Green function.
We will calculate a two-particle Green function for multi-band supercon-
ductors. The Hamiltonian describing the systems of interacting electrons
and phonons of a crystal is written in the form
 1 
1 υ2 υ2 υ1
H= ευk a+υ υ
kσ akσ + Vq a+υ
k1 σ1 ak2 σ2 ak2 +q,σ2 ak1 −q,σ1
N
k,a,υ q,k1 ,σ1 ,υ1 ,k2 ,σ2 ,υ2

1  
υ
+√ χsq a+υ s
k,σ ak−q,σ Qq + Ωqs b+s s
q bq , (A.1)
N k,σ,υ,q,s q,s

where a+vkσ are creation operators for the electron with the momentum k in
band υ, avkσ are annihilation operators for the electron with the momentum
k in band υ, b+s q and bsq are creation and annihilation operators for the
phonons with the momentum q, εvk is the energy of the electron with the
impulse k, Ωqs is the energy of the phonons, Qsq = b+s s
q + bq , s is the number
of phonon branches, V −q = Vq , Xq∗s = Xqs are the Fourier components
of the Coulomb interactions of the electron and their coupling constant
with the lattice phonons, respectively. We can see that both constants are
independent of the spin index of an electron. Let us introduce new operators
for electron and phonon systems by the rules

avk = eS Ak,v e−S , bsq = eS βs,q e−S , (A.2)

where S is the anti-Hermitian operator (S + = −S).

193
194 Modern Aspects of Superconductivity: Theory of Superconductivity

In the following unitary transformation it will be more convenient to

rewrite the term He−p in the form

+v μ
He−p = χv,μ
s,q ak ak−q Qq (χs,q = χq δv,μ )
s v,μ s
(A.3)
k,q,s,v,μ

where we have united two indices and so υ = (υ, σ) and μ = (μ, σ ) are the
complex indices which characterize the number of crystals in the electron
zone and the spin of the electron. The Hamiltonian describing the system
of interacting electrons and phonons of the crystal after transformation by
a unitary operator is written in the form98–102
  2 
 1  χsq 
H= εk −
v
A+k,v Ak,v
N s,q Ωs,q
k,v
   
1   χsq 2
+ Vq − 2 A+ +
k,v Ak  ,v Ak  −q,v Ak−q,v
2N Ωs,q
q,k,v,k  ,v s

+ Ωs,q β+
s,q βs,q + higher-order term. (A.4)
s,q

The unitary transformation gives rise to renormalization of the electron

energy (first term) and renormalization of the Fourier component of the
Coulomb electron–electron interaction. To calculate the density of electron
states we have to study the Green function for the case with the approx-
imation t → t − 0. The two-particle Green function can be written as
follows:
  
k2 , v2 ; k, v 
G2 t−t 
k + q, v; k2 − q, v2
8 9

= −iT Ak+q,v (t)Ak2 −q,v2 (t)A+
k2 ,v2 (t − 0)A+
k,v (t ) (t → t − 0).
(A.5)

The equation for the Green function can be found from the equation of
motion:
3   4
∂8 
9
 ∂ 
i −iT Ak,v (t)Ak,v (t ) = δ(t − t ) + −iT
+ +
Ak,v (t) Ak,v (t ) ,
∂t ∂t
(A.6)
 Appendix A: Two-particle Green Function for Multiband Superconductors 195

where
∂Ak,v 8 9 8 9
 
i = [Ak,v · H] = −iT Ak,v (t)A+ k,v (t ) = −iT Ak,v (t)A+
k,v (t )
∂t
1 
= ε̃vk Ak,v + Ṽq A+
k2 ,v2 Ak2 +q,v2 Ak−q,v
N
q,k2 ,v2

1 
= ε̃k,v Ak,v − Ṽq Ak+q,v Ak2 −q,v2 A+
k2 ,v2 , (A.7)
N
q,k2 ,v2

with
 
1  χsq 2
ε̃vk = εvk − , (A.8)
N s,q Ωs,q
 
 1 
ε̃k,v = ε̃k +
v
Ṽq=0 − n Ṽq . (A.9)
v
N q
2

Therefore, the equation for the Green function (A.6) is changed, after we
insert (A.7):
∂ 8 
9
i −iT Ak,v (t)A+ k,v (t )
∂t
8 9
= δ(t − t ) + ε̃k,v −iT Ak,v (t)A+
k,v (t 
)

1  8 9

− Ṽq −iT Ak+q,v (t)Ak2 −q,v2 (t)A+
k2 ,v2 (t − 0)A+
k,v (t ) .
N
q,k2 ,v2
(A.10)
Such a two-particle Green function satisfies an equation of the Bethe–
Salpeter type (we do not split this function into two one-particle Green
functions of the Gorkov type). The solution of this equation according to
the Bogoliubov–Tyablikov method gives rise to the following expression for
the Fourier component of the two-particle Green function:
   .
k2 ,υ;k1 ,μ  f (k1 , μ; k2 , υ; ω) σ,σ ϕ(μ, υ; σ, σ )
G2 ω ∼ , (A.11)
k1 + q, μ; k2 − q, υ 1 − V K(k1 , μ; k2 , υ; ω)
with
μ υ
1  1 − nk1 +q − nk2 +q
K(k1 , μ; k2 , υ; ω) = , (A.12)
N q ω − εk1 +q,μ − εk2 +q,υ
196 Modern Aspects of Superconductivity: Theory of Superconductivity

where V is the Fourier component of the Coulomb interactions of the elec-

tron, ϕ(μ, υ; σ, σ ) is some function depending on the frequency ω, and the
momenta k1 and k2 of interacting electrons, f (k1 , μ; k2 , υ; ω) is expressed in
terms of functions, nμk1 +q and εk1 +q,μ are the number of filling and energy
of electrons respectively,
ϕ(μ, υ; σ, σ ) = δσσ δσ σ − δμυ δσσ δσσ , (A.13)
and σ and σ are the spins of the first (σ) and second (σ ) electrons.
The superconducting gaps are given by the zeros of the denominator of
(A.11).99, 102 Let us study the particular case where k1 = k2 = k0 + k and
εk0,μ = εμ corresponds to the extremum of the zone. Then, expanding the
energy by momentum k ± q in series up to terms of the second order, we
can obtain
(k ± q)2 (k ± q)2
εk±q,μ = εμ + = εf + Δμ + (μ = 1, 2, . . .), (A.14)
2mμ 2mμ
where mμ is the effective mass of electron in the μ energy band of the
crystal, Δμ is a parameter which indicates the position of the exremum of
the μ band relative to the Fermi level, and εk±q,μ is the energy of electrons.
The sum in the denominator (A.11) is reduced to the following expression
(we consider only one zone):
1  1 − nk+q − nk−q
K(k, k , ω) =
N q ω − εk+q − εk−q
 Δ
dε
= 2N (εf )
0 ω − 2(εf + Δ1 ) − 2E − 2ε
  
 Δ 
= 2N (εf ) − ln 1 −  , (A.15)
a
where such assignments have been used: ε = q 2 /2m; E = k2 /2m; a = ω −
2(εf +Δ1 )−2E; m∗1 = m1 /m and m are the reduced effective mass of electron
in the crystal energy zone and the mass of free electrons, respectively.
√ 
N (εf ) = 2π m∗1 m∗1 ε (1 − nk+q − nk−q ) εf , (A.16)
  −1
εk − εf
nk = exp +1 . (A.17)
T
The equation for the superconducting gap has the following form:
  
 Δ 
1 − V N (εf ) −ln 1 −  = 0. (A.18)
a
 Appendix A: Two-particle Green Function for Multiband Superconductors 197

Fig. A.1 The effect of temperature on the K(k, μ; k, υ; ω) function (E = k2 /2m = 0) for
different structures of the energy zones: (a) Δ1 = −1, m∗1 = 1; (b) Δ2 = 0.2, m∗2 = −2;
(c) Δ1 = −1, m∗1 = 1, Δ2 = 0.2, m∗2 = −2; curves 4, T = 2 K; curves 3, T = 10 K; curves
2, T = 50 K; curves 1, T = 100 K; arrows 5–7 correspond to different 1/V values (arrow
5, V = −0.07; arrow 6, V = −0.06; arrow 7, V = −0.05). All energy values are taken as
arbitrary values, i.e. V = V /M , where M is a scale factor (M = 1 eV, for convenience;
m∗i = mm
i
, where m is the free electron mass).

Fig. A.2 The temperature dependence of the superconducting gap. Dashed line —
theory; continuous line — BCS; a axis, b axis — experiment.

The results of some numerical calculations are given in Fig. A.1. We

will study theoretically the situation close to the Ba2 Sr2 CaCu2 O8 crys-
tal because it has a high Tc . The results of some numerical calculations
of the temperature dependence of superconductivity are given in Fig. A.2.
For illustration, the experimental dependences of the superconducting gap
obtained for the Ba2 Sr2 CaCu2 O8 crystal are presented in Fig. A.2 together
with theoretical curves of the BCS type and our calculation. The obtained
dependences are not similar to the standard BCS ones. It is clearly seen that
198 Modern Aspects of Superconductivity: Theory of Superconductivity

our calculation predicting the maximum in the T dependence of the super-

conducting gap gives the curve which is closer to the experimental results.
In our approach we take into account that all electron bands (the numerical
calculations show that only bands located near the Fermi are important)
contribute to the superconductivity.
 Bibliography

1. Parks, R. D. (ed.) (1969). Superconductivity (Marcel Dekker, New York).

2. Bednorz, J. G. and Müller, K. A. (1986). Possible high-Tc superconductivity in the
Ba–La–Cu–O system, Z. Phys. B 64, pp. 189–193.
3. Schrieffer, J. R. (1983). Theory of Superconductivty (Benjamin, New York).
4. Feynman, R. P. (1982). Statistical Mechanics (Addison–Wesley, Redwood City).
5. Anderson, P. W. (1987). The resonating valence bond state in LaCuO and super-
conductivity, Science 4, pp. 1196–1198.
6. Anderson, P. W. (1959). Theory of dirty superconductivity, J. Phys. Chem. Solids
11, pp. 26–30.
7. London, F. and London, H. (1935). The electromagnetic equations of the supracon-
ductor, Proc. R. Soc. London A 147, p. 71.
8. Ryder, L. H. (1996). Quantum Field Theory, 2nd edn. (Cambridge University Press).
9. Ginzburg, V. L. and Landau, L. D. (1950). On the theory of superconductivity, Zh.
Eksp. Theor. Fiz. 20, p. 11064.
10. Tinkam, M. (1975). Introduction to Superconductivity (McGraw-Hill, New York).
11. Nambu, Y. (1960). Quasi-particles and gauge invariance in the theory of supercon-
ductivity, Phys. Rev. 117, p. 648.
12. Takahashi, Y. (1976). Field Theory for Solid State Physics (in Japanese)
(Baifuhkan).
13. Nagao, H., Nishino, H., Shigeta, Y. and Yamaguchi, K. (2000). Theoretical studies
on anomalous phases of photodoped systems in two-band model, J. Chem. Phys.
113, pp. 11237–1244.
14. Kondo, J. (2002). Theory of multiband superconductivity, J. Phys. Soc. Jpn. 71,
p. 1353.
15. Fetter, L. A. and Walecka, J. D. (1987). Quantum Theory of Many-Particle Systems
(McGraw-Hill).
16. Abrikosov, A. A., Gorkov, L. P. and Dzyaloshinski, I. E. (1975). Methods of Quantum
Field Theory in Statistical Physics (Dover).
17. Aono, S. (2003). Molecular orbital theory: spinor representation, J. Phys. Soc. Jpn.
72, pp. 3097–3105.
18. Kimura, M., Kawabe, H., Nishikawa, K. and Aono, S. (1986). EPR observations of
trivalent titanium in orthophosphate single crystals, J. Chem. Phys. 85, pp. 1–6.
19. Salem, L. and Longuet-Higgins, H. C. (1959). The alternation of bond lengths in
long conjugated chain molecules, Proc. R. Soc. London A 255, p. 172.
20. Peierls, R. E. (1955). Quantum Theory of Solids (Oxford University Press).

203
204 Modern Aspects of Superconductivity: Theory of Superconductivity

21. Kimura, M., Kawabe, H., Nakajima, A., Nishikawa, K. and Aono, S. (1988).
Possibility of superconducting and other phases in organic high polymers of poly-
acene carbon skeletons, Bull. Chem. Soc. Jpn. 61, pp. 4239–4224.
22. Kubo, R. and Tomita, J. (1954). A general theory of magnetic resonance absorption,
Phys. Soc. Jpn. 9, pp. 888–919.
23. Doi, Y. and Aono, S. (1974). Theory of line shape in nuclear magnetic resonance,
Prog. Theor. Phys. 51, pp. 1019–1029.
24. Hebel, L. C. and Slichter, C. P. (1959). Nuclear spin relaxation in normal and super-
conducting aluminum, Phys. Rev. 113, pp. 1504–1519.
25. Fibich, M. (1965). Phonon effects on nuclear spin relaxation in superconductors,
Phys. Rev. Lett. 14, pp. 561–564.
26. Maki, K. (1967). Gapless Superconductors in Superconductivity, ed. Parks, R. D.
(Marcel Dekker, New York).
27. Uchino, K., Maeda, A. and Terasaki, I. (1959). Nature of the High Temperature
Superconductor (in Japanese) (Baifuukann).
28. Aono, S. (1998). Superconductivity as symmetry breaking classical treatment, Bull.
Chem. Soc. Jpn. 71, pp. 49–56.
29. Scalapino, D. J. (1995). The case for dx2 −y 2 pairing in the cuprate superconductors,
Phys. Rep. 250, pp. 329–360.
30. Gorkov, L. P. (1959). Microscopic derivation of the Ginzburg–Landau equations in
the theory of superconductivity, Sov. Phys. JETP 9, pp. 1364–1375.
31. Negele, J. W. and Orland, H. (1987). Quantum Many-Particle Systems (Addison-
Wesley, Reading, Masscchusetts).
32. Kawabe, H., Nishikawa, K. and Aono, S. (1994). Path integral approach to correla-
tion energy, Int. J. Quantum Chem. 51, pp. 265–283.
33. Feynman, R. P. (1948). Space–time approach to non-relativistic quantum mechanics,
Rev. Mod. Phys. 20, pp. 367–387.
34. Hubbard, J. (1959). Calculation of partition functions, Phys. Rev. Lett. 3, pp. 77–78.
35. Das, A. (2006). Field Theory: A Path Integral Approach (World Scientific,
Singapore).
36. DeWitt, B. (1984) Supermanifolds (Cambridge University Press).
37. Ferraz, A. and Nguyen, A. V. (1995). Supersymmetry and electron–hole excitations
in semiconductors, Phys. Rev. B 51, pp. 10548–10555.
38. Yang, C. N. (1962). Concept of off-diagonal long-range order and the quantum phases
of liquid He and of superconductors, Rev. Mod. Phys. 34, pp. 694–704.
39. Dunne, L. J., Murrell, J. N. and Brändas, E. J. (1997). Off-diagonal long-range
order, eta-pairing, and Friedel oscillations in high Tc cuprate superconductors and
the ground state of the extended Hubbard model, Int. J. Quantum Chem. 63,
pp. 675–684.
40. Ginzburg, V. L. and Kirzhnitz, D. A. (1982) (eds). High-Temperature Super-
conductivity (Consultants Bureau, New York).
41. Fröhlich H. (1954). On the theory of superconductivity, Proc. R. Soc. London A
223, p. 296–305.
42. Bogolyubov, N. N. (1958). A new method in the theory of superconductivity, Zh.
Eksp. Teor. Fiz. 34, pp. 58–65.
43. Josephson, B. D. (1974). The discovery of tunneling superconductivity, Rev. Mod.
Phys. 1074, pp. 251–254.
44. Maroushkine, A. (2004). Room-Temperature Superconductivity (Cambridge
International Science, Cambridge).
 Bibliography 205

45. Davydov, A. S. (1990). High-Temperature Superconductivity (Naukova Dumka,

Kiev).
46. Davydov, A. S. and Kruchinin, S. P. (1991). Interlayer effects in the newest high-T
superconductors, Phys. C 179, pp. 461–468.
47. Davydov, A. S. and Kruchinin, S. P. (1991). Dependence of the superconducting
transition temperature on the number of cuprate layers in a unit cell of high-
temperature superconductors, Sov. J. Low Temp. Phys. 17, 10, pp. 634–635.
48. Timusk, T. and Statt, B. (1999). The pseudogap in high-temperature superconduc-
tors: an experimental survey, Rep. Prog. Phys. 62, pp. 61–122.
49. Plakida, N. M. (1995). High-Tc Superconductivity: Experiment and Theory
(Springer), p. 306.
50. Brusov, P. (1999). Mechanisms of High Temperature Superconductivity (Rostov State
University Publishing), p. 685.
51. Abrikosov, A. A. (1998). Fundamentals of the Theory of Metals (North-Holland,
Amsterdam), p. 380.
52. Akhiezer, A. I. and Pomeranchuk, I. Ya. (1959). On the interaction between con-
ductivity electrons in ferromagnetics, Zh. Eksp. Teor. Fiz. 36, p. 819.
53. Akhiezer, A. I., Bar’yakhtar, V. G. and Peletminskii, S. V. (1968). Spin Waves
(North-Holland, Amsterdam), p. 369.
54. Little, W. A. (1964). Possibility of synthesizing on organic superconductivity, Phys.
Rev. 134, p. 1416.
55. Alexandrov, A. S. and Ranninger, J. (1981). Bipolaronic superconductivity, Phys.
Rev. 23, 4, pp. 1796–1801.
56. Pekar, S. I. (1951). Studies on the Electron Theory of Crystals (Gostekhizdat,
Moscow), p. 168.
57. Brusov, P. (1999). Mechanisms of High Temperature Superconductivity (Rostov State
University), p. 685.
58. Ausloos, M. and Kruchinin, S. (eds.) (1999). Symmetry and Pairing in
Superconductors: Proc. NATO ARW (Kluwer, Dordrecht), p. 410.
59. Annett, J. and Kruchinin, S. (eds.) (2002). New Trends in Superconductivity: Proc.
NATO ARW (Kluwer, Dordrecht), p. 435.
60. Kruchinin, S. (eds.) (2003). Modern Problems of Superconductivity: Proc. Yalta
Conference; Mod. Phys. Lett. B 17, 10–12, p. 393–724.
61. Sigrist, M. and Rice, T. M. (1995). Unusual paramagnetic phenomena in granular
high-temperature superconductors — a consequence of d-wave pairing, Rev. Mod.
Phys. 67(2) pp. 503–513.
62. Laughlin, R. B. (1998). Magnetic induction of dx2 +y 2 + idxy order in high-Tc super-
conductors, Phys. Rev. Lett. 80, p. 5188.
63. Scalapino, D. J. (1999). The case for d pairing in the cuprate superconductors, Phys.
Rep. 250, pp. 331–370.
64. Annet, J. F. (2003). Superconductivity, Superfluids and Condensates (University of
Bristol, Oxford University Press), p. 183.
65. Volovik, G. E. and Gorkov, L. P. (1985). Superconducting classes in heavy-fermion
systems, Sov. Phys. JETP 61, p. 843.
66. Mineev, V. P. and Samokhin, K. V. (1998). Introduction to the Theory of
Unconventional Superconductivity (Moscow Fiz.-Tekh. Inst. Press, Moscow), p. 141.
67. Van Harligen, D. J. (1995). Phase-sensitive test of the symmetry of pairing state
in the high-temperature superconductors — evidence for D symmetry, Rev. Mod.
Phys. 67, 2, pp. 515–535.
206 Modern Aspects of Superconductivity: Theory of Superconductivity

68. Tsuei, C. C. and Kirtley, J. R. (2000). Pairing symmetry in cuprate superconductors,

Rev. Mod. Phys. 72, 4, pp. 969–1015.
69. Pines, D. (1991). Spin fluctuations and high temperature superconductivity in the
antiferromagnetically correlated oxides: YBaCuO, LaSrCuO, Physica C 185–189,
pp. 120–129.
70. Monthoux, P., Balatsky A. V. and Pines, D. (1991). Toward a theory of high-
temperature superconductivity in the antiferromagnetically correlated cuprates
oxides, Phys. Rev. Lett. 67, 24, p. 3448.
71. Balatsky, A. V., Vekhter, I. V. and Zhu, J. X. (2008). Impurity-induced state in
conventional and unconventional superconductors, Rev. Mod. Phys. 78, pp. 373–433.
72. Izyumov, Yu. A. (1999). Spin-fluctuation mechanism in high-Tc superconductivity
and symmetry of order parameter, Usp. Phys. 169, 3, p. 215.
73. Morya, T. and Ueda, K. (2000). Spin fluctuations and high-temperature supercon-
ductivity, Adv. Phys. 49, 5, pp. 556–605.
74. Levi, B. G. (1993). In high-Tc superconductors, is d-wave the new wave? Phys. Today
5, pp. 17–20.
75. Manske, D. (2004). Theory of Unconventional Superconductors (Heidelberg,
Springer).
76. Maple, M. B. (1986). New type superconductivity in f -electron system, Phys. Today
4, p. 72.
77. Miyake, K. and Schmitt-Rink, S. (1986). Spin fluctuation mediated even party pair-
ing in heavy fermion superconductors, Phys. Rev. B 34, 9, pp. 6554–6556.
78. Kruchinin, S. P. (1995). Functional integral of antiferromagnetic spin fluctuations
in high-temperature superconductors, Mod. Phys. Lett. B 9, pp. 205–215.
79. Kruchinin, S. P. and Patapis, S. K. (1996). Thermodynamics of d-wave pairing in
cuprate superconductors, J. Low Temp. Phys. 105, pp. 717–721.
80. Kruchinin, S. P. and Patapis, S. K. (1997). Specific heat of antiferromagnetic spin
fluctuations in cuprate superconductors, Physica C 282–285, pp. 1397–1398.
81. Kruchinin, S. P. (2000). The pseudogap of antiferromagnetic spin fluctuations in
cuprate superconductors, Physica C 341–348, pp. 901–902.
82. Kruchinin, S. P. (2005). Condensation energy for spin fluctuation mechanism of pair-
ing in high-Tc superconductors, in Proc. NATO Advanced Research Workshop “New
Challenges in Superconductivity: Experimental Advances and Emerging Theories”,
ed. Ashkenazi, J. (Kluwer, Dordrecht), pp. 145–150.
83. Popov, Y. N. (1987). Functional Integrals and Collective Excitations (Cambridge
University Press), p. 216.
84. Conwall, J., Jackiw, R. and Tomhoulis, E. (1974). Effective action for composite
operators, Phys. Rev. D 10, 8, pp. 2428–2444.
85. Weinberg, S. (1993). Effective action and renormalization group flow of anisotropic
superconductors [preprint UTTG 18], pp. 1–17.
86. Shen, Z. X., Dessan, D. S., Wells, B. O., King, D. M., Spicer, W. E., Arko, A. J.,
Marshall, D., Lombarolo, L. W., Kapitulnik, A., Dickinson, P., Doniach, S., Carko,
J. Di, Loesser, A. G. and Park, C. H. (1993). Anomalously large gap anisotropy in
the a–b plane of Bi2 Sr2 CaCu2 O8+γ , Phys. Rev. Lett. 70, p. 1553.
87. Moler, M. B., Barr, D. I., Urbach, J. S., Hardy, W. H. and Kapitulnik, A. (1994).
Magnetic field dependence of the density of states of YBaCuO as determined from
the specific heat, Phys. Rev. Lett. 73, p. 2744.
88. Momono, N., Ido, M., Nakano, T. and Oda, M. (1994). Low-temperature electronic
specific heat in LaSrCuZnO, Physica C 235–240, pp. 1739–1740.
 Bibliography 207

89. Loram, J. W., Mirza, K. A., Cooper, J. R. and Tallon, J. L. (1998). Specific
heat evidence on the normal state pseudogap, J. Phys. Chem. Solids 59, 10–12,
pp. 2091–2094.
90. Revaz, B., Genond, J. Y., Junod, A., Naumdier, K., Erb, A. and Walker. E. (1998).
d-wave scaling relations in the mixed-state specific heat YBa2 Cu3 O7 , Phys. Rev.
Lett. 80, 15, pp. 3364–3367.
91. Hasselbach, K., Kirtley, J. R. and Flouquet, J. (1994). Symmetry of the gap in
superconducting UReSi, Phys. Rev. B 47, pp. 509–512.
92. Muhrer, G. and Schachinger, E. (2000). Free energy of a d-wave superconductor in
the nearly antiferromagnetic Fermi liquid model, J. Low Temp. Phys. 88–120, 1–2,
pp. 65–88.
93. Dorbolo, S., Ausloos, M. and Houssa, M. (1998). Electronic specific heat of super-
conductors with van Hove singularities: effects of a magnetic field and thermal fluc-
tuations, Phys. Rev. B 57, pp. 5401–5411.
94. Dobrosavljevic-Crujic, L. and Miranovic, P. (2003). Low temperature specific heat
in anisotropic superconductors, Physica C 397, pp. 117–122.
95. Bogolyubov, Jr., N. N. and Kruchinin, S. P. (2003). Modern approach to the calcu-
lation of the correlation function for superconductivity model, Mod. Phys. Lett. B
17, 10–12, pp. 709–724.
96. Nagao, H., Kruchinin, S. P., Yaremko, A. M. and Yamaguchi, K. (2002) Multiband
superconductivity, Int. J. Mod. Phys. B 16, pp. 3419–3428.
97. Kruchinin, S. P. and Nagao, H. (2005) Two-gap superconductivity in MgB, Phys.
Particles Nuclei 36, Suppl. 1, pp. 127–130.
98. Kruchinin, S. P. and Yaremko, A. M. (1998). Many zone effect in cuprate supercon-
ductors, Supercond. Sci. Technol. 11, pp. 4–8.
99. Nagao, H., Kawabe, H., Kruchinin, S. P., Manske, D. and Yamaguchi, K. (2003).
Theoretical studies on many band effects superconductivity by using renormalization
group approach, Mod. Phys. Lett. B 17, 10–12, pp. 423–431.
100. Nagao, H., Kawabe, H., Kruchinin, S. P. and Yamaguchi, K. (2003). Supercon-
ductivity in two-band model by renormalization group approach, Int. J. Mod. Phys.
B 17, pp. 3373–3376.
101. Yaremko, A. M., Mozdor, E. V. and Kruchinin, S. P. (1996). Coupled states of
electron–phonon system, superconductivity and HTSC of crystals, Int. J. Mod. Phys.
B 10, pp. 2665–2674.
102. Kruchinin, S. P. and Nagao, H. (2006). Multi-gap superconductivity in MgB2 , in
Proc. NATO Advanced Research Workshop “Symmetry and Heterogeneity in High-
Tc Superconductors”, ed. Bianconi, A. (Kluwer, Dorderecht), pp. 43–53.
103. Nagao, H., Yaremko, A., Kruchinin, S. P. and Yamaguchi, K. (2002). Many band
effects in superconductivity, in Proc. NATO Advanced Research Workshop “New
Trends in Superconductivity”, (eds.) Annett, J. and Kruchinin, S. P. (Kluwer,
Dordrecht), pp. 155–167.
104. Bardeen, J., Cooper, L. N. and Schrieffer, J. R. (1957). Microscopic theory of super-
conductivity, Phys. Rev. 108, pp. 1175.
105. Anderson, P. W. and Zou, Z. (1988). “Normal” tunneling and “normal” transport:
diagnostics for the resonating-valence-bond state, Phys. Rev. Lett. 60, p. 132.
106. Kampf, A. P. (1994). Magnetic correlations in high temperature superconductivity,
Phys. Rep. 249, p. 219.
107. Emery, V. J. (1987). Theory of high-Tc superconductivity in oxides, Phys. Rev. Lett.
58, p. 2794.
208 Modern Aspects of Superconductivity: Theory of Superconductivity

108. Hirsch, J. E. (1985). Attractive interaction and pairing in fermion systems with
strong on-site repulsion, Phys. Rev. Lett. 25, p. 1317.
109. Zhang, F. C. and Rice, T. M. (1988). Effective Hamiltonian for the superconducting
Cu oxides, Phys. Rev. B 37, p. 3759.
110. Fradkin, E. (1991). Field Theories of Condensed Matter Systems (Addison Wesley).
111. Nagaosa, N. and Lee, P. A. (1990). Normal-state properties of the uniform
resonating-valence-bond state, Phys. Rev. Lett. 64, p. 2450.
112. Baskaran, G., Zou, Z. and Anderson, P. W. (1987). The resonating valence bond
state and high-Tc superconductivity — a mean field theory, Solid State Commun.
63, p. 973.
113. Fukuyama, H. and Yoshida, K. (1987). Critical temperature of superconductivity
caused by strong correlation, Jpn. J. Appl. Phys. 26, pp. 371–373.
114. Yamamoto, S., Yamaguchi, K. and Nasu, K. (1990). Ab initio molecular-orbital study
on electron correlation effects in CuO6 clusters relating to high-Tc superconductivity,
Phys. Rev. B 42, pp. 266–272.
115. Nagao, H., Kitagawa, Y., Kawakami, T., Yoshimoto, T., Saito, H. and Yamaguchi, K.
(2001). Theoretical studies on field-induced superconductivity in molecular crystals,
Int. J. Quantum Chem. 85, pp. 608–618.
116. Kimura, M., Kawabe, H., Nakajima, A., Nishikawa, K. and Aono, S. (1988).
Possibility of superconducting and other phases in organic high polymers polyacene
carbon skeletons, Bull. Chem. Soc. Jpn. 61, p. 4239.
117. Torrance, J. B., Bezinge, A., Nazzal, A. I. and Parkin, S. S. (1989). Disappearance of
high temperature superconductivity induced by high carrier concentrations, Physica
C 162–164, p. 291.
118. Ord, T., Kristoffel, N. and Rago, K. (2003). MgB2 superconductivity properties in
a two-gap model, Mod. Phys. Lett. B 17, pp. 667–673.
119. Suhl H., Matthias, B. T. and Walker, R. (1959). Bardeen–Cooper–Schrieffer theory
of superconductivity in the case of overlapping bands, Phys. Rev. Lett. 3, p. 552.
120. Moskalenko, V. A. (1959). Superconductivity metals with overlapping energetic
bands, Fiz. Metalloved 8, p. 503.
121. Antipov, E. and Abakumov, A. (2008). Structural design of superconductors based
on complex copper oxides, Phys. Usp. 51, p. 180.
122. Müller, K. A. (2006). The search for new high temperature superconductors,
Supercond. Sci. Technol. 58, pp. 1–3.
123. Bennemann, K. H. and Ketterson, J. B. (2008). Superconductivity (Springer,
Heidelberg), vol. 1.
124. Nagamatsu, J., Nakamura, N., Muranaka, T., Zentani, Y. and Akimitsu, J. (2001).
Superconductivity at 39 K in magnezium deboride, Nature 410, p. 63.
125. Canfield, J. P. C., Bud’ko, S. L. and Finnemore, D. K. (2003). An overview of the
basic physical properties of M gB2 , Physica C 385, 1–2, pp. 1–3.
126. Binnig, G., Baratoff, A., Hoenig, H. E. and Bednorz, J. G. (1980). Two-band super-
conductivity in Nb-doped SrT iO3 , Phys. Rev. Lett. 45, p. 1352.
127. An, J. M. and Picket, W. E. (2001). Superconductivity of MgB2 : covalent bonds
driven metallic, Phys. Rev. Lett. 86, p. 4366.
128. Kortus, J., Mazin, I. I., Belashenko, K. D., Antropov, V. P. and Boyer, I. L. (2001).
Superconductivity of metallic boron in MgB2 , Phys. Rev. Lett. 86, p. 4656.
129. Kondo, J. (1963). Superconductivity in transition metals, Prog. Theor. Phys. 29,
p. 1.
130. Kristoffel, N., Ord, T. and Rago, K. (2003). MgB2 two-gap superconductivity with
intra- and interband couplings, Europhys. Lett. 61, pp. 109–115.
 Bibliography 209

131. Szabo, P., Samuely, P., Klein, T., Marcus, J., Fruchart, D., Miraglia, S., Marcenat C.
and Jansen, A. (2001). Evidence for two superconducting energy gaps in M gB2 by
point-contact spectroscopy, Phys. Rev. Lett. 87, p. 137005.
132. McMillan, W. L. (1968). Transition temperature of strong-coupled superconductors,
Phys. Rev. 167, p. 331.
133. Konsin, P., Kristoffel, N. and Örd, T. (1988). The interband interaction as a possible
cause of high-temperature superconductivity, Phys. Lett. A 129, pp. 339–342.
134. Combescot, R. and Leyronas, X. (1995). Coupling between planes and chains in
YBa2 Cu3 O7 : a possible solution for the order parameter controversy, Phys. Rev.
Lett. 75, pp. 3732–3735.
135. Konsin, P. and Sorkin, B. (1998). Electric field effects in high-Tc cuprates, Phys.
Rev. B 58, pp. 5795–5802.
136. Nagao, H., Nishino, M., Shigeta, Y., Yoshioka, Y. and Yamaguchi, K. (2000).
Theoretical studies on superconducting and other phases: triplet superconductiv-
ity, ferromagnetism, and ferromagnetic metal, J. Chem. Phys. 113, pp. 721–732.
137. Kondo, J. (2001). Superconductivity of the two-dimensional Hubbard models with
a small U., J. Phys. Soc. Jpn. 70, p. 808.
138. Nagao, H., Mitani, M., Nishino, M., Shigeta, Y., Yoshioka, Y. and Yamaguchi, K.
(1999). Theoretical studies on anomalous phases in molecular systems with external
field: possibility of photo-induced superconductivity, Int. J. Quantum Chem. 75,
pp. 549–561.
139. Nagao, H., Nishino, M., Mitani, M., Yoshioka, Y. and Yamaguchi, K. (1997).
Possibilities of charge- and/or spin-mediated superconductors and photo-induced
superconductors in the intermediate region of metal-insulator transitions, Int. J.
Quantum Chem. 65, pp. 947–957.
140. Nagao, H., Mitani, M., Nishino, M., Shigeta, Y., Yoshioka, Y. and Yamaguchi, K.
(1988). Possibility of charge-mediated superconductors in the intermediate region of
metal–insulator transitions, Int. J. Quantum Chem. 70, pp. 1075–1084.
141. Bogoliubov, N. N. and Bogoliubov, N. N., Jr. (1986). An Introduction to Quantum
Statistical Mechanics (Nauka, Moscow) (transl. into English 1994).
142. Bogoliubov, N. N. and Shirkov, D. V. (1959). Introduction to the Theory of Quantized
Field (Interscience, New York).
143. Mattuck, R. D. (1976). A Guide to Feynman Diagrams in the Many-Body Problem
2nd edn. (McGraw-Hill, New York).
144. Kimura, M., Kawabe, H., Nishikawa, K. and S. Aono (1988). Possibility of supercon-
ducting and other phases in organic high polymers of polyacene carbon skeletons II:
screened electron–electron interaction, Bull. Chem. Soc. Jpn. 61, pp. 4245–4252.
145. Kimura, M., Kawabe, H., Nishikawa, K. and Aono, S. (1986). Superconducting and
other phases in organic high polymers of polyacenic carbon skeletons I: the method
of sum of divergent perturbation series, J. Chem. Phys. 85, pp. 3090–3096.
146. Kimura, M., Kawabe, H., Nishikawa, K. and Aono, S. (1985). Superconducting and
other phases in organic high polymers of polyacenic carbon skeletons II: the mean
field method, J. Chem. Phys. 85, pp. 3097–3100.
147. Kimura, M., Kawabe, H., Nakajima, A., Nishikawa, K. and Aono, S. (1988).
Possibility of superconducting and other phases in organic high polymers of poly-
acene carbon skeletons II: screened electron-electron interaction. Bull. Chem. Soc.
Jpn. 61, pp. 4239–4244.
148. Bogoliubov, N. N. and Tyablikov, S. (1959). Retarded and advanced Green func-
tions in statistical physics, Dokl. Acad. Sci. USSR 126, p. 53; Sov. Phys. Dokl. 4,
p. 589.
210 Modern Aspects of Superconductivity: Theory of Superconductivity

149. Richarson, R. W. (1963). A restricted class of exact eigenstates of the pairing-force

Hamiltonian, Phys. Lett. 3, pp. 277–279.
150. Matveev, K. A. and Larkin, A. I. (1997). Parity effect in ground state energies of
ultrasmall superconducting grains, Phys. Rev. Lett. 78, pp. 3749–3752.
151. Ralph, D. C., Black, C. T. and Tinkham, M. (1997). Gate-voltage studies of discrete
electronic staes in aluminum nanoparticles, Phys. Rev. Lett. 78, p. 4087.
152. Schechter, M., Imry, Y., Levinson, Y. and von Delft, J. (2001). Thermodynamic
properties of a small superconducting grain, Phys. Rev. B 63, pp. 214518–214534.
153. Sierra, G. (2000). Conformal field theory and the exact solution of the BCS
Hamiltonian, Nucl. Phys. B 572, pp. 517–527.
154. Stenuit, G., Michotte, S., Govaerts, J. and Piraux, L. P. (2003). Vortex matter in
lead nanowires, Eur. Phys. J. B 33, pp. 103–110.
155. Kasumov, A. Yu., et al. (1999). Supercurrents through single-walled carbon
nanotubes, Science 284, pp. 1508–1511.
156. Nagao, H., Kawabe, H. and Kruchinin, S. (2006). Nanosize two-gap supercon-
ductivity, in Proc. NATO ARW “Electron Correlation in New Materials and
Nanosystems”, eds. Scharnberg, K. and Kruchinin, S. (Springer), pp. 117–127.
157. Kawabe, H., Nagao, H. and Kruchinin, S. (2006). Exact solution of two-
band superconductivity, in Proc. NATO ARW “Electron Correlation in New
Materials and Nanosystems”, eds. Scharnberg, K. and Kruchinin, S. (Springer),
pp. 129–139.
158. Nagao, H. and Kruchinin, S. P. (2008). Kondo effect coupled to superconductivity
ultrasmall grains, in Proc. NATO ARW Electron Transport in Nanosystems eds.
Bonca, J. and Kruchinin, S. (Springer), pp. 105–115.
159. Kruchinin, S. P., Dzezherya, Yu. and Annett, J. (2009). Imteractions of nanoscale
ferromagnetic granules in a London superconductor, Supercond. Sci. Technol. 19,
pp. 381–384.
160. Ralph, D. C., Black, C. T. and Tinkham, M. (1995). Spectroscopic measurements
of discrete electronic states in single metal particles, Phys. Rev. Lett. 74, p. 3241.
161. Smith, R. A. and Ambegaokar, V. (1996). Effect of level statistics on superconduc-
tivity in ultrasmall metallic grains, Phys. Rev. Lett. 77, p. 4962.
162. von Delft, J., Zaikin, A. D., Golubev, D. S. and Tichy, W. (1996). Parity-affected
superconductivity in ultrasmall metallic grains, Phys. Rev. Lett. 77, pp. 3189–3192.
163. Yamaji, K. and Shimoi, Y. (1994). Superconducting transition of the two-
chain Hubbard model indicated by diagonalization calculations, Physica C 222,
pp. 349–360.
164. Black, C. T., Ralph, D. C. and Tinkham, M. (1996). Spectroscopy of the supercon-
ducting gap in individual nanometer-scale aluminum particles, Phys. Rev. Lett. 76,
pp. 688–611.
165. Braun, F. and von Delft, J. (1998). Fixed-N superconductivity: the crossover from
the bulk to the few-electron limit, Phys. Rev. Lett. 81, pp. 4712–4715.
166. von Delft, J. and Braun, F. (1999). Superconductivity in ultrasmall metallic grains,
Phys. Rev. B 59, pp. 9527–9544.
167. Gladilin, V. N., Fomin, V. M. and Devreese, J. T. (2002). Shape of nanosize super-
conducting grains: does it influence pairing characteristics?, Solid State Commun.
121, pp. 519–523.
168. Jankó, B., Smith, A. and Ambegaokar, V. (1994). BCS superconductivity with fixed
number parity, Phys. Rev. B 50, pp. 1152–1161.
 Bibliography 211

169. Matveev, K. A. and Larkin, A. L. (1997). Parity effect in ground state energies of
ultrasmall superconducting grains, Phys. Rev. Lett. 78, pp. 3749–3752.
170. Richardson, R. W. and Sherman, N. (1964). Pairing models of Pb206, Pb204 and
Pb202, Nucl. Phys. 523, pp. 253–268.
171. Richardson, R. W. (1977). Pairing in the limit of a large number of particles,
J. Math. Phys. 6, pp. 1802–1811.
172. Richardson, R. W. (1966). Numerical study of the 8–32-particle eigenstate of the
pairing Hamiltonian, Phys. Rev. 141, pp. 949–955.
173. Sierra, G., Dukelsky, J., Dussel, G. G., von Delft, J. and Braun, F. (2000). Exact
study of the effect of level statistics in ultrasmall superconducting grains, Phys. Rev.
B 61, pp. 11890–11893.
174. Kouwenhoven, L. and Glazman, L. (2001). The revival of the Kondo effect, Phys.
World 14, 1, pp. 33–38.
175. van der Wiel, G., De Franceschi, S., Fujisawa, T., Elzerman, J. M., Tarucha, S. and
Kouwenhoven, L. P. (2000). The Kondo effect in the unitary limit, Science 289,
pp. 2105–2108.
176. Inoshita, T. (1998). Kondo effect in quantum dots, Science 281, pp. 526–527.
177. Sasaki, S., De Franceschi, S., Elzerman, J. M., van der Wiel, W. G., Eto, M.,
Tarucha, S. and Kouwenhoven, L. P. (2000). Kondo effect in an integer-spin quantum
dot, Nature (London) 405, pp. 764–767.
178. Inoshita, T., Shimizu, A., Kuramoto, Y. and Sakaki, H. (1993). Correlated elec-
tron transport through a quantum dot: the multiple-level effect, Phys. Rev. B 48,
pp. 114725–114728.
179. Izumida, W., Sakai, O. and Shimizu, Y. (1998). Kondo effect in single quantum dot
systems study with numerical renormalization group method, J. Phys. Soc. Jpn. 67,
p. 2444.
180. Yeyati, A. L., Flores, F. and Martin-Rodero, A. (1999). Transport in multilevel
quantum dots: from the Kondo effect to the Coulomb blockade regime, Phys. Rev.
Lett. 83, pp. 600–603.
181. Pasupathy, A. N., Bialczak, R. C., Martinek, J., Grose, J. E., Donev, L. A. K.,
McEuen, P. L. and Ralph, D. C. (2004). The Kondo effect in the presence of ferro-
magnetism, Science 306, pp. 86–90.
182. Matsubayashi, D. and Eto, M. (2007). Spin splitting and Kondo effect in quantum
dots coupled to noncollinear ferromagnetic leads, Phys. Rev. B 75, p. 165319.
183. Buitelaar, M. R., Nussbaumer, T. and Schönenberger, C. (2002). Quantum dot in
the Kondo regime coupled to superconductors, Phys. Rev. Lett. 89, p. 256801.
184. Ueda, A. and Eto, M. (2006). Resonant tunneling and Fano resonance in quantum
dots with electron–phonon interaction, Phys. Rev. B 73, pp. 235353–235365.
185. Eto, M. and Nazarov, Y. V. (2000). Enhancement of Kondo effect in quantum dots
with an even number of electrons, Phys. Rev. Lett. 85, pp. 1306–1309.
186. Reich, S., Leitus, G., Popovitz-Biro, R. and Schechter, M. (2003). Magnetization of
small lead particles, Phys. Rev. Lett. 91, p. 147001.
187. Yamaji, K. and Shimoi, Y. (1994). Superconducting transition of the two-
chain Hubbard model indicated by diagonalization calculations, Physica C 222,
pp. 349–360.
188. Combescot, R. and Leyronas, X. (1995). Coupling between planes and chains in
YBa2 Cu3 O7 : a possible solution for the order parameter controversy, Phys. Rev.
Lett. 75, pp. 3732–3735.
212 Modern Aspects of Superconductivity: Theory of Superconductivity

189. Cox, D. L. and Maple, M. B. (1995). Electronic pairing in exotic superconductors,

Phys. Today 48, 2, p. 32.
190. Hewson, A. C. (1993). The Kondo Problem to Heavy Fermion (Cambridge,
University Press).
191. Yoshimori, A. and Sakurai, A. (1970). Functional integral approach to the bound
state due to the s–d exchange Interaction, Suppl. Prog. Theor. Phys. 46, pp. 162–181.
192. Eto, M. and Nazarov, Y. V. (2001). Mean field theory of the Kondo effect in quantum
dots with an even number of electrons, Phys. Rev. B 64, p. 85322.
193. Fulde, P. and Ferrell, A. (1964). Superconductivity in a strong spin-exchange field,
Phys. Rev. 135, pp. A550–A563.
194. Larkin, A. and Ovchinnikov, Y. (1965). Inhomogeneous state of superconductors,
Sov. Phys. JETP 20, p. 762.
195. Lyuksyutov, I. F. and Pokrovsky, V. L. (2005). Ferromagnet superconductor hybrids,
Adv. Phys. 54, pp. 67–136.
196. Ryazanov, V. V., Oboznov, V. A., Rusanov, A. V., Veretennikov, A. V.,
Golubov, A. A. and Aarts, J. (2001). Coupling of two superconductors through
a ferromagnet: evidence for a π-junction, Phys. Rev. Lett. 86, pp. 2427–2430.
197. Frolov, S. M., Van Harlingen, D. J., Oboznov, V. A., Bolginov, V. V. and
Ryazanov, V. V. (2004). Measurement of the current-phase relation of superconduc-
tor/ferromagnet/superconductor Josephson junctions, Phys. Rev. B 70, p. 144505.
198. Bell, C., Loloee, R., Burnell, G. and Blamire, M. G. (2005). Characteristics of
strong ferromagnetic Josephson junctions with epitaxial barriers, Phys. Rev. B 71,
p. 180501.
199. Ruotolo, A., Pepe, G. P., Bell, C. Leung, C. W. and Blamire, M. G. (2005).
Modulation of the dc Josephson current in pseudo-spin-valve Josephson multilayers,
Supercond. Sci. Technol. 18, pp. 921–926.
200. Lange, M., Van Bael, M. J., Van Look, L., Raedts, S., Moschalkov, V. V. and
Bruynseraede, Y. (2002). Nanostructured superconductor ferromagnet bilayers, in
New Trends in Superconductivity, eds. Annett, J. F. and Kruchinin, S., NATO
Science Series II: Mathematics, Physics and Chemistry 67, pp. 365–373 (Kluwer,
Dordrecht).
201. Aladyshkin, A. Yu., Silhanek, A. V., Gillijns, W. and Moshchalkov, V. V. (2009).
Nucleation of superconductivity and vortex matter in superconductor–ferromagnet
hybrids, Supercond. Sci. Technol. 22, pp. 1–48.
202. Milosovec, M. V. and Peeters, F. M. (2005). Field-enhanced critical parameters in
magnetically nanostructured superconductors, Europhys. Lett. 70, pp. 670–676.
203. Fetter, V. M. and Larkin, A. I. (2002). Spin glass versus superconductivity, Phys.
Rev. B 66, p. 64526.
204. Andreev, A. F. (1998). Superfluidity, superconductivity and magnetism in meso-
scopics, Usp. Phys. 168, 6, pp. 656–663.
205. Milosevic, M. V., Yampolskii, S. V., Peeters, F. M. (2002). Magnetic pinning of vor-
tices in a superconducting film: the (anti)vortex–magnetic dipole interaction energy
in the London approximation, Phys. Rev. B 66, p. 174519.
206. de Gennes, P.-G. (1966). Superconductivity of Metals and Alloys (Addision-Wesley,
Advanced Book Programme).
207. Van der Wal, C. (2000). Quantum superposition of macroscopic persistent-current
states, Science 290, p. 773.
208. You, J. Q. and Nori, F. (2005). Superconducting circuits and quantum information,
Phys. Today 12, pp. 42–45.
 Bibliography 213

209. Nakamura, Y., Pashkin, Y. A. and Tsai, J. S. (1999). Quantum oscillations in two
coupled charge qubits, Nature 398, pp. 786–788.
210. Makhlin, Y., Schon, G. and Shnirman, A. (2001). Quantum-state engineering with
Josephson-junction devices, Rev. Mod. Phys. 73, pp. 357–400.
211. Chiorescu, I., Nakamura, Y., Harmans, C. J. P. M. and Mooij, J. E. (2003). Coherent
quantum dynamics of a superconducting flux qubit, Science 299, pp. 1869–1871.
INDEX

Index Terms Links

π junction 88 167

acoustic phonons 74 79
anharmonic model 72 76
antiferromagnetic 62 64 70 71
88 93 101 107
111 138 193
anyonic superconductivity 78

basis functions 81 82
BCS theory 27 32 44 51
64 65 72 78
79 88 89 104
114 116 135 137
140 142 145 161
163
bipolaronic mechanism 72 78
Bose condensation 80
Bose particles 78
boson field 44 47 48 53
55 59

This page has been reformatted by Knovel to provide easier navigation.

Index Terms Links

carbon nanotubes 136 181

charge qubit 188
chemical potential 4 11 18 20
25 29 33 90
93 98 111 123
125 132 138 143
146 152 162
condensation energy 104 137 141 144
145 154 161
continual model 93
Coulomb blockage 188 189
critical temperature 21 24 27 28
32 51 52 61
70 74 89 98
104 107 117 121
127 192
CuO2 planes 62 71

Debye energy 72 138 160 161

deformation energy 23
dielectric properties 78
d symmetry 81 85 90 91
104 106

This page has been reformatted by Knovel to provide easier navigation.

Index Terms Links

effective interaction 111 123 129 138

142 144
exact solution 68 115 135 145
146 152 156 163
197
exchange interactions 166
exciton mechanism 75

ferromagnetic 74 109 137 166

171
ferromagnetic granules 174 199
FFLO 167
flux qubit 187
free energy 95 101 102 107
170

gap equation 20 21 24 25
50 125 140 141
Gauss theorem 170 201
Ginzburg–Landau equation 58 59
Green functions 67 94 113 114
139 197

This page has been reformatted by Knovel to provide easier navigation.

Index Terms Links

heat capacity 89 103

heavy fermions 82 88 104
high-temperature superconductors
(HTSCs) 61 62 65 66
71 72 73 78
80 82 90 91
104 109 117 118
137 138 191
HOMO 25
Hubbard 76
Hubbard model 90
Hubbard–Stratonovitch transformation 44 46 47 59

instability 55 90
interacting 15 36 39 42
168
interaction constant 138 142 144
interferometer 185

Jahn–Teller effect 77
Josephson junction 3 84 106 182
184
jump of the heat 89 104 107

This page has been reformatted by Knovel to provide easier navigation.

Index Terms Links

Knight shift 84
Kondo effect 68 115 157 158
160 165 166 197

Lagrangian 3 7 44 53
55 56 94
length scale 168 169
level spacing 135 141 161 165
166
London equation 168
London superconductors 166 168 199
LUMO 25

magnetic resonance 28 31 32
magnons 74 90
mechanism of pairing 62 63 72 74
77 78 80 93
106 126 127
mesoscopic superconductivity 135
Mg1-xAlxB2 121 122
mixed state 81 82
multiband superconductivity 61 108 135 190

This page has been reformatted by Knovel to provide easier navigation.

Index Terms Links

Néel temperature 70 71
Nambu spinor 4 7 29 39
Nambu–Goldstone theorem 7
nanosize two-gap superconductivity 137 138 141 143
NMR 31 84 91 92
Noether theorem 7

order parameter 2 44 160 166

organic superconductors 75

parity gap 137 141 143 155

path-integral 44 46 57 137
139 145
Peierls instability 21
penetration depth 62 84 91 136
166 173 178 200
phase diagram 69 70 108 112
121 127 128 130
133 134
phase difference 88 184
Pippard case 168
plasmon mechanism 78
polyacene 21
polymeric 21 25 75 76

This page has been reformatted by Knovel to provide easier navigation.

Index Terms Links

propagator 10 11 16 17
19 33 37 41
42 53
pseudogap 70 71

quantum computer 181 189

qubit 181 189

Rabi oscillation 188

relaxation 28 29 31 33
36 43 84
renormalization equation 127 128 132
renormalization group 108 121 127 128
130
Richardson solution 135
RKKY 168 173 174
room-temperature superconductors
(RTSCs) 1 61 64 108
116 117
RVB 1 74 75

self-energy 33 37 38 42
170
size effect 136
spin dynamics 41 92

This page has been reformatted by Knovel to provide easier navigation.

Index Terms Links

spin fluctuations 62 80 88 101

107 108 110 111
138 193
spin waves 74 89
spin-orientation phase transitions 174
spinor 3 4 6 12
27 29 39 48
SQUID 185 189
superconducting grains 137 143 157 158
161 185
supersymmetry 44 56 58
susceptibility 33 34 36 71
90 174
symmetry breaking 3 8 53
symmetry group 81
symmetry of pairing 63 107 110 137
193

theory of superconductivity 1 2 32 43
62 80 88 108
thermodynamics 62 88 89 93
107 111 138 193
Tl-based 67 68
two-dimensional 59 76 77 81
84 90 101 118
119 174
two-gap superconductivity 65 66 108 112
113 119 127 137
143 157 194 196
This page has been reformatted by Knovel to provide easier navigation.
Index Terms Links

two-level 182 184 189

ultrasmall grains 68 115 142 145

158 161 163 197

vertex correction 128 131

yttrium ceramics 66 67

This page has been reformatted by Knovel to provide easier navigation.