The±~liyMC& of the

Standard Model
and

T. Morii • C. S. Lim • S. N. Mukherjee

 The Physics of the
Standard Model and Beyond
This page is intentionally left blank
 The Physics of the
Standard Model
and Beyond

T. Morii
Kobe University, Japan

C. S. Lim
Kobe University, Japan

S. N. Mukherjee
Banaras Hindu University, India

^P World Scientific
NEW JERSEY • LONDON • SINGAPORE • BEIJING- SHANGHAI • H O N G K O N G • TAIPEI • CHENNAI
Published by
World Scientific Publishing Co. Pte. Ltd.
5 Toh Tuck Link, Singapore 596224
USA office: Suite 202, 1060 Main Street, River Edge, NJ 07661
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.

THE PHYSICS OF THE STANDARD MODEL AND BEYOND

Copyright © 2004 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 981-02-4571-8

This book is printed on acid-free paper.

Printed in Singapore by Mainland Press
To our wives, Harue, Yoko and Reba
This page is intentionally left blank
 PREFACE

Great advances in particle physics have been attained in the last several decades. In
the development of the new generation of accelerators and detectors, the quantity
and quality of the data on various interaction processes have been vastly increased
and our knowledge of particle physics became surprisingly fruitful. Moreover, many
interesting and new ideas were proposed and developed successfully in quantum field
theories for describing strong, electromagnetic and weak interactions. Those new
ideas were concentrated into a simple and beautiful theory known as the standard
model in the late 1960's which was formulated in the framework of non-Abelian
gauge theory and was extremely successful in describing a wide range of existing
phenomena of elementary particles. However, it is also believed that the model is not
the ultimate theory; it has many arbitrary parameters which cannot be predicted by
the theory and fundamental problems, such as the hierarchy problem, which should
be solved in some physics beyond the standard model, "New Physics". Thus it is a
general consensus that to search for the evidences for physics beyond the standard
model is the most urgent issue in the particle physics of the 21st century.
The aim of the present textbook is to give a unified description of the struc-
tures and interactions of elementary particles, by discussing the underlying theo-
ries, namely the standard model of elementary particles and physics beyond the
standard model. Especially, concerning the description of the physics beyond the
standard model, we will select the hot topics of our current interest, including the
issues inspired by various experiments as much as possible.
This book has emerged out of the introductory lectures we gave in the graduate
course or series of lectures given in various places over several years. Therefore,
this book is written primarily for graduate students and can be used by those who
wish to major in theoretical and experimental high energy physics. Throughout this
textbook, it is assumed that the readers have already gone through a basic course
in quantum field theory and particle physics. We hope that this book is also useful
for the researchers in the field of particle and nuclear physics.
There are two main features of this textbook. First, we have aimed at a concise
description but at the same time we have paid attention so that the basic concepts

vii
vm Preface

are clearly mentioned. For such a purpose each section is guided by pedagogical
arguments. Second, we also attempt to provide excitement in particle physics,
focusing on the important experimental observations (old and new) and a variety
of nice ideas for their interpretation.
The organization of the book is as follows. In Chapter 1, we give a brief in-
troduction of the standard model for the elementary particles to describe their
structures and interactions. This is followed by Chapter 2 with a description of
Fermi theory of weak interactions and its limitation. Preparation for the under-
standing of electroweak gauge theory is done in Chapter 3, which covers symmetry,
symmetry breaking and the gauge theory. Glashow-Weinberg-Salam (GWS) the-
ory of electroweak interactions is covered in Chapter 4. In Chapter 5, quantum
chromodynamics (QCD), i.e. the theory of strong interactions, is discussed briefly.
After these chapters, we will come to the description of physics beyond the standard
model. In Chapter 6, we discuss a current hot topic, neutrino physics, i.e. neutrino
masses and neutrino ocsillations. Then, in Chapter 7, we deal with supersymmetric
theories as the typical example of physics beyond the standard model. We further
attempt to discuss other representative topics in the physics beyond the standard
model; precision tests of electroweak radiative corrections and new physics search
through these analyses in Chapter 8 and flavor physics and CP violation, which
are also sensitive to the presence of new physics, in Chapter 9. Each chapter is
accompanied by a few problems, hoping that solving them will be of some help in
the understanding of the main text. The topics we chose depend on our preference.
However, we believe that those topics must be of general interest and instructive
not only for students who are going to enter the field of theoretical and experimen-
tal particle physics but also even for researchers in high energy physics. Finally,
appendices for some basics are given for the reader's convenience.
During this project, we have learned a lot from many excellent books and re-
views; some of them (Bjorken and Drell, 1964; Abers and Lee, 1973; Jauneau, 1977;
Quigg, 1983; Cheng and Li, 1984; Halzen and Martin, 1984; Rudaz, 1986; Aitchi-
son and Hey, 1989; Wess and Bagger, 1992; Nagashima, 1999) were good guides in
preparing this book.
Unfortunately, because of limited space and time, we could not help neglecting
some of the important topics, such as the full contents of QCD, discussion on the
grand unified theories (GUT), detailed phenomenological analyses of supersymmet-
ric theories, and so on. Fortunately, concerning these topics, there already exist
many excellent textbooks or reviews and we refer the reader to some of them: for
example, for comprehensive discussion of QCD, the textbooks by Yndurain (Yn-
durain, 1992) and by Muta (Muta, 1998), for discussion of GUT, the book by Ross
(Ross, 1985) and a review by Langacker (Langacker, 1982), for more details on su-
persymmtry including its phenomenology, the books by Weinberg (Weinberg, 2000)
and by Ross (Ross, 1985), and so on. Furthermore, in this textbook the references
are far from complete but rather limited only to the ones directly related to the
 Preface IX

discussions extended in each chapter. We apologize to all authors whose papers we

have neglected in spite of their great contributions to the fields.
In carrying out this project, we are deeply indebted to many people, including
our teachers, collaborators, students and colleagues who taught and encouraged
us a lot. One of the authors (S. N. M.) is grateful to Professor Amitava Roy
Chaudhuri for useful comments and to Dr. Sudip Sanyal for research collaboration.
The financial assistance obtained from the Department of Science and Technology,
New Delhi, India, is also gratefully acknowledged. It took a long time to complete
this project; we started at the end of 1993 and finally came to an end in the fall of
2003 after a long struggle. The project was interrupted many times by unexpected
difficulties which we came upon in the course of this project, such as the terrible
earthquake in Kobe in 1995, the passing away of the mother of one of us (T. M.) in
2002, and so on. In spite of the terrible delay of our work, H. T. Leong, editor of this
book, has encouraged us constantly with a warm heart and it is our great pleasure
to sincerely thank him for his patient and continuous support for publication of this
book. It is also our great pleasure to acknowledge the aid of our graduate students,
S. Oyama, K. Mawatari, K. Sudoh, K. Hasegawa and T. Nagasawa, for their self-
sacrificing help in drawing and arranging figures and for useful comments. Finally,
we are greatly indebted to our families for their warm support and encouragement
throughout this long time.

T. Morii, C. S. Lim and S. N. Mukherjee

Kobe, Japan and Vanarasi, India

November 2003

Corrections to this book

Misprints and corrections to this book will be presented on the website at

http://www.phys.sci.kobe-u.ac.jp/~lim/bookstb.html. We should be grateful if the
reader who would find additional errors or have other comments could kindly send
them to [email protected].
This page is intentionally left blank
 CONTENTS

Preface vii

Chapter 1 INTRODUCTION 1
1.1 Elementary Particles in the Standard Model 1
1.2 Interactions Among Fundamental Particles 4

Chapter 2 WEAK INTERACTION 7

2.1 The Fermi Theory of Weak Interaction 7
2.1.1 Parity Violation in Weak Interaction 8
2.1.2 Road to Current-Current V-A Interaction 9
2.1.2.1 Helicity and chirality 11
2.1.2.2 Observation of electron helicity in the /?-decay 13
2.1.2.3 Determination of neutrino helicity 13
2.1.2.4 Angular correlation between e + and v 14
2.1.2.5 V-A interaction 15
2.1.3 Lepton Current Universality 17
2.1.4 Pion Decays 20
2.1.5 Cabibbo Currents 22
2.1.6 Difficulties in the Fermi Theory 25
2.1.7 Unitarity Violation 26
2.1.8 Non-renormalizability 28
2.2 Intermediate Weak Boson Model 28

Chapter 3 S Y M M E T R I E S A N D T H E G A U G E THEORIES 33
3.1 Global Symmetries and Noether's Theorem 33
3.2 Local Gauge Symmetries and Gauge Fields 36
3.2.1 Quantum Electrodynamics — £/(l) model 37
3.2.2 Yang-Mills Gauge Theory — SU(2) model 40
3.3 Spontaneous Symmetry Breaking and Goldstone Bosons 44
3.4 Higgs Mechanism 51

xi
xii Contents

Chapter 4 T H E S T A N D A R D MODEL OF ELECTROWEAK

INTERACTIONS 55
4.1 Fermions in the GWS Model 55
4.2 SU(2)L x U(1)Y Invariant Lagrangian 59
4.3 Spontaneous Breaking of SU(2)L X U(1)Y Symmetry 61
4.4 Charged and Neutral Currents, Comparison with Effective Fermi
Theory 65
4.5 Addition of More Leptons 70
4.6 Extension to Quarks 74
4.6.1 Quark Mass Matrix 76
4.6.2 Flavor Mixing 78
4.6.2.1 The case of charged current interaction 78
4.6.2.2 The case of neutral current interaction 83
4.7 Anomalies 84

Chapter 5 QUANTUM CHROMODYNAMICS 89

5.1 Evidence for Colors 90
5.2 QCD Lagrangian and the Strength of Color Forces 92
5.3 Running Coupling Constants 96
5.3.1 Renormalized Charge in QED 96
5.3.2 Running Coupling Constant in QCD 102

Chapter 6 NEUTRINO MASSES A N D NEUTRINO

OSCILLATIONS 105
6.1 Type of Fermions and Fermion Masses 105
6.2 Neutrino Masses 108
6.2.1 Possible Types of Neutrino Masses 108
6.2.2 The Mechanism of Neutrino Mass Generation Ill
6.3 Flavor Mixing and Neutrino Oscillation (in the Vacuum) 114
6.3.1 Flavor Mixing 115
6.3.2 Neutrino Oscillation (in the Vacuum) 117
6.4 Resonant Neutrino Matter Oscillation 123
6.5 Neutrino Oscillation in the Three Generation Scheme 130
6.5.1 Vacuum Neutrino Oscillation in the Three Generation
Scheme 132
6.5.2 Resonant Matter Oscillations of Neutrinos in the Three
Generation Scheme 137
6.6 Atmospheric and Solar Neutrino Oscillations 139
6.6.1 Atmospheric Neutrino Oscillation 140
6.6.2 Solar Neutrino Oscillation 142
 Contents xiii

Chapter 7 SUPERSYMMETRY 147

7.1 Supersymmetry and the Hierarchy Problem 147
7.2 Two-Component Notation 149
7.3 SUSY Algebra and SUSY Group 152
7.4 Superfield Formulation 155
7.5 Chiral Superfield 160
7.6 Wess-Zumino Model 162
7.7 Vector Superfield 167
7.8 SUSY QED 171
7.9 SUSY Yang-Mills Theories 173
7.10 Minimal Supersymmetric Standard Model (MSSM) 174
7.11 Some Phenomenological Predictions of SUSY Gauge Theories 179
7.12 Spontaneous SUSY Breaking 184

Chapter 8 PRECISION T E S T OF ELECTROWEAK

RADIATIVE CORRECTIONS A N D
N E W PHYSICS 191
8.1 The Meaning of Precision Test of Electroweak Radiative Corrections . . 191
8.2 Decoupling and Non-decoupling 194
8.3 Oblique Corrections and S, T, U Parameters 201
8.4 Global Symmetries 209
8.5 Operator Analysis 211

Chapter 9 FLAVOR P H Y S I C S A N D CP VIOLATION 215

9.1 The Interest in Flavor Physics and CP Violation 215
9.2 Flavor Symmetry and FCNC Rare Processes 216
9.3 Rare Processes in Kaon System 219
9.4 CP Violation in Kobayashi-Maskawa Model 232
9.5 CP Violation in the Neutral K System 237
9.6 CP Violation in the Neutral B System 241

Appendix A Notation and Useful Relations 249

A.l Four Vectors 249
A.2 7 Matrices 250
A.3 Trace Theorems 252
A.4 Right- and Left-Handed Dirac Spinors 252
A.5 Dirac Equation in the 75-Diagonal Representation 253

Appendix B Cross Sections and Feynman Rule 257

B.l Cross Sections 257
B.2 Decay Width 259
B.3 Feynman Rule 260
xiv Contents

Appendix C Basics of the Group Theory 269

C.l Group and Representation 269
C.2 SU(n) Group and Lie Algebra 270
C.3 Representation in SU(n) Group 272

Appendix D C, P and T Transformation 275

D.l Parity or Space Inversion P 275
D.2 Time Inversion T 276
D.3 Charge Conjugation C 277
D.4 CP Transformation 278

Appendix E The Quark Model 281

E.l Isospin Symmetry 281
E.2 Quark Model and SU(3) Symmetry 283
E.3 Representations of Mesons and Baryons 286

Bibliography 291

Index 295
 Chapter 1
INTRODUCTION

"From the earliest times, man's dream has been to comprehend the complexity of
nature in terms of as few unifying concepts as possible"*

Stars which are twinkling in a night sky have brought us various dreams from
ancient times. People have been attracted in their mysterious appearance and have
looked for the physical law of the cosmos; how was the cosmos born?, what are the
most fundamental building blocks of matter constructing the cosmos?, how do those
fundamental particles interact with each other or one another? and so on. There
have been many questions to be answered. Particle physics is the most powerful
tool for investigating these fundamental questions.
In this introductory chapter, we make a simple explanation on fundamental
particles in the standard model and on the characteristic properties of their inter-
actions.

1.1 Elementary particles in the standard model

Elementary particles are the most fundamental building blocks of matter. Entity of
elementary particles has been changed in the long course of development of physics.
By the end of the last century, people have found that matter is made of molecules
and/or atoms, atoms are made of nuclei and electrons, nuclei are made of nucleons,
i.e. protons and neutrons, and finally nucleons are made of quarks. Elementary
particles of our own time are quarks (together with leptons such as electron). In
the development of big accelerators in the last 50 years, many (more than 300)
particles called hadrons were discovered in addition to protons, neutrons and pions.
There are two kinds of hadrons; baryons (proton, neutron etc.) with the baryon
number B = 1 and mesons (pions, kaons, etc.) with B = 0.
The quark model proposed by Gell-Mann and independently by Zweig in 1964

•Address by Abdus Salam at UNESCO Celebration of the Centenary of Einstein's birth, Paris
May 1979.

1
2 INTRODUCTION

(Gell-Mann, 1964; Zweig, 1964) classifies all existing hadrons surprisingly well based
on the internal symmetry of SU(3) for hadrons composed of relatively light three
quarks, and successfully explains static properties of those particles (see Appendix
E). In the quark model, baryons are composed of three quarks (qqq) as p = (uud),
n = (udd), A = (uds), etc. and mesons are composed of a quark q and an anti-quark
q as 7r+ = (ud), n~ — (ud), K+ = (us), K~ = (su), etc. Nowadays six different
quarks u, d, s, c, b, t are known to exist and thus, it is said that quarks possess 6
degrees of freedom called "flavor". A quark flavor can change into another quark
flavor through weak interactions mediated by charged weak bosons W±, which are
predicted to appear in the electroweak standard model as discussed later in Chapter
4. In addition to flavor, quarks have another degree of freedom called "color" as
briefly discussed in Chapter 5. The interaction between quarks due to the color
"charge", which is nothing but the strong interaction, is mediated by gluons and is
described by quantum chromodynamics(QCD). QCD is the gauge theory with color
SU(3) symmetry. While the flavor symmetry is broken by the difference of quark
masses, largely for heavy quarks, the color symmetry is an exact symmetry.
Another kind of elementary particles called leptons (electron e, muon LI, tau
r and their corresponding neutrinos ve, v^, vT) exist in Nature. Leptons are free
from the strong interaction and have no color degrees of freedom, i.e. leptons are
colorless. Among them, neutrinos possess only weak interactions, while e, fx and r
have the both of weak and electromagnetic interactions. Productions and decays
of leptons are successfully described by the electroweak standard model, i.e. the
SU(2)L x U(1)Y gauge theory of electroweak interactions, discussed in chapter 4.
The elementary particles in the standard model are as follows;

*-* ( ; ) • {'.)• ( ! ) • <">

{
photon 7,
weak (gauge) bosons W^,Z°, (1.3)

gluons g,
Higgs bosons H. (1.4)

Quarks and leptons are fundamental building blocks of matter. All of them are
fermions and have spin | , whose quantum numbers are summarized in Table 1.1
and Table 1.2. It is interesting to note that both quarks and leptons are paired into
three doublets, the members of each doublet participating in the charged current
weak interaction processes together. The repetition of the doublets is stated as there
are three generations of quarks and leptons. The corresponding particles in different
 Elementary particles in the standard model 3

Q /s S c B T mass
u +* +^ 0 0 0 0 1.5 ~ 5 MeV
1
d -* 2
0 0 0 0 3 ~ 9 MeV
s —3 0 -1 0 0 0 60 ~ 170 MeV
c 0 0 +1 0 0 1.47 ~ 1.83 GeV
b -i 0 0 0 -1 0 4.6 ~ 5.1 GeV
t +1 0 0 0 0 +1 174.3 ± 3.2 ± 4.0 GeV

Table 1.1 Quarks (Q: electric charge, 73: 3rd component of isospin, 5: strangeness, C: charm-
ness, B: bottomness, T: topness. These quantum numbers change their signs for anti-quarks.)

Q Le w LT mass
e -1 +1 0 0 2* 0.511 MeV
Ve 0 +1 0 0 <3eV
M -1 0 +1 0 =* 105.66 MeV
Vy. 0 0 +1 0 < 0.19 MeV
T -1 0 0 +1 =* 1777.0 MeV
vT 0 0 0 +1 < 18.2 MeV

Table 1.2 Leptons (Q: electric charge, L e : electron number, L M : muon number, L r : tau number.
These quantum numbers change their signs for anti-leptons.)

generations, for instance u, c, t have exactly the same quantum numbers. The only
property to distinguish different generations is the difference of the masses of quarks
and leptons, depending on the generation. The electric charges of upper and lower
components of quark doublets are + | and — | , respectively, while those of lepton
doublets are 0 and — 1, respectively. At the first sight, such charge "quantization"
seems to be miraculous. The quantization, however, is known to be well-suited to
the anomaly-free condition for the standard model as explained in Chapter 4.
Gauge bosons having spin 1 are mediators of interactions between quarks or
leptons. Interaction strength depends on which gauge bosons propagate between
quarks or leptons. Electromagnetic, weak and strong interactions are mediated by
photons 7, weak bosons W±, Z° and gluons g, respectively. A photon does not
couple to itself, while gluons and weak bosons have self-couplings, i.e. can couple to
themselves. This is due to the fact that the electromagnetic interaction is described
by the Abelian gauge theory, while the strong and weak interactions are by the
non-Abelian gauge theories. Among those gauge bosons shown in (1.3), a photon
7 and gluons g are massless and hence their interaction ranges are infinite, though
actually the argument on the strong interaction does not go so straightly because of
the non-Abelian nature of color interactions and quarks are confined inside hadrons
with the range ~ O(10~ 15 m). On the other hand, the weak bosons W±, Z° are
4 INTRODUCTION

massive and their interactions are very short-range, such as —-— ~ C(10 _ 1 8 m).
The Higgs boson with spin 0 is introduced for the Higgs mechanism to work,
which is operative in the theories with spontaneous symmetry breaking of local
gauge symmetries as will be discussed in Chapter 3. In the Higgs mechanism a
larger symmetry is spontaneously broken into a smaller symmetry through the vac-
uum expectation value of the Higgs field and accordingly (a part of) gauge bosons
become massive. The masses of not only gauge bosons but also all quarks and
leptons are originated from the spontaneous symmetry breaking in the standard
model, though neutrino masses are a little bit controversial. The real understand-
ing of the mechanism of the spontaneous breakdown and the Higgs mechanism is
still extremely challenging problem to be solved in field theories. Obviously, the dis-
covery of the Higgs boson in experiment will provide a crucial hint to the problem.
However, so far there is no evidence of the production of the predicted Higgs boson
in high energy reactions. To uncover the Higgs boson is the most important issue
to finally establish the standard model or even to search for the physics beyond the
standard model, "New Physics". This hopefully can be performed in the early time
of this century.

1.2 Interactions among fundamental particles

It is well known that there are 4 characteristic interactions among fundamental

particles: 1. Electromagnetic interaction mediated by massless photons (m 7 = 0)
with spin=l, 2. Weak interaction mediated by massive weak bosons (mw — 80.4
GeV/c 2 , mz — 91.2 GeV/c 2 ) with spin=l, 3. Strong interaction mediated by
massless gluons (mg = 0) with spin=l and 4. Gravitational interaction mediated
by massless gravitons (mo = 0) with spin=2. (see Table 1.3.)
Among these interactions, the gravitational interaction is usually out of game for
particle physics because it is extremely weak compared with other interactions and
has no meaningful effect on any reactions of those particles (,unless the energies of
interacting particles are extremely high). For example, the ratio of the gravitational
force to the Coulomb (electromagnetic) force between 2 protons at the distance
10~ 15 m is about 1 0 - 3 6 .
The electromagnetic interaction mediated by a photon 7 has a long history of
investigation and now it is known to be described by quantum electrodynamics
(QED) which is the gauge theory having the Abelian U(l) symmetry. QED is
beautifully formulated in the framework of quantum field theory and is renormaliz-
able, i.e. various divergences originated from the loop integrals in the higher orders
of perturbation theory can be renormalized into physical masses and wave func-
tions of particles. Because of smallness of the coupling constant a = f^ ~ j | y , the
perturbation works well for QED.
As will be described in Chapter 2, the theory of weak interactions for weak
processes originally formulated by Fermi, was developed in 1950's and excellently
 Interactions among fundamental particles 5

Interaction coupling strength mediator spin

1. Electromagnetic " — 4TT - 137 photon 1
2. Weak G f S 1.16 x 10~ 5 GeV- 2 weak boson 1
3. Strong as — 4ir - u.i gluon 1
4. Gravitational GN = 6.71 x 10- 3 9 (GeV/c)- 2 graviton 2

Table 1.3 Particle interactions

described by the current-current interaction with V — A currents. It works well

for low energy processes. Unfortunately, the theory is not renormalizable in spite
of its small coupling constant. This is due to the fact that the Fermi coupling GF
has the dimension of [mass] -2 . Thus the Fermi interaction should be regarded as
the effective model for weak processes working only in the low energy region. In
the dedicated study of weak interaction physics in 1960's, many theoretical difficul-
ties in the weak interaction were surmounted. A beautiful renormalizable theory
was finally formulated, based on the unified picture of weak and electromagnetic
interactions, in the framework of non-Abelian gauge theory with SU(2)i, x U(l)y
symmetry (the subscript L means the fields participating in the interaction are
left-handed and Y denotes the weak hypercharge), which is now called the elec-
troweak standard model. One of the main themes of this textbook is to describe
the structure and physics of this model.
The strong interaction is mediated by gluons which have color charges. Since
quarks have also color charges, gluons can couple to quarks. The field theory for the
strong interaction is formulated in the non-Abelian gauge theory with SU(3)C color
symmetry and is called quantum chromodynamics(QCD). The coupling constant of
QCD has conspicuous behavior for a variation of momentum transfer square Q2, as
2

briefly described in Chapter 5. The strong coupling constant as(Q2) = | j "runs"

as Q2 varies. On one hand, as(Q2) becomes small for large Q2 region as realized in
hard scattering such as deep inelastic scattering, where quarks and gluons behave
as free particles, implied by the word "asymptotic-free", and in such regions the
perturbation theory works well. On the other hand, for small Q2 region as realized
in the static state of bound quarks inside hadrons, as(Q2) becomes large and in this
region the perturbative treatment is not reliable, where quarks are confined inside
hadrons (in color singlet states). This is called the "confinement" phase. QCD must
be the theory for describing the dynamics of quarks and gluons in all Q2 regions
from "asymptotic-free" to "confinement" phases.
Weak and electromagnetic interactions are formulated by the gauge theory with
SU(2)L x U(1)Y symmetry and furthermore, the strong interactions are described
by the gauge theory with color SU(3)C symmetry. Hence, one can naturally expect
that all these interactions of elementary particles must be described by the gauge
theory with some internal symmetry G, that is, the Lagrangian has to be invariant
under the gauge transformations of G. The simplest example is to take the symme-
6 INTRODUCTION

try group G to be a direct product of each symmetry, G — SU(3)cx SU(2) £,xf/(1)Y •

The resultant theory is called the standard model of particle physics. The impor-
tant principle in the formulation is that (l)the theory is the gauge theory, (2)it
must be renormalizable and anomaly-free and (3)the symmetry breaking must oc-
cur spontaneously.
Unfortunately, the standard model has many problems to be solved and many
theorists believe that it is not the ultimate theory. To solve these remaining prob-
lems and go further, we must look for evidences for the physics beyond the standard
model, "New Physics". The discussion of the New Physics is another main theme
of this textbook and will be discussed in detail from Chapter 6 to Chapter 9. The
issues discussed in these Chapters are selected from the challenging hot topics of
our current interest.
 Chapter 2
W E A K INTERACTION

The theoretical description of weak interaction processes was first worked out by
Fermi in 1933 for the /3-decay of nuclei (Fermi, 1933). The following two decades saw
the refinement of the explicit form of the current-current interaction first suggested
by Fermi. The discovery of parity violation by Wu et al. in 1957 (following the
theoretical work of Lee and Yang) led Feynman and Gell-Mann and also Marshak
and Sudarshan to the vector minus axial vector V — A structure of the weak current
in 1958 (Feynman and Gell-Mann, 1958; Marshak and Sudarshan, 1958). In 1961,
Gell-Mann and Neeman independently introduced the 517(3) symmetry of strong
interaction (Gell-Mann and Neeman, 1964) and in 1963, Cabibbo (Cabibbo, 1963)
used the hypothesis that the weak currents of hadrons have definite SU(3) transfor-
mation properties to describe the decay of strange particles in the current-current
coupling scheme.
In this Chapter, after touching on the parity violation in weak interaction pro-
cesses, we describe the development of weak interactions following the above histor-
ical chronology. We also discuss some important subjects in the Fermi theory, i.e.
the lepton current universality, pion decays and Cabibbo currents. Unfortunately,
the Fermi theory is not complete. It has serious fundamental difnculies, i.e. unitar-
ity violation and non-renormalizability. These difficulties will be discussed in the
subsequent section. Finally, we will give some discussions on the intermediate weak
boson model, which is an alternative model of the Fermi theory, as the preliminary
discussion of the Standard Model.

2.1 The Fermi Theory of Weak Interaction

The weak interaction among elementary particles was first discovered in the /?-decay
of nuclei. Basically it is realized as a decay of neutron n in the nucleus into a proton
p, an electron e~ and an anti-neutrino 9,

n -> p + e~ + v. (2.1)

7
8 WEAK INTERACTION

The life of a free neutron due to the /?-decay is about 900sec which is extremely
longer than that of other particles induced by strong interaction like r(p° ->
7r+7r-) ~ 10~24sec or electromagnetic interaction like r(7r° -> 27) ~ 10~ 18 sec.
Since the nuclear /^-decays share common nature with other weak processes, they
are classified into weak interaction processes, though the neurton life-time is much
larger than that of typical weak decays like r(A° -> pK~) ~ 10~ 10 sec.

2.1.1 Parity Violation in Weak Interaction

One of the most conspicuous properties of weak interactions is parity violation.
Before 1956 when Lee and Yang (Lee and Yang, 1956) proposed an idea of parity
violation in particle physics, people had considered that all physical processes were
invariant under space inversion and thus parity was a good quantum number for all
interactions. After the experimental discovery of parity violation by Wu et al. in
1957 (Wu et al, 1957), much progress was made for determining the weak interaction
type which was finally established as V - A interactions. The story of this progress
is exciting and will be discussed in the next section. But before coming into this
history, here we would like to touch on the story of discovery of parity violation.
It is not in the /3-decay of neutron but in the K+ decay that Lee and Yang
suggested first an idea of parity violation. K+ meson decays into n+Tr° (Br =
21.1%) and also even into 7r+7r+7r~ (Br = 5.6%), where Br denotes the branching
ratio. Since a 7r meson is pseudoscalar, its intrinsic parity is —1. The spin of the
K meson and •K meson is 0 in either case. Therefore, for 7r+7r° decays, the orbital
angular momentum of 7r+7r° is 0 because of total angular momentum conservation
and thus the parity of the 7r+7r° system becomes + 1 . On the other hand, for the
7r+7r+7r_ system, the total orbital angular momentum of this system is given by the
sum of the orbital angular momentum between two ir+ and the one between the
center of mass of two 7r+ and the remaining ir~. The sum should be 0 because it
must be equal to the spin of the K meson which is 0. This can be realized only
when the magnitude of these two orbital angular momenta are equal. Therefore, the
parity of this system due to the orbital angular momentum becomes + 1 . However,
since the intrinsic parity of n meson is —1, the intrinsic parity of the three pion
system is ( - 1 ) 3 = - 1 and hence the parity of 7r+7r+7r_ becomes —1, after all. Thus,
the parity of 7r+7r° is + 1 , while that of n+n+ir~ is —1.
Then, if parity is conserved in these decays, we must have two independent K
mesons, say 9+ and r + ; one is of even parity and another is of odd parity. In fact,
in those days the decays were understood as 9+ —> ir+ir° and T + —> 7r+7r+7r_. It,
however, seems to be too accidental that these two particles 8+ and T + have the
same mass, the same life-time, the same spin and so on. This problem was known
as the r - 8 puzzle in those days. To solve the puzzle, Lee and Yang suggested
that the parity is violated in weak decay processes. If this is the case for any weak
 The Fermi Theory of Weak Interaction

y*
mirror

60
Co

Fig. 2.1

interaction processes, the effect should appear even in the /?-decay of nucleus. They
proposed to observe the correlation between the spin J of 60 Co and the momentum
pe of elecron produced in the following /3-decay,

Co( Jp = 5+) -» 6 0 Ni(4+) + e~ + v.

60 c (2.2)

First, align the spin J of 60 Co in a +z direction perpendicular to a fixed x — y

plane and then set a mirror in parallel to the x — y plane at some point z (Fig.
2.1). The direction of spin J of 60 Co in the mirror image is the same as that of the
real 60 Co because of axialvector nature of J. Then, when an electron is produced
in a direction, forming an angle 9 relative to J, the electron in the mirror image is
produced in a relative angle 7r — 6 because of vector nature of pe as shown in Fig.
2.1. Therefore, if the process is parity invariant, the probability for finding electrons
produced with the angle 9 and TT — 0 must be the same. Wu and her collaborators
observed that the probability for finding electrons produced with 9 = IT was much
larger than the one for the case of 9 — 0. This was the first discovery of parity
violation.

2.1.2 Road to Current-Current V —A Interaction

In 1933, Fermi proposed an idea that the light leptons e~ and v are emitted from
the neutron n in the /?-decay (Fig. 2.2(a)) like a photon j emitted from radioactive
nuclei, and treated the process based on quantum field theory. In quantum electro-
dynamics, the radiative process p -> p + 7 (Fig.2.2(b)) where p is a proton in the
10 WEAK INTERACTION

Fig. 2.2

nucleus is described by

e j H # = e[upVp]#, (2.3)

where e is the electric charge of proton and up and A^ are the Dirac spinor of
proton and the photon field, respectively. The factor j ^ = u P 7 M u p is called an
electromagetic current of proton. To describe the /?-decay process n —> p + e~ +i>,
Fermi replaced A>* and the electromagnetic current jjf by a 4-vector weak lepton
current jfv_^e\ — u e 7 M u„ made of e~ and v fields and the weak nucleon current
j(p-->p) _ Up^^-Un of p and n fields, respectively, as

Gj(^P) J(^e) = G [fip7M«n] [0e7M«v] , (2.4)

where G is the weak coupling constant being much smaller than the electromagnetic
coupling constant e. (Actually GE2 <C e2 for moderate energies E.) That is, the /?-
decay interaction is described by the product of two weak currents; nucleon current
j)j n-> and lepton current ffv^e\, in which 4 fermions p, n, e~ and v couple at
the same space-time point. There is no propagator connecting these two currents
and this is apparently different from electromagnetic processes where, for instance,
proton current and electron current are mediated by a photon with its propagator.
In addition to a big difference in magnitude of G and e, it is remarkable that the
weak nucleon and lepton currents are charged,* while the electromagnetic currents
are neutral.
Taking account of the parity violation in weak interaction, Fermi's original idea
for the /3-decay n —> p + e~ +9 can be generalized by writing the weak interaction
Hamiltonian for this process as (see, for example, Jauneau, 1977)
G
Hw =
^2~5~ [^P0^] [i'eOiil + CJ7 5 )VV] + h.C, (2.5)

where Cj can be taken to be ± 1 since the parity violation is maximum and actually

"Long later, in 1973 the weak neutral current was also discovered with neutrino elastic scattering
v + e~ —> v + e~ and deep inelastic scattering v + N —> v + X and P + N —>P + X a s predicted
by the Glashow-Weinberg-Salam(GWS) theory.
 The Fermi Theory of Weak Interaction 11

Ci becomes - 1 because the neutrino is left-handed, as discussed later in 2.1.2.3. Gi

is the coupling constant for the type i(= S, V, T, A, P) corresponding to the specific
Lorentz structure of interactions and Oi is given by

Os = 1, Scalar(S')
Ov = In, Vector (V)

Or = ^(7M7,-7,7M)> Tensor(T) (2.6)

OA = 757M> Axialvector(A)
Op = 75. Pseudoscalar(P)

Assuming the time reflection invariance in the /?-decay, we can take Gi to be real.
Then, the remaining task is to determine experimentally which type of interaction
among (2.6) works in the /3-decay.

2.1.2.1 Helicity and chirality

Let us start with the plane wave of a fermion with momentum p, energy E and
mass m. The equation of motion of this particle is given by the Dirac equation, in
the natural unit, h = c — 1,

(a-p + (3m)tp = Eip, (2.7)

where a and /? are 4 x 4 matrices and related to 7-matrices as 7 0 = /3 and 7 = /?a.

For a massless particle with m = 0 as in the case of neutrino*, we have

a- pip = pi[>. (2.8)

In the 7°-diagonal representation, 75 = I I (see (A.25)) and by introducing

. , . ., / ff 0 \ . _ / 0 3 \ _ -. ^.
a 4 x 4 matrix a' = I _ , one can write a = _ as a = 75CT = a 75,
\ 0 aJ \ a 0 J
where 3 is the Pauli matrix. Now using 7! = 1, one can obtain the following
equations from (2.8)
3' • p 3' • p
i> = 75^, ihi> = ip, (2.9)
p p

t Recently, due to observation of neutrino oscillation, it has been strongly suggested that neutrinos
are massive. But in those days neutrinos were considered to be massless and here we treat
massless neutrinos. In any case, neutrino mass has to be extremely small even if it is massive
and it is a good approximation to consider the massless neutrinos.
12 WEAK INTERACTION

Then, by taking addition and subtraction of two equations in (2.9) and introducing
the helicity operator defined by h = (<? • p)/2p, we obtain from (2.9)

htpR = -i>R, hxpL = - - V L , (2.10)

where ipR = PR^ with PR = (1 + 7 5 )/2 and VL = PL^ with PL = (1 - 7 5 ) / 2

and they are called right- and left-handed fermions, respectively. PR and PL are
projection operators for right- and left-handed fermions, respectively. (2.10) shows
that Vfl a n d ipL are eigenstates of helicity with eigenvalues h = + 1 and h = — 1,
respectively*, /i = + 1 means that the particle spin is parallel to the direction of
its momentum p and h — — 1 means that the particle spin is anti-parallel to its
momentum p.
75 is called chirality operator and the following eigenvalue equations hold for
any fermions with or without masses

l5i>R = 4>R, Ib^L - ~i>L- (2.11)

Comparing (2.10) with (2.11) we see that helicity h is the same as chirality 75 for
massless fermions. This is true even in the 75-diagonal representaion. (see Appendix
A.5.)
Now let us consider a free electron with nonzero mass m. We are interested
in the effect of the projection operators PR,L on the Dirac spinor of the electron,
which is the solution of (2.7)

(2.12)

an
where iV is the normalization factor and x = ( n ) d I ) represent that the
electron spin is parallel and anti-parallel to the z-axis, respectively. With (2.12),
we obtain

1+
^>), fc^f/'-^V (2,3)
Now let the electron momentum p be along the z-axis, for simplicity. Then, we can
calculate the expectation value of helicity h^ of electron as

p(e) = <^R(L)\h^\j>R(L)> = ±£i (214)

•* Though according to the definition of helicity operator, the face value of helicity eigenvalue for
fermions with spin | i s / i = ; o r / i = - 1 , the two times value of it is conventinally used and we
follow this convention in this textbook.
 The Fermi Theory of Weak Interaction 13

where the sign +(-) corresponds to IJ)R(L)- As § equals to v (in the natural unit
c = 1), with v being the velocity of electron, P^ approaches to ± 1 for relativistic
electron with v ~ 1. In other word, PR(L) picks up approximately the h^ = +1(—1)
state for a high energy electron(see also Appendix A.5).

2.1.2.2 Observation of electron helicity in the /3-decay

In 1957, Frauenfelder et al. observed the longitudinal polarization of electron P^
in (2.14) in the /3-decay 60 Co —>60 Ni + e~ + P, which was measured to be —v, i.e.
chirality of electron was approximately left-handed (Frauenfelder et al, 1957). There
are several methods to observe the longitudinal polarization of electron. Many
experiments were performed in those days and the conclusion was that electrons
have negative helicity (left-handed) while positrons, i.e. anti-particles of electrons,
have positive helicity (right-handed).
If electrons are left-handed, we can replace Ve —> ^-^-ipe and also 4>e —• ^ ^ - ^
in the Fermi interaction (2.5). Then the lepton current factor xpeOi(l + Cij^ip,,
becoms
- 1 + 75
r/>e0»(l + Cj 75 )VV -> ^e—»—0i(<1 + c»7s)Vv

= •IpeOi — - ( 1 + CiJ5)lp„

= &0<(1=F*)^V, (2-15)
where we used the relation 7^75 = —757^. The sign in the right-hand side of (2.15)
shows that — is for i = V,A and + for i = S, T, P. This equation is very important
in order to determine the type of Oj. If neutrino helicity is measured to be —1, one
can say that the interaction type must be V or A or a combination of them, and Cj
becomes — l(c» = + 1 is not allowed). On the contrary, if it is + 1 , the interaction
type should be S, T, P or a combination of them, and Cj become +l(cf = —1 is not
allowed). Therefore, the next step is to know neutrino helicity which is necessary
to determine the type of weak interactions.

2.1.2.3 Determination of neutrino helicity

In 1958, Goldhaber et al. directly measured neutrino helicity in their exquisite
experiment (Goldhaber et al., 1958). To measure neutrino helicity, they studied
the A'-capture decay of 152 Eu to 152 Sm; 152 Eu first absorbs an electron in its atom,
which is called A'-capture since an electron is absorbed from the if-orbit, and decays
into an excited state of 152 Sm* and a neutrino v, and then finally 152 Sm* comes to
its ground state by emitting a photon 7. After all, the total reaction is as follows;
152
Eu(J p = 0-) + e" -> 152 Sm*(l-) + v -> 152 Sm(0+) + 7 + v. (2.16)
14 WEAK INTERACTION

In this process, the produced neutrino cannot be detected because it is neutral and
has no electromagnetic interaction.
Now, let us consider a photon which runs only in the direction parallel to
152
Sm(0 + ) in the final state. If we take the direction of running neutrino as +z-axis,
we see that only two cases of spin alignment are allowed in this process:
152 152
Eu((T) + e" -> Sm(0+) + 7 + 1 /
(i) 0 -\ 0 - 1 +\ (2.17)
(ii) 0 + | 0 + 1 -|

If /i(") = + 1 , i.e. Jzv' = + | , then Jz = - 1 for the produced photon because

of the angular momentum conservation. It means that the produced photon is
right-circularly polarized, since it is running along — z direction. On the contrary,
if /i(") = — 1, i.e. Ji' = — | , then the produced photon must be left-circularly
polarized. The key point of this experiment was that they determined neutrino
helicity h^ by measuring the photon circular (left- or right-) polarization. The
observed photon was left-circularly polarized, which led to h^v' = — 1. This results
in a = — 1 as seen from (2.15) with maximum parity violation. Prom this result,
the lepton current in (2.5) becomes a V — A type, i.e. ^ 7 ^ ( 1 — T s ) ^ -
From the above observation, we can say that the allowed types of the nucleon
current for /?-decay processes are also Oi = V and A because of Lorentz invariance
of the interaction Hamiltonian (2.5). Historically, it has been well known that in
the nuclear /?-decay there are two types of transitions, i.e. (l)Fermi transition and
(2)Gamow-Teller transition, depending on the spin of the lepton pair e~v which
can be singlet(5 = 0) and t r i p l e t ^ = 1) states, respectively. In the nonrelativistic
approximation which works for ordinally /3-decay processes, one can easily find
that Fermi transitions occur for 0% — S, V, while Gamow-Teller transitions do for
Oi = T, A. Moreover, Oi = P contributes to Fermi transitions only in the order of
•% and thus can be neglected in the nuclear /3-decay. The process discussed here
is the Gamow-Teller transition and hence the above result tell us that Oi = A is
allowed but T is not allowed for the nucleon current of this process.

2.1.2.4 Angular correlation between e+ and v

Another interesting experiment for studying the interaction type of /?-decays was
measurement of angular correlation between e + and v in the decay 35 A —>35 CI +
e + + v which was observed by Allen et al. (Hermannsfeldt et al, 1957). It is known
that this process is Fermi transition (spin of the lepton pair e+u being 0) and only
Oi = V or S contribute to this transition (Kistner at al, 1956). In those days before
the experiment by Goldhaber et al. (1958) which established the neutrino helicity
to be /i(") = — 1, people were interested in which one of V and S is chosen by
Nature.
Let us take the direction of the emitted positron e + to be the positive z-axis and
consider two possible cases; e + and v are emitted (l)in parallel with 6 — 0, where
 The Fermi Theory of Weak Interaction 15

6 is an angle between momenta of e + and v, and (2)in anti-parallel with 6 = •n.

Since the helicity of e + is + 1 , the angular momentum of e + is Jz — + | . Then, if
the helicity of neutrino is negative, i.e. h^ = — 1, we can expect, from (2.15) and
the angular momentum conservation, that the case (1) is possible only for Oi = V
but is not allowed for S and the case (2) does not occur for the Fermi transition
with Oi = V. That is to say, it is forbidden for e+ and v to be emitted in parallel if
Oi = 5. On the contrary, if h^ = + 1 , the case (2) can occur and this time Oi = S
is possible and V is not allowed from (2.15). Allen et al. observed that the case (1)
was realized.
From those many experimental observations, it was concluded that the weak
interaction of the /?-decay is given by a combination of V and A as follows;
/-<
Hw =
~2~ [VV7M^I] [$e7 M (l - 7s)Vv]

+ -y [i>plslM [^e7M(l - 7s)Vv] + h.c. (2.18)

2.1.2.5 V — A interaction
Experimentally, Gy was determined from the decay 1 6 O(0 + ) -» 1 4 N(0 + ) + e + + v.
In this process, the spin of initial and final nucleons does not change and the spin
sum of e + and v is 0. Thus, the process is Fermi transition and hence only Gy in
(2.18) can contribute to this decay and then from this decay width one can obtain
the value of Gy; when we define Gp = Gy/y/2, we obtained

Gp = 1.147 x K T 5 G e V - 2 « l(T 5 /™p, (2.19)

where mp is the mass of proton being the order of 1 GeV. Using this value, GA can be
obtained, in principle, from the /?-decay of neutron, where the Fermi transition and
the Gamow-Teller transition coexist. There was a long history of this measurement
and now we have gA = \GA/GV\ — 1.26. Unfortunately, the relative sign of Gy
and GA, cannot be determined from unpolarized experiments. To determine it, one
must study the polarized experiment.
Now, let us consider the /?-decay of neutron whose spin is polarized to the +z-
axis and let electron and anti-neutrino momenta be given by pe and pp, respectively.
Since the mass difference between neutron and proton is tiny, the produced proton's
momentum is very small and thus, it is a good approximation to treat the nucleon
current nonrelativistically. First consider the first term of (2.18), i.e. the Fermi
transition term. Using the nucleon wave function which is of the same form as (2.12)
with almost equal mass mp w mn « m for proton and neutron, we see that 'ipp'yoipn
is of 0(1), while 'ippjipn is of O(^) in the nonrelativistic approximation. Thus,
we can neglect the contribution of 'ipp^ipn and the remaining term is only "tpp^o^Pn
which can be written as xPXn in the nonrelativistic approximation using the Pauli
spinor Xp(n) °f proton(neutron). It has no spin flip operator a and therefore, in
this case only two patterns of the spin and momentum arrangement for particles
16 WEAK INTERACTION

it it H + a- it it + ij. + it
+ I
n P e-
(a) (b)

it —• a- + it + it
n p

(c)

Fig. 2.3 Spin arrangement in the neutron /?-decay. Long arrows show the momenta of e~ and Pe

participating in this /3-decay can be allowed as shown in Fig. 2.3(a) and (b). Next,
let us consider the second term of (2.18), i.e. the Gamow-Teller transition term.
This time the same wave function of nucleons leads to that 1/^775V"n is of 0(1), while
V'P7o75'0n i s of 0(%)- Thus, we can neglect V>p7o75Vv»- Only ^,775 VVi contributes
to this transition. ^775^/>n results in x\<?Xn in the nonrelativistic approximation.
ForCT3term in xl^Xn, the spin and momentum arrangement of particles are same
as shown in Fig. 2.3(a) and (b). However, o\ and 02 terms contribute to spin flip
and thus we have the spin and momentum arrangement as shown in Fig. 2.3(c).
After all, in the nonrelativistic approximation we have the following interaction
Hamiltonian for the /?-decay of neutron.

Hw - Gv [xlXn] [vtlv]
+ GA [xl^Xn] [vtffriv] + h.c, (2.20)
where ^ 2 L t/'i/ i n (2-18) is replaced by r\v being the two-component neutrino wave
function with helicity h^ — — 1 and r]e is also the two-component Weyl spinor
of electron with helicity h^ = — 1 (see (A.53) for the notation of two-component
Weyl spinors). Now, let us calculate the matrix element for the process presented
in Fig. 2.3(c) where the neutron spin flips. The spin flip is originated from the
following terms in (2.20),

= 2 [x^+Xn] [vla-riv] + 2 [xPcr-Xn] [vt<r+riv] , (2.21)

where a± — (a\ ± icr 2 )/2 and actually only the 2nd term of (2.21) can contributes
to the spin arrangement presented in Fig. 2.3(c). We can parametrize this matrix
element as

M ( c ) = 2GAF, (2.22)
 The Fermi Theory of Weak Interaction 17

where F is a factor containing all dynamics of this decay process except for the
coupling constant GA-
Next, let us consider the case of Fig. 2.3(a). This time, the 1st term of (2.20)
and the 173 term in the 2nd term of (2.20) contribute. In Fig. 2.3(a), the electron
spin being in parallel to the neutron spin is along the +z-axis. Hence 03 works as
+ 1 in the matrix element and we can write the matrix element for this process as

M ( a ) = (Gv + GA) F. (2.23)

For the case of Fig. 2.3(b), since the electron spin is along the —z-axis, we obtain
the following matrix element,

Mw = (Gv - GA) F. (2.24)

Then, we can calculate the probability for an electron to be emitted in parallel to

the neutron spin as shown in Fig. 2.3(b), as

PtftPe t ) = \M(b)\2 = \GV - GA\2\F\2. (2.25)

To the contrary, the probability for an electron being emitted in anti-parallel to the
neutron spin, which is given by the sum of Fig. 2.3(a) and (c), can be calculated as

P{31& I) = \M^\2 + |MW| 2 = \GV + GA\2\F\2 + 4\GA\2\F\2. (2.26)

If Gv « GA, then P(S t Pe t ) w 0> which means that an electron cannot be emitted
in parallel to the neutron spin. If Gv ~ -GA, then P(a \ pe f) « -P(<? t Pe I),
which means that the number of electrons being emitted in parallel and anti-parallel
to the neutron spin is almost same. The latter case was observed in experiment
and then it was finally established that the weak interaction Hamiltonian for the
/?-decay was described by the V — A current-current interactions as summarized as
follow;

Hw = n= $j»7„(l - 9A-Y6)il>n] [ & V ( 1 - Tfe)^] + h.c, (2.27)

where the Fermi constant Gp is given by (2.19) and gA — -GA/GV — 1-26.

2.1.3 Lepton Current Universality

The discovery of muon neutrino as a new lepton distinct from an electron neutrino
led to the concept of fi — e universality. A muon decays into an electron as n~ —¥
e~ + ve + v^ with its life-time rM = 2.2 x 10 _6 sec. At first sight, pT could decay
electromagnetically into e~ by emitting a photon as /x~ —> e~ + 7, because this
process is kinematically allowed. However, there has been no observation for this
18 WEAK INTERACTION

process so far.§ Existence of the particular decay mode [i~ —> e~ + i>e + v^ and
non-existence of the decay mode fi" —> e~ + 7 led that in Nature there are additive
conserved lepton numbers: the electron number Le (Le = +1 for e~~, ve and Le = — 1
for e + , 9e) and the muon number LM (LM = +1 for [i~, i/M and LM = - 1 for / J + ,
PM). e~ interacts with ue alone and ^~ interacts with v^ alone. Later, the third
lepton members, the tau r and its neutrino vT (tau neutrino) were also dicsovered
with additive conserved tau number LT (LT = + 1 for r~, uT and LT = — 1 for T+,
vT). Now we have three families of leptons as

The universality of weak interactions is expressed by writing the total weak current
of leptons as a sum of an electron, a muon and a tau current with equal weight as,

= i>e7 M (l-75)e + i/ M 7 M (l-7 5 )/i + £ ; r7M( 1 -7S)T, (2.29)

which can be rewritten as

JW = (ue 9» P r ) 7 M (l - TS)V ( /* ] , (2.30)

where the wave function of each particle is written by the symbol representing the
particle itself and V is a 3 x 3 diagonal matrix given as follows;

(2.31)

The explicit form of V shows the lepton universality.

Now let us concentrate on the muon decay jU~(pM) -» e~{pe) + De{kVe)JtvIJi{kUix),
where momentum of each particle is given in parenthesis. The decay amplitude is
given by

M = ^ [ « ( * „ > , ( ! - 7S)«(P/.)] [«(p e )V(l - 7 6 ) « ( ^ . ) ] , (2-32)

where u and t; denote the spinors for particles and anti-particles, respectively. The
weak coupling constant GM can be determined from measured value of the life-time
of muon. Since the muon decay involves no complication due to hadronic currents,
we can accurately determine the value of GM.

§Only upper bound of the branching ratio for this decay mode, being extremely small, is known:
Br(n -» e + 7) < 1.2 x 1 C T n .
 The Fermi Theory of Weak Interaction 19

The decay rate (see (B.15) is given by

1
dT=—\M\2dR, (2.33)

where dR is the Lorentz invariant phase space (see (B.14) given by

dR (2 )4j(4)(
=m ( ^ (2^X - ^ -» - ^ - **•>
X 2
- ^ ^ ^ ( ^ - ^ - ^ J ^ - p e - A ^ ) ) - (2.34)
(2TT)5 2 £ e 2w„.

In (2.34), 0 is the step function, 6(x) = 1 for x > 0 and = 0 for x < 0. \M\2
denotes the probability of spin-averaged for initial particles and spin-summed for
final particles,

spin
1 G2
= 2~2~ ^ N M 7 P ( 1 ~ Tto)u(ft»)u(P/*)7<r(l - 7te)«(*^)]
spin

X £ [S(Pe)7 P (l - 75)^(^ e )D(^j7 < T (l - 7fc)«(Pe)]

spin

= -fTr[^7p(l-75)(^-^)7„(l-75)]
x Tr^7"(l-75)^7'(l-75)]
= 64G2(Av.-pM)(^-pe), (2-35)

where the electron mass was neglected because it is very small as me < m M /200. In
the rest frame of muon p = (mM, 0,0,0), we have {kUt -p^ik,,^ -pe) = \u)Vi,m^rri^ —
2mliuje) and thus we can write the decay rate as

x 6(ml - 2rnflEe - 2 m ^ „ , + 2EecjVf(l - cos/9)), (2.36)

where 6 is an angle between produced electron and anti-neutrino. Using the relation,
dzped3kVe = 4-!rE2dEe •2-noj2tdiJjVedcos6', and integrating over cos# using 6 function,
we obtain
G2
dT = -ArdEedujUcuVc{ml - 2m M w,J. (2.37)

Prom the condition — 1 < cos# < 1, the energy u>„e and Ee are constarined as
m M /2 — Ee < OJVC < m M /2 and 0 < Ee < m M /2, respectively. Then, the energy
20 WEAK INTERACTION

spectrum of electron is obtained by integrating wVc over this region

dT _ G\mlEl /, 4Ee
(2.38)
dEe 47r3 V 3 m,,
The total muon decay rate is obtained by integrating by Ee

1 Gimi [*•* , / 4 Ee \ Giml

where the maximum of electron energy is E™ax = m M /2. Using TM = 2.2 x 10 6 sec
and mM = 105 MeV, we get

G p = 1.166 x 10- 5 GeV - 2 » W~5/m2p. (2.40)

Comparing (2.40) with (2.19), we see that GM for muon decays is almost equal to Gp
for nuclear /3-decays and thus the weak coupling constant looks universal. Therefore,
the origin of the weak interaction seems to be same. In fact, by taking account of
radiative corrections (Roos and Sirlin 1971; Beg and Sirlin, 1982), we see that the
values of GM agrees with the one of G$ within only about 2%. However, it is very
important to notice this small difference between couplings in lepton currents and
hadron currents, which led Cabibbo to the splended idea of the so-called Cabbibo
mixing currents discussed later.
Before coming to the discussion of Cabibbo currents, we just consider the decay
of the third lepton r. The decay width of r is similarly calculated as in the muon
case discussed above. This can be done by just replacing mM and GM by mT and
G r , respectively. Then, using the relation
GM a mM 5Br(T~ ^e-9evT)
T T M } { (
- G T mT> B r V - > e - P e I V ) ' '
and the experimental data of mT = 1776.99 MeV, mM = 105.658357MeV, rM =
2.19703 xl0- 6 sec, r r = 2.906 xl0- 1 3 sec, BT(T~ ->• e~vevr) = 0.1784 and Br{n~ ->
e~~veVn) = 100%, we can obtain Gr/Gf,. = 1.001. The result confirms again the
universality of the weak interaction.
In summary, the near equality of coupling constant involved in the weak leptonic
decay of fi, r leptons and in the /3-decay, i.e. GM « GT w G/3 shows that there is
a universal weak interaction constant G in all weak processes. In other words, the
value of weak charge is universal and therefore, it is customary to represent this
universal coupling constant G by Gp called the Fermi constant.

2.1.4 Pion decays

Can we understand the meson decays, too, by the V — A interaction? the answer
is YES. We can explain them also by the V — A interaction. Let us consider the
 The Fermi Theory of Weak Interaction 21

pion decay as an example. A pion can decay into a muon(Br « 100%) or an

electron(5r = 1.230 x 10~4) as

ir~ -» H'+Vf,, (2.42)

-> e~ + Pe. (2.43)

For these decays, there are two remarkable experimental findings; (l)the observed
li~ in the process (2.42) was right-handed. This observation results in the conclusion
that the v^ is also right-handed, since in the rest frame of ir~~ being spinless, \T
and Up run into opposite direction with equal momentum and because of angular
momentum conservation, two decay particles must be right-handed. This result is
cosisitent with the (1 — 75) factor in the lepton current. (2)The observed ratio of
branching ratios of (2.43) to (2.42) is extremely small, i.e.

r ,r- -> e- + v )
( e w x lo_4
r(7r- -> n~ + i/M)
This is completely contradictory to the prediction of the phase space effect. The
origin of the strong suppression must be dynamical. Actually, this is due to the
V — A structure of the lepton current. In fact, since ve is right-handed, the emitted
electron must be also right-handed because of angular momentum conservation. But
this is the "wrong" helicity state for light electron, because a fast moving electron
should be left-handed as described in 2.1.2. This is not the case for a muon; a muon
emitted in the pion decay is rather heavy and has the right-handed component, i.e.
the "right" helicity state.
Let us go to a quantitative discussion. Assuming the V — A interaction, the
matrix element for these decays ir~(q) —>• £~(p) + vt(k) where I — fj, or e and
momenta of TT~, £~, Di are given in parentheses, is written by

M=
71 < 0|J*~|7r~(<?) > ^ f r ) ^ 1 - -*)""«(*)' ( 2 - 45 )
where < 0| J* \ir~(q) > contains all strong interaction effects, being called the pion
decay constant /„•, expressing the effect of the pion decay into vacuum, and must
be vector or axialvector so that this factor can contract with the lepton current to
make M be Lorentz scalar. Since a ir~ is spinless, available 4-vector is q alone.
Then we can write

<0\Jf\7r-(q)>=qiifn. (2.46)

Substituting (2.46) into (2.45) and using q = p+ k, we can write the amplitude as
Q
M = - ^ f e +W-[^(P)7 M (1-75K,(A;)],
Q
= -7^m,ifvui[p){\--f5)vvt{k), (2.47)
22 WEAK INTERACTION

where in going from the 1st line to the 2nd line, we used the Dirac equation for a
lepton £ with mass me and the one for a neutrino vt with zero mass. Then, using
the standard technique taken in the calculation of the muon decay width described
in the previous section, we obtain the result

r(7r--*r+P,) =^ / > w m ? ( l - ^ ) • (2-48)

Then, we have the ratio

Y{,-^e-+Ve) _fmeYfml-ml\2^128xl0_^ ( J M g )
r(7i— ->• fi- + i/p) V m M/ Vmf-mM

which fantasitcally agrees with the experimental result of (2.44). Furthermore, from
the observed life-time of TT - , T = 2.6033 x 10 _8 sec, we can estimate the value of the
decay constant f„ as

U ~ 0.92m,,. (2.50)

which is an order of the pion mass. v

Above observation suggests us that the V — A interaction works well even for
meson decays.

2.1.5 Cabibbo Currents

The universal Fermi interaction of the V — A current-current form is quite successful
in describing observed leptonic weak decays. However, the situation is expected to
be less clear in the treatment of semileptonic and hadronic weak decays because
of hadronic strong interaction effects. There are many such weak decay transitions
of hadrons including strangeness. Experiments show that the strangeness non-
conserving (AS ^ 0) weak decays are relatively suppressed as compared to the
strangeness conserving (AS = 0) weak decays.
A modification of the weak current that explains the observed suppression of
AS = 1 transitions compared to AS = 0 transitions, was proposed by Cabibbo
in 1963 (Cabibbo, 1963). Instead of introducing new couplings to accommodate
strange particle decays, he tried to keep universality by modifying the hadronic
current. He assumed that the total weak current of hadrons flows into the AS = 0
and AS = 1 branches, keeping the total weak current of hadrons

jM=aJ^+bj£\ (2.51)

where Jj?' and J^ 1 ' are currents corresponding to AS = 0 and AS = 1 transitions,

respectively. J° and J^1' are normalized so that the strength of the corresponding
transitions are given by the coefficients a and b. Assuming that the total weak
 The Fermi Theory of Weak Interaction 23

hadronic current remains unchanged, we have

|a| 2 + |6| 2 = l. (2.52)

It is customary to write a = cos# c and b = sin# c , where 9C is called Cabibbo angle.

The normalization condition (2.52) is then automatically satisfied and the weak
current of hadrons is written by a mixing of the strangeness conserving (AS = 0)
current and the strangeness changing (AS = 1) current

J<fc> = cos ec j(°> + sin 9C jM. (2.53)

To find the value of the Cabibbo angle 0C, we compare the decay width of n~ ->
fi~ + Vy, and K~ -> \iT + v^. The decay width for TT~ ->• \i~ + v^ is given by (2.48)
in which G is replaced by cos6cG by taking account of the Cabibbo mixing effect
presented in (2.53). The decay width for K~ -> fi~ + v^. is also given by replacing
mn, f„ and G by TTIK, JK and sin# c G, respectively. Then, we have

m
T(K~ ->/i- + v») singg / J ^ C1 ~ ^ )

Assuming the 51/(3) symmetry for meson decay constants of fx = fn and using
the experimental result(« 1.335) for the lhs of (2.54) and the values of mM, mn,
TTIK presented in the particle data table (Particle Data Group, 2002), we obtain

t a n 0 c ~ 0.275, (2.55)
which leads to sin# c « 0.26. More detailed analyses were carried out with the
results, sin0 c = 0.220 ± 0.002 from K~ ->• TT° + e~ + 9e (Braun et al., 1975) and
sin^ c = 0.231 ± 0.003 from A ->• p + e~ + ve (Bourquin et al., 1983). Because of
the small values of 9C, those decays whose amplitudes are proportional to cos# c ,
are known as Cabibbo-favored decays, while those with amplitude proportional to
sin0 c are Cabibbo-suppressed decays.
Let us now understand Cabibbo's proposal at the quark level. The /3 decay
process n —> pe~ue at a quark level is d —> ue~ve (Fig.2-4 (a)): one of the d quarks
in the neutron (ddu) transforms into a u quark with remaining d and u quarks as
spectators. In contrast, in the decay process A0 —> pe~ue in which A0 has quark
content uds, the strange quark in A0 transforms into a u quark (Fig.2-4 (b)). Again
this involves charge changing weak currents but in this case there is also a change of
strangeness at the baryon vertex and hadronic current is therefore called strangeness
changing or AS = 1 weak current. The quark current has the same V — A structure

Jjr9,)= <T7,.(l-7fc)9. (2-56)

Furthermore, we retain the universality for all particles such as quarks as well as
leptons and use the same coupling constant for A S = 1 and AS — 0 decays, except
for the Cabibbo angle. In weak interactions with charged curents, leptons can only
24 WEAK INTERACTION

Gcos6,

AS = 1

(b)

Fig. 2.4

be transformed into their partners in the same doublet; e.g. e «-» ve for

and (j, f> i/,, for )• Similarly, we can group the quarks into families like

But for quark transitions, they are observed not only within a
family but, to a lesser degree, from one family to another via Cabibbo mixing.
Therefore, for charged currents of quarks, the "partner" of the flavor eigenstate u
is not just the flavor eigenstate d but a linear combination of d and s, which is
called d'. Similarly, the "partner" of the c quark is another linear combination of
d and s which is orthogonal to d' and is called s'. The coefficients of these linear
combinations can be written by using a single mixing parameter 6C as

d' = cos8c d + sin9c s,

s' = —sin6cd + cos9cs. (2.57)

(2.57) can be written in the matrix form as

cos 0C sin 6C
U with U= (2.58)
— sin6r cos6r
 The Fermi Theory of Weak Interaction 25

U is unitary and performs the rotation of quark states, I 1 and I 1, into

"rotated" states, ( I and I , 1. The transitions s f+ u as compared t o d f > u

are therefore suppressed by a factor of sin2 6C : cos2 6C « 1 : 20, in accordance with

the data.
If we add the third generation of quarks I I, the 2 x 2 matrix of (2.58) is re-
placed by a 3 x 3 matrix which was originally introduced by Kobayashi and Maskawa
in 1973 (Kobayashi and Maskawa, 1973) and is called the Cabibbo-Kobayashi-
Maskawa matrix (CKM matrix),

d' \ I Uud Uus Uub\ ( d\

s' = Ucd Ucs Uch ]( s . (2.59)
V J V Utd Ut. Ua J \b J
The probability for a transitions from a quark q to another quark q' is proportional
to |f/99'|2, the square of the matrix element. The diagonal elements of this matrix
describe transitions within a family; they deviate from unity by only a few percent.
It was found from the analysis of various meson decay data that the values of matrix
elements Uus and Ucd were nearly one order of magnitude smaller than those of Uud
and Ucs. Furthermore, transitions from the third to the second generation (t -> s,
b —> c) are further suppressed compared to transitions from the second to the first
generation. The average values of \Uud\, \UUS\, \Ucd\ and \UCS\ obtained from the
analysis of experimental data (Particle Data Group, 2002) are

\Uud\ = 0.9735 ± 0.00108, |17„,| = 0.2196 ± 0.0023,

\Ued\ = 0.224 ±0.016, \UC3\ = 1.04 ±0.16. (2.60)

The observation that strangeness non-conserving weak interactions are relatively

suppressed as compared to the strangeness conserving weak interactions led to fur-
ther extension of the concept of universality to involve weak hadronic currents.
After all, the Cabibbo theory established the quark-letpon universality and sloved
the puzzle on the slight discrepancy between the value of Gp (in (2.19)) determined
from the nuclear /3-decay and the one of GM (in (2.40)) from the /j-decay; in fact,
Gp = G,j, cos0c with cos# c » 0.975.

2.1.6 Difficulties in the Fermi Theory

As we have seen in the previous sections, the Fermi theory of weak interactions,
whose definitive form was established by Feynman and Gell-Mann as the current-
current form with V - A currents and which was extended by Cabibbo to processes
including hadrons such as A 5 = 0 and AS = 1 decays, works very well in describing
26 WEAK INTERACTION

Fig. 2.5

phenomena of the observed charged current weak interactions. The Fermi theory of
its simple and elegant form could not only explain many experimental data of decays
but also matches well with the physical principle such as universality, algebraic
properties embodied in Cabibbo currents, and so on.
However, it is now also well known that the Fermi theory cannot be a complete
theory, even if it works well as phenomenology. When we apply it to scattering
processes, the lowest order approximation violates the unitarity bound. Further-
more, unlike the quantum electrodynamics for electromagnetic interactions, the
Fermi interaction is not a renormalizable theory and thus we cannot manage those
higher-order contributions to yield a finite outcome.

2.1.7 Unitarity violation

To see how the Fermi interaction violates the unitarity bound, let us consider the
neutrino-electron scattering (see Fig. 2.5)

ve[k) + e"(p) -)• e-(p') + ue(k'), (2.61)

whose amplitude is given by

M = ^ | [u„e(A')7,.(l - 7*)«e(p)] M P V U - T f c K . W ] , (2-62)

from which, by neglecting the electron mass me and following the standard technique
for calculating cross sections, we can obtain the differential cross section as

dQ 64?r2s 4TT2 V ;

where s = (k+p)2 — 4E2 = 4fc2 (E(k) is the CMS energy (momentum) of the ve+e~
system). Integrating over the solid angle fi, we obtain the total cross section for
this process

a = ^£Ls. (2.64)
7T
 The Fermi Theory of Weak Interaction 27

Fig. 2.6

On the other hand, as is well known, every scattering cross section can be decom-
posed into partial waves. For spinless particles, we have in general

1 oo
da 2 -^(U+VftPticosO) (2.65)
dQ, = \f(0)\ 2ik
1=0

where fi is the partial wave amplitude for an orbital angular momentum I. For
the pointlike Fermi interaction, only /o contributes to ve — e~ scattering (2.61) and
there is no angular dependence (with P0(cos9) = 1). Then, we expect to have

da 1
AE2 I/of-
(2.66)
dn
However, since unitarity requires \fi\ < 1 for every partial wave, we obtain the
upper bound for the diffrential cros section as

da_ _ 1 _
(2.67)
dn ~ AE2'
Then, the total cross section is also bounded similarly as

da da -K
—raSJ = ATT—T < —2r . (2.68)
' • / dCl dQ ~ E

Therefore, the prediction of (2.64), being in the lowest order of the Fermi interaction,
violates unitarity bound at some high energy

E>- 4=A/^«370GeV. (2.69)

y/2\/ GF

One might suppose that this is due to neglection of higher-order corrections of

Fermi interactions. But this is not the case. For example, if we take into account
the 2nd order diagram shown in Fig. 2.6, we will find that the calculation leads to
a divergent result. We cannot cure this disease from those higher order corrections.
Rather, we can see that the root lies in the Fermi interaction itself.
28 WE A K INTER A CTION

Fig. 2.7

2.1.8 Non-renormalizability
Another serious difficulty in the Fermi theory is that it is not a renormalizable
theory, which is very different from the case of quantum electrodynamics(QED),
being renormalizable. As is well known, we have divergent integrals originating
from higher-order diagrams even in QED as well. However, in QED we can remove
those divergences by renormalizing the mass and charge into the observed physical
values. On the other hand, in the case of the Fermi interaction, if we try to follow
the renormalization procedure analogously as in QED and go to the next higher
order diagram, we will encounter a more serious divergence, which needs further
renormalization constants. Then, when we consider all higher order diagrams, we
need an infinite set of renormalization constants. That is, the theory cannot be
renormalized. (see, for example, Aitchison and Hey, 1989)
By taking the reaction ue + e - -» ve + e~, let us see what is going on. The
lowest order diagram for this reaction is given in Fig. 2.5, which contributes to
the amplitude with the order of GF- The 2nd order correction comes from Fig.
2.6. There we have two fermion propagators, each one behaves as 1/k for large
internal fermion momentum fc, and thus due to these internal fermion propagators
we have a loop integral contribution to the amplitude with GF f^r, which is di-
vergent. Similarly, the 3rd order diagram (Fig. 2.7) contributes to the amplitude
with G3F(f^r)2, which is more seriously divergent. After all, we have a series of
expansion with GF f^r> where the higher order terms yield more and more severe
divergence. This is because GF has a dimension of [mass] - 2 ; each term has a new
GF in the expansion series, which must be compensated by a divergent factor / ^ r
with [mass]2 dimension in each term to keep the dimension of the amplidude.

2.2 Intermediate Weak Boson Model

One of the interesting ideas to attempt to take away those difficulties is the in-
termediate weak boson exchange model. The weak coupling constant G(= GF)
appearing, for instance, in (2.27), (2.32) and (2.62) has a dimension of [mass] -2
(see (2.40)). This fact suggest us that we can develop further an analogy between
the weak interaction and the electromagnetic interaction by postulating that the
 Intermediate Weak Boson Model

Fig. 2.8

weak interaction is mediated by a massive weak boson, just as the electromagnetic

interaction is mediated by a massless photon. For example, assuming the W boson
exchange between two charged lepton currents (Fig. 2.8), let us write the decay
amplitude for fi~ —> e~ + ue + v^ as

1-75 9 _ 1-75
Af = - >1P—IT- u
u (2.70)
q2~Mw LV2 U e 7 f f U
W 2 "
instead of (2.32), where g is the dimensionless coupling constant and q represents
the momentum of an intermediate W boson with mass Mw- This form is quite
analogous to the electron-proton scattering amplitude mediated by a massless pho-
ton

M - - [eiipj^Up] — [-euej^Ue]. (2.71)

We see that a photon propagator between two electromagnetic currents with charge
e is replaced by a W-boson propagator between two charged currents with coupling
constant A=. Since the Fermi interaction works as a contact interaction at zero
range, the W boson must be very heavy, unlike a massless photon. Furthermore,
at low energy where q2 is much smaller than M ^ , the Fermi interaction works very
well and thus (2.70) should reduce to (2.32) with

GF
(2.72)
v/2 8M£/
From (2.72), the fact that the weak interaction is weak with small GF is considered
to be not because the coupling g is small, but because Mw is large with g w e. This
suggests that the electromagnetic and weak interaction might be formulated into a
unified theory. In fact, later this idea was beautifully realized in the non-Abelian
Gauge theory called Glashow-Weinberg-Salam(GWS) theory, which is the one of
main subjects of this textbook and will be discussed in detail in the next Chapters.
But before coming into the GWS theory right now, it is instructive to learn a few
things about the intermediate weak boson model. One is related to the problem
30 WEAK INTERACTION

W+

Fig. 2.9

of unitarity bound. As discussed above, the Fermi interaction, i.e. the contact
interaction leads to violation of unitarity bound. Then, can the intermediate weak
boson model cure this disease? The answer is NO. For example, let us consider the
process, in the lowest order, (Fig. 2.9)

MP) + V»{P') -> W+(k,X) + W-(k',X'), (2.73)

which is predicted to exist in the intermediate weak boson model. In (2.73), the
momentum of each particle is given in parentheses, and A and A' denote the polar-
ization of W+ and W~, respectively. The invariant amplitude for this process is
given by

1 - 7s ($-# + mM) „l-75

M<A'V e
9 \2^*(A')
(k')e<x\k)v(j/)r
u

W> 2 (p-ky-ml 1 7
2 'MP),
(2.74)
where e^x\k) and e'A \k') are the polarization vector of W+ and W~, respectively.
From this amplitude, one can calculate the differential cross section at high energy
as
da U
F „;„2 QT?2
(2.75)
dtt 8^Sin 9E
'
where E and 6 are the energy and the scattering angle of the incoming u^ in the
CMS, respectively. Integration over the solid angle results in the total cross section

a = I2n" (2.76)

where s = AE2. The cross section increases with s and hence violates the unitarity
bound at some high energy. Therefore, the disease of unitarity violation cannot be
cured even in the intermediate weak boson model.
Another difficulty, i.e. the non-renormalizability of the Fermi interaction also
cannot be solved by the intermediate weak boson model. To see this apparently, let
us consider an example of scattering presented in Fig. 2.10. In this figure, (a) is the
 Intermediate Weak Boson Model 31

(a) (b)

Fig. 2.10

two photon exchange correction to the QED process, e + + e~ —> e+ + e~, while (b)
is the two W boson exchange correction to the weak process, ve + e~ —» ve + e~.
For the photon exchange diagram (a), since the photon propagator behaves as 1/k2
for large internal momentum k, we have a loop integral J^r for this amplitude,
which is finite. On the other hand, for the W boson exchange diagram (b), since
the W boson is massive and its propagator has a longitudinal component
1 i fZ^Kv.
(2.77)

in which k dependence vanishes at large k, the corresponding loop integral becomes

fjjr> which is quadratically divergent and the situation does not change from the
case of the Fermi interaction, i.e. the contact interaction. After all, the intermediate
weak boson model cannot be renormalized to get a finite result.
After a long struggle, an extremely elegant and consistent theory without theo-
retical dificulties of unitarity violation and non-renormalizability was discovered for
the lepton family by Weinberg and Salam in 1967 (Weinberg, 1967; Salam, 1968).
The theory was extented to quark families successfully and, combined with quantum
chromodynamics(QCD) which is the fundamental field theory of strong interactions,
was developed as the standard model of the particle physics. But before going into
the Glashow-Weiberg-Salam(GWS) theory, we need some preparations, which will
be discussed in the next Chapter.

Problems
2.1 Show that the coupling constant d in (2.5) must be real if the interaction
Hamiltoniam (2.5) for the /?-decay is time-reflection invariant.
32 WEAK INTERACTION

2.2 Using (2.13), prove (2.14).

2.3 In the nonrelativistic approximation, one can see that the Fermi transition for
/?-decay processes occurs for O, = V and A, and the Gamow-Teller transition does
for Oi = A and T, while Oi — P can contribute to the Fermi transition only in the
order of ^ and thus can be neglected. Show these results using the nonrelativistic
expression for the Dirac spinor of nucleons.

2.4 Starting with (2.45), derive (2.48).

2.5 Stating with (2.62), drive (2.63).

2.6 Starting with (2.74), drive (2.75).

 Chapter 3
SYMMETRIES AND T H E GAUGE
THEORIES

One of the most fundamental principles in particle physics is that interactions among
fundamental particles are described by symmetry principles. The invariance of the
Lagrangian under certain symmetry transformations leads to a set of conservation
laws. In addition to space-time symmetries such as Lorentz invariance, parity invari-
ance, time reversal invariance etc., the internal symmetries such as isospin, flavor,
color etc. have been also known. The development of our understanding about
the fundamental interactions is based, to a large extent, on our understanding of
such underlying symmetries of Nature. It is remarkable to know that there is a
connection between exact symmetries and conservation laws and to see that the
requirement of local gauge invariance can serve as a dynamical principle to guide
the construction of interacting field theories.
In this Chapter, we first discuss the relation of the global symmetry to the
Noether's theorem, which connects the invariance of the Lagrangian under a con-
tinuous symmetry transformation to a conservation law and the conserved quantum
numbers. Then the local gauge theory will be developed with an example of quan-
tum electrodymanics(QED), being an Abelian gauge theory with U(l) symmetry.
A non-Abelian Yang-Mills gauge theory is also discussed with an example of SU(2)
symmetry. Furthermore, an idea and an important result of the spontaneous sym-
metry breaking(SSB) and the Goldstone theorem will be discussed. Finally, we will
discuss the Higgs mechanism, which is realized in the mechanism of SSB with local
gauge symmetries.

3.1 Global Symmetries and Noether's Theorem

In the field theory, fields are used to describe the fundamental particles. Dynamics
among those particles are given by the Lagrangian density* C((p(x),d,j.(f>(x)) which
depends on the field <f>{x) and its derivative d^^x). Fields are arranged so as to
realize various symmetries mentioned above. The Lagrangian formalism provides

"The Lagrangian density is often called, simply, Lagrangian. We follow the same usage.

33
34 SYMMETRIES AND THE GAUGE THEORIES

a systematic way of identifying these symmetries and extracting the constants of

motion associated with these symmetries.
Let us consider a system composed of a set of n independent fields* (j>a (x) (a =
1,2, •••,n) and let the Lagrangian of this system be invariant under the trans-
formation of a certain group G. The equation of motion for <pa(x), called the
Euler-Lagrange equation,

is generated from the requirement of least action,

6S = 8 f d4x£(4>0(x), dMx)) = 0. (3.2)

We define the "momentum" fields which are canonically conjugate to field variables
(f>a(x) as

na(x) = , (3.3)
0{Oo<Pa)
and assume finally the canonical commutation relations to quantize the system

[<t>a{x,t),Mv,t)} = [^a(x,t),n(y,t)) = 0, (3.4)

[4>a(x,t),nb(y,t)] = iSabS3(x-y). (3.5)

Now, we are interested in the variation of £ under the transformation <j)a -t (t>a+Scf>a.
The result turns out
dC 8C
8£ - -^rWa a + a/fl j . ^^{d^<f>a)
d<j)a d(dp<j>ay
d£L_f) dC
»»7S^TT)J*» + 9 » ( 5 ( | ^ T ' 5 * .

- a>(mk>H')- (36)
-
by using the Euler-Lagrange equation (3.1). When the transformation parameters
are independent of the space-time point x, it is called the "global" transformation.
As an example, we introduce a transformation under a special unitary group G =
SU(n) as
6<j>a = -ieXbMx) {4>a -> 4>'a = e~i(>iTL<Pb), (3-7)
using x-independent infinitesimal parameters 01 and a n n x n matrices 2", where
a,b = 1,2, • • •, n and i = 1,2, • • •, n 2 — 1. Here the summation over over i and also
tFor simplicity, we consider the case of boson fields for the moment. For the case of fermion fields,
we need no change in the following discussion if commutation relations (3.4) and (3.5) below are
replaced by anti-commutation relations.
 Global Symmetries and Noether's Theorem 35

over b is implied. Tl is called generators of the SU(n) group. Then, when we define
the current

ffl5-^* <3 8)
'
(3.6) can be written as

SC = 6idll3^(x). (3.9)

Thus, if the transformation (j)a —• </>a + 6^>a leaves C invariant, i.e. 8C = 0, we have
the current conservation law,

d^(x) = 0. (3.10)

By integrating this equation over all space,

Jd3x^tJio + V-j^=0, (3.11)

and by assuming for jl(x) to vanish at spatial infinity, we are led to, by using the
Gauss's theorem, the charge conservation law,

? = "• <3-12'
where the "charge" is defined by

Q ' ( t ) = f d3xji0(x,t). (3.13)

One can see that a symmetry of the Lagrangian under a field transformation implies
conservation laws; this is called the Noether's theorem (Noether, 1918) and the
current jlfi(x) is called Noether currents.
In quantum theory, we have the Heisenberg equation

dJ^x)=i[P^r{x)], (3.14)

where PM is the four-momentum operator. If the current is conserved, one can

obtain, by integrating over all space,

- ^ = i[H,Q'} = 0, (3.15)

with H = Po, which is the Hamiltonian. Therefore, it turns out that Ql commute
with Hamiltonian H and the states with Q' (i — 1,2, • • •, n2 — 1) degenerate with the
same energy so as to make a multiplet. Using the canonical commutation relations,
one can easily show that the charges Ql satisfy the following relation, which is the
Lie algebra,

[Qi,Qj) = ifijkQk, (3.16)

36 SYMMETRIES AND THE GAUGE THEORIES

since the generators Tl of SU(n) always satisfy the same Lie algebra,

[T*,T']=ifijkTk, (3.17)

where /f^ is the structure constant of the Lie group and totally antisymmetric with
respect to i, j and k. That is, the charges Q' satisfy the same Lie algebra as the
generators Tl of the group SU(n) do. Furthermore, for an infinitesimal # ' ( < 1),
one can derive the following relation, using canonical commutation relations of (3.4)
and (3.5),

[Qi,d>] = -Ti<f>. (3.18)

This relation shows that Q1 is the generator of an infinitesimal transformation

0 _• j ' = e&Q* fa-is'Q* = e-ieiTi <j>, (3.19)

as can be seen by comparing the following relation

eieiQi4>e-if)iQi =<!> + i6i[Qi,(f>] + 0(62), (3.20)

with

e-"'1* 4> = <j> - iO'Tcf) + 0(62). (3.21)

In summary, the Lagrangian is a fundamental object describing the dynamics

among elementary particles. It has various symmetries which are represented by
transformation groups. The Lagrangian should be constructed to be invariant under
those transformations. The invariance of the Lagrangian under a global transforma-
tion specified by a unitary group results in a set of conserved charges <5l(Noether's
theorem) and these charges become a generator of the Lie algebra of the group.

3.2 Local Gauge Symmetries and Gauge Fields

The properties of strong, weak and electromagnetic interactions appear to be un-

related at low energies. For example, they have quite different coupling constants.
However, there is a possibility that at some extremely high energy the coupling
constant may converge to a single value and that interactions among elementary
particles could be explained in terms of single unified field. Glashow, Weinberg
and Salam made a major breakthrough along the path to unification by unifying
the weak and electromagnetic interactions. The most significant step in this direc-
tion is the realization that all fundamental interactions are invariant under local
gauge transformations and the hope is that gauge theories will provide a basis for
a comprehensive unification of fundamental interactions.
The standard model of particle physics refers to three quantum gauge theories
which describe electromagnetic, weak and strong interaction of elementary particles.
All these theories are renormalizable and are based on certain symmetries, which
 Local Gauge Symmetries and Gauge Fields 37

are classified into two ways, (1) manifest or unbroken and (2) hidden or sponta-
neously broken. The electromagnetic interaction is unbroken local U(l) symmetric,
the weak interaction is spontaneously broken local SU(2) x U(l) symmetric and
the strong interaction is unbroken local SU(3) symmetric. The non-Abelian gauge
theories of weak and strong interaction is the generalization of quantum electrody-
namics(QED), which is the Abelian local C/(l) gauge theory.

3.2.1 Quantum Electrodynamics-U(1) model-

Let us begin with the Lagrangian C describing the field of a single free fermion with
mass m. In the natural unit(/i = c = 1), it is expressed as
C(x) = ${x) ( r f d „ - m) tp{x). (3.22)
from which the Dirac equation is obtained by taking variation for V>(a;);
(tya M - m)i/>(x) = 0, (3.23)
where ip(x) is a 4-component spinor field of the fermion at point x in four dimen-
sional space-time and du = d/dx11 (n = 0,1,2,3). 7^ are the 4 x 4 Dirac matrices
and the summation over /i is implied. It is understood that m in the mass term is
multiplied by a 4 x 4 unit matrix I.
First consider a global transformation for this system. As discussed in the
previous section, it is described by a phase change of the fermion field
V>'(x) = e-i$ip(x) = Ueip(x). (3.24)
The Lagrangian (3.22) is invariant under this phase transformation where 6 is a
constant everywhere in space-time. That the Lagrangian is invariant under the
global transformation means that the phase 6 of the fermion field is not observable.
Clearly, transformation (3.24) is unitary, i.e. Ug~Ug = 1 and it is Abelian, i.e.
UeiUe^ = Ug^Ue^. That is to say, transformation (3.24) is global, unitary and
Abelian. This transformation is denoted by U(l).
Now, let us study what happens when this global symmetry is made local. It is
well known that the electromagnetic dynamics has a local gauge symmetry larger
than a global symmetry. A local gauge transformation is defined as

V>'(x) = e-*Mil>(x), fix) = eie^4>(x), (3.25)

by introducing a z-dependent phase parameter 6{x) varying locally from point to
point in space-time. Then, the kinetic energy term is no longer invariant under this
transformation because an additional term being proportional to dll9{x) appears
from the derivative of the transformed fermion field. The new Lagrangian under
this transformation becomes

£-*>£' = ^'(a;)(*ya M - m)tp'(x)

38 SYMMETRIES AND THE GAUGE THEORIES

= j>(x)(i'fdll - m)ij>{x) + {p{x)^xP{x)d^6{x)

l,
= C + j (x)dlie(x), (3.26)

where j M (x) = •0(x)7MV(2;) is the vector current carried by the fermion. Hence, the
Lagrangian is not invariant under the local transformation. Only if d^9(x) = 0, i.e.
6(x) is independent of x, C is invariant. Then, if we make the following replacement
in (3.22)

«9M -> £>M = df, - ieA^x), (3.27)

by introducing some vector field A^, we get instead

C-tC + epAp, (3.28)

where e is the electromagnetic charge of the fermion which is negative for an electron.
Thus, if we define a new Lagrangian by replacing 9M in (3.22) by £>M of (3.27)

= $(i'fdll-m)il) + eihllrl>Alt, (3.29)

then this C becomes invariant under (3.25) if at the same time we make the replace-
ment

AM -> A^ = AM - -d»0{x) or 5AIM = --d»6{x), (3.30)

which precisely cancels the additional unwanted term in (3.26). -DM defined above is
called a "covariant derivative", a terminology borrowed from the general relativity.
From the above observation, one can prescribe the procedure for getting the
gauge invariant Lagrangian in a little different way; starting from an original La-
grangian (3.22) which possesses a global symmetry and making the replacement
(3.27), we can get a new Lagrangian

C = # 7 " £>„ - m)ip, (3.31)

which is invariant under the local gauge transformation by requiring

(D^ix))1 = e-ie^D^(x) or 6{D^) = -i6(x)D^(x). (3.32)

The requirement (3.32) leads to (3.30), where 6{x) is the ^-dependent infinitesimal
phase parameter. Then, electromagnetic dynamics is made invariant by introducing
a spin 1 vector (gauge) boson field A^ called the photon through the covariant
derivative, which is called the "minimal coupling". It is very important to know that
in the gauge invariant theories the interaction between gauge bosons and particles
(fermions and/or bosons) is uniquely determined only through the minimal coupling.
The Lagrangian which is invariant under local U(l) gauge transformation is,
therefore, given by (3.31). However, this is not the complete Lagrangian for de-
scribing the whole system. We have to add the kinetic energy term and mass term
 Local Gauge Symmetries and Gauge Fields 39

of the electromagnetic field Afi(x), which should be also gauge invariant. As for the
mass term, the local gauge invariance necessarily leads to massless photon because
2

the mass term ^A^A1* violates the local gauge invariance, unless m = 0. It is
well known that the kinetic energy term of the electromagnetic field is given by
— \FllvF,iV, where the electromagnetic field-strength tensor FM„ is defined by
F^ = d ^ - dvA„. (3.33)

The coefficient - 1 is necessary for the requirement that the Euler-Lagrange equa-
tion is just the Maxwell's equation. The kinetic energy term is obviously gauge
invariant because SF^ = 0 under the transformation (3.30) as easily checked by
directly introducing (3.30) into (3.33). It is also seen in a little different way; the
covariant derivative satisfies

[Dlt,Dv]iJ> = -ieFllvil>. (3.34)

From (3.32), it follows that ([£>„,£>„] V)' = e-ie(-x\[D^,Dv]ip) and thus F'^ip' =
e-HMF^i/j = F^', that is, F^ = FM„.
Therefore, the complete gauge invariant Lagrangian for the system of an electron
and a photon takes the following form,

CQED = $(x)(i'fDll - m)V(x) - ^F^F^. (3.35)

The field theory with CQBD is called quantum electrodynamics(QED). The gen-
eralization of classical electrodynamics to describe quantum effects had culminated
in the development of QED as the theory of interaction of photons and electrons.
In QED, the electromagnetic interactions are mediated by massless photons.
Photons are quanta of the electromagnetic fields, being massless with no charge and
with spin 1, and they do not interact amongst themselves. Theory of gauge fields
represents a class of theories which share and generalize characteristic properties
of the Maxwell's theory of electromagnetic fields. There are two characteristics of
Maxwell's theory: (1) Forces created by gauge fields, i.e. photons, are long-ranged
and obey inverse square of distance as realized as Coulomb force. (2) Force is
proportional to the quantum number of its source and that quantum number, i.e.
the charge, is conserved.
Typical physical processes due to the electromagnetic interactions are shown
in Fig. 3.1: (a) a fermion emitting or absorbing a photon, (b) annihilation of a
fermion-antifermion pair to produce a photon, (c) a fermion scattered by another
fermion in a two step process in which a photon emitted by the first fermion is
absorbed by the second.
The principle of gauge invariance has come to be recognized as most powerful
guiding light for our understanding of not only QED but also probably all interac-
tions.
40 SYMMETRIES AND THE GAUGE THEORIES

f r
(c)

Fig. 3.1

3.2.2 Yang-Mills Gauge Theory-SU(2) model-

In 1954, Yang and Mills (Yang and Mills, 1954) proposed that the 17(1) gauge theory
of electromagnetic interactions could be generalized to the non-Abelian gauge the-
ories which are invariant under any non-commutative continuous symmetry group.
They chose the isospin of 5(7(2), which was familiar at that time. Therefore, the
original Yang-Mills non-Abelian gauge theory is an extension of the 17(1) Abelian
gauge theory of QED to the internal isospin SU(2) symmetric theory. The formal-
ism can be easily generalized to more general cases of SU(n) with n > 3.
We have seen that the electromagnetic fields are described by an antisymmetric
tensor F^v, which is made of a vector potential AM. Now we add an isospin index
to Ap and F^ and ask the following question; Under what condition is a theory
invariant under a space-time coordinate and isospin dependent phase transforma-
tion?
Likewise in (3.25), one can write an x-dependent SU(2) transformation (phase
rotation) as

ip(x) -¥ ip'(x) = Uip{x), (3.36)

<fl <(s)
U = e- ^' , (3.37)

where the summation over i is implied, g is the coupling constant associated with
the group SU(2). T* (i = 1,2,3) are three 2 x 2 Pauli matrices and 6l (i = 1,2,3)
are three real parameters corresponding to three generators T1/2, (i = 1,2,3) of the
 Local Gauge Symmetries and Gauge Fields 41

group SU(2).

Tk

= ieijkY ( 3 - 38 )

where the summation over k is implied, eijk, being the totally antisymmetric Levi-
Civita symbol with £123 = 1, is the structure constant of the group SU(2). Obvi-
ously these three generators do not commute, i.e. SU(2) is non-Abelian, hence the
gauge theory in question is called non-Abelian. ip is both a spinor and an isospinor.
Thus, instead of a single fermion field, we start with a doublet Dirac field

^ = ( tl ) ' ^ = (^ ^
$a = rl>t'f. (a =1,2)

Originally, Yang and Mills chose " 0 = 1 ) doublet where p and n denote the

proton and neutron, respectively. However, one can also choose ip = I 1 doublet
as well where u and d denote the up-quark and down-quark, respectively. Here, the
coupling strength g is a real constant to be determined from experiment (analogous
to e in QED). We insist that the Lagrangian is invariant under transformation (3.36)
and accordingly we encounter a number of additional complications, due to the fact
that there are now three orthogonal symmetry motions which do not commute with
one another. To construct a Lagrangian that is invariant under (3.36), we must
again define a covariant derivative that transforms in a simple way. In analogy of
QED, it is natural to introduce three independent gauge field potentials acting in
different directions, namely, A^(x) for i = 1,2,3 and require that operation on ip is
only through the covariant derivative

D„ = <9M - igA„, (3.39)

where AM = X^=i T ^ M = f • AM is a matrix-valued gauge field. Likewise in QED

discussed above, the covariant derivative should satisfy

(D^)' = D'^1 = U(D^), (3.40)

which leads to

D ; = UDpU-1. (3.41)

Then, if we demand for AM to transform as

A'll = UAllU-1--(dllU)U-1i (3.42)

42 SYMMETRIES AND THE GAUGE THEORIES

the covariant derivative (3.39) satisfies t h e expected transformation property,

= (dnUW + Udrf-ig uk^u-1 - -(d^u)u-1 Utf>

= U(dpil> - igA^ip)
(3.43)

For an infinitesimal transformation,

U = l-i6 + O(02),
U'1 = l + i6 + 0(62), (3.44)

where 6 = £ ? = 1 $6' = f • 0, (3.42) results in

A^ — Ap I 0,A„ -V- (3.45)

Using (3.38), we obtain

4=4+ey*#4--0M0i, (3.46)

or

J 4 = etffc^"4 - iaMfl*. (3.47)

To get t h e gauge field-strength tensor like t h e electromagnetic field-strength tensor

Fpv of (3.33) a n d its transformation rule under an infinitesimal transformation
(3.44), we generalize (3.34) as

[Dll,Dv]ip = -igFliVil), (3.48)

w t n e
where F M „ = J2i=i ^Fftv = f 'Fpv ^ g a u g e field-strength tensor F£„ defined
by

K» = 9 " 4 ~ fl"4 + aeijk^Al (3.49)

From the relation

(3.50)

which follows from (3.40), one can easily derive

F ^ ' = Uf^ (3.51)

or

f'^^uf^u-1, (3.52)
 Local Gauge Symmetries and Gauge Fields 43

which leads to the following transformation relation of the field-strength tensor

under an infinitesimal transformation (3.44),

8F%, = eijk6^F^. (3.53)

Note that F^„ transforms nontrivially unlike the case of QED where F^v is invariant
because of no structure constant in an Abelian group. In analogy of QED, we
construct the kinematic term for the gauge field A1^ as

-iTr(FM„ • f"") = ~\J2 MjF^F^) = -\FlvF^ (3.54)

i,j=l

where the 2nd equality is due to the relation Tr(Y-y) = \&x* • One can easily see
that this term is invariant under the gauge transformation as shown in the following;

8{FlvF^v) = 2SF^Fi>"/
= 2ey i f e ^F ( J„F""'= 0, (3.55)

where the last equality is due to the totally anti-symmetric property of Sijk • There-
fore, we can adopt (3.54) as the kinetic term for the gauge field A1.
In summary, we can write down the gauge invariant Lagrangian for a fermion
with mass m in the SU(2) symmetric world as

C = CF + CG, (3.56)
CF = $(x)(i'fDll-m)il>(x),
CG = - ^ ( F ^ - F ^ ^ - ^ F ^ ,

where dp and CG are represent the Lagrangian for the fermion and the kinetic
term of gauge fields, respectively. The covariant derivative is defined by (3.39) with
3 gauge fields A1 (i = 1,2,3). Transformation property of the gauge field A* is
given by (3.46) or (3.47). The gauge field tensor F^v is defined by (3.49) and its
transformation rule under 5(7(2) gauge transformation is given by (3.52). Change
under an infinitesimal transformation (3.44) is given by (3.53). In (3.56), the mass
term of gauge fields does not appear again because it violates the gauge symmetry.
Some comments are in order for (3.56). (1) Contrary to QED, the kinetic term
of gauge fields JCQ contains 3 gauge boson interactions with the same coupling
constant g as the one of the gauge field to fermions. It contains also a 4 gauge
boson interaction with the coupling g2 as follows;

CG = -^Md'A™ - PA*) - geijkAlAid»Ak" - 9

leijk£UmAlAkvA^Am".
(3.57)
(2) The Yang-Mills theory, i.e. the non-Abelian gauge theory is not the free theory
even without matter fields because it contains self-interactions among gauge fields.
This is very different from the Abelian gauge theory like QED, where there is no
44 SYMMETRIES AND THE GAUGE THEORIES

self-coupling of photons. (3) As long as we demand the gauge invariance alone, we

can add renormalizable Yukawa interactions such as,

CY = GY${x)ip(x)4>(x) (3.58)

in which the scalar field <j>(x) is introduced for Cy to be invariant under the gauge
transformation. There is no principle to determine the form of the Yukawa term
except for the requirement of gauge invariance. Gy is also totally unconstrained;
there is no other way to determine it from experiments.
Unfortunately, this theory (3.56) is not useful for weak interaction because it
gives identical coupling to right- and left-handed fermions and leads to the parity
conservation. To preserve gauge invariance, it is essential to have massless gauge
bosons A^, which should give weak interactions of infinite range like electromagnetic
case, being contrary to the real physics. Only if the gauge symmetry is broken by the
inclusion of mass term, it becomes possible to achieve agreement with experiment.

3.3 Spontaneous Symmetry Breaking and Goldstone Bosons

Nature seems to possess various types of symmetries such as geometrical (Lorentz

invariance, parity and time reversal invariance etc.) and internal (isospin, flavor
and color, etc.) symmetries, discrete and continuous symmetries and so on. Some
of them are exact symmetries and others are approximate symmetries. Here we are
concentrated on the symmetry and its breaking in the physical world. In the field
theory, dynamics of the physical world is described by the Lagrangian. There are
two ways to discuss the symmetry breaking in the field theory; (l)One way is to
add a symmetry breaking term to the symmetric term by hand as

L> — I^sym "•* ^breaking- \o.o&)

Examples are seen, for example, for the case of an approximate symmetry such as
the isospin SU{2) or flavor SU(3) symmetry and so on. This type of symmetry
breaking is useful when the symmetry breaking term is small and the perturbative
treatment is meaningful. In this case, C recovers the exact symmetry for vanishing
^breaking- However, this case is rather artificial because there is no fundamental
principle to determine the exact form of CbreaHng- (2) Another way is the one called
a hidden or a spontaneous symmetry breaking(SSB), where the Lagrangian remains
symmetric under certain group transformation while the physical vacuum is made
non-invariant. It is well-known that there are many examples of the SSB both in
classical and quantum physics such as bent rods under a strong force, infinite fero-
magnets, crystal lattices, superconductors, etc. Here we are interested in the field
theory model of the SSB.

• Goldstone theorem
 Spontaneous Symmetry Breaking and Goldstone Bosons 45

Let us consider the U(l) global invariant Lagrangian composed of a complex

scalar field cj)(x) and <j)*(x),

£ = aM^*a"^-V(^), V(<f>*cf>)=m<f>*<l) + \(<t>*<j))2. (3.60)

The system is equivalent to the one described by the following Lagrangian composed
of 2 real fields 931 and v?2 which are related to (j> and </>* as $ = (<pi + i<p2)/'v/2 and
4>* - (Vi - i<P2)/V2,

c = -d^xd^tpi
-d^d^i + -8^28^2
-i - v(<pl + ft). (3.6i)

(3.61) has the 0(2) symmetry, that is, the Lagrangian of (3.61) is invariant under
the following 0(2) transformation,

/>* W t f W c o s * - s i n 0 W ^
Smd C0S
V ¥>2 / V ¥>2 / V ° J \ <P2 J

and one can see that the 17(1) symmetry is equivalent to the 0(2) symmetry. The
potential V{(j)*4>) = V{<p\ + <p\) must have the following properties for the theory
to be meaningful; V is (1) at most the 4th order of the fields in order to ensure
renormalizability of the theory and (2) bounded below for change of (4>*(j>)1/2 = |^|,
so that the theory has a stable ground state. A typical example of V is given in
(3.60) with A > 0.
In quantum field theories, particle excitations of a field are defined as quantized
fluctuations of the field about its lowest energy state, i.e. the vacuum state. The
constant value of the field corresponding to the lowest energy state is called the
vacuum expectation value (VEV), i.e. (0|(/>|0) = <f>o- To find the particle spectra,
we expand the potential about its minimum corresponding to the lowest energy
state as

V((pi,<p2) = V(<poi,<p02)+ ^ (^—1 (fa-foa)

0=1,2 V a < ^ Q / o

+
O 5Z ( a,„ a,„ ) (<Pa. ~ <P0a)(<Pb ~ <P0b) + • • • , (3-63)

w h e r e <j>0 = (<poi,<p02) is t h e V E V of (j> = (ipU(p2), i.e. if0a = {0\<pa\0) (a = 1 , 2 ) .

Since the potential V has its minimum at (j> = cf>0, the 2nd term of the r.h.s. of
(3.63) is zero. The factor ( d t j ) V h ) = m a6 m the 3rd term is called the mass
matrix which is diagonalized to generate the particle spectrum.
Now we have 2 possible cases of V. (1) One is the case where the vacuum is
unique and is called "Wigner phase". In this case, only one vacuum state (<^oi =
0,^02 = 0) is realized as shown in Fig. 3.2(a). Let us take the parameters m 2 and
46 SYMMETRIES AND THE GAUGE THEORIES

vm

+ V2 *"¥>2

Fig. 3.2

A for the potential in (3.60) to be positive, m 2 > 0 and A > 0,

2 ^
(3.64)
V{fp\ + Vl) = ^ ( V i + <P%) + 4 (</>? + ¥>2)2,

and require the following condition for the vacuum,

f -g— J = m V o i + A</?oi (Voi + fli) = 0.

f—1 = mVo2 + ^(^02(^01 + fm) = °- (3.65)

Then, we are led to the unique vacuum </>oi = <Po2 = 0. The mass matrix becomes
diagonal in this case

m2 0
™-ab = (3.66)
0 m2

which means that <$\ and </?2 have the same mass m as already seen from (3.64).
(2) Another one is the case where the vacuum is not unique and called "Nambu-
Goldstone phase". This case is realized, for example, for continuously or infinitely
degenerated vacuum states with <p0i ^ 0 and/or ip02 ^ 0, as shown in Fig. 3.2(b),
for the potential with m 2 = -fi2 (fi2 > 0) and A > 0,

Vivl + vl) = - ^ ( y ? + y!) + -M + V2)2 (3.67)

*
The minimum of V is realized by requiring

= -/x 2 Voi + V o i ^ o i + V02) = 0,

\0Vi/<
 Spontaneous Symmetry Breaking and Goldstone Bosons 47

•r— ] = - ^ V o 2 + Ay>02(Voi + V02) = °> ( 3 - 68 )

which leads to the condition

¥&+¥&=«2 = ^ or (0V)o = W 2
= y = ^- (3-69)

In other words, all points on a circle with radius v = yfJJP/X in the (</>i,y2) plane
correspond to the minimum of V, that is, the vacuum state is no longer unique but
is 0(2) symmetric. One can choose any point as the physical vacuum. From (3.67),
we obtain

d2v = {-ii2 + X{ip\ + ¥>!)) + 2A^ 2 ,

0 = (-/i2 + A(^+vl))+2A^, (3.70)

J
2
d2v = 2Xifiitp2-
8<pid(p2

Then, if we choose a point (</>oi = v, tpo2 — 0) as the physical vacuum, we obtain

the mass matrix as

m = (3 71)
«{ 0 oj- -
Therefore, we find that <p[ = (fi — v corresponds to a massive particle with mass
m 2 = 2Au2, while ip'2 = tpz is massless. ip2 is called a "Goldstone boson". Actually
using these new fields, we can rewrite the Lagrangian (3.61) as

£ = 5CW1)2 + \(Wi? ~ l(2Xv2)ip'2 + \v<p'M2+lp'2) - ±(<p'2 +ip'2)2. (3.72)

This Lagrangian (3.72) has no longer 0(2) symmetry, though the original La-
grangian (3.61) has it explicitly. That is to say, the symmetry of the original
Lagrangian has been broken by breaking the symmetry of vacuum. This is called a
hidden symmetry or spontaneous symmetry breaking(SSB).
In summary, starting from the Lagrangian which has a global symmetry, taking
the negative parameter m2 = —/J? (fi2 > 0), and breaking the symmetry of the vac-
uum states by choosing a particular point among symmetrically degenerate vacuum
states, we found that a massless particle, called the Goldstone boson, appeared.
This mechanism is called the Goldstone theorem. (Goldstone, 1961; Goldstone,
Salam and Weinberg, 1962; Bludman and Klein, 1962) In general, when a global
symmetry is spontaneously broken, a number of massless Goldstone bosons appear
depending on the symmetry properties.

• Useful parametrization for the C/(l) model

48 SYMMETRIES AND THE GAUGE THEORIES

In literature, another and more useful parametrization of the field is often used.
Introducing the two real fields p(x) and 0(x), we can write the complex field fax)
in (3.60) as

fax) = ^ p ( i ) e t f ( * ) / « , (3.73)

where v is a constant given by (3.69). Then, substituting

dli4>=^=e»<"(dllp+X-pdlt0), (3.74)

into (3.60), we have

£=\{9,pf+2^w)2 - vtfy ( 3 - 75 )
To find the mass of the particle corresponding to the excitation of the would-be
radial field p{x), we expand it as p(x) = v + r)(x) and obtain

£ = \(d,V)2 + |(0 M 6!) 2 + ^(d,0)2 + ^(d.0)2 - V(p2), (3.76)

with

V(P2) = \{2fW + W + \^ - \p2v2. (3.77)

It is noted that there is no quadratic term of 0 in this Lagrangian. Prom this La-
grangian, one can easily find that we have a massive 77 field with mass m,, = \J2p?
and a massless Goldstone boson 0.

• Extention t o the SU(2) model

In the SU(2) model, the field is given as a doublet, 0 = 1 ) , composed of

2 complex fields, fa = {ip\ + i^ 2 )/%/2 and fa = (</>3 + itpi)/y/2. Then, the global
SU(2) invariant Lagrangian is given as

£ = d^d^-Vitffa, (3.78)
V = -pVtf + W W . (3-79)
with p? > 0 as before. Now, by using 4 real fields H(x) and £*(a;) (i = 1,2,3), we
write 0 as

* = 7f"'™/2'(»+V))' (3 80)
'
 Spontaneous Symmetry Breaking and Goldstone Bosons 49

where the VEV v is defined as ^ + <pl2 + V03 + V04 - ^ 2 =

1r or
(^V)o = |^o| 2 =
£ = f j . Then, putting

1 0
dM<£ = —eiT't'/2v (3.81)
v2 aM# + ;TW v+H

into (3.78), we get

£ - I |(0 V 7 ) -v^ ( 0 u + tf)-

— i
~2

{(^)^T^(JE)}-V^^- <3'82>
which leads to

C = ^Hd^H + ^d^d^eiv + H)2 V((v + Hf), (3.83)

using the relation rV-' = 6** +ieijkTk. Prom this Lagrangian, one can see that the 3
fields £* (i = 1,2,3) have no mass terms. They are massless Goldstone bosons, while
H is massive with mass ma = v 2 / ? - The original SU(2) symmetry disappeared.
It is remarkable to note that the number of Goldstone bosons is equal to the number
of generators breaking the symmetry of the vacuum state. In an example of SU(2)
model, the number of Goldstone bosons are three, £l (i = 1,2,3), corresponding to
the 3 generators, r* {i = 1,2,3).
This is generalized for cases with larger symmetries. Now, let us consider an
SU(n) symmetric world with n component fields,

f2

(3.84)

\ fn j

The SU(n) group has TV = n 2 - 1 generators T i (i = 1,2, • • •, N) and under this

group, <p transforms as follows;

(f> -> (j)' = -tfl'T* (3.85)

e <t>,
or under an infinitesimal transformation,

6(f) = - z ^ T V , (3.86)

where the summation over i is implied. Tl are the matrices with nxn components.
Suppose that the vacuum is not unique but is symmetrically degenerate. Then,
50 SYMMETRIES AND THE GAUGE THEORIES

choose a particular vacuum state,

¥>2

<i>o = (3.87)

Under an infinitesimal transformation, a change of this vacuum state is Scf>o =

—i6lTl(j)Q. Then, one can see that the vacuum is not invariant under operation of
generators T* if TVo ^ 0. Suppose that M(< N) generators T* (i = 1,2, • • •, M)
are such generators, while other Tl (i = M + 1, M + 2, • • •, AQ leave the vacuum
invariant as T'fo = 0. Then, by introducing a new field $ — 4> — <Ao, we expand
the potential V around the vacuum as before,

T. T, ^fdV\ , 1 " { d2V \ , ,

(3.88)

with
82V \
and mlab
h = (3.89)
\8<P*)o ~ \dtpadw)0

The potential should be invariant under SU(n), i.e. SV = ^2a J^-<fy>a = 0 which,
by using (3.86), results in
dV
T 6 = 0 for
E^r > ' * = l,2,---,iV. (3.90)
d<Pa
a,b

Then, by differentiating this equation by <pc, we obtain

d2v dV
E dipcdifa 7 > 6 + E d>fia
a.b
T l
—0 (3.91)

which leads to, at the vacuum <f> = fa,

Em-Tai»(^)o=0, for i= l,2,---,N. (3.92)

a,6

As described above, TVo ^ 0 for i = 1,2, • • - , M . Therefore, Tlip0 span a M-

dimensional subspace in the n dimensional space spanned by (f> of (3.84) and thus,
m2ca has M zero eigenvalues. In other words, there exist M Goldstone bosons or to
each generator which breaks the symmetry, a Goldstone boson appears.
For more detailed discussion on the spontaneous symmetry breaking and Gold-
stone bosons, see, for example, reviews by Abers and Lee (Abers and Lee, 1973) or
Coleman (Coleman. 1975).
 Higgs Mechanism 51

3.4 Higgs Mechanism

In the previous section, we discussed the spontaneous symmetry breaking under the
global symmetry, i.e. the Goldstone theorem. Here we go further by extending the
global symmetry to the local gauge symmetry.

• The U(l) model

First, let us consider an example of the scalar electrodynamics for a scalar com-
plex field 4> = ipi + i<f2, which is local U(\) invariant. The Lagrangian is given
by

C = -^F^ + iD^nD^-V (</>», (3.93)

V{<j>*<j>) = -fPt + XWitf, (3.94)

with /i 2 > 0, where FM„ = d^A,, — <9„AM is the field-strength tensor for photon
fields AM and D^cf) is the covariant derivative D^cj) = (9M — ieA^cp as before. The
Lagrangian (3.93) is invariant under the following f/(l) gauge transformation,

4> - > </>' = e-ia(-x)(f>, (3.95)

A^ -> A'll = Alt-ldlla(x). (3.96)

Now let us minimize the potential V as before with \(j>0\2 = \ = ^ . Then, if

we parametrize the field <f>{x) as

4>(x) = ±(v + T,{x))eieW°, (3.97)

where r](x) and 6(x) are real fields. We can define a new set of fields by taking a
particular gauge transformation with a(x) = 6(x)/v, which is called the "unitary
gauge",

<Kx) -> 4>W = e-i9lx)/v<t>(z) = -j=(v + v(x)), (3-98)

A^x) -> Bll{x) = Ali{x) d,fi{x). (3.99)

Under this unitary gauge transformation, we have

D^{x) -> D'^'ix) = (dM - i e B M ) - ^ ( « + ij(a:))) (3.100)

and

F^{A) = d»Av - d„j4„ -> F^(B) = <9M5„ - 3VB^. (3.101)

52 SYMMETRIES AND THE GAUGE THEORIES

Here we can easily show Fflv(B) — Fllv{A) by substituting (3.99) into (3.101), that
is, the field-strength tensor FMI/ is gauge invariant as it should be. Substituting
(3.98) ~ (3.101) into (3.93), one can rewrite the Lagrangian as follows;

£ = \\d,r1-ieB^v + r1)\2-^(v + r,)2-j(v + r])4-\F^(B)F^(B)

= Id^r, - A 2 " \Fllv{B)F^{B) + \{evfB^

+ \e2B^r]{r] + 2v) - Xvrf - ^r? 4 . (3.102)

We can see that this Lagrangian describes a massive vector boson B with mass
TUB = ev and a massive scalar 77 with mass mn = ^2fi2. 77 is called "Higgs boson".
Here we have no Goldstone boson which has gone out of the Lagrangian (3.102).
In summary, by extending the symmetry of the Lagrangian from the global to
local one, we found that the massless Goldstone boson 6 disappeared and a massive
gauge vector boson B and a massive scalar boson 77 called Higgs boson came out.
This is called the Higgs mechanism. (Higgs, 1964, 1966; Englert and Brout, 1964;
Guralnik, Hagen and Kibble, 1965; Kibble 1967) The Goldstone boson 6 was eaten
up by the gauge boson B and became the longitudinal component of it. It is re-
markable to note that in the Higgs mechanism the degree of freedom is conserved,
that is to say, starting from 2 real scalar fields ( y i , ^ ) or (r],9) plus 2 polarization
states of massless photons A^, we finally got one real massive scalar field 77 and one
massive vector boson B^ with 3 polarization degrees of freedom

• The SU{2) model

Let us next consider the non-Abelian SU(2) model by generalizing the C/(l)
x
model. We have again the complex doublet field (f> = I I. Then, the gauge
invariant Lagrangian with SU(2) symmetry is given by

£ = (£> M 0)t(D"0 - i j ^ F * " " - V(<j>U), (3-103)

with

£>„</> = (d^-igjA^, (i = 1,2,3) (3.104)

Ku = d^Ai-dvAi + geijkAiAku, (3.105)

2
V(tf<j>) = -|/ ^ + A(^)2. (M 2 >0) (3.106)
Now, introducing the new real fields H(x) and fl(a;) (i = 1,2,3), let us parametrize
the field <j>(x) as
 Higgs Mechanism 53

Then, by taking the unitary gauge as before, we define the new fields as

4>{x) -> <j>\x) = U(x)4>(x) = ^ ( v + °H{x) ) . (3-108)

AM -> Bli = U(x)AllU-1--(dliU)U-\ (3.109)

with
U(x) = e-iTi(t^/2v, (3.110)
where the summation over i is implied. This transformation leads to

D (3 nl)
»+ - v>»+y=^ -"T^TS (t,+°H(X) ) - -
Flv{A)Fi,lv{A) -> F^(B)F^(B) = F^(A)Fi"l/(A), (3.112)
with
F}V(B) = dMB* - duBi + geijkBlBt. (3.113)
Then the Lagrangian becomes

£ = {D^'^D^y - JF;,(5)F^(5) + / i V V ) - H<t>'V')2- (3-114)

Prom this Lagrangian we can see that the three C(x) (i — 1> 2,3) fields disappeared.
Where did these fields go? We can find the answer by writing C in terms of the
component fields of <f>'. First, let us write down the covariant derivative term,

\{D^)ia{D^)'a = \d»Hd»H + g2BiiiB^(^j (j) ^ ^

,2
= ^Hd^H+V-BlB^iv + H)2. (3.115)

Then, we finally obtain the following Lagrangian,

£ = ^Hd^H-ti2H2-\F^(B)FiliV(B) + ^-BifiBi'i
v
9 D i ltl
n i u T r m . . , TT\ \..TT3 3
+ ?-BlB H(2v + H)~ XvH - ^-H*
ui
H - —. (3.116)
4 ~1
A triplet of massive vector fields B^ (i = 1,2,3) with mass TUB — \gv and a single
massive scalar, i.e. Higgs boson H with mass mjj = y/2fj,2 appeared. We found here
again that the Goldstone bosons f' (i = 1,2,3) were eaten by the gauge bosons Bl
(i = 1,2,3) to make their longitudinal components. This is the Higgs mechanism in
the non-Abelian 517(2) gauge theory. It is again interesting to note that the number
of degrees of freedom is conserved in the Higgs mechanism, i.e. the 3 Goldstone
bosons £* (i = 1,2,3) become the longitudinal components of the respective gauge
fields which lead to appearance of 3 massive vector bosons Bl (i = 1,2,3). The
54 SYMMETRIES AND THE GAUGE THEORIES

discovery of the spontaneous symmetry breaking (SSB) and the Higgs mechanism in
the non-Abelian gauge theories made a great breakthrough toward the unification
of electromagnetic and weak interactions, which will be discussed in detail in the
next Chapter.

Problems
3.1 Show that when the generators Tl satisfy the Lie algebra (3.17),

[Ti,T']=ifiJkTk,
the "charge" Ql also satisfy the same Lie algebra (3.16),

[Qi,Qi] = ifiJkQk,
where Ql is defined by (3.13).

3.2 Using the canonical commutation relations of (3.4) and (3.5), prove the relation
(3.18).

3.3 In the SSB of the global 0(2) symmetric theory described in 3.3, discuss what
happens if we take the vacuum to be (y>oi = 7k, <A)2 = 775) instead of (<p0i = v,
(p02 = 0), and show that both cases are physically equivalent.

3.4 Derive (3.116).

3.5 Consider the global 0{ri) symmetric Lagrangian,

c = \dll<t>T{x).d^-v{<s>T•</>), with n0 T -</o = ^ T - 0 + ^ T - < « 2 ,

where

4> =
\ <Pn I
After the spontaneous breaking of this 0(n) symmetry by taking m 2 — —fx2 (fi2 >
0), show that one can make the vacuum to be invariant under 0(n — 1) transfor-
mation, that is, 0(n) breaks into 0(n — 1). Furthermore, show that the number of
Goldstone bosons is equal to the number of broken generators of the original 0(n)
group.
 Chapter 4
T H E STANDARD MODEL OF
ELECTROWEAK INTERACTIONS

In this Chapter, we describe the Glashow-Weinberg-Salam(GWS) model of elec-

troweak interactions (Glashow, 1961; Weinberg, 1967; Salam, 1968). It is a non-
Abelian gauge theory with SU(2)L X U(1)Y gauge symmetry accompanied by the
Higgs mechanism. It is the first successful model toward the unified theory of ele-
mentary particle interactions. The model is extremely successful in particle physics
phenomenology without serious discrepancy with almost all existing data; only one
exception at present might be an evidence of massive neutrino established with re-
cent observation of neutrino oscillation. The discovery of W± and Z° bosons with
expected masses and a weak neutral current mediated by a massive neutral vector
boson Z° is a great triumph of the model. Here we first consider the model for
one lepton family of an electron e and the corresponding neutrino ue. Then, the
realistic case of more lepton families is described. Extension to the quark sector
is also discussed, where the non-diagonal quark mass matrices and the Cabibbo-
Kobayashi-Maskawa(CKM) matrix are introduced. These embody the standard
model of the electroweak interaction.

4.1 Fermions in the G W S Model

Let us start with the discussion on the one family of leptons, an electron e and
its neutrino ve. ve is considered to be massless in this model, while e is massive
with a small mass, me ~ 0.5MeV. Both of them have spin 1/2. In this section, we
assign these particles to the appropriate representations of SU(2)L X U(1)Y gauge
symmetry, the gauge symmetry of the GWS model.
As we have seen in chapter 2, in the weak processes such as

\i~ -> e~ + ue + Vp, (4.1)

-
•n" -> M +^> (4.2)
n -» p + e~+ue, (4.3)

only left-handed leptons and right-handed anti-leptons take part in and the decay

55
56 THE STANDARD MODEL OF ELECTROWEAK INTERACTIONS

amplitudes of these processes can be written down in terms of charged currents

J M (x) = J p (a;) + = PeL(x)j^eL(x)

= ^(^(i-ls^o;), (4.4)
Jn(x)* = J»(x)~ = eL(a;)7MJ/eL

= 2e(x)'yll(l-j5)ve(x). (4.5)

This suggests that BL and veL could be arranged to simply make a doublet associated
with SU(2) group. Let us introduce a lepton doublet composed of the left-handed
components of fermions,

L=^(::) = (:t). (4.6)

Then, by using this doublet and 2 x 2 matrices in the so-called weak isospin space

_ r i - H r W o n
(4.7)
2 ~ ^ o o y'
Tl-iT2 ( 0 0 ^
T
(4.8)
- 2 " V 1 0 ) '

one can rewrite the charged currents (4.4) and (4.5) as follows;

J+ = L^T+L, (4.9)
J
H = LJ^T-L. (4.10)

(Note that in this Chapter T1 are used to denote Pauli matrices, in order to make
clear that they stand for the weak isospin, not the ordinary spin.) These forms
suggest that the weak currents make an SU(2) group by introducing an additional
neutral current,

= ^elu^e ~ ^eL^eL. (4.11)

Then, we have 2 charged and 1 neutral currents, Jjf and J3, which couple to
the weak bosons W^ and A3^, respectively, just as the electromagnetic current
J^m(x) couples to photon Afl(x), as we will see later. Existence of J ± and J3
suggests that we have weak isotopic triplet currents of SU(2), belonging to the
adjoint representation,

Jl(x) = ISY^L = LJXL, (1 = 1,2,3). (4.12)

 Fermions in the GWS Model 57

It is easily shown that the corresponding charges Tl (i = 1,2,3) defined by

r = fr0{x)d3x, (4.13)

satisfies the SU(2)L algebra

[T\Tj] =ieijkTk, (4.14)

with e'jk being the totally antisymmetric Levi-Civita tensor with e 123 = 1. In
(4.12) the subscript L of SU(2)L means that the weak isospin currents are composed
of only the left-handed Weyl fermions. In all of these currents, the right-handed
component of electron eR does not appear. It has no interactions with any other
particles and thus it should be a singlet under SU(2)L transformation, i.e.

R=\{l + 7s)e^eR. (4.15)

As mentioned above, in this Chapter we assume ve to be massless and thus ve has

no right-handed component. At a first glance, one might think of identifying the
neutral current J 3 of (4.11) with the electromagnetic one, J^m- But it is impossible
because of the following reasons; (l)the neutral current J^(x) has no right-handed
component, while the electromagnetic current J^m(x) for an electron has both left-
handed and right-handed components as follows, essentially because parity is a good
symmetry in the electromagnetic interactions;

Jtm(x) = - g 7M e = SLJ^L - eR^eR, (4.16)

(2)the "charge" of neutrino which couples to A 3 is not zero but opposite to the
one of electron as shown in (4.11), while it should be zero for the electromagnetic
interaction, and (3)contrary to the electromagnetic current, the neutral current J 3
maximally violates parity. These arguments suggest that gauge symmetry SU(2)L
should be enlarged. Let us note that, though (4.11) leads to a wrong relation saying
that the sum of electric charges of ve and e~ vanish, due to the property Tr r 3 = 0,
it gives the correct difference of the charges, i.e. \ - (— ~) = 1 — 0, This implies that
J^ m should have a piece generated by a diagonal 2 x 2 generator with non-vanishing
trace. Namely the gauge group should be enlarged so that it contains a new U(l)
symmetry. Therefore, we also need another gauge field B^ associated with the new
U(l) symmetry. This new C/(l) group should be independent of SU(2)L group and
thus its generator should commute with the generators of SU(2)L, Tl (i = 1,2,3).
The gauge group is thus extended to the direct product of SU{2)i and C/(l). Then,
how can we practically realize it? We want to keep the form of the electromagnetic
current for a fermion ip with charge Q to be

j;mW=HW. (4-17)
where Q is the charge for the fermion ip. For the case of an electron, the eigenvalue
of Q is Q = —1. From (4.17), we can define the generator of U(l)em for an electron,
58 THE STANDARD MODEL OF ELECTROWEAK INTERACTIONS

lepton family Q (T,T3) Y

"e,*Vr 0 (*,+*) -1

ez,,A*L, TL -l (I _ I ) -1
V2' 2'

efi,Mfl> T
R -l 0 -2

Table 4.1 Quantum number assignment of lepton families in the GWS model.

i.e. the charge of an electron as

Q = ! J(x)e0md3x = - f el0ed3x = - f (e[eL + eReR)d3x. (4.18)

(The reader should not be confused by this notation, though the same Q as in (4.17)
is used here.) However, since this generator Q does not commute with Tl denned
by (4.13), [/(l)em and SU(2)i cannot be simultaneous symmetries of the model.
Then, we look for a new U(l) symmetry, called £/(l)y, so that Q is given by a
linear combination of the generator T3 of the SU(2)L group and the generator of
this new U(1)Y group. In doing so, it is interesting to note the relation

Q-T3 = J' d3x(-\v\LveL ->eLeL eRCR (4.19)

which shows that each element in an SU(2)L doublet (ve e)lL has the same eigenvalue
— | and the eigenvalue of en is — 1. Furthermore, Q — T3 commutes with Tl{i =
1,2,3), i.e. [Q -T3, T'] = 0, that is, Q — T3 and Tl can be simultaneous symmetries
of the model. Thus, it is reasonable to define this new generator of the U(1)Y group
as Q - T3 or

Q = T3 + ^-. (4.20)

The eigenvalue of Y is called weak hypercharge; the name of this quantum number
is originated from the fact that (4.20) is of the same form as Nakano-Nishijima-
Gell-Mann relation (Nakano and Nishijima 1953; Gell-Mann 1953), established in
1950s in successfully describing the hadron classification, in which Y was called
hypercharge. These quantum numbers for an electron e and its neutrino ve are
summarized in Table 4.1. (In this table, the quantum numbers of muon and tau
families are also given,)
In summary, the GWS model is the SU(2)L X U(1)Y gauge theory and we have,
as matter fields, a left-handed doublet L and a right-handed singlet R of SU(2)
group which are represented as

L = R = eR. (4.21)
 SU(2)L x U(l)y Invariant Lagrangian 59

The model starts with the Lagrangian constructed with L and R, which is invariant
under the direct product of SU(2)L and U(1)Y groups:

SU(2)L : L^L' = e-ia'ix)Iz-L, R^R' = R,

U(1)Y : L->L' = e^^L, R -> R' = ei0(-x)R, (4.22)

where a1 (i = 1,2,3) and j3 are group parameters for weak isospin and weak hyper-
charge operators, respectively. Since we are considering the local gauge invariant
Lagrangian, a1 and /? are z-dependent.

4.2 SU(2)L X U(1)Y Invariant Lagrangian

In the previous section, we introduced the representation of an electron and its

neutrino with respect to the SU(2)L X U(1)Y symmetry, which is given in (4.21).
The gauge invariant Lagrangian with SU(2)L X U(1)Y symmetry for these fermions
is constructed as

CF = Li^id^-igyA^+'-g'B^L
+ Ri^id^ + ig'B^R, (4.23)

where Al^(i = 1,2,3) and B^ are gauge boson fields associated with SU(2)L and
U(1)Y, respectively, g and g' are the gauge coupling constants corresponding to
SU(2)L and I/(l)y, respectively. Here the explicit forms of the covariant derivatives
for L and R come out from the general form

A* = d» ~ ig\ • \ - ig'-B^ (4.24)

by taking account of Y = — 1 for L and Y = — 2 for R, respectively (see Table

4.1). R is a singlet of SU(2)L and hence does not couple to A*. In (4.23), the
fermion mass term, which connects L and R fields, does not appear because it
violates SU{2)i x £/(l)y invariance. Therefore, all fermions, i.e. an electron and
its neutrino, are massless at this stage.
The kinetic term of the gauge fields which should be added to CF is given by

£G = -\F1^F^ - i^B"", (4.25)

with

K» = d^Ai-dvAl+geijkAiAt, (4.26)
£„„ = dy.Bv-dvB„, (4.27)

where F ^ ( i = 1,2,3) and B^ are field strength tensors of gauge fields correspond-
ing to SU{2)L and U(l)y, respectively. As is well known, the mass terms of these
gauge bosons do not appear because of the local gauge invariance.
60 THE STANDARD MODEL OF ELECTROWEAK INTERACTIONS

In order to make fermions and gauge bosons massive and make the real world,
we need spontaneous breakdown of gauge invariance, i.e. the Higgs mechanism,
described in the previous chapter. Since we are living in the U(l)em symmetric
world with massless photon, we need to have the following breaking,

SU{2)L x U(1)Y —• U(l)em. (4.28)

To realize this symmetry breaking, we introduce the scalar fields, called Higgs
bosons, which give rise to the Higgs mechanism. Since we start with 4 gauge
bosons(3 associated with SU(2)L and 1 with U(l)y) and finally want to have 1
massless photon associated with U(l)em, we need scalars with at least 4 degrees of
freedom. The simplest example of such scalars, which is called the minimal model,
is an SU(2) doublet of 2 complex scalar fields whose weak hypercharge is Y$ = + 1 ,

<A = ( £o ) . (4-29)

where (p+ and tp° are positively charged and neutral complex scalar fields, respec-
tively. The Lagrangian for these scalars is given by

£s = (D^(D'i<f>)-V(^cl>), (4-30)
with

D^ = (0„ - igT- • 1 M - %
-g'B^)<j>, (4.31)

where the explicit form of the covariant derivative is due to Y$ = + 1 . The potential
term V((j>^<j>) being gauge invariant is given by

V{^4>) = m2tf<t> + X^U)2, (4.32)

where m 2 and A are real constant parameters. A should be positive to ensure the
stable vacuum. Further higher power terms of </>V are not allowed in order for the
theory to be renormarizable.
We can also add the coupling terms between fermions and scalars, called Yukawa
interaction terms, which are SU(2)L X U(1)Y gauge invariant and are to provide
the electron mass after the spontaneous symmetry breakdown,

CY = -Ge{L<j>R + R$L) + h.c, (4.33)

where Ge is called Yukawa coupling constant and cannot be determined within the
GWS model itself. One can easily check the SU(2)L X U{1)Y invariance of Cy using
the values of hypercharge of L, R and <>/ defined above.
A full set of SU{2)L X U(1)Y gauge invariant Lagrangian of the GWS model is,
thus, given by the sum of pieces presented above,

£ = £F + £G + £S + £Y. (4.34)
 Spontaneous Breaking of SU(2)i x t/(l)y- Symmetry 61

Strictly speaking, the gauge fixing term with gauge parameter £ (of "i?f gauge")
and the associated term for Faddeev-Popov ghosts also should be included, in the
quantization procedure of the theory. (See, for example, the books by Peskin and
Schroeder (Peskin and Schroeder, 1995) or Pokorski (Pokorski, 2000).)

4.3 Spontaneous Breaking of SU(2)L X U(1)Y Symmetry

As shown in Fig. 3.2(b), the potential V($((>) of (4.32) with positive A and negative
m 2 = —/i2(/x2 > 0) has a minimum at the value of <j> determined by
2 I 2~

tH = \<S>? = y , with <> = y y - (4.35)

Then, as described in Chapter 3, spontaneous symmetry breaking occurs when the
scalar doublet <j> of (4.29) develops a vacuum expectation value

</>o = (0\4>\0) = (v/y/2)

{v ^ J• (4.36)
It should be noted that though T 3 and Y do not annihilate the vacuum cp0

*** - K ;-°.) ( JU )->•

Y<j>0 = <t>o, (4-38)
and thus they are broken generators, i.e. e~la T <fro 7^ <j>o a n d e~l/3^<po ^ </>o, t h e
electric charge operator Q is not a broken generator,

O0o = (T3 + l^ ^o 0== ( JJ no)(,/°v^

L././5 =0. (4-39)
i.e. e~'EQ(j>o = (po, where e is an arbitrary parameter. Therefore, even after the
symmetry breaking, there remains a symmetry associated with the charge operator
Q of U(l)em being compatible with our real world.
Now, it is convenient to parametrize the scalar doublet with 4 degrees of freedom
in terms of the fields denoting the shifts from the vacuum state 4>o,

f+ \ _ „W-«72v ( °

Here the original 2 complex scalar fields <p+ and <p° in (4.29) are replaced by 4 real
fields, &(i = 1,2,3) and H, where & are so-called Goldstone bosons being absorbed
into the longitudinal components of W± and Z° bosons as described later and H is
a Higgs boson. (4.36) leads to zero vacuum expectation values for all of these fields

<0|&|0) = (0|ff|0>=0. (» = 1,2,3) (4.41)

62 THE STANDARD MODEL OF ELECTROWEAK INTERACTIONS

Here we can rewrite the Lagrangian in the "unitary gauge", where 3 Goldstone
bosons & disappear by being 'eaten up' by gauge bosons, W± and Z°, and thus,
physical particle spectra and their interactions become apparent. By applying the
unitary SU(2) transformation
[/(£) = e~ir-(/2v^ ( 4 4 2 )

one can come to the real world induced in the unitary gauge. Then, we can define
the new fields in our real world as

L' = U($L, (4.44)

A; = UiOA^UiO'1 - l
-(d„U(0)U\0, (4-45)

with x = I I a n d AM = A^ • ^, where the new fields transformed from the

original ones are presented with a prime. R and BM remain unchanged under this
SU(2) transformation,

R' = R, (4.46)
Bl = B». (4.47)

The Lagrangian is invariant under this transformation and one can rewrite each
piece as

CF = L'i^^-ig^-A^+'-g'B'JL' + R'i^idv+ig'B'jR1, (4.48)

£G = -IF^F'^-^B'^B'^, (4.49)
Ca = (D^YiD^y-V^'W), (4.50)
CY = -Ge(L'(f)'R' + R'^L') + h.c. (4.51)

Now let us discuss the physics described by this Lagrangian realized in the
unitary gauge. First we consider the scalar sector. The scalar fields generate masses
of gauge bosons and those of quarks and leptons via the Higgs mechanism. Cs is
explicitly written as

CS = {DI14>)'{D^)'-V{^4>'), (4.52)

with

(D^)' = (d.-igyA^-'-g'B'^'

= {d.-ig^-A'^-'-g'B'^v + ^x. (4.53)

 Spontaneous Breaking of SU(2)i x C/(l)y Symmetry 63

The first term of (4.52) contains the mass-squared term for weak gauge bosons
which is originated from the quadratic terms of gauge fields as shown in the follow-
ing,

= ^(g2^ • A " + gpBlB"* - 2gg'B'flA'3n

v2
(g2A'^A^ + g2A'2A'^ + (gA« - g'B'tf), (4.54)
where in turning from the 1st line to the 2nd line, we used the formula T'T^ =
Sij + ielikrk. Now let us introduce charged boson fields W± defined by
Aa -jz i A'2
W± = 4l*±k (4-55)

Then the sum of the 1st and 2nd terms of (4.54) can be written as \g2v2W+W~,i.
±
It means that the charged vector bosons W are massive with the mass

Mw = -gv. (4.56)

The remaining term which is described by neutral fields can be written as

which can be diagonalized into

j(Z> ^) ( 9" +/2 l ) ( % ) = jtf +9 a

W +0'V4", (4-58)

by an orthogonal transformation
Z ^ \ _ / cos6w -sm9w \ / A*
(4.59)
Ap J \ sin8w cos#jy 7 V B'„
M

where 9\v is called the weak mixing angle or Weinberg angle. The diagonalization
leads to
<?'
tanfliy = — (4.60)

or
sin^w = , •, cos#w = . (4.61)

Prom (4.58), we see that the neutral Z boson becomes massive with the mass

Mz = \vy/g2 + g'2, (4.62)

64 THE STANDARD MODEL OF ELECTROWEAK INTERACTIONS

and another neutral boson AM is massless and hence can be identified with the real
photon. Note that in the GWS model the mass of Z° boson is related to the one of
W ± bosons as

MZ = 2*%-. (4.63)
cos 6 w
One can see that the masses of W*1 and Z° are quite large; in fact by using (4.83),
(4.84) and (4.89) below, we can estimate them to be
2
l i p 1 OQ
1 / 2
Mw = ^ = x(^^) —--^-GeV>37GeV, (4.64)
2 2 \J2GF sinflw siatiw
Mz = -¥%- ~ . !fa GeV > 76GeV. (4.65)
cos Uw sin 20w
The values of Mw and Mz are obtained if sinfljy is determined experimentally.
Actually, the value of sin2 6w is obtained in experiment to be around 0.23, leading
to Mw ^ 80GeV and Mz ~ 90GeV.
The potential term (4.32) of scalars becomes, after symmetry breaking,

= - ^ + \{2f)H2 + \vH* + \ H \ (4.66)

From (4.66), we see that the mass of the physical Higgs boson H can be identified
with

MH = > / V , (4.67)

whose value cannot be predicted, based on some principle, in the GWS model. After
all, in the unitary gauge the Lagrangian Cs results in (up to a constant term)

= Id.Hd^H - \M2HH2 - XvH3 - ^ff4

2 2 4
2
+ ?-{H +2Hv) —L-Z^Z" + 2W+W-»
COS2 t/w
+ M^W+W-^ + ^M^Z^Z". (4.68)

Let us consider next the Yukawa interaction term Cy in the unitary gauge

CY = -Ge{L'(j)'R' + R'^L') + h.c.

= -Ge (e'L^(v + H)e'R + €R±=(V + H)e'^ + h.c.

= _^e'e'-%Fe'e'. (4.69)
 Charged and Neutral Currents, Comparison with Effective Fermi Theory 65

Since only physical degrees of freedom appear in the unitary gauge, we can identify
e' to be physical electron and thus we suppress the prime ' hereafter. Then, the first
term of this equation corresponds to the mass term of an electron with the mass

me = ^ v . (4.70)
V2
It is interesting to note that the electron mass is proportional to the vacuum expec-
tation value v of the Higgs boson as well as in the case of weak gauge boson masses.
The second term shows an interaction term of an electron to the Higgs boson with
the coupling constant

% =^ . (4.71)

It is remarkable that the coupling constant is proportional to the electron mass.

In fact, the Higgs coupling to fermions in the Feynman rule (see Appendix
B.3) is given by -i^- = - i ^ ^ - ^ - = -i{^2GFfl2ms from (4.56), (4.89) and
(4.83), where GF is the Fermi coupling constant and mf is the mass of the fermion
coupled to a Higgs boson. Since v ~ 246GeV (see (4.85)), the Higgs coupling to
ordinary quarks (u, d, s, c, b) or leptons (ue, e, u^, [i, uT, T) is extremely small and
one can neglect the Higgs interaction effects on these particles. However, if there
exist extremely heavy fermions such as the top quark, 4-th generation quarks (Q)
or leptons (L), the Higgs coupling to those particles becomes large and cannot be
neglected. (Inazawa and Morii, 1988; Strassler and Peskin, 1991) It might be even
larger than gluon coupling to those extremely heavy quarks.

4.4 Charged and Neutral Currents, Comparison with Effective

Fermi Theory

The fermion part CF of (4.48) in the unitary gauge is written as

CF = L'i^d^L' + R'i^d^B! + gJ^ • A'» + ^J^B'11, (4.72)

2
where

JM = L%^L', (4.73)
Jj = -L'jliL'-2R'lliR'. (4.74)

In (4.72), the 1st and 2nd terms result in the conventional kinetic terms of an
electron and its neutrino

L'i-y^L' + R'i^d^R' = ei^d^e + v^^d^L, (4.75)

where the prime for electron and neutrino fields is dropped in the right hand side.
66 THE STANDARD MODEL OF ELECTROWEAK INTERACTIONS

Fig. 4.1

Prom (4.72), one can pick up the charged current interaction which is given by

Ccc = 9{JlA'^ + J2Al2») = fyj;W-» + J+W+"), (4.76)

where

J± = Jl±Ul. (4.77)

J * are explicitly expressed as

J+ = J^+ Ul = L'^T+L' = DeLllieL = -P e 7 M (l - 7s)e, (4.78)

J» = Jl ~ iJl = L'^T-L' = SL-y^UeL = -S7 M (1 - 7 5 )i/ e . (4.79)

Now, it is interesting to compare these charged currents with current x current

interactions in the effective Fermi theory discussed in Chapter 2. Let us consider
the lowest order process of ve + e —> ue + e in the low energy, which occurs through
an exchange of W^ boson (Fig. 4.1). The corresponding Feynman amplitude is
given by

M = -9-J+» , „ . Mw J~v, (4.80)

2 K
2 q -M^+ic
which reduces to
M i j+ j (4 8i)
=- 4^ " ^ -
in the limit of q2 /M^ -> 0. The same amplitude can be obtained from the effective
Fermi theory as

CeM = ~ • <U+(iJ; = ~ (Pe7M(l - 7s)e) (e 7M (l - 7fe)"e) • (4-82)

 Charged and Neutral Currents, Comparison with Effective Fermi Theory 67

Therefore, one can see that the Fermi theory of weak interactions is the low energy
effective theory of the GWS theory. Comparing (4.81) with (4.82), we obtain

% = sk- <«3>
By using

GF = 1.16639 x l ( T 5 G e V - 2 (4.84)

and (4.56), (4.83) and (4.84), one can determine the vacuum expectation value of
<f> as

v = (V2GF)~* ^ 246GeV, (4.85)

which is called weak scale.

From (4.72), by using (4.59) one can also extract the neutral current piece

CNC = gW + lg'JlB'x
JY
= (gsm9wJl+g'cos9w-±-)A>i
JY
+ (gcoS9wjZ-glsm9w^-)Z». (4.86)

Recalling (4.20), we have

JY
m
JZ = 4 +It-ir- ( 4 - 87 )
2
By using this relation, the current coupled to A^ in (4.86) becomes
JY
gsm6wJl +g'cos6w-^- = g'cos9W J°m + {gsin9w - g' cos9w)J%. (4.88)

The second term vanishes from (4.60). The first term can be identified with the
interaction of an electron to photon and hence we can identify the coupling constant
to be electric charge e, i.e.

e = g' cos 9\y = g sin 9w, (4.89)

or

4 = 4 + 4r (4-9°)
e g g
The current coupled to Z^ becomes

gcos9wJl - g'sm9w-*- = — ^ - ( J ^ - sin2 6 ^ ™ ) , (4-91)

Z COS U\y
68 THE STANDARD MODEL OF ELECTROWEAK INTERACTIONS

and thus we define a new neutral current coupled to Z p as

J* = J*-sin2 0wJ<m. (4.92)

In general, if a fermion family is given by

L=
(f) ' Rf = '*> Rf' = &> (493)
Jjf is explicitly written, by using (4.11) and (4.17), as

Jl = 4-sin2ewj;m
= L 7 / , y L - sin2 OwiQfftnf + Q'ff'^f)

= O{/L7M/L + ^RJRI^R + afL TLI^'L + "R / R 7 / . / * > (4-94)

where Qf, Q', are electric charges of / , / ' (in the unit of e), respectively, and

a{ = - - Qf sin2 6W, a{ = ---Q'fsm26w, (4.95)

aR = -Qfsm26w, afR = -Q'fsm26w. (4.96)

Or, j j is often written in literature as

J? = H(Cv - C{l5)f + I'l^cC - CA\)f, (4.97)

with

Cfv = \(afL + aR) = ~Qfsm26w, (4.98)

C'A = \(4-4) = \, (4-99)

Cv = \(afL+afR) = -±-Q'fsm26w, (4.100)

CfA = \{4-«'R) = -\- (4-101)

Similarly as in (4.81), the lowest order amplitude for the low energy neutral
current processes, e.g. ve + e —t ve + e, shown in Fig. 4.2, is given by

M = - i j v . , , 9 , „ JuJZ,i- (4-102)
2!M|cos20w M
where the factor ^ comes from the fact that j£ J^ M is the square of the identical
current. Thus, the corresponding effective Lagrangian for neutral current processes
can be written as

C^ = -^-U^Jz\ (4.103)
 Charged and Neutral Currents, Comparison with Effective Fermi Theory 69

Fig. 4.2

defining QL = 8M a fosi e • Hence, we see that the total effective Lagrangian at low
energy can be described by

£e// =
" 7 § ' 4(J^+J~"+ pJZjZ,i)> (4 104)
-
where
n„N
G M Ml.
^
p
GF M | cos2 6W 1. (4.105)

It should be noted that p = 1 is the inevitable consequence of the fact that the
Higgs scalars belong to the doublet representation of SU(2)L- It is no longer the
case for Higgs scalars with higher dimensional representations of SU(2)L, such as
triplet. To see this explicitly, let us consider the case with higher representation of
Higgs multiplet. Then, we can only replace the Higgs doublet <j> in Cs of (4.52) by
a new Higgs scalars <j>'h(x) of higher multiplet. The gauge boson mass term can be
written as well as in (4.54)

Cmass = ^x[(gfh • A',, + 9'^BJ{gfh • 4 + g'^-B^xn, (4.106)

where Th and Y/, are appropriate generators corresponding to SU(2)L and U(l)y,
respectively, and Vh is the vacuum expectation value of this Higgs scalar. \h rep-
resents a similar vector as in (4.43), in which only one nonzero component is 1 and
others are zero. For this multiplet we have again Qh — Th3 + ^-, as before, where
Qh is charge of h and Thz is the 3rd component of 7/,. Now, let us put Qh = 0
for one component of <j)'h{x). For this component, we have ^ = —Thz- Then, the
vacuum expectation value Vh of this field generates the masses of gauge bosons, ex-
cept the photon, in the same way as the case of the doublet Higgs scalars described
before. The resultant mass term is

V
Cmass = ^x\{^2fh-A^fh-A^ + g'\^fB'^ + 2g9'fh-A'^-B'^xh
70 THE STANDARD MODEL OF ELECTROWEAK INTERACTIONS

r.2,,2 9
= -^iAnAn,T2i+Al2Al2,T22)

9
+ -^f
L
(A*A*»1& + ^ W " - 29-TtsA«B'»
\ 9 9
= ^(A^A^ni + A^A^Ti2) + ^£^-ZlT2h3, (4.107)

where in the second and the third lines, Xh^liXh (* = 1,2,3) has been written
simply as T ^ . Prom this equation, we have

P
(4.108)
Mfcos^ ZhvlTL'
with

TL = \{Tl + T2h2) = \{fl - Tl3) = \ (lh(Ih + 1) - T2h3) , (4.109)

where //, is the (magnitude of) weak isospin of h. Therefore, p can take any value,
in principle. When all Higgs multiplets belong to the same representation, as in the
case of the supersymmetric extension of the standard model, i.e. the "Minimal Su-
persymmetric Standard Model" (MSSM), YLiv\ c a n c e l °ut between the numerator
and the denominator of (4.108) and we have

iw+D-na (4110)
HZ

One can easily check p = 1 for Ih = \ and T/j3 = ± | , as in the case of the standard
model or MSSM. In other words, the GWS model with minimal Higgs doublet
necessarily predicts p = 1 at the classical (tree) level, though the p deviates from 1
at the quantum level, as we will see in Chapter 8.

4.5 Addition of More Leptons

So far we have considered only one lepton family, i.e. electron e and its neutrino ve.
The GWS model can be easily extended for the case with more lepton families. Let
us first consider, for simplicity, the case of 2 lepton families, electron (1st generation)
and muon (2nd generation) families, represented as

i e = ( M , Re = eR; Lll=(V*\ , i?M = /ifi. (4.111)

(1st generation) (2nd generation)

The Lagrangian for this fermion sector is given by

CF = CP+CP, (4.112)
 Addition of More Leptons 71

where both terms have the same expression as (4.23) by replacing L and R with
those given in (4.111), respectively. The Lagrangian for gauge field and scalar boson
sector remains unchanged. However, as for the Yukawa interactions, new terms are
allowed without contradicting with the requirement of SU{2)i x £/(l)y invariance,

Cy — -GeeLe(pRe - GftfiL^Rft - GeiiLe4>Rn - G^L^Re + h.c. (4.113)

Notice that here we require only the local gauge invariance and do not assume
separately the electron number and muon number conservation to start with. After
the spontaneous symmetry breaking, one can easily obtain the fermion mass term
from Cy just as we did before for an electron family.

^, = -{meeeLeR + mllfl,iJ,Lij,R + metieLfiR + mfiefiLeR + h.c.)

-^fiL)(y Ze")(eR.)+h.c (4-114)

m^e mm J \ [i-R

where my = Gij-75 (i,j = e or (j).

Note that the mass matrix

iHp.e.
ee '"'efj,
efi
M = (4.115)
\ mMe mMM )
is, in general, complex and not necessarily hermitian and is not diagonalized by use
of one unitary matrix. In general, it can be mathematically shown that an arbitrary
n x n complex matrix M can be diagonalized by a bi-unitary transformation as

WMV = Md, (4.116)

where Md is a diagonal matrix and both of U and V are unitary matrices, which
are the unitary matrices to diagonalize the hermitian matrices MM^ and M^M:
W(MM*)U = V t ( M t M ) V = MdM^ Redefining the phase of \x and e fields
appropriately, one can always make the elements of M& be real and positive as

WMV-*.-(* I), (4.117)

where me and mM correspond to the masses of electron and muon, respectively.

To follow this program, let us diagonalize the M of (4.115) by applying the
unitary matrices,
COs9 s i n
U=(COS6L -sinM V=( » - M. (4.118)
\ smffL COSUL J \ sinflii cosflfl /
To be precise, each of U and V may have a few phases, in addition to the mixing
angle, 6L or 8R. These phases, however, turn out to be physically irrelevant (see the
discussion in the subsection 4.6.2 below). Thus here we simply ignore those phases.
72 THE STANDARD MODEL OF ELECTROWEAK INTERACTIONS

Then the £}™ass> c a n De rewritten as

4mo") = -(eLjlL)M( eR
R \+h.c.

6
= -{e'Lp!L)Md( 'f )+h.c. (4.119)

where

(e't)=Ul(eL) = ( °°?'j Sin L eL

° )( ), (4-120)
V ML / V Pi / V -sin^L COS0L J \ fiL J
sVl( <*) = ( ™<>n d*«R ) ( e " ) . (4.121)
\ m J V -sin^H cos0 R J \ HR J
Notice that the fields with prime (e^ #,/•*£ «) which are called mass eigenstates are
not equal to the fields without prime (eL,R, HL,R) which are called current/weak
eigenstates. The observed states are mass eigenstates. Therefore, when we write
weak charged currents interaction in terms of the observed states, the interaction
is no longer diagonal in the generation space and we potentially have an intergen-
erational mixing. The detail will be discussed later in section 4.6 when we discuss
the quark sector.
However, here we just point out that in the case of massless neutrinos the mixing
angle 9i can be rotated away and thus, the mass eigenstates of leptons become equal
to weak eigenstates. This can be seen considering the lepton doublet (4.111). By
using (4.120), the weak eigenstates are written as

L6 = ( Ue ) = ( UeL
)
\ e )h \ cos6Le'L-sm6Ln'L J'

L _ ( "/• ^ _ ( V»L \ (4.122)

M
V M) L V sin6» L ei,+cos6>zX J

As is easily shown, under the following rotation between the lepton doublets
/ Le \ / L'e \ _ / cos a sina \ / Le \
(4.123)
\ L^ J \ L'p ) \ -sina cos a J \ L„ J '
we have

LehaDaLe + LprfDcLn = l'tilaBaL'e + V^-fDJJ^ (4.124)

Then, if we choose a = 9L , we can write

L'e = COS8LLS +sin^x,L M = I , J' (4.125)

n • a T , a T ( - s i n 0 L i / e i + cos0z,i/ML \ . .
LJ, = -sm8LLe + cos0z,LM = I , )' (4-126)
 Addition of More Leptons 73

Here ve and v^ are degenerate in their masses, as they are both massless, and
therefore we can redefine the observed neutrino fields as

v'eL = cosOLveL +smOLi>nL, (4.127)

V'^L ~ -sin8Li>eL +cos8Lv^L, (4.128)

and so we end up with the weak doublets, which are still diagonal in the base of
mass eigenstates. The mass eigenstates are now written as

D (4129)
-{"i)L' ">~{$)L-
and there is no flavor mixing between electron and muon in the weak charged
current interactions. Thus, if we define individual lepton numbers as ne (electron
number) = + 1 for e'~, v'e; ne = - 1 for e'+, v'e and nM(muon number) = + 1 for fj,'~,
v
'\i'i nn = _ 1 f° r M' + J v'tn respectively, then it is naturally derived that both of the
electron number ne and the muon number nM are conserved separately. The GWS
model automatically leads to the conclusion that the decay mode ^ -¥ e^ + 7 is
strictly forbidden. It is remarkable to note that the flavor conservation was not put
in by hand but originated from the massless nature of the neutrinos in the GWS
model. In other words, if neutrinos are massive, one can no longer redefine the
neutrino fields as in (4.127)-(4.128) and the weak eigenstates of neutrinos are given
by a linear combination of the mass eigenstates v\ and 1/2, which leads to neutrino
oscillations, as we will see in Chapter 6. (In the quark sector which will be discussed
later in section 4.6, the mixing angle cannot be rotated away because quarks are
massive with different masses.)
Let us next discuss the Higgs boson couplings to electrons and muons. Prom
(4.113), one can extract them in the unitary gauge as

m m
CHU = -~(eL H) ( - <» ) ( * « ) + h.c. (4.130)

When we go to the mass eigenstate by using (4.120) and (4.121), CHU turns into

e
CHU = -~{e'L ~A)Md ( 'f ) +h.c.

= (mee'e' + mlip,'n'). (4.131)

This shows that (i) Higgs couplings do not mix the lepton flavors and (ii)they are
proportional to the masses of leptons, which couple to the Higgs.
In conclusion, there is no lepton flavor number violation in the standard GWS
model; the electron number ne and muon number n^ are strictly conserved. Exten-
sion of the model to tau lepton family (uT, r) is straightforward and the tau number
74 THE STANDARD MODEL OF ELECTROWEAK INTERACTIONS

nT is also conserved. At present we have 3 generations of leptons;

( : - ) , - ( ; - ) , - ( ; - ) , - <«->
In the standard GWS model, ne, n^, and nT are separately conserved. Therefore,
the processes such as fi~ -> e~7, fi~ -> e+e~e~ and \i~ + Z —> e _ + Z, etc. are
forbidden at the all orders of perturbation in the GWS model. Therefore, if some
of these processes are discovered, it will clearly signal the presence of some physics
beyond the standard model, such as MSSM.

4.6 Extension to Quarks

The well-established observation of left-handed charged weak currents of hadrons

suggests that the left-handed components of quark fields should be constructed into
a doublet similarly to the case of leptons. Now we have the following 3 generations
of quarks.

(1st generation) I j , UR, dR,

(2nd generation) I 1 , CR, SR, (4.133)

b
(3rd generation) ( . ) , *fl> R,

The GWS model can be easily extended so that it can incorporate these quark
families, though there are several differences between quarks and leptons; (i) quarks
have three color degrees of freedom, whereas leptons are colorless. However, since
the electroweak interactions are color-blind, one can simply suppress the color index
of quarks in the GWS model, unless otherwise mentioned, (ii) all quarks are massive,
while neutrinos are considered to be massless in the GWS model. Therefore, we
have 2 right-handed singlets such as UR and <1R for each generation, compared to
the case of lepton families where we have only one right-handed singlet such as
eR. Thus, the Lagrangian must contain an additional term for the up-type right-
handed singlets like UR, in order to form their mass term, (iii) charge of quarks are
different from that of leptons and thus, keeping the relation, Q — T3 + y , weak-
hypercharges of quarks are also different from those of leptons. The charge Q, weak
iso-spin (T,T3) and weak-hypercharge Y of these quarks are summarized in Table
4.2.
To write the Lagrangian for quarks in a compact form, we introduce the following
notation

O i < = f n ' l ' U*> -°™> (i = 1 ' 2 ' 3 ) ( 4 - 134 )

 Extension to Quarks 75

quark family Q (T,T 3 ) Y

UL,CL,tL T
3 (§,+§) +1
1
d,L,SL,bL 3 V2' 11 +3-
UR,CR,tR ^3 0 +1
dR,sR,bR 1 2
3
0 3

Table 4.2 Quantum numbers of quark families.

where i stands for the generation: U\ = u, U2 = c, U3 = t, D\ = d, £>2 = s, D3 = b.

Based on Table 4.2, we can explicitly write down the Lagrangian for these quarks
with covariant derivatives as

CF = Y^QLih,1{d^-ig1--All-l-glBlt)QLi

3
+ ^tW^-i^Egt/m
3
+ ^2DBii'f(dlt + '\<"
Ti-g'BJDm. (4.135)
i=l

The terms for gauge fields CQ and scalar fields Cs remain unchanged from the case
of leptons. As for the Yukawa coupling, we can write the following SU{2)L X U{1)Y
invariant terms

CY = - YsFljQLitDRj + V^QL^URj + /i.e.), (4.136)

where 1^ • and T\j are Yukawa couplings which are, in general, unconstrained
complex parameters. 0 whose weak-hypercharge is —1, i.e. Yi = —1, is defined
from 0 as
0*
4> = {TI4>* = (4.137)
-<p-

where 0* is the complex conjugate of 0 given by (4.29). Inclusion of 0 in the 2nd

term of (4.136) is allowed because the fundamental representation and its conjugate
are equivalent in SU(2) group (see Appendix C3), though this is not the case for
any other SU(N) groups with N > 3. This may be easily seen as follows; Since 0+0
is SU(2)L invariant, so is (0*0)' = 0'0* = 0 ' ( - * T 2 ) ( J T 2 0 * ) = 0*e0 (e: Levi-Civita
tensor). On the other hand, anti-symmetric product of arbitrary two doublets is
also SU{2)L singlet, which means that 0 behaves as a doublet.
76 THE STANDARD MODEL OF ELECTROWEAK INTERACTIONS

After spontaneous symmetry breaking, moving to the unitary gauge, we have

0->-L( w + ff)( J ) , (4.138)

4>^-j=(v + H)[ J ) , (4.139)

and we can see that the 1st and 2nd terms of (4.136) yield the mass matrices of the
down-type and up-type quarks, respectively.

Qu4>DRj -> (ULi DLi)—=- f 1 J DRj

= -^DLiDRj + ^=DLiDRj, (4.140)

QLi4>URj -> (tfL« DLi)^^f ( I ) t/flj

= -^UUURJ + -^ULiUnj. (4.141)

Here, all fields in the right-hand side are ones redefined in the unitary gauge

4.6.1 Quark mass matrix

As described above, we can extract the quark mass matrices from Ly of (4.136),

£(">*) = -Y^DLiM^DRi -Y^ULiM($URi + h.c. (4.142)

with
M{D) _ „(£>) _^_ M(U) _ r(U) V_ ,. 14o^

which are, in general, complex-valued matrices.

By applying the bi-unitary transformation to each term of (4.142) as we did in
the lepton sector, we can diagonalize the mass matrices, MSD^ and MSU\ sepa-
rately.

tfDMiD)VD = Af<D> (4.144)

{u)
U}jM Vu = M^ (4.145)

where UD,U and VD,U are unitary, and M^D,U^ are diagonal matrices. Thus we have
moved to the base of mass eigenstates:

D'L = tfDDL, (4.146)

D'R = VDDR, (4.147)
 Extension to Quarks 77

U'L = U\jUL, (4.148)

U'R = VlUR, (4.149)

where we used the matrix notation with

/ DLI
\ / D'L1 \
DL2 D'L2
f
DL D'L = (4.150)

V I \ /
etc. and the fields with and without prime denote the mass eigenstates and the
weak eigenstates, respectively. Then, we have

£(™Q) = -D'LM^D'R - D'RM^D^D'L - U'LM^U'R - U'RM^U^U'L. (4.151)

D
Though M ' ) and AfW) are diagonal matrices, they are complex, in general. How-
ever, if we parametrize the i-th diagonal element, using the absolute value m\ ' '
and phase a\ ' , as

<»> = m i V , (4.152)
M.(u)
ia
= m) 'e ' , (4.153)
we can absorb all these phase factors into new quark fields induced by a chiral U(l)
transformation, which keeps physics unchanged.
(D)
D'Li -»• e" /2D',Li' (4.154)
,-la^/2D, (4.155)
D'Ri
U'Li e*°r/'U'u, (4.156)
U'Ri ->*-*>#" fiV'n. (4.157)

Then, finally the mass terms read as

£(mo) = -D'm^D' - U'm^U', (4.158)

Du
where m^ ' ^ are real and diagonal,
md 0 0 mu 0 0
m^ = 0 ms 0 m{u) = | 0 mc 0 (4.159)
0 0 mb 0 0 mt
and D' and U' are down-type and up-type quarks in the mass eigenstates, respec-
tively.
d'
D' = U' = (4.160)
V t'
78 THE STANDARD MODEL OF ELECTROWEAK INTERACTIONS

In (4.159), the matrix element rrn represents the quark mass of the i-th flavor.
In the above discussion on the GWS model, we have implicitly assumed that
we have only electroweak interaction as the gauge interaction. However, if we take
strong interactions of quarks into account, £( m «) must have another piece induced
by the QCD anomaly of the chiral U{1) current; this is related to the problem called
"strong CP problem". (See, for example, the review by Peccei (Peccei, 1989).)

4.6.2 Flavor mixing

4.6.2.1 The case of charged current interaction
Now let us discuss the charged weak current interactions, which can be extracted
from the first line of (4.135)

4 c = ^ E ULi-fDLiW* + ^ E DLi-fUuW-, (4.161)

where the sum of i is taken for all generations. For a while, we let the number of
generations, which we denote by n, be arbitrary, for generality. In terms of mass
eigenstates defined in (4.146)-(4.149), we can write this CQQ as

4 Q c = -jfilA'fUDD,LW+ + ^D'LUI^ULU'LW;, (4.162)

where D'L etc. are column vectors defined in (4.150), though now they are n-
component vectors. Then, defining a flavor mixing matrix V, which is unitary,

V = U\jUD, (4.163)

we can write

4°c = ^U'LrVD'LWt + -!LD'LV^»U'LW;,

= JL-U'-fQ. - ^)VD'W+ + h.c. (4.164)

2\/2
Here we are interested in how many physically meaningful independent parameters
exist in V. V is a n x n unitary matrix with n 2 real parameters. (Because of the
unitarity condition of V, there are n 2 relations among 2n 2 parameters of V. Then
the number of remaining parameters is n 2 .) As is well-known, a real unitary matrix
is an orthogonal matrix with n(n—1)/2 parameters. Then, V has n{n —1)/2 angles
and n 2 — n(n — l ) / 2 = n(n + l ) / 2 phases. However, some of these phases can be
absorbed into the new quark fields redefined by phase transformation; for example,
when we consider

U'L-fVD'L = ULir(VuD'L1 + Vl2DL2 + --- + VlnD'Ln)

+ U'L2r(V2iD'L1 + V22D'L2 + --- + V2nD'Ln)
 Extension to Quarks 79

+
+ U'Lnr(VniD'L1 + Vn2D'L2 + --- + VnnD'Ln), (4.165)

the n components of the 1st column (Vn, V21, • • •, V n i)' of the mixing matrix V can
be made real by writing Vn = \Vn\eiai and redefining the up-type quark fields as
U'Li -> e"*< U'Li (i = 1,2, • • •,n). Similarly, we can make the (n — 1) components
of the 1st row(Vi2, V13, • • •, Vin) real in the redefinition of D'Li -¥ e~*a< D'Li (i =
2,3, •••,n). (Note that Vn = |Vn| has already been made real by the redefini-
tion of U'L1.) After this procedure, (In — 1) phases are removed from V without
changing physical consequences. After all, remaining parameters are characterized
by n(n - l ) / 2 angles and n(n + l ) / 2 — (2n - 1) = (n - l)(n - 2)/2 phases.

• Four quark case (n = 2)

Here the quark mixing matrix V can be parametrized by only one flavor mixing
angle 6C, called Cabibbo angle. There is no phase (("~ 1 H"~ 2 ) = 0 for n = 2)
and hence we have no CP violation in this case because of the absence of complex
coupling constants. The charged weak current interaction can be written as

d
Ccc = ^=(« c)r^~^V ( s ) W+ + h.c. (4.166)

with

v _ ( Vud Vus \ _ ( cos6c sin6c \ (Aim)

{ }
V Vcd Vcs J V -sin^c cos^ c ) •

Here and hereafter, the mass eigenstates of quarks are simply denoted by u, d, etc.,
suppressing the prime '.
This means that the left-handed quark doublets are written as

o(1)-
(:),• ° H : ) , -
L <"«>
where
dc \ _ ( cos6c sin# c \ ( d
(4.169)
sc ) L V -sin^c cos^ c ) \ s , L

and the weak charged current processes are due to the transition u «-» dc or c •(-> sc.

• GIM mechanism

In 1960s, only one quark doublet was known from the analysis of the ft decay of
neutron and A which occur via weak charged current transition u «-» dc between
80 THE STANDARD MODEL OF ELECTROWEAK INTERACTIONS

the members of the doublet,

Q{
? = (l)L = {dcosecls^ec)L < 4 - 17 °)
At that time, the orthogonal combination of CIL and SL, i.e. sc = scos6c — dsm9c,
was left as a SU(2)L singlet. Then, when we apply this to neutral currents, we have

QL^I^QL^ = -^LlnUL - cos2 9cdLi^dL - sin2 OCSL^SL

- cos0csm6c(dLJnSL + SLlndL)), (4-171)

where the final term corresponds to a strangeness changing neutral current. (We can
check that the part of the neutral current, proportional to the electric charge, does
not contain such strangeness changing neutral current.) If this term really exists, we
necessarily predict a rather large decay width of the flavor changing neutral current
transition such as K^ -> fi+fj,~ or K+ -> n+uu, since flavor changing sdZ and dsZ
vertices appear already at the classical (tree) level. But this prediction is completely
in disagreement with experimental results, Br(K^ —• fi+fJ.~) = (7.25 ± 0.16) x
lfT 9 , or Br(K+ -> ir+Dv) = (1.6+Jj) x 1 0 - 1 0 . (See Chapter 9 for some detailed
discussions on these flavor (strangeness) changing neutral current processes.)
In 1970, Glashow, Iliopuolos and Maiani(GIM) (Glashow, Iliopuolos and Maiani,
1970) proposed a new mechanism to solve this problem by introducing the 2nd quark
doublet which contains the 4-th quark, which is now called the charm quark c,

Q = (4 1?2)
™ ( sc ) L = ( S cos0 c - dsm9c ) L ' -

which produces additional neutral current,

Q(L}7MT^2) =
2^Ll,iCL ~ C ° s 2 e<:^SL + sin2 e
cdu^dL
+ cosdcSmOcidLjfiSL+ sLj^dL)). (4.173)

Summing up (4.171) and (4.173), we find that the neutral current finally becomes
flavor diagonal;
3 3 1
Q(L1»\Q(L+Q(LIJ-^Q(L = ^{uL^uL+cLlllcL-dLllldL-sLlliSL), (4.174)

and we have no flavor changing neutral current at all at the tree level.
Then, how the flavor changing neutral current transitions are possible, even
though it is extremely suppressed? GIM proposed a wonderful idea that it will occur
at the quantum level via, for example, the higher order charged current interaction
shown in Fig. 4.3 (for the K\ -> /u,+ fi~ decay). The contribution of the diagram
 Extension to Quarks 81

sin0 c
-* f\AAAA| •-
W+
K° u ..
W~
^ W W V •
COS#c

(a)

COS0C
Y W V W
W+
K° c ,i

w-
— sin 6C
(b)
Fig. 4.3

2 2
(a) to the decay amplitude behave, ignoring the powers of small jjX- a n d
j$f", a s

<74 sin 0ccos6,

A(K°L -* p+p~) <x GFCtsmdccosOc, (4.175)
M*w

where a = e2/47r is the fine structure constant. But this contribution, though
suppressed by the factor a, is still too large to account for the decay width of
the "rare process". However, the contribution of the 2nd doublet introduced by
GIM, as shown in the diagram (b), yields a contribution, — g s ' n A / f 2 C Q S c > which is
just opposite to that of (a). Thus the amplitude exactly vanishes at the leading
m2
order of the perturbation in the powers of -rrf-
M
(q = u,c), and therefore the non-
w
m2
vanishing contribution stems from the 1st order of TI§-,
m yielding a suppression factor
w
mc—jn 4
3 x 10~ . This mechanism of suppressing flavor (strangeness) changing
ATST
neutral current is called G I M mechanism. Hence, t h e amplitude behaves as

i/T,n 4- - \ g 4 s i n # c c o s 0 c m2 — m2 _, . . „, , •>-.,.
A(K°L-+ii+v )<x^ -| '-• \ 2 " «G2Fsm6ccosec(m2c-ml). (4. 176)
Mlw MW
Therefore, if mc = mu, K% -> fi+n~ decay is strictly forbidden. Gaillard and Lee
(Gaillard and Lee, 1974) predicted the mass of the introduced c quark as the value
to reproduce, via the above formula, the experimental value of Br(K^ —> fx+fi~),
before the discovery of the c quark at the collider experiments. The predicted
value was quite consistent with the value (mc « 1.5GeV) determined from the
spectroscopy of charmonium. More detailed discussion of these rare processes in the
82 THE STANDARD MODEL OF ELECTROWEAK INTERACTIONS

full six quark (3 generation) Kobayashi-Maskawa model will be given in Chapter 9.

It is remarkable that the GIM mechanism was proposed before discovery of J/ip in
1974.

• Six quark case (n = 3)

Here V has 3 angles and 1 phase. Because of the existence of 1 phase, we

can expect the CP violation in this case. The extension of the model to the 3
generation scheme, in order to accommodate the observed CP violation in KL decay,
was first proposed by Kobayashi and Maskawa (Kobayashi and Maskawa, 1973).
They generalized the Cabibbo mixing matrix, (4.167), to the six quark case of
3 generations and thus the proposed 3 x 3 mixing matrix V is called Cabibbo-
Kobayashi-Maskawa(CKM) matrix. It is a remarkable fact that the proposal by
Kobayashi and Maskawa was made before the discovery of the c quark in the 2nd
generation! The GWS model extended to the Kobayashi-Maskawa 3 generation
model embodies the standard model of electroweak interaction nowadays. In this
case the charged weak current interaction becomes

Ccc = fy*5 t)r^Y^V I s \w+ + h.c. (4.177)

There are several pamaretrizations of the CKM matrix V, which are physically
equivalent. Using 3 flavor (generation) mixing angles, 6\, 62, 03, and a CP violating
phase, 5, one example is given a la Kobayashi and Maskawa by

V = R1(e2)R3(61)C(0,0,6)R1(e3), (4.178)

where Ri is a rotation matrix around the axis i

Ri(0i) = I 0 a Si I, R3(0i) = I -si a 0 I (4.179)

Ci

with Ci = cos 6i and Si — sin 6i and

C(0,0,S)= 0 1 0 . (4.180)

V is explicitly written as
S1C3 S1S3
V = I -sic2 C1C2C3 - s2s3ei5 C1C2S3 + s2c3eiS | , (4.181)
-C1S2C3 - c2s3etS -C1S2S3 + c2c3e%s
where the values of the 3 angles and the phase are not predictable within the
framework of GWS model; they must be extracted from experiments. A phase 8
 Extension to Quarks 83

cannot be removed by the redefinition of quark fields and leads to the CP violation
of the processes including S. It is noted that the phase S disappears if any of #i,
62, 93 vanishes; If 63 = 0, Ri(93) = 1. Then, the phase S can be absorbed into

the newly defined fields D' = C(0,0,S)D, where D = l s , i.e. by "re-phasing"

VbJ
of the quark fields. For the case of 02 = 0, using the relation R3(9i)C(0,0,S) =
C(0,0,<5)i?3(#i), the phase S can be also removed by redefining the new field U' =

C(0,0,-6)U with U = c ). For 9X = 0, we have V = Rx(92)C{0,0,6)^(63)

which leads to

V = 0 a b , (4.182)

where the 2 x 2 submatrix in V can be made real by suitable re-phasing, just as in

the case of n — 2.
Another parameterization of V is the so-called "standard" parameterization
which is taken by the Particle Data Group (Particle Data Group, 2002) and is
characterized in terms of 3 angles 8u, #23, #13 and a phase £13 as

( -S12C23 -
C12C13

Ci2S23Sl3etl513 C12C23 -
S12C13

Sl2S23Sl3elc513
si3e~iSl3

S23C13
\

, (4.183)
S12S23 - Ci2C23Si3e'Sla -C12S23 - Sl2C23Sl3e"513 C23C13 /
where Cij — cos 6ij and Sij = sin8ij (i,j = 1,2,3).
A convenient way to approximate V has been proposed by Wolfenstein (Wolfen-
stein, 1983); it is a parameterization by expanding each element of V in the powers
of the sine of Cabibbo angle, A = S12 — 0.22. In the approximation up to the order
of A3, it is written as
/ l-A2/2 A AX3(p-iri) \
V = -A 1 - A 2 /2 AX2 , (4.184)
\A\3(l-p-irj) -AX2 1 /
where A, p and rj are real parameters of the order one, and r] corresponds to the
CP violating phase. This is often used in phenomenological analysis because this
parameterization shows clearly the hierarchical structure of the CKM matrix.

4.6.2.2 The case of neutral current interaction

Let us next consider the situation of neutral current interactions,

4 $ = eJrA" + 7^h-(4 ~ sin2 faJF)2"' ( 4 - 185 )

COS V\v
84 THE STANDARD MODEL OF ELECTROWEAK INTERACTIONS

where

Je™ = lulltU-±Dj„D,

4 = \(UL1VUL - DL^DL). (4.186)

These currents remain flavor diagonal, even in the base of mass eigenstates, because
of the unitarity of Uu, KD, VU and VD- For example, one can easily see

U^U = ULI^UL + UR^UR,

= U'LuWUvU'L + U'RV^VUU'R,

= U'^U'L + U'R^U'R,
= U'-ypU'. (4.187)

Therefore, we have no flavor changing neutral current at the tree level. The kinetic
terms of quarks are also unchanged when we move to the mass eigenstates. Thus,
we have no CP violation solely due to the weak neutral current and electromagnetic
interactions.
What about Higgs boson couplings to quarks? From (4.133), (4.140) and (4.141),
one can extract the Higgs-quark interactions in the unitary gauge as
H -
Cjjqq = (muuu + mddd + msss + rribbb + mccc + mitt), (4.188)
which is also flavor diagonal. Then, the Higgs couplings to quarks in the standard
model with one Higgs doublet do not generate CP violation. It should be noted
that this is not the case in the models with multi-Higgs, such as the models of spon-
taneous CP violation discussed by Lee and also by Weinberg (Lee 1974; Weinberg
1976).

4.7 Anomalies

Renormalizability is an essential principle for the theory to be meaningful at the

quantum (loop) level. It is realized by sophisticated cancellation mechanism implied
by gauge invariance, which is related to the current conservation called Ward-
Takahashi identities (Ward, 1950; Takahashi, 1957). However, in field theories
there can be the case that a conservation law holding at the tree level is violated
at the quantum level by the contributions of loop diagrams. This is known as
anomalies. A typical example is the triangle anomaly which is ofFen called the
Adler-Bell-Jackiw(ABJ) anomaly or the 75 or axial anomaly (Adler, 1969; Bell and
Jackiw, 1969; Bardeen, 1969). The "gauge anomaly", the anomaly of local gauge
symmetry, is originated from the (linear) divergence of a fermion loop diagram of
triangle shape with 2 vector and 1 axial vector currents at three vertices, as shown
in Fig. 4.4. If this anomaly remains, the renormalizability of the theory is spoilt
 Anomalies 85

(VWW^

Fig. 4.4

and the model becomes meaningless at the quantum level (Georgi and Glashow,
1974; Gross and Jackiw, 1972; Bouchiat, Iliopoulos and Meyer, 1972). In general,
anomaly cancellation is not trivial for the models with parity violation, containing
both vector and axial vector currents, in contrast to the case of e.g. QED where
only vector current exists. Thus, it is very interesting to examine if the anomaly
cancellation works for the GWS model which has both vector and axial vector
currents.
Let us consider, in general, the Lagrangian describing an interaction of left-
handed and right-handed fermions with gauge bosons A1^,

C = $L-f(d» - igTlA^L + fo7"(0M - igT^A^n, (4.189)

where the summation over i is implied. VL and tpR are column vectors composed
of left-handed and right-handed fermions, belonging to some representations of the
gauge group, and T£ and TR are the corresponding generators of the gauge group
G satisfying

^L,R^L,R\ — ifijkTLtR, (4.190)

for each of TlL and TR, where fak is the structure constant of G. Then, the currents
coupled to gauge bosons A1^ are given by

= ^ 7 M ( 1 - 7 5 ) ^ + ^ ( 1 + 75)7^- (4.191)

Now it is known that the triangle anomaly associated with the triangle diagram
with external gauge bosons A1, Ai, Ak is proportional to
Aijk = Aijk _ Aijk ^ (4192)

where the contributions of the right-handed and left-handed fermions are given by

AfL = Tr ({2£ > L ,2£ i L }7* L ) , (4.193)

and do not depend on the fermion masses. The anti-commutator in the above
equation comes from the exchange of A1 and A^ gauge bosons, which are assumed
86 THE STANDARD MODEL OF ELECTROWEAK INTERACTIONS

to couple to the vector currents at the two vertices. Then, the condition for being
anomaly free is

Tr ({TA,:4}T*) - Tr ({rL,T[}T£) = 0. (4.194)

For the vector {TlL = T^) or vectorlike (TlL = [7 -1 Tp£7 with a unitary matrix
U) gauge theories*, the anomaly free condition (4.194) is automatically satisfied.
Typical examples of the vector theories are QED and QCD, both conserving parity
symmetry. Another possibility to satisfy (4.194) is the case
Aijk = Aijk = 0 ( 4 195)

This is realized, for example, for the gauge group of SU(2);

f c
^' =Tr({^,y}y)=0 (4-196)

due to {T1,^} — 26ij. This property essentially comes from the fact that the
representation of SU(2) is "real", and therefore the contributions of a Weyl fermion
ip and (V0C! w ith c standing for the charge conjugation, should be the same, while
their contributions to the anomaly should have opposite signs, as their chiralities
are different.
Then, how about the GWS model? In the GWS model with only one family
of leptons, matter fields are the left-handed doublet ( f e i , e Z)* a n d a right-handed
singlet e^ and the generators for the gauge group are given by

T[ = j , Y ; TR =
¥' (4J97)

=
It is evident that T r M ^ ^ - l V ) 0 a s shown above. Then, for the model to be
anomaly free we must impose

Tr ( { £ , y } y L ) = \SijTrYL = ^ T r ( 2 Q - r 3 ) = S^TrQ = 0, (4.198)

and

TrYj? - TrT£ = 0. (4.199)

(Note that another possible anomaly associated with a diagram with currents ac-
companied by single r% and two weak-hyper charge generators Y simply vanishes,
thanks to the relation Trr 4 = 0 . ) Here for right-handed singlets, Q = ^f- and thus
we have

TrFJ = 8TrQ 3 oc TrQ, (4.200)

•Currents are written as Jf, = I ? L 7 / I ^ 1 V L + ^Rl^T^ipR — i>Llv.T)J)L + •tpR*ipUTlLU-1T(>R =

$7 M T i i/', where ip = ipL + U~lipR and T* = T [ .
 Anomalies 87

since Q 3 = Q for leptons, and for left-handed doublets, we see

TryL3 = Tr(2Q-r3)3
= 8TrQ 3 + 61V (Q(r 3 ) 2 ) - 6Tr(Q 2 r 3 ) - Tr ((r 3 ) 3 )
oc TrQ, (4.201)

where we have used relations (T 3 ) 2 = 1, I r ( r 3 ) 3 = 0 and Tr(Q 2 r 3 ) = TrQ for

leptons. After all, for the GWS model to be anomaly free, it is necessary to have
TrQ = 0. (4.202)
This condition is not satisfied in the GWS model only with leptons and should be
satisfied by extending the model so that it includes quarks with 3 color degree of
freedom. For example, for the 1st generation model of leptons and quarks with Nc
color degree of freedom

{V:)L^ {ud)L>uR'd«> (4 203)

-
the condition of anomaly cancellation

TrQ = 0 - 1 + Nc(l - i) = 0 (4.204)

can be satisfied only for iVc = 3, where 0, —1, § and — | represent the charges
of ue, e, u and d, respectively. (Though the statement that (4.199) is equivalent
to (4.202) was made for the pure leptonic model, it turns out to hold even after
the inclusion of the quarks.) The lepton anomaly can be cancelled by the quark
anomaly only when the color degree of freedom is 3. Therefore, the standard model
with G = SU(3)C x SU(2)L X U(1)Y suggests that leptons and quarks are closely
related to each other. The real unification of quarks and leptons in the multiplets
of gauge group is realized in the "Grand Unified Theories (GUT)" to unify strong
and electroweak interactions, such as SU(5) or 5O(10) gauge theories.

Problems
4.1 Prove the commutation relations [Q — T3, Tl] = 0 for i = 1,2,3, where Q is the
electric charge operator defined by (4.18) and Tl are weak isospin operators denned
by (4.13).

4.2 Show that the Yukawa interaction defined by (4.33) is invariant under SU(2)L X
U(\)Y gauge transformation.

4.3 Show that in the GWS model, the electric charge operator Q is not a broken
generator, though weak isospin operators T% violate the symmetry of the vacuum
88 THE STA NDA RD MODEL OF ELECTR O WE A K INTERA CTIONS

given by (4.36).

4.4 Derive (4.68).

4.5 Show that arbitrary n x n complex matrices can be diagonalized by bi-unitary

tarnsformations.

4.6 The lowest order decay amplitude of W -> li>i (I = e, /z, r) is given by

M = ^ = f i ( p ) 7 , . ( l - 7sM9)e"(*, A),

where e'x(fc,A) is the polarization vector of the W boson with helicity A. Then,
neglecting the lepton masses, calculate the decay width for this decay which is
given as

4.7 The interaction between Z° and a fermion / is given by

£Z y rZ rrfj.
NC ~ a—Ju " i
COS0ty ^
where jjf is given by (4.97).
(l)Neglecting the fermion mass, calculate the decay width of Z° ->• / / in the tree
level, which is given as
v{z
° ~* ff) =2^0K)2 + {CIA)2) '
where C ( , C{ are given in (4.98)~(4.101).
(2) Since the top quark is heavier than Z°, Z° cannot decay into a top quark pair.
Then, assuming sin2 6w — 0.22, estimate the branching ratio of Z° -> e + e~.
 Chapter 5
Q U A N T U M CHROMODYNAMICS

Quantum chromodynamics (QCD) is the theory of strong interactions which de-

scribes the dynamics of quarks and gluons. QCD is the non-Abelian gauge theory
with SU{3) color symmetry. In 1930s, the idea of strong interactions was first in-
troduced by Yukawa to explain new strong forces those days, called nuclear forces,
between nucleons mediated by pions. Before long, many hadrons, in addition to nu-
cleons and pions, were discovered and it was revealed that the interactions among
those hadrons were very complicated and did not seem to be fundamental.
Now, we know that all hadrons are composite particles of quarks which were
first introduced by Gell-Mann and Zweig in 1964 (Gell-Mann, 1964; Zweig, 1964)
to explain the spectroscopy of hadrons those days. The quarks were established
as the fundamental constituents of hadrons in the development of the quark model
in 1960s and 1970s; baryons are composed of three quarks qqq and mesons of a
quark q and an antiquark q. The quarks are bound inside hadrons through strong
interactions mediated by gluons with the coupling of color charges gs\, where
A1 (i = 1,2, •••,&) are the 3 x 3 Gell-Mann matrices as explicitly presented in
Appendix (C.4). The situation in hadrons is quite similar to the case of positronium
where the electromagnetic Coulomb forces between a positron and an electron are
mediated by photons with the coupling of electric charge e. It is well-known that
interactions between electrons (or positrons) and photons are beautifully bescribed
by quantum elecrodynamics(QED), i.e. the gauge theory of U(l) symmetry called
U(l)em, which is invariant under local Abelian U(l) phase transformation. QCD
which describes the strong interactions between quarks and gluons is also the gauge
theory but this time it is not an Abelian theory but a non-Abelian theory of SU(3)
color symmetry, described as SU(3)C, which is invariant under local non-Abelian
SU(3) transformation in color space. Notice that the color being in 3 species as
i?(red), B(blue) and G(green), is unrelated to another attribute of quarks, i.e. the
flavor, which is in 6 species (u, d, s, c, b, t). In contrast to the flavor symmetry
like the SU(3)f symmetry among u, d and s quarks which is rather largely broken,
the color SU(3)C symmetry is unbroken and exact as well as the U(l)em symmetry
in QED. Therefore, just like photons in QED, gluons as gauge bosons mediating

89
90 QUANTUM CHROMODYNAMICS

strong interactions are also massless.

Since there exist no colored hadrons in Nature, we assume that all observed
hadrons must be colorless, i.e. in the color singlet states. This seems to be quite sim-
ilar to the case of isospin states for two-nucleon systems. As is well known, though
we can expect, at a glance, to have two different isospin states J = l(isotriplet) and
I = O(isosinglet) for the two-nucleon systems composed of a proton and a neutron,
actually only 7 = 0 state is allowed to exist as a deuteron. However, one should
notice that there is a big difference between hadrons and deuteron; though in the
case of deuteron, a proton and a neutron which are constituents of deuteron exist
separately in Nature, colored quarks which are constituents of hadrons do not exist
separately in Nature but are confined inside hadrons as colorless states. This is the
so-called confinement problem which should be explained in QCD.
In this chapter, after describing the evidence of the color degree of freedom, we
discuss the coupling strength of color forces between two quarks and also between a
quark and an antiquark. Then, we discuss the running coupling constants in QED
and QCD cases and discuss the difference between QED and QCD.

5.1 Evidence for colors

There are several evidences for existence of the color degree of freedom, which is
actually in 3 species. The color was first introduced to solve the difficulty in the
relation of spin and statistics in the baryon spectroscopy. In the quark model,
baryons are made of three spin \ quarks. Without color space, the wavefunction
of a baryon is described by the product of space-, spin- and flavor-wavefunctions.
Then, consider the A + + (1232) with spin | . It is the ground state of the uuu
bound system and hence its wavefunction is totally symmeric under interchange
of any u quark pairs in space-, spin- and flavor-wavefunctions. However, since the
A + + (1232) is a fermion with spin | , its total wavefunction must be antisymmetric
under interchange of any u quark pairs. This difficulty can be solved by introducing
an antisymmetric color space wavefunction, where the color space is a new internal
space.
A more direct evidence for color to be in 3 species comes from the experimental
data on e + e~ annihilations at high energies. Based on the quark parton model, the
ratio

„ a(e+e~ —> hadrons)

R=—T————, s (5.1)
a{e+e~ —> fi+[i )

is predicted to be R = Y^,q eq, where eq is the electric charge of the quark q in quark
pairs produced in the e+e~ annihilation. Beyond the s quark pair production
threshold but for not so high energies, only u, d and s quarks contribute to this
 Evidence for colors 91

Fig. 5.1

ratio R and yield

R = el + e\ + e\ = - , (without color)
R = 2. (with 3 colors) (5.2)

Similarly, for higher energies beyond the c quark pair production threshold, R be-
comes

R = el + e\ + e2s + e\ = -r-, (without color)

10
R = (with 3 colors) (5.3)

and for more higher energies beyond the b quark pair production, it becomes
11
R = el + e2d + e2s+e2c+e2b (without color)
9'
R = —. (with 3 colors) (5.4)

Experimental data prefer the 3 colors for any cases.

Another important observation suggesting the color number to be 3 is the decay
rate of n° -> 77. The process proceeds through the coupling of the pion to a quark
loop as shown in Fig. 5.1, in which the u and d quark loops contribute to this decay
kinematically. The calculated result is given by

T(n0^11) = N2(e2u-e2d)2^y m" (5.5)

327T/,'

where Nc is the number of colors and fn ~ 130MeV is the pion decay constant.
Then, we can predict the rate to be
Vj<86 e V
-° _->i -vV\
r(7r° 77) - J ' ( W i t n 0 U t C O l ° r (NC ~ 1)) (5.6)
"^ " " I 7.i
.65 eV. (with 3 colors (Nc = 3))
The measured rate is (7.74 ± 0.55) eV which agrees with iVc = 3.
Further example is the branching ratio of the r lepton, Br = r(r-%o») • ^
T~ lepton decays into kinematically allowed lepton pairs, (e~ De), {fi~ D^), and
quark pairs, (d u), (s u), through the diagram shown in Fig. 5.2. Among these
channels, the contribution of (s u) is very small because it is a Cabibbo-suppressed
92 QUANTUM CHROMODYNAMICS

le-\l»-\l d\l s \

\ v* \ n u j
MM

Fig. 5.2

process and hence can be neglected. Since the coupling is same for all these channels,
we can simply predict the above branching ratio as

Br (without color)
3'
Br — -. (with 3 colors) (5.7)
5
The experimental data is Br = (17.84 ±0.06)% which is again in agreement with 3
colors.

5.2 Q C D Lagrangian and the strength of color forces

The QCD Lagrangian describing the interaction between quarks q and gluons A1^
(t = l,2,---,8) is given by

C= q(ip-m)q--FJtvFi^ (5.8)

where the summation over i (i = 1,2,•••,8) is implied. The quark field q is

given in both the Dirac space and the color space with three color components

as q , where the superscripts, R, G and B, denote red, green and brue,

respectively. The covariant derivative is given by

. A
D» 19s Al. (5.9)

where gs is the strong coupling constant and A* are the 3 x 3 Gell-Mann matrices
presented in Appendix (C.4). The summation over i (i — 1,2, • • •, 8) is implied. F^
are the field strength tensor for gluon fields A^ and given by

0MAt-0„4+0./yfc4Aj, (5.10)

where fak are the structure constant of the color SU(3)C group.
 QCD Lagrangian and the strength of color forces 93

Here we are interested in the strength of color interactions between two quarks
and also between a quark and an antiquark. Using the explicit expression of A1
presented in (C.4), the interaction between quarks and gluons given in (5.8) is
written as
A'
M
2

+ ^ ( < A V - « V ? < X + qRYqBGl + f-ffGl

+ fl"qBGl+qB^qGGl

+ += (qRl»qR + fl»qG ~ 2qB^qB)Gl) , (5.11)

where

GJ, = ( 4 - iAD/V2, Gl = (Al + iAl)/V2, G* = 4

G^(Al-iAD/V2, Gl = {A*+iA*)/y/2, (5.12)
Gl = (Al-iAl)/V2, Gl = (Al + iAl)/V2, G« = A«

represent redefined gluon fields. As can been seen here, the gluons come in 8 different
color combinations, RG, GR, ^(RR - GG), RB, BR, GB, BG, ^(RR + GG-
2BB). This means that the gluons belong to a color SU(3)C octet. For example,
G* changes the color of the quark from G(green) to i?(red). Another combination,
i.e. the color singlet -h=(RR + GG + BB), does not mediate color charge.
Now we are concerned with the interactions between two quarks. First let us
consider the interaction between i?(red) and i?(red) which arises from 2 diagrams
due to G^ and G® exchange as shown in Fig. 5.3(a) and (b). Then, the coupling of
this interaction is obtained from (5.11) as

The interaction between B (green) and B (green) arises only from G® exchange as
shown in Fig. 5.3(c) and the coupling becomes

The interaction between i?(red) and G(green) arises from 3 diagrams shown in Fig.
5.3(d), (e) and (f) and the coupling becomes
2

~±)'<*$)•(-*$)<%)'-U$)'-
.N/6%/2 ^
94 QUANTUM CHROMODYNAMICS

„R

££
^fe) W ^

(b)

2
fe))WOT ^K^fe)
(c) (d)

^
^fe)WMW-^fe)
(e) (f)

Fig. 5.3

As seen in the above examples, all color interactions between two quarks (color-
color interactions) are equal as expected from exact color symmetry. In other words,
two quark interactions are blind to the difference of colors.
What about the interactions between a quark and an antiquark (color-anti-color
interactions) in the color-singlet state, -h=(RR + GG + BB), i.e. between q and q
in a meson? Here, since each color is in equal weight in the color-singlet state of a
meson, it is enough to consider, for example, the case of S(blue) and -B(anti-blue)
whose interaction arises from 3 diagrams shown in Fig. 5.4(a), (b) and (c). (The
situation is same for between .R(red) and i?(red) and also between G(green) and
G(green).) Then, the coupling of this interaction becomes

yfcy/2
(5.16)
Here we should notice that the coupling between antiquark and gluon is in the
opposite sign to the one between quark and gluon because of vector nature of gluons,
 QCD Lagrangian and the strength of color forces 95

-^ ($) V o W ^ (^)

*S«T

Fig. 5.5

just as in QED mediated by a photon where the charge of positron (anti-particle

of electron) is opposite to the electron charge. We have the same contribution of
RR and GG as the one of BB obtained in (5.16). However, when we consider the
color-singlet state of mesons, each of initial and final states in Fig. 5.4 has a factor
4 j . Then, taking into account of these factors as 3 x ( 4y J = 1 , the coupling for
qq interactions in a meson becomes the same as given in (5.16).
The similarity of QCD and QED is in many cases useful in calculating processes
in the tree approximation. Let us take one such example, a process, q -> q + g (Fig.
5.5). The matrix element of this process is given as

M= ffsuV(y) uaAl„ (5.17)

which is essentially same for the QED process, q —• q + 7 except for the coupling
constant and the color matrix factor, a and b denote the color indices. Squaring
the above matrix element, one can extract the color-dependent factor as

lo> 2 2 39°
(5.18)
96 QUANTUM CHROMODYNAMICS

<Z3> ces> 0

Fig. 5.6

using the relation IV y y = 4, where the summation over i (i = 1,2, • • • ,8) is

implied. In (5.18), a factor | is due to averaging over the initial quark's color states
(=3). Then, the transformation from QED process to QCD process can be done
simply by replacing a by | a s in the transition probability for the corresponding
QED process.

5.3 Running coupling constants

5.3.1 Renormalized charge in QED

In QED, an electron shows many guises; an electron virtually emits photons which
produce electron-positron pairs and those pairs emit further photons and so on.
Hence, an electron turns out to be surrounded by many virtual electrons and
positrons and due to the Coulomb attractive force, positons become closer to the
original elecrton (which we call the "bare" electron) and thus the vacuum is polar-
ized as shown in Fig. 5.6. Because of this vacuum polarization, the effective charge
(observed charge) of an electron decreases as we go away from the bare electron,
2

that is to say, the observed charge e or the fine-structure constant, a = f^, is not
a constant but depends on the distance r from the bare charge or the momentum
transfer Q2 in scattering by a test charge. In fact, the value of a(Q2) decreases
with decreasing Q2 (or increasing r) and thus the vacuum polarization leads to the
charge screening as shown in Fig. 5.7.
 Running coupling constants

<*(Q2)

a(fi2) - ^

large Q small Q2
(small r) (large r)

Fig. 5.7

Fig. 5.8

To understnad this property in QED, let us consider the electron-electron scat-

tering, e~(A;i) + e~(k2) -> e.~{k[) + e~(k'2). The lowest order amplitude, being in
0(a), for this process is given by (see Fig. 5.8)

iM = [ieu(k[)j>iu(k1)]^^[ieu(k'2)Yu(k2)}, (5.19)

where q = fci — k[ is the vertual photon momentum. However, to get an exact am-
plitude for this scattering, we need to calculate higher order corrections containing
all order of a. The first order correction to the above amplitude comes from the
diagram of Fig. 5.9, in which the photon propagator contains an electron loop, and
according to the standard Feynman rule, it is written as
l
-iM = (-l)[ieu(fci)7'Ju(Jfci)] - 9 UP
98 QUANTUM CHROMODYNAMICS

q-p

Fig. 5.9

d 4 pTr ( . e r ) ^ + ^ ( . e 7 A ) i « - ^ + m)
4
TT)
(2TT> y m' (P-Q)2 m*
-ig\v
[ieu(k'2)Yu(k2)}, (5.20)

where p is the loop momentum of the internal fermion(electron) and m denotes the
electron mass. The factor (—1) is originated from the fact that this diagram con-
tains one fermion loop. It should be noticed that the internal momentum p should
be integrated from 0 to oo and leads to the divergence of the amplitude. (Exact
calculation results in only a logarithmic divergence, though one might expect it
to be quadratic because of the apparent form f£°pdp of the integral in (5.20).)
This divergence can be removed in QED by the renormalization technique as dis-
cussed below. Note that though the amplitude (5.20) is of the 0(a2), this one-loop
correction to the photon propagator is of the 0(a).
Now, one might worry about additional contributions of the same 0(a) origi-
nated from other diagrams depicted in Fig. 5.10, where (a) and (b) are the self-
energy terms of electrons and (c) is the vertex correction. However, to our surprise,
QED resuts in exact cancellation of the sum of the contribution of (a) and (b) and
the one of (c). Actually this cancellation occurs in every order of perturbation.
This is the fundamental property of QED and known as the Ward-Takahashi iden-
tity. Because of this cancellation, it suffices to evaluate only the modification to the
photon propagator.
The addition of (5.20) to (5.19) can be considered as modification of the photon
propagator as
-iguv -jg»v
'-IP*
+
 Running coupling constants 99

e e e e
7 / \ 7.

(a) (b)

(c)

Fig. 5.10

(5.21)
g2 V ^2
q2
where

V(<?2) p T r [(*e7/x) ^ _ m 2 (*e7-) w m)

(5.22)
(2TT)4 y (<?- m2
After a lengthy calculation, we find that the one-loop correction to the photon
propagator can be written as

V = -ig^q2i{q2) + (5.23)

with
q2(l-x)
Z
(5.24)
37T Jm2 P IT J0 V m"
The dots in (5.23) contain terms proportional to q^ and vanish when the prop-
agator is coupled to external electron currents. Substituting (5.23) in (5.21), we
obtain
-i9tii> -igyv
-¥ (i-/(s 2 ))- (5.25)

Therefore, we find that in the 0(a) the modification of the photon propagator
can be made just by mutiplying a factor (1 — I{q2))- Explicit derivations of these
100 QUANTUM CHROMODYNAMICS

formulas are beyond this textbook and can be found, for example, in the books
by Bjorken and Drell (Bjorken and Drell, 1964) or Jauch and Rohrlich (Jauch and
Rohrlich,1976). Here, we just comment on some important points of results. I(q2) is
infinite as mentioned above. But the infinity arises only from the first term of (5.24)
which is logarithmically divergent and independent of q2. One can estimate this
integral by introducing the cut-off parameter A for the upper limit of the integral
and then by approaching it to oo. Actual calculation will be done for the two
extremes, i.e. (-q2) » m2 and (-q2) <C m2.
Let us first consider the case of (—q2) » m2. In this case, log ( l - q x£i~ J ^
log(^j-) and then, we can evaluate I(q2) as

Thus, for large (—q2), the photon propagator is changed only by mutiplying a factor
(1 — i^F l ° g ( ^ i ) ) ' The multiplicative factor from loop corrections of all orders in a
can be also evaluated as

1
- £log (-^fi)+ ° ( Q 2 ) = 1
—nj\> (5 27)
-
and thus, the full amplitude becomes

-iM = [ieu(M)y*«(fci)]^^ f * , , | [ieu(k'2)7liu(k2)], (5.28)

q 1+ lo
V £ s(^)/
which suggests that the effective charge or fine-structure constant cc e // is given in
terms of a = f^ by

a a (5 29)
°" = / » \ - -

So far, we considered the value of a = | ^ to be a constant, which is the coupling at

the vertex of a "bare" electron and photon. We call it as the bare or unrenormalized
charge/coupling and write it as ao hereafter. On the other hand, the observed
coupling, usually called the fine-structure constant (w 137)) is different from this
and called the renormalized charge/coupling, which is, in practice, given by a e / /
in (5.29) and we write it as a(Q2) hereafter, where Q2 — —q2. Then, we rewrite
(5.29) as

«W2) = ^VTTV- ( 5 - 3 °)
l + f?l°g(&)'
 Running coupling constants 101

where a(Q2) and ao represent the renormalized and bare charge/coupling, respec-
tively.
Now, suppose that we can get the result, a(fi2) = ^§7, in experiment at some
value of Q2 = fj,2. Then, using the formula (5.30) for a(p,2), one can obtain

a(/z 2 )
a(Q2) = (5.31)
^loe
Notice that in (5.31) there is no dependence on the cut-off A and «o- &{Q2) depends
only on the finite measurable quanitities. Since a(Q2) depends on Q2, a(Q2) is
called the "running coupling constant". (5.31) shows the charge screening, that is
to say, d{Q2) decreases (increases) with decreasing (increasing) Q2. (See Fig. 5.7)
Here we introduced a parameter (i with the dimension of mass to avoid the infinity of
the amplitude by renormalizing the charge, p is called the renormarization mass(or
scale). Different choice of fi corresponds to different renormalization schemes but
final result does not depend on the choice of fi. This is because the dependence of
the amplitude on A is absorbed into a(fJ.2).
Now, let us move to the oppsite extreme of small (—q2). This limit is relevant
in the Coulomb scattering of an electron by a static nucleus terget with charge Ze.
In the limit (—q2) -¥ 0, we see log(l - q2x(l — x)/m2) ~ -q2x(l - x)/m2 and thus,
(5.24) reduces to

/(s2) = f l o g ( 4 ) + T f 4
z l
(5.32)
3ir \m J 157r m
where we again introduce the cut-off parameter A for upper limit of the first term
of the integral (5.24). Note that the divergent term, i.e. the first term of this result
is of the same form as (5.26), just by replacing m2 by — q2.
Then, to order a 2 , the matrix element for the Coulomb scattering of an electron
by a charge Ze is calculated by replacing the factor [ieU(k'2)j'/u(k2)] in (5.20) by
-if with j " = (p = Ze,j = 0)

[A2 qz
= [i e u(Jfei)y u (*i)]-^ •1 a a
-iM 1 g Hi")
"37 ° U ? 157r m2
2 ,22
—% a (A \ a q
ZTT \m2 J 157rm2 (-iZe). (5.33)
= [ieu(fc|[)7ou(A:i)]-2-

When we write this result

-iM as= [ieflu(A;i)7ou(A;1))]-5-(-iZefl), (5.34)
q
with
1/2
a
eR = e 1 , fA' (5.35)
1 - 5 - log —j
37r V mz
102 QUANTUM CHROMODYNAMICS

Fig. 5.11

(a) (b)

Fig. 5.12

we can easily check that to 0(a2), (5.35) and (5.34) are equal, en is interptered as
the measured or renormalized charge | £ = a = ^ . Here the infinity of the matrix
element coming from the cut-off A —> oo is again absorbed into the observed charge
en by renormalizing the bare charge/coupling.

5.3.2 Running coupling constant in QCD

As shown in (5.8), gluons have self-interactions because of the non-Abelian nature of
QCD. This nature produces a drastically different behavior of the running coupling
2

constant as(Q2) = | £ from the one in QED discussed above. The basic quark-gluon
interaction is given in Fig. 5.11. Concerning the one-loop corrections, there are two
diagrams as shown in Fig. 5.12(a) and (b). (a) shows the correction from a quark-
loop which contributes equally to every quark flavor because of flavor independence,
(b) shows the gluon-loop correction originated from the self-interaction of gluons.
Apart from the color factor, the contribution from the quark-loop diagram (a) is
 Running coupling constants 103

essentially the same as the one in QED; it suffices to replace the factor ° ^ ^ by
"ST f° r e a c n flavor by taking into account of the color factor. The crucial difference
between QCD and QED arises from the gluon-loop diagram (b). This diagram give
rise to another numerical factor, -|^Q: s (/x 2 ), whose sign is opposite to the one from
the quark-loop. Combining the contribution from both the quark-loop (a) with n/
flavors and the gluon-loop (b), one can obtain the QCD running coupling constant
as(Q2) just by replacing the factor ^^- in QED by ^ ^ ( f n / - 11) as

As described before, we know that there are n/ = 6 flavors in Nature. Putting

nf — 6 into (5.36), we see that the sign of the 2nd term in the denominator of
(5.36) is opposite to the QED case in (5.31). Therefore, the running coupling
constant as(Q2) shows antiscreening; as(Q2) decreases with increasing Q2 as known
as "asymptotic freedom". In contrast, for small Q2, as(Q2) becomes large. In fact,
the demominator in (5.36) becomes 0 at some value of Q2 — AQCD where

A2QCD = »2e <»-'«/>-«•'>. (5.37)

By using this AQCD, we can write as(Q2) as

(33-2 n / )log^

Here we introduced a free parameter AQCD with mass dimention by removing

the renormalization mass parameter [i. We cannot determine the value of AQCD
theoretically in QCD. It is extracted from experiment and was determined to be
2 2
AQCD ^ 200MeV for Q ~ lOOGeV . With nf = 5 flavors (u, d, s, c, b) taking
part in scattering processes, the value of as becomes as ~ 0.2 which justifies the
perturbatuve calculation of QCD. The behavior of as(Q2) is depicted in Fig. 5.13.

In summary, for large values of Q2 much larger than A2QCD, the effective cou-
plings between quarks and gluons becomes small and thus, the strong interaction
can be treated perturbatively. In this region, quarks and gluons behave as free
particles ("asymptotic freedom"). In fact, this is the region of the deep inelastic
scattering scattering where the parton picture works well. On the other hand, for
small Q2 region like Q2 w A"QCD, the quark-gluon coupling becomes large and the
perturbative calculation cannot be justified. Because of the large coupling con-
stant, all quarks and gluons are confined in hadrons ("confinement"). AQCD is
the scale which separates the world of confinement scale(hadrons) and asymptotic
freedom (free quarks and gluons). QCD is the color 5(7(3) gauge theory of strong
interactions and is the fundamental field theory describing the dynamics of quarks
104 QUANTUM CHROMODYNAMICS

as(Q2)

a. ~ 0.2

Q2 = lOOGeV2
large Q small <32
(small r) (large r)

Fig. 5.13

and gluons, though the whole understandings on the nonperturbative confinement

problem still incomplete.

Problems
5.1 T~ can decay into pairs of (e~ ue), (fi~ D^), (du), and (s u); among these decay
modes, the contribution of (s u) is very small because it is a Cabibbo-suppressed
process and can be neglected. Neglecting the masses of all decay particles, estimate
the branching ratio, Br = r(^%aii)T to be jr as given in (5.7).

5.2 In the tree level, the color-dependent coupling factor for the process q -> q + g,
where q and g represent a quark and a gluon, respectively, is given by (5.18). Then,
estimate the factor for g —> q + q.

5.3 Consulting with other textbooks, for example, the one by Bjorken and Drell
(Bjorken and Drell, 1964), derive (5.24).
 Chapter 6

N E U T R I N O MASSES AND N E U T R I N O
OSCILLATIONS

6.1 Type of Fermions and Fermion Masses

When P.A.M. Dirac first invented the spinor to describe the electron, it was a 4-
component complex vector. Fermions described by such 4-component spinors are
called Dirac fermions. Though the electron has a finite mass, it may be worthwhile
considering a massless fermion. In this case a new feature arises: a state with h = 1
(h being the helicity of the particle) and a state with h = — 1 never mix with each
other during the propagation of the fermion. This can be understood intuitively by
considering a inertia frame, observing the fermion co-moving toward the direction of
the fermion's momentum. If the helicity h of the fermion is 1 in that frame, in any
frame the helicity should be the same, just because the massless fermion is moving
with the light velocity and any observer never can pass through it to invert the
direction of momentum and therefore the helicity of the fermion. It is well-known
that for massless fermion, the eigenstates of helicity and chiral fermions with definite
eigenvalues of 75, i.e. Weyl fermions ipn and ipL, are identical. We easily know that
each of chiral fermion forms an irreducible representation of Lorentz transformation
SL(2,C), as [75, £""] = 0 (E"" = £[7",7"] are 6 generators of 517(2, C)). In such
a sense, Weyl fermions are more fundamental fermions than Dirac fermions. Let
us recall that fermions to start with in the Standard Model are Weyl fermions,
essentially because the weak interaction maximally breaks parity symmetry.
The statement above becomes more explicit if we use the chiral representation
of 7 matrices, where 75 is a diagonal matrix (see also Appendix A):

0 <r" \
* = Uo» 0 )>
0 ) M
a* = {-I,Oi), (6.2)
a" = (-1,-*), (6.3)
I 0 \
75 = i-f-f-fr =[ 0
n -r / ; . (6-4)

105
106 NEUTRINO MASSES AND NEUTRINO OSCILLATIONS

where / is a 2 x 2 unit matrix and <7j (i = 1,2,3) are Pauli matrices. In the
chiral representation a right- and a left-handed Weyl fermions are described by
2-conponent complex spinors rja and | a :

V* = ( ^ ) , (6.5)

i>L = [f>)> (<*," = 1,2). (6.6)

The Lorentz generators have, as we expect, a form of block diagonal

a"" = *"*", a^v=aliav, (p?v). (6.8)

For a massive fermion moving with a speed lower than the speed of light, the
helicity may change depending on the velocity of the inertia frame, observing the
fermion: the chirality is no longer preserved for massive fermions. Accordingly the
fermion mass term connects Weyl fermions with different chiralities:

m(ipRipL+'*pL'4>R)- (6-9)
Thus the fermion suitable to describe the propagation of the massive fermion is the
combination of these two Weyl fermions, i.e. a Dirac fermion I[>D, a full 4-component
complex spinor:

TpD = 1>R + IpL = I J J, (6.10)

Vfl = WD, i>L = L<pD, (i?=l±^, L = i ^ ) , (6.11)

The free Lagrangian for the Weyl fermions is now rewritten in terms of a Dirac
spinor tpjj:

•4>Ri$ipR + il)Li$il)L ~ m(ipRtpL + IPL^R) = V»B(*# - m)ipD- (6.12)

Namely, ipR and tpi are "chiral-partners" to form a full 4-component Dirac spinor.
Actually the chiral partners need not to be two independent Weyl spinors. We
should note that the action of "charge conjugation" for Weyl fermions changes
their chiralities; e.g. for (ipR)c = -i^2(^-^-tp)*,

R(ri>Ry = o, L(ijRy = wRy, (6.13)

i.e. (ipR)c = (IPC)L- Or, in terms of 2-component notation of Weyl fermions,

^ = ( "o ) "> ^Rr = { t ) ' (6 14)

'
 Type of Fermions and Fermion Masses 107

where (r)a)* - rj&, rj"1 = e^rjp, e12 = - e 2 1 = 1. Thus chiral partner of a Weyl
spinor can be the charge conjugation of the Weyl fermion itself. Combining these
chiral partners the following full 4-component spinors are possible to be formed:

V-Ml = fa + ( W = ( J£ ) ' (6-15)

V<M2 = i>L + ( W = ( h J • (6-16)

By construction these 4-component spinors tpMi,M2 are self-conjugate under the

charge conjugation, (V>MI,M2)C = •0MI,M2, and are called Majorana spinors. Though
Majorana spinors are 4-component spinors, the number of independent complex de-
grees of freedom is 2, i.e. the same as that of a Weyl spinor. Thus as far as kinetic
term is concerned, where there is no mixing between chiral partners, there is no dif-
ference between Weyl and Majorana fermions and a Dirac fermion is equivalently
describable in terms of either 2 Weyl or 2 Majorana spinors:

tpDi^D = IpRiQlpR + IpL^L = -^l>M\i$i)M\ + V'M2*#M2)- (6.17)

In other words, for massless fermions, the propagators are just the sum of those for
independent Weyl fermions.
Once fermions get masses, the type of fermion has real physical meaning. Clearly
Weyl fermion is irrelevant in this case, since the mass term connects the chiral
partners. What determines the type of the full 4-component spinor, i.e. Dirac or
Majorana, is the choice of chiral-partners or in other words, the type of fermion
mass term. In the case of ordinary mass term for Dirac fermion (Dirac type mass
term),

-moipoi'D = -mD(tpLipR + h.c.) = mD(£aria + h.c), (6.18)

the chiral-partners are two independent Weyl fermions £ a and r)a. We, however,
need not to take independent 2-componet spinors £ and rj as chiral partners and
they can be identical. Namely if we take as chiral partners the charge conjugation
of each Weyl fermion the following mass terms are also possible,

-^mR((ipRyi>R + h.c.) = - m / i ( i | a i h + /i.c.), (6.19)

-\mL{{tpLY^L + h.c.) = \mL^a^a + h.c). (6.20)

These mass terms clearly violate fermion number (lepton number for, e.g., neutri-
nos) and are called Majorana type mass terms, while we can assign a conserved
fermion number for the case of Dirac mass term. If we adopt the Majorana type
mass terms listed above, the fermions suitable to describe the free Lagrangian should
108 NEUTRINO MASSES AND NEUTRINO OSCILLATIONS

be Majorana fermions ipMi and ipM2'-

ipRi$i>R - -mR((ipR)cipR + h.c.) = -^M1(^-mfi)i/>Mi, (6.21)

ipLi$il>L-2Tni>({i>L)cipL + h.c.) = -tpM2{i^-mL)it)M2- (6.22)

6.2 Neutrino Masses

6.2.1 Possible Types of Neutrino Masses

As we have seen in the previous section, the Majorana mass terms violate fermion
number. Actually the mass terms violate all possible U(l) global symmetries the
Weyl fermions may have,

i/fR-te^R, ^L->e**^L, (6.23)

(j>L and <pR being real transformation parameters. Thus the Majorana masses are not
allowed for charged particles like electron, since if it is the case U(l)em symmetry
is violated, i.e. the charge conservation breaks down. The Majorana mass term is,
therefore, allowed only for electrically neutral particles, such as neutrinos or neutral
gauge fermion in supersymmetric theory, such as photino. In this chapter we focus
on the case of neutrinos.
Neutrinos of course may have ordinary Dirac masses, in addition to the possible
Majorana masses. Let vR and vi be right- and left-handed neutrinos. It is inter-
esting to note that even if we consider the most general mass term for vR and VL
including the Dirac mass term,

Cm = --mR{vR)cvR - -mL(vi)cvL - mDvRvL + h.c, (6.24)

the free Lagrangian for the neutrino still can be written in terms of two Majorana
neutrinos, as we now confirm. The Dirac mass term m o in Cm causes a mixing
between vR and VL- We thus have to "diagonalize" the whole mass term. The mass
term can be neatly written by use of matrices as

where the property VRVL — {VL)C{VR)C has been used. Thus, the mass term, as the
whole, can be written in the form of Majorana type masse term, even though there
is Dirac mass term as well. This is based on the observation that vi, and {yR)c
are both left-handed Weyl fermions and have no essential difference, though they
have opposite lepton numbers, L = 1 and L = — 1. Thus every mass term should
be generally written in the form of Majorana mass term, -(V'IL) C V'2L + h.c; in the
case tpiL = 1P2L the mass term is "genuine" Majorana mass term, while in the case
of Dirac mass term ipn ^ ip2L- We also find that the 2 x 2 mass matrix above
 Neutrino Masses 109

is symmetric complex matrix, not a hermitian matrix, in general. This property is

due to a relation {VH,)CVJL — [uji,)cViL and will be valid for more general cases with
arbitrary numbers of generations. Though such symmetric complex matrices are
known to be diagonalized by use of a unitary matrix in general, as we will see in the
following section, here we consider the case where mass parameters rriR, TTIL, TJID
are all real, for brevity. The eigenvalues ms, ma and the corresponding eigenstates
us, ua of the mass matrix

Mv = TTIL mo
(6.26)
mD mR
are given as

m. ^{{mR + mL) + \J{mR - mL)2 + lm2D}, (6.27)

ma -^{-(mR + mL) + \J{mR - mL)2 + 4m|,}, (6.28)

c
sin6vUL + cos6u(vR) , (6.29)
c
i{cos6vvL - sm9v(vR) }, (6.30)

where
2m,D
tan2#„ = (6.31)
ma — mi.
Note that the eigenvalues of Mv obtained by an orthogonal transformation are ms
and —ma, not ma. We, however, have changed the sign, — ma -> m 0 , by putting
the extra factor i for the eigenstate ua, so that the eigenvalues ma and ma get
degenerate for the case of pure Dirac mass, mR = TTIL = 0, and both eigenvalues
become positive for the case of "seesaw" scenario. The mass term now has been
diagonalized:

Cm = - g 7 7 1 * (vs)CVs ~ 2 m a {Va)CVa + h.C. (6.32)

Thus, including the kinetic term, the free Lagrangian for the neutrino is now written
in terms of two Majorana neutrinos obtained by combining vs and va with their
anti-particles as chiral partners, Ns = vs + (us)c and 7Va = va 4- (^o)c,

Cv = l{K(i9 ~ ms)Ns + Wa(iQ - ma)Na}. (6.33)

In the case of a scalar field, charge conjugation is equivalent to the complex

conjugation of the field. So a self-conjugate Majorana fermion (tpc = ip) corresponds
to a real scalar (0* = 4>), and the general mass matrix for two Majorana particles
just corresponds to a general mass-squared matrix for two real scalars, </>i and ifo,
110 NEUTRINO MASSES AND NEUTRINO OSCILLATIONS

Actually for the case of scalar fields what matters is the mass-squared matrix. To
make the correspondence clear, let us take

m m mD L +
MM =( y ° , ^ TR) ) , (6-35)
instead of M„ itself. It has been well-known that when the masses of two real scalar
fields are degenerate, m\ = m2, = m 2 , mi 2 = 0, they can be equivalently described
by a complex scalar <f> = (l/-\/2)(<£i + i<fo),
2
£ct>m = - m (f)*(j>. (6.36)

The same thing happens for fermions; for the case of pure Dirac mass, TUL =
TUR = 0, MvMl has degenerate eigenvalues m|> and two Majorana neutrinos are
equivalently described in terms of Dirac particle,

Ns-iNa IT
VD -j=— = vR + vL (ev = -), (6.37)

whose free Lagrangian is written as

£v = ~v5{i$ - mcJi'D. (6.38)

The factor 1/2 appearing in the free Lagrangian for Majorana neutrinos (6.32) just
corresponds to the same normalization factor in the free Lagrangian for real scalars.
Hence, just as in the case of scalar fields, such normalized Majorana fields Ns and
TV,, should have the same propagator as the one for a Dirac field. For instance,
for the Majorana fermion ipMi in (6-21) the propagator without chirality flip, and
therefore without lepton number violation, is given (in momentum space) as usual,

R • (OIV'MIV'MI |0) • L = (OlV'flV'fllO) = R • 2 L. (6.39)

p — mR
This suggests that both ipMi and ipM2 are properly normalized fields. Thus, gener-
ally the propagator for a Majorana field tpM with a mass m is given as

WMV^IO) = -T-^— (6.40)

p— m
(0|VMV^|0> = J^-C-1- (6.41)

A characteristic feature of Majorana field is that the propagator, apparently vi-

olating the fermion number, (0|T/>MV'M|0)> exists in addition to the ordinary one,
0 2
(0|V'MV'M|0). In (6.41), t stands for the transpose and C = i'y ^ is the charge
conjugation matrix. The propagator of the type (0|I/'MV'M|0) is needed in the cal-
culation of e.g. neutrino-less double /3-decay.
 Neutrino Masses 111

6.2.2 The Mechanism of Neutrino Mass Generation

Recent observations of neutrino oscillations strongly suggest that neutrinos have
finite masses. These data on neutrino oscillations, however, do not tell us the type
of neutrino masses, i.e. Dirac or Majorana. Though Majorana masses violate lep-
ton number, the smallness of neutrino masses prevent us to confirm or exclude the
Majorana property by use of the data of the processes sensitive to lepton number
violation, whose typical example is the neutrino-less double /?-decay. Thus, in the
consideration of the scenario of neutrino mass generation, we do not have a priori
any idea of relative strengths of Dirac and Majorana masses. Therefore, let us now
consider three typical (extreme) cases, depending on the relative strengths of these
masses. We will find that the smallness of neutrino masses can be naturally ac-
counted for in the third possibility, so-called seesaw scenario.

(A) Pure Dirac

We first consider the case that neutrinos have only Dirac masses, TUR = mx, = 0,
and neutrinos are Dirac particles. As we have already seen above, in this case a
Dirac neutrino may be regarded as the superposition of two Majorana neutrinos Ns
and Na with degenerate masses, ms = ma = mo (see (6.27), (6.28)). The mixing
angle is "maximal", i.e. dv = IT/4, as can be seen in (6.31). This maximal mixing,
however, does not lead to any neutrino oscillation, though it mimics the flavor or
generation mixing; as we will discuss, neutrino oscillation necessitates not only a
mixing angle but also a mass (-squared) difference. In this pure Dirac case, we thus
need flavor mixings among different generations to realize the neutrino oscillations,
as will be discussed in the next section. It is interesting to note that though each of
the two Majorana neutrinos Ns, Na does contribute to the lepton-number violating
neutrino-less double /3-decay, their contributions, obtained by use of the propagator
(6.41), exactly cancel with each other, due to the factor i in (6.30). This should
be the case, since in pure Dirac case the lepton number should be preserved. In
this case, however, we have no special reason to expect that the Dirac masses of
neutrinos should be much smaller than the Dirac masses of corresponding charged
particles, such as electron. So a serious problem of why neutrinos masses are so
small remains to be solved.

(B) Pseudo-Dirac
We next consider what happens if we put small Majorana masses, TUR, TUL -C mo-
The neutrinos are still almost Dirac particle and are called pseudo-Dirac neutrinos
(Wolfenstein, 1981). The mixing angle is still almost maximal, i.e. 8U ~ 7r/4.
One remarkable feature is that in this case there appears a slight difference of mass
eigenvalues m1-m2a~ 2mD(mR + mi). Together with the almost maximal mixing,
this mass-squared difference leads to a neutrino oscillation even if there is only 1
generation (Kobayashi and Lim, 2001). In fact a neutrino oscillation VL —> {VR)C
occurs with almost maximal mixing. The probability for the neutrino produced as
112 NEUTRINO MASSES AND NEUTRINO OSCILLATIONS

VL at time 0 with energy E to be observed as (VR)C at time t, in the simplified 1

generation scheme, is known to be given as

Piu^iu^^^r^^h (6.42)

One interesting feature of the neutrino oscillations of pseudo-Dirac neutrinos is

that the maximal mixing, strongly suggested by the Super-Kamiokande data on the
atmospheric neutrino oscillation, is almost automatically realized. Such neutrino
oscillations, however, are those into "sterile" states (VR)C without electro-weak in-
teraction, which are not favored by the data of Super-Kamiokande experiments.
Again, the problem of smallness of neutrino masses cannot be naturally solved in
this scenario.

(C) Seesaw (Yanagida, 1979; Gell-Mann-Ramond-Slansky, 1979)

We finally consider another extreme case where Majorana masses are much larger
than the Dirac mass. We, however, find that the Majorana mass for left-handed
neutrino, TUL, cannot be large, while the Majorana mass for right-handed neutrino,
TUR, can be sufficiently large. This difference essentially comes from the difference
of the transformation property of each Majorana mass term under SU(2)L X U(1)Y,
the gauge group of the standard model. We immediately know that the right-handed
Majorana mass term mn(uR)cUR is gauge invariant under SU(2)L X U(l), as VR is
a SU(2)L singlet neutral particle. On the other hand, the left-handed Majorana
mass term mi(i/i) c i/i, is not gauge invariant, as UL belongs to a SU{2)i doublet,
and this bare mass term is not allowed in the Lagrangian. In fact, we know that the
product of the fields {VL)CVL should belong to either singlet or triplet of SU(2)L
(doublet x doublet = singlet + triplet). Actually it cannot be the singlet, since
(VLYVL behaves as h = 1. Thus (VL)CVL should behave as a component of 51/(2)/,
triplet, and to form a gauge invariant (renormalizable) Yukawa interaction, we need
a Higgs HT belonging to SU(2)L triplet, so that TUL ~ f(Hr) with / being the
Yukawa coupling. The VEV (HT), however, cannot be large. If the VEV is larger
than or comparable to the VEV of ordinary SU(2)L doublet Higgs, the famous
relation, well-established experimentally,

M
r- - ^ ~1, (6.43)
~ M%cos2ew

no longer holds at the classical (tree) level. On the other hand there is no good
reason to expect that m,R, being SU(2)L invariant, should be small. Rather, it will
be natural to expect that it is much larger than the VEV of the doublet Higgs,
i.e. than the weak scale M\y, since it is quite possible that the TUR actually comes
from the energy scale of some "new physics" beyond the standard model, which
is expected much higher than Mw- The above argument suggests that mi -C
m c <C m,R. Here, for brevity, let us assume that HT is absent and therefore
 Neutrino Masses 113

TTIL — O (mo "C TTIR). In this case the neutrino mass matrix reduces to

m D
Mv={ ° ) . (6.44)
\ mD mR )
One immediately knows that one mass eigenvalue of M„, m8 ~ TUR under mo -C
rriR, and another eigenvalue

ma ~ ^ < mi?, (6.45)

mfi
since the product of two eigenvalues is given by det M„ = —m2D. (Strictly speaking
2

this means ma ~ — ^ ^ , with opposite sign, but the extra minus sign can be absorbed
by putting a phase i to the eigenstate va, as is seen in (6.30).) These eigenvalues
may be obtained from (6.27) and (6.28), as well. We thus have obtained a naturally
small (<C TUD) neutrino mass ma by making TUR much larger than mp. The relation
ms • ma ~ m2D suggests the terminology seesaw. Namely, when one eigenstate Ns
gets heavy another eigenstate iV0 is "lifted". We know that under mo <C TUR the
mixing angle 6V is small

6V ~ ^ ^ « l (6.46)
mR
Ns ~ VR + {VR)\ Na~i{vL-(vL)c}. (6.47)

As the heavier Majorana neutrino iVs decouples from lower energy processes, only
Na ~ i{uL — {VL)C} will remain and there will be no neutrino oscillation in one
generation scheme. Thus, as in the case of pure Dirac scenario, to have neutrino
oscillation, flavor or generation mixing is inevitable.
One may wonder, in the absence of the triplet Higgs HT, which operator can
provide the small Majorana mass m Q , which is effectively the Majorana mass for
the left-handed neutrino, as iV0 ~ I{VL — {VL)C}- It turns out that though there
is no renormalizable operator in the original Lagrangian relevant for the Majorana
mass, a higher dimensional (d = 5) "irrelevanth effective operator,

^ 0 W • ( "L ei ) eaa C ( ^ ) , (6.48)

plays the role, where <j> — (<p+, <p0)1 is the Higgs doublet and e is 2x 2 anti-symmetric
matrix (e12 = —e21 = 1) and C is the charge conjugation matrix. The coefficient
c is a dimensionless coupling and if we regard c = f^ (fn'- Yukawa coupling in
the Dirac mass term), M = TUR and replace ip° by the VEV v/\/2, we get a left-
f2 v2 m2
handed Majorana mass, ~ ^ — ~ j ^ 2 - , in accordance with the result obtained by
the diagonalization of the mass matrix. In fact, the diagram shown in Fig. 6.1 is
known to provide (a part of) the effective higher-dimensional operator, when the
momentum is ignored compared with mR in the propagator of the internal line.
 NEUTRINO MASSES AND NEUTRINO OSCILLATIONS

/
,+
/
/
-4 (
VR

VL VL

Fig. 6.1

When M goes to infinity the effective operator vanishes. Thus the essence of
seesaw may be understood as the decoupling phenomenon of a heavy particle with
mass M, which is right-handed Majorana neutrino Ns in the case of seesaw mech-
anism. The importance of the argument in terms of the effective operator is that
the M needs not to come from the right-handed Majorana neutrino, but may be
attributed to any heavy 51/(2) singlet field with mass M.

6.3 Flavor Mixing and Neutrino Oscillation (in the Vacuum)

If we include right-handed neutrinos to accommodate neutrino masses, the lepton

sector is described by 3 doublets and 6 singlets of SU(2)L'

VeL VTL
VeR, VvR,VTR,eR,m,TR- (6-49)
e-L V-L TL

Without loss of generality, the charged leptons, e^, e^ etc., may be understood as
their mass eigenstates, and their partners in the doublets, vei, etc. are states emitted
by weak interaction processes and are called "weak eigenstates". Suppose the mass
matrix for neutrinos in the base of weak eigenstates is diagonal. (For brevity we
here assume neutrinos have only Dirac masses), Then, the generation number or
quantum number of flavor such as electron number Le (Le = 1 for e~, ve, Le = 0 for
others), muon-number LM etc. are strictly conserved. If, on the other hand, the mass
matrix is non-diagonal, the flavor changing processes such as ue —> v^ or /J, —> e'y
will become possible. The phenomenon such as ve -> */M, where a neutrino of some
specific flavor is transformed into a neutrino of another flavor, is called "neutrino
oscillation", since the probability of the transition oscillates as the function of time.
Even if we switch on flavor mixing, provided the neutrino masses are all vanishing
or, more generally, degenerate, the amplitudes of flavor changing processes will
exactly vanish. This can be explicitly checked in the formula of neutrino oscillation
probabilities, as we will see later (see for instance (6.69) for Amjj = 0). This fact
also may be easily understood from a symmetry argument. When neutrino masses
are all degenerate, there appears a global symmetry C/(3) among 3 generations or
 Flavor Mixing and Neutrino Oscillation (in the Vacuum) 115

flavors in neutrino sector, and by use of Noether's theorem, we easily find that
there should be 3 conserved additive quantum numbers, which may be interpreted
as Le etc. Or, more intuitively, if neutrino masses are all degenerate there would
be nothing to distinguish neutrino flavors, and we may always perform a unitary
transformation, so that all flavor mixing angles vanish in the new base. Thus to
realize flavor changing processes we need both flavor (or generation) mixing and
non-degenerate neutrino masses. This is in complete similarity with the content of
GIM mechanism in quark sector.
It should be emphasized that now there are increasing evidences for neutrino os-
cillations claimed by the experiments to detect the solar and atmospheric neutrinos,
as will be discussed later in this chapter. The phenomenon of neutrino oscillation
is quite interesting by its own right. In addition, theoretically, it is expected to
play a central role in the search for and the establishment of "new physics", since it
clearly indicates non-vanishing neutrino masses and therefore some physics beyond
the standard model.
A nice comprehensive review on the neutrino masses and neutrino oscillations is
given, e.g., in the recent textbook by Fukugita and Yanagida (Fukugita-Yanagida,
2003).

6.3.1 Flavor Mixing

We now discuss the flavor mixing, the origin of neutrino oscillation, for two typical
scenarios, (A) pure Dirac, and (B) seesaw.

(A) Flavor mixing in the pure Dirac scenario

Just as in quark sector, if neutrinos have only Dirac masses, the mass term for weak
eigenstates (veL, U^L, VTL) and {veR, u^R, UTR) is given as

Cm = {mD)a017^Zvm + h.c. (a, 0 = e, n, T). (6.50)

Though a Dirac fermion can be decomposed into two Majorana fermions with de-
generate masses as we have already seen in the previous section, we do not take
this picture, and the mass matrix is 3 x 3 matrix just as in the quark sector. As
we have learned in the quark sector (see (4.144), (4.145)), the matrix m o can be
diagonalized by a bi-unitary transformation,

/ mi 0 0 \
U^mDV = I 0 m2 0 , (6.51)
\ 0 0 m3 /

where U and V are unitary matrices. The left-handed eigenstates of mass matrix,
i.e. mass-eigenstates, vn, V2L, V3L with Dirac masses, m\, rri2, m.3 are related to
116 NEUTRINO MASSES AND NEUTRINO OSCILLATIONS

weak eigenstates as

( VeL
VeL \\I (1 VlL
v1L \
V»L \=u. 1 V2L (6.52)
1
VTL ) V VZL
Accordingly the charged current interaction of left-handed leptons is written in
terms of U

Cc = ^={el W TZ)U1A V2L J • W-", (6.53)

and the unitary matrix, corresponding to the Kobayashi-Maskawa matrix in quark

sector, is called Maki-Nakagawa-Sakata (MNS) matrix (Maki, Nakagawa and Sakata,
1962). In the SU(2)L X U(1)Y gauge theory, only left-handed neutrinos have weak
interactions, and as we will see in the following section, only neutrino oscillations
without chirality-flip are important. These lead to the conclusion that not mp itself,
but only the combination mDTn'D is relevant for the physical processes, including
neutrino oscillation. Since

•m,DnvD = U dia,g{m\,m\,m\) U\ (6.54)

the unitary transformation of right-handed neutrinos V never appears in the phys-

ical observables.

(B) Flavor mixing in the seesaw scenario

Now with 3 generations the seesaw mechanism should be implemented in the 6 x
6 mass matrix. In the base of ip = (veL, V^L, VTL, (^efl)c, (^Vfl)c, (^TRYY, the 6
x 6 mass matrix m„ takes the form of

mL m
fV (6.55)
mo rriR J
where 3 x 3 matrices TUL, TUR, mo denote left-handed, right-handed Majorana
mass matrices and Dirac mass matrix, respectively. As we have already seen in the
simplified 1 generation scheme, the mass matrix m„, as the whole, is a symmetric
matrix, m^, = m„ (mL = mL,mR = TUR). TO consider the seesaw scenario, we
assume that mi = 0. The mass matrix with TTIL = 0 is easily achieved in the
standard model, by introducing right-handed neutrinos and their Yukawa couplings
and bare mass terms,

-fap\v^L l~L)4>vpR-{mR)a0{vaRYupR, (6.56)

where {mD)a0 = f%* • ^ , 4> = iT2<\>* (see (4.137)) and ^ is the VEV of the
neutral Higgs ip°. To account for the seesaw mechanism we understand that all
matrix elements of TUR are much larger than those of mo- Then, as we have seen in
 Flavor Mixing and Neutrino Oscillation (in the Vacuum) 117

the 1 generation scheme, the mutual mixings among uaL and {vaR)c (corresponding
to 0V in (6.29), (6.30)) are small, being suppressed by the ratios of elements of mo
to those of TUR. Just as in the 1 generation scheme, the mass matrix is made into the
form of block-diagonal by a suitable unitary transformation with the small mixing
angles,

/ %I -imjjmjf 1 \ m ( H_ ™>D™~R \
\ mRlm*D I J " \ —im*R 1rriD I J
„ ( m^rofc)-1^ 0 \ , 5?)
y
- V 0 m*R J' •
Thus again eigenstates with masses of the order of TUR, decouple from the low
energy effective theory, and the mass matrix for the remaining neutrinos, which are
approximately uaL, may be regarded as

mvL = mtD(mR)~1mD, (6.58)

which is again a symmetric matrix.
It is mathematically proven that a symmetric complex matrix can be diagonal-
ized with real eigenvalues by a unitary transformation,

\]tmviJ\] = diag(mi, m 2 , m 3 ). (6.59)

Again what matters in the neutrino oscillation is not m^i itself, but

TtvvlmVL = U d i a g ( m i , m \ , m \ ) U^. (6.60)

It is quite important to note that the right-hand-side of the equation is exactly
the same as the corresponding one in pure Dirac scenario, (6.54). Thus as far as
our concern is neutrino oscillation, to discriminate the scenarios of neutrino mass
generation, pure Dirac or seesaw, is impossible. Note, however, if we observe a
process with chirality-flip, VL —> VR or vi —> (^L) C , it is in principle possible to
discriminate these two scenarios, since the final states UR and (VL)C have opposite
lepton numbers and only (VL)C has weak interaction. The typical and interesting
example with chirality flip and lepton number violation is the neutrino-less double
/?-decay. If the process is observed, it will clearly indicate the Majorana property
suggested by the seesaw scenario, though the decay amplitude is suppressed by the
small Majorana neutrino masses of vi,, and the detection is not easy.

6.3.2 Neutrino Oscillation (in the Vacuum)

Now let us move to the discussion on the neutrino oscillation in the vacuum. (Neu-
trino oscillation in the presence of matter will be discussed in the next section.)
The neutrino oscillation is known for long time to be the most natural framework
to explain the lower than expected capture rates of solar and atmospheric neutrinos,
i.e. the puzzles of solar and atmospheric neutrinos, which we will discuss later in
118 NEUTRINO MASSES AND NEUTRINO OSCILLATIONS

this chapter. We have argued that neutrino oscillation is possible, for pure Dirac or
seesaw scenarios, only if there are more than 1 generation of leptons. Strictly speak-
ing even in 1 generation scheme, there exists an "oscillation" of the type "VL -> VR"
with chirality flip. This seems to be able to explain, just with 1 generation, the puz-
zle of e.g. solar neutrino, since the transformed state VR has no weak interaction
(a "sterile" state) and escapes the detection. However, unfortunately this is not
the case. As the neutrino masses are known to be quite small, neutrinos are highly
relativistic under the real situation of experimental detection, i.e. mv -C Ev. Since
for the relativistic particle the chirality is nearly equal to the helicity and therefore
is almost preserved, the probability of the transition UL —• VR is quite small,

P{VL -» vR) ~ fe2 « 1. (6.61)

The fact that the chirality flip is negligible suggests that in the consideration of
neutrino oscillation the spin degree of freedom is not relevant. (If we take the ef-
fect of background magnetic field into account, which may be suitable for neutrinos
propagating inside the sun or supernovae, the spin-precession of the neutrinos due
to the anomalous magnetic moment may take place without the chiral suppression
factor {mv/Ev)2 (Voloshin et al., 1988; Lim and Marciano, 1988; Akhmedov, 1988),
and this statement does not hold. We, however, ignore here the effect of the mag-
netic field.) Thus the relevant equation of motion for neutrinos is the Klein-Gordon
equation, rather than the Dirac equation. We have already discussed that when
the chirality flip can be ignored there is no essential difference, concerning neutrino
oscillations, in the two cases of pure Dirac and seesaw. Thus we now assume that
neutrinos are pure Dirac particles. The mass eigenstates vn, (i = 1,2,3) obey the
following Klein-Gordon equation
( • + m2) Vi = 0, (6.62)

whose solution with definite momentum and (positive) energy is just an ordinary
plane-wave solution

Vi = e-ipi»-x" = e~iEit • eip's, (6.63)

2
where Ei — \Jp* + m and the 3-momentum p is taken to be common for all V{
for convenience. For relativistic neutrinos, \p\ 3> rrn, the approximation Ei ~ |p| +
(m2/2\p\) ~ |p| + (m2/2E) (E: the average neutrino energy) is valid. Furthermore,
the factor e~1^1 • el$'3 is an overall phase factor, common for all neutrinos, and
does not affects the physical observables. Thus if we regard the matter waves of
neutrinos as Vi(t) = e-i^mil'1E); they obviously satisfy the following equation of
motion
m1 0 0
IE
0 2E 0 (6.64)
0 0 mi
TJ7
 Flavor Mixing and Neutrino Oscillation (in the Vacuum) 119

As neutrinos are emitted and observed as weak-eigenstates, not as mass-eigenstates,

it will be useful to consider the time-evolution equation in the base of weak-
eigenstates. The (6.52), (6.54) tells us that the differential equation is obtained
by a unitary transformation with the MNS matrix U:

(6.65)

As we have seen above this expression holds for both of pure Dirac and seesaw
scenarios. This is why we cannot discriminate these two types of neutrino mass
generation just from the data of neutrino oscillations. The above evolution equation
is readily solved to be

exp(-^mI/m+i)

= U (6.66)

\
where a relation exp(— jgmumlt) = exp(—; m
-2^ diag
2
= Uexp(-^m diagt)U* ( m d i a g = diag(mf,mi,m|)) has been used.
Suppose a neutrino is born as va at time 0. The initial condition is then given
as va(0) = 1, others = 0. Under this initial condition, the probability amplitude
m 2
to detect VQ at time t becomes u0(t) = £V Upi e - *"^"' U*{. Thus the probability
for the neutrino born as va at time 0 to be observed as vp at time t is given by a
formula (see Fig. 6.2)

P{ya ^u0) = |£ U0i e - ^ < U*ai\2

= I£ u
Pi e~ U*ai\ (6.67)

2 _ 2
In the second line of (6.67) Am-j rrv- — m\(Am ! = 0) and an overall phase
factor has been ignored. This formula clearly shows the fact that to get neutrino
oscillation both flavor mixing described by U and the mass (-squared) differences
Am 2 are necessary. In fact, if A m ^ = A m ^ = 0 then P(va -> v$) = 0 for
a 7^ P due to the unitarity (UU^)pa = 0. More intuitively, the neutrino oscillation
120 NEUTRINO MASSES AND NEUTRINO OSCILLATIONS

_.m?

» » » > » •

Fig. 6.2

is purely quantum mechanical effect and may be understood as a sort of "beat"

of (coherent) matter waves describing mass eigenstates V{. The beat, of course,
needs slight differences of frequencies, i.e. slight mass differences of i/; (for a fixed
momentum p t o realize the coherence). Though we have supposed that there are 3
generations, actually the above formula is valid for arbitrary number of generations.
One remark here is concerning CP violating phases. In the case of pure Dirac
neutrinos the physically meaningful degrees of freedom of CP violating phases in
the MNS matrix U is \n-l)\n-2) ( n . t n e number of generations), just as in the ar-
gument of Kobayashi-Maskawa matrix in the quark sector. In the case of Majorana
neutrinos suggested by the seesaw scenario, however, it is known that the argument
needs some modification. Namely, the re-phasing of (left-handed) neutrino states
in order to eliminate some phases in U defined in (6.59) is no longer possible, while
the re-phasing of charged leptons goes just as in the quark sector. This is because
the Majorana mass term for mass-eigenstates Vi, —\mi (VIL)0 ViL, is not invariant
under the re-phasing vn -t e1^' vu,, as the mass term violates U(l) symmetries,
especially the lepton number, in general. Thus the number of physically meaning-
ful CP-violating phases is modified into n 2 - n(.n~1) - n = n^n~1), The difference
"'"2~ ' — '"~ '2'n~ ' — n — 1 denotes the number of newly added CP phases for
Majorana neutrinos and these phases are called Majorana-phases. Thus even in
2 generation we have CP violation, in principle. Such Majorana-phases, however,
does not appear in neutrino oscillation processes. This is simply because the dif-
ference between pure Dirac and seesaw (or Majorana) scenarios does not manifest
itself in the neutrino oscillations. The essence of this statement is that even if there
may appear a phase e 2tl ^mj by the re-phasing of neutrinos, what matters in the
oscillation probability is the combined factor (e2t<)>imi)(e2l<t>imi)* = m 2 , not the
mass itself. Recall that the neutrino oscillation we are considering is the one with-
out chirality-flip. This suggests that the Majorana phases are really meaningful in
the processes with chirality flip and the violation of lepton number, whose typical
example is neutrino-less double /3-decay. In fact, the factor i appearing in (6.30) is
a sort of Majorana phase, and we have seen there that the phase played a role to
cancel the amplitude of the neutrino-less double /3-decay in the pure Dirac scenario.
Though there exist 3 generations, in reality, for illustrative purpose let us discuss
for a while the simplified 2 generation scheme, assuming that va = ve, i/M and
Vi = v\, V2- We may assume without loss of generality that the mixing matrix U is
 Flavor Mixing and Neutrino Oscillation (in the Vacuum) 121

real orthogonal matrix, as there does not appear any CP violating phase for n = 2:
( cos 6 sin 6 \
TT . „_
U= [ . 6.68
v
\ -sin 6 cos 6 J '
In this case, the mass-squared difference A m ^ is unique. Then (6.67) tells us that
the transition probability of the process ve —> v^ reads as

P(„ e -> v) = sin220 s i n 2 ( - ^ I * ) . (6.69)

As we expected, P(ue ->• i/^) = 0 for 8 = 0 or A m ^ = 0. On the other hand the

"survival probability" of ve is given as

P(i/ e ->• i/e) = cos40 + sin4<? + 2cos 2 0sin 2 0cos(-—^t)

= l-sin220sin2(^|^). (6.70)

These probabilities satisfy a relation

P(ue -> ue) + P{ve -> i/„) = 1, (6.71)

which reflects the conservation of probability.

For an arbitrary number of generations, the CPT theorem implies

P{ya -> vp) = P(vp -> P£). (6.72)

If CP is an exact symmetry, we further get

P(va -*• vf>) = Pfc -> Pji). (6.73)

If T is an exact symmetry, we get

P(va -»• i//3) = P ( ^ -)• i/ a ). (6.74)

(6.73) is obtainable from (6.72) and (6.74) as the CP-invariance is equivalent to
T-invariance, under the CPT theorem. In the two generation case there is no CP
violating effect, and all of these relations hold. (These relations are spoilt, in general,
once the matter effects, discussed in the next section, are taken'into account.) For
instance,
P(Vli -> Ve) = P(yt -> !/„), (6.75)

and therefore the unitarity for ve, (6.71), and the corresponding relation for v^
imply

P(v» -> "„) = P(ye ->• i/e). (6.76)

In realistic experimental situation, it is quite possible that the wave length of
the neutrino oscillation is much smaller than the uncertainties of the positions of
122 NEUTRINO MASSES AND NEUTRINO OSCILLATIONS

production and detection points of the neutrinos. If it is the case, the relevant
quantities are time-averaged (the average over the positions is equivalent to the
time-average) transition or survival probabilities. For instance, in the 2 generation
scheme, we have the time-averaged probabilities denoted by P,

P~{ve -> vj = ^sin 2 20, (6.77)

P~(ve -)• ve) = 1 - isin226>. (6.78)

Though the survival probability (6.70) itself can be sufficiently small for 6 ~ 7r/4,
the time-averaged probability has a lower bound,

P(i/ e -> ve) > \- (6.79)

This can be easily generalized to the scheme of arbitrary number of generations n:

P > e -> ue) > - . (6.80)

n
To prove this inequality, we first notice that when time-average is taken, 6.67 pro-
vides us

P(va -+vp) = Yd \U0i\2\Uai\2. (6.81)

i

In particular, the averaged survival probability of va is simply given as

i> Q ->^) = ^|£U4. (6-82)

Then the inequality is easily proven by use of the unitarity condition of the MNS
matrix, £ ] f |£/ tt j| 2 = 1. The square of this unitariry relation yields

£ |t/ a i | 4 + 2 Y, \Uai\2\Uaj\2 = 1. (6.83)

i i<j

On the other hand, we have a trivial relation

J^QUail2 - \Uaj\2)2 > 0, (6.84)

which gives

(n - 1 ) ( J 3 \Uai\4) - 2 ^ \Uai\2\Uaj\2 > 0. (6.85)

i i<3

Summing up (6.83) and (6.85), and dividing by n, we get what we need

4
Y>-| >-- (6-86)
 Resonant Neutrino Matter Oscillation 123

The inequality P{ve -*ve)>\ is the simplest example in the 2 generation scheme.
We further note that the equality in the above relation, i.e. £)i \Uai\A — ^ is real-
ized when the equality in the (6.84) is met: \Uai\2 — \Ua2\2 ~ ••• = \Uan\2, namely
in the case of "maximal mixing".

6.4 Resonant Neutrino Matter Oscillation

We have seen in the previous section that the time-averaged survival probability
never gets lower than 1/n (n : the number of generations). Thus, in the 3 genera-
tion scheme it is impossible for the vacuum oscillation (neutrino oscillation in the
vacuum) to account for data which indicates the survival probability lower than
1/3, such as the one of the pioneering solar neutrino experiment by R. Davis and
collaborators. Furthermore to realize the minimum value 1/3, a fine tuning of the
mixing angles is necessary so that \Uai\2 = \Ua2\2 = \Ua3\2 is realized.
Such difficulty may be overcome, once we invoke to a resonant neutrino os-
cillation due to the interaction of neutrinos with matter inside the medium, e.g.
the sun. Before discussing the detail of the "resonant matter oscillation" it will
be useful to note that the matter effect (the interaction of neutrinos with matter)
leads to an additional potential energy V(x) only for ve. The V{x) then causes
the change of the frequency of the matter wave of ve, depending on the coordinate
x, and therefore time t (x ~ t), for a highly relativistic light neutrinos, while the
frequency of the matter wave of another flavor, e.g. v^ in 2 generation scheme,
stays a constant. Thus it is quite possible that at some point, or equivalently at
some time, the frequencies of these two matter waves just coincide, and a resonance
phenomenon occurs. If we start from the matter wave of ue, at the resonance point
almost complete transition of the matter wave from ve to v^ is possible, even for
small mixing angle 8, as far as some suitable condition is met. It is analogous to the
case of two tuning-folks of musical instrument, the length of one of these two folks
being variable. When the lengths of two folks coincide the vibration of one folk
may be resonantly transferred into that of another folk, almost completely. One
important thing is that for the transfer to be efficient the variation of the length
should be slow enough, i.e. an adiabaticity condition should be met. The role of
the mixing angle 6 is played by the air (!), which mediates the sound connecting
the vibrations of the two tuning folks (see Fig. 6.3).
Now let us confirm the above intuitive argument really holds by use of math-
ematical formulae. First let us calculate the potential V(x). Since neutrino os-
cillation is caused by the interference effect (beat) of coherent matter waves with
a fixed direction of momentum ft, the potential V(x) should be attributed to the
elastic forward scattering of neutrinos off electron, proton and neutron inside the
medium, due to the weak interaction. V(x) changes the dispersion relation between
|p| and E of neutrinos, and therefore the index of refraction of neutrinos. Such elas-
124 NEUTRINO MASSES AND NEUTRINO OSCILLATIONS

•* • •* •

resonance
point
air

Fig. 6.3

p,n,e

'ei"\ii p,n,e
(a) (b)

Fig. 6.4

tic scattering with protons and neutrons are possible only through neutral current
processes (Fig. 6.4(a)). In the case of neutrino interaction with electron, however,
in addition to the neutral current process, only i/e has a charged current process
(Fig. 6.4(b)). The matter effect due to the neutral current process via Z-exchange,
Vn(x) is universal for all of three neutrinos. This Vn(t) provides an overall phase
factor exp(—if Vn(t')dt') for all neutrinos. The universal phase has no physical
consequence, since neutrino oscillation is caused by superposition of multiple matter
waves with slightly different frequencies, i.e. by the "beat" of matter waves. The re-
maining charged current process only for ve seems to be a scattering process, rather
than providing a potential. It, however, turns out that it can be interpreted as a
process to give the potential. The charged current process, Fig. 6.4(b), provides an
effective 4-Fermi contact interaction Hamiltonian (for neutrino energies E C Mw)

~T= • Velnil ~ 7s)e • e7"(l - ja)ve. (6.87)

The (V - A) x (V — A) type 4-Fermi coupling is known to be rewritten by use of

Fierz transformation as

- | • Ve1»{l ~ 7 s H • e7"(l - 7 5 )e, (6.88)

 Resonant Neutrino Matter Oscillation 125

which is practically equivalent to the neutral current process. Let us recall that in
QED, a background electric Coulomb potential V(x) interact with electron as e^e x
V(x). Similarly, the above 4-Fermi interaction may be regarded as an interaction
of the left-handed ve with "Coulomb" type "static" potential Vc{x)

ueLj0i/eL • Vc(x), (6.89)

Vc(x) = V2GFNe(x), (6.90)

where e7°e is identified with the electron number density Ne(x), and other com-
ponents, ejle (i = l,2,3),e7 M 7 5 e have been neglected as they are expected to be
proportional to the expectation values of velocity or spin of the electron and may
be neglected for static and un-polarized medium of electron. Let us again recall
that under the influence of 4-potential A^ = (V, A) the dispersion relation be-
tween momentum and energy of ve is modified into E — Vc = y(p — A)2+m2 or

E = y (p - A)2 +m2 +VC. Thus concerning the static un-polarized medium, only
uei gets additional potential energy Vc, and corresponding change of the energy.
Thus the time evolution equation, "Schrodinger equation" in the base of (ve, v^, vT),
(6.65) is accordingly modified into
0
°\ / v
a(t) 0 e
0
Amj|
| • C/+ + 0 0
° HW""r
0 0 0/
2E
(6.91)
where the matter effect has been indicated as a(x) = ^/2GpNe(x), instead of Vc,
and the replacement mf —> Am-] has been done ignoring an overall phase associated
with ml/2E.
Though we should treat the realistic 3 generation scheme, such treatment is com-
plicated, generally speaking. As we will discuss in the next section, fortunately there
is an approximate reduction formula, valid under the hierarchy of mass-squared dif-
ferences A m ^ <3C A m ^ , which enables us to reduce the problem in 3 generation
scheme to that in an effective 2 generation scheme. Therefore, we will deal with the
matter oscillation in 2 generation scheme with a set of generic parameters of one
mass-squared difference and one mixing angle (Am2,8), in this section. Such argu-
ment will be useful, even in the realistic 3 generation scheme, since the reduction
formula is valid.
In the simplified 2 generation scheme, by use of (the simplified version of) (6.91)
and (6.68) the time-evolution equation can be explicitly written (after subtracting
a common factor ^ 21 sin 2 fl at the diagonal elements) as

Afue\_f V2GFNe(t) $*£sm26

(6 92)
' s U J n V * - #~»J-UJ- -
Since the "Hamiltonian" H(t), the 2x2 matrix, is time dependent, the above dif-
126 NEUTRINO MASSES AND NEUTRINO OSCILLATIONS

ferential equation can not be solved analytically, unless the iVe (t) has some specific
time dependence. However, in an appropriate circumstance, we may still get a sim-
ple analytic formula for the neutrino oscillation. Namely, as we have seen in the
analogous case of the tuning-folk, as far as the time dependence of a time variant fre-
quency is mild (the exact meaning of this will be defined shortly), almost complete
conversion of one matter wave to the another should be possible. The condition of
the "mild time dependence" is called "adiabaticity". The resonance phenomenon,
which occurs when the two frequencies become identical, should happen in this sys-
tem of two neutrinos when the two diagonal elements of H(t) becomes the same,
namely at the point where the following condition is met,

V2GFNe{t) = ^rcos29 = V2GFNer. (6.93)

2E
To see these things explicitly, we will move to a (time-dependent) base of neutrino
states, where the Hamiltonian H(t) is diagonalized by a time-dependent unitary
transformation:
Um(t)*H(t)Um(t) = diag(E1(t),E2(t)), (6.94)

:;) = *-<«>•(£)• <6'96>

where the time-dependent energy eigenvalues and mixing angle are given by

£i,2(0 = \ (V2GFNe(t) + ^-cos2^

I Am2 Am2 \
± ^(y/2GFNe(t)-—a>820)* + {—sw2e)*\, (6.97)

^fsm26
tan20m = -r-2 — 7= • (6.98)
^£cos26-V2GFNe(t)
Fig. 6.5 shows how the mixing angle inside the matter 6m changes as Ne varies.
We see that the resonance point, satisfying Ne(t) — Ner, is the point where the
conversion of matter wave is most efficient and therefore the mixing 0m is maximal,
i.e.

0m = \- (6.99)

We may also say that the 6m changes most rapidly around the resonance point, and
the "resonance region" can be defined as to be a region where |tan20 m | > 1.
Actually, although we have diagonalized H(t), the evolution equation is still a
coupled differential equation, since the unitary transformation by Um(t) is time-
dependent. That is why we cannot solve the differential equation analytically.
 Resonant Neutrino Matter Oscillation 127

Fig. 6.5

Namely, we get a residual term with off-diagonal elements in the evolution equation
in the base of {vmi, vm2)1,

.d E1(t) 0 0
!-— (6.100)
dt 0 E2(t) + 0
Then what is a merit of working in this base? Well, in this base the time evolution
in the adiabatic case, the case where the adiabaticity condition is met, is easily
solved. In fact if the condition

|^m|«|£2-E1| (6.101)

is always satisfied, the residual term in the Hamiltonian can be safely ignored, and
the system is approximately diagonalized:

Bk(t) 0
(6.102)
'dt Vm2 0 E2(t)
The adiabaticity condition will be the most non-trivial at the resonance point, where
the change of 0m is maximal and the difference of energy eigenvalues is minimum,
\E2 — E\\ —
= 4rg-sin20. Thus the adiabaticity condition may be written as
tan20 E
» (6.103)
d logJVc Am 2 sin20'
dx

where 6m has been written in terms of — " 1d^g" ' ',. which should be evaluated at
the resonance point. The physical meaning of this adiabaticity condition is that
the wave length of the "beat" at the resonance point, i.e. ~ 1/|£^2 — E\\ =
E/(Am2sm28), should be much smaller than the width of the resonance region,
i.e. ( A ^ s i n 2 0 ) / ( v / 2 G F M , | r e s ) = t a n 2 0 / ( f L ! o g ^ | r ( ; s ) _
128 NEUTRINO MASSES AND NEUTRINO OSCILLATIONS

The above differential equation is easily solved to yield

uml(t) = exp(-t f EiWdt') • vml(0), (6.104)

Jo
Vma(t) = exp(-t / E2(t')dt')-i/m2(0). (6.105)
Jo
Thus the survival probability is given as

P(ve -> ve)

(i/e|Texp(-t / H(t')dt')\ue)
Jo
e x p ( - i /„' E1(t')dt')
( cos9m(t) sm6m(t) )
exp(-t /„' E2(t')dt')
cos0m(O)
sin0m(O)

cos0m(i)cos0m(O) e x p ( - i / Ex(t')dt')
Jo

+ sin#m(£)sin0TO(O) exp E2(t')dt') (6.106)

Jo
If an average is taken over the detection time t, the interference term may be ignored
and the formula simplifies into

P(ye -> i/e) = cos 2 0 m (i)cos 2 0 ro (O) + sin 2 ^ m (i)sin 2 6l m (0). (6.107)

If the matter effect is ignored, i.e. if Ne = 0 and 6m(t) = 9m(0) = 9, this reduces
to (6.78) for the vacuum oscillation.
We now realize that even if the mixing 9 is small, almost complete conversion
of a solar neutrino, starting as ue, into v^ at the solar surface is possible, so called
MSW effect (Mikheyev and Smirnov, 1985; Wolfenstein, 1978). Namely, setting
9m(t) ~ 9, 9m(0) — f and assuming iVe(0) 3> Ner, the above formula is simplified
into

P(ve ->• ve) ~ sin 2 ^, (6.108)

which can be sufficiently small even for small mixing 9, in clear contrast to the case
of (6.78), due to the resonance phenomenon. Such effective adiabatic conversion
mechanism between two states, having level crossing, is visualized in Fig. 6.6.
In Fig. 6.6, the dashed lines indicate the behaviors of the two diagonal elements
of H(t) as the functions of Ne. The solid curves correspond to the two eigenvalues
i?i(£) and E2(t). If ve is produced at a point with iVe(0) 3> Ner, ue state is approx-
imately on the upper solid curve, as the state ve is approximately an eigenstate of
H(0) with dominant energy due to the matter effect. As the matter density and
 Resonant Neutrino Matter Oscillation 129

E
i i

s ^ v e

"m2 > ^

-^—'
Am2 * ^ S"

2E v ^T«^ =-^
* "M
^J>^^ "ml
^ ^ 1
1
1
^s^ve
. 1 —• JVe
Ne/

Fig. 6.6

therefore iVe decreases, e.g. as ve traverses from the center toward the surface of
the sun, the state evolves along the upper solid curve, as long as the adiabaticity
is met, and after passing through the resonance (level crossing) point, it will follow
the upper dashed line, to end up as the i/M state.
In the case that the adiabaticity condition is not satisfied, the above formula
should be modified. In this case at the resonance point or at the level crossing,
there should be a "jumping" between vm\ and vm2 states, i.e. between the two
solid curves. Let the probability of the jumping be Pjump- Then the two states
interchange by a probability PjUmp and remain to be the same by a probability
1 — Pjump- Now the survival probability reads as

P(ue -> i/e) = (1 - P jump )(cos 2 6> m (t)cos 2 0 m (O) + sin20m(*)sin2<9m(O))

+ (sin 2 0 m (i)cos 2 0 m (O) + cos 2 0 m (£)sin 2 0 m (O))

9 + (9 Pjump oJ)cos20m(£)cos20m(O). (6.109)

Setting 6m(t) ~ 0, 0TO(O) ~ | , the formula reads as

P{ve -> ue) ~ sin20 + Pjump cos20 (6.110)

This is based on the reasonable assumption that the jumping takes place only
near the resonance point. Then it will be a meaningful way of doing to replace
the matter effect by its linear approximation so that they coincide at the resonance
point. Such a replacement makes it possible to use the Landau-Zener formula for a
level crossing (Landau, 1932; Zener, 1932), leading to

7rAm2sin220
• jump = exp(- -), (6.111)
d log
4Ecos20 dx
 NEUTRINO MASSES AND NEUTRINO OSCILLATIONS

log Am 2

log (sin2 20/ cos 20)

Fig. 6.7

where the derivative —J* " e should be taken at the resonance point.
In this way, substantial conversion of ve is possible by the MSW effect for solar
neutrinos, as the typical example of the resonant matter oscillation. When the
survival probability P{ve -> ve) is fixed by a experimental data, it will determine a
curve in the 2-dimensional space of the parameters, ( ^ f ,Am 2 ). When P(ue -¥
ue) < 1/2, the curve is known to draw a closed triangle (roughly speaking) in the
log-log plot of the parameters shown in Fig.6.7, which we call MSW triangle.
The three sides of the triangle have their own physical meaning. Namely, the
horizontal size represents the parameters, for which the level crossing starts to
occur, Ne(0) ~ Ner = A 92y/2G
T
^ | ^° SE^ 6> with E being an average neutrino energy. For
F

E = 10(AfeV) and JVe(0) = 100{g/cc), Am2 10 _ 5 (eV 2 ). The vertical line is

given by a relation sin 2 0 = observed survival probability, i.e. by use of (6.108).
For the parameters on this line, all neutrinos with different energies experience the
level-crossing and satisfy the adiabaticity condition. The diagonal line is given by a
2
relation P,„ m „ = constant, i.e. by Am • . ^ 1 ^ . = constant. This line separates the
jump
adiabatic (upper) and non-adiabatic regions, since the factor Am 2 • 4gn0g2g inside
Pjump determines the adiabaticity of the transition. This MSW triangle plays an
important role when we derive the allowed parameter region from the data on solar
neutrinos, as we will see briefly later in this Chapter.

6.5 Neutrino Oscillation in the Three Generation Scheme

So far the concrete formulae for the probabilities of neutrino oscillations by use
of the mixing angle and the mass-squared difference have been given only for the
simplified two generation scheme, for illustrative purpose. Needless to say, to make
the analysis realistic, we have to derive formulae in the full three generation scheme,
in terms of two mass-squared differences of neutrinos,

Am^, Am^, (6.112)

 Neutrino Oscillation in the Three Generation Scheme 131

and three mixing angles and one CP violating phase,

012, 023, 013, S, (6.113)
which appear in t h e MNS m a t r i x as follows,

C12C13 s 1 2 ci3 sue"15

l5 ,(5
U = I -S12C23 - C i 2 S 2 3 S l 3 e C12C23 - Sl2S23Sl3e S23C13
lS <5
S12S23 - Ci2C23S\3e -C12S23 - Si2C 2 3Si3e* C23C13

f 1 0 0
0 C23 «23
V 0 -S23 C23

23^13^12,

where Sij and Cij represent sin#,j and cos0jj, respectively. (This parameterization
just corresponds to (4.183) for the KM matrix in the quark sector.) Such formulae,
however, are complicated even for the case of vacuum neutrino oscillations, being
described by the 6 parameters A m ^ , A m ^ , ^ , 0 2 3 , 0 1 3 , 5 . Thus given data on the
neutrino oscillations provide 5-dimensional hyper-surfaces in the parameter space,
which of course cannot be shown graphically. Even if we get the allowed values of the
6 parameters numerically, it may be practically impossible to get some meaningful
physical information on the parameters from them. To be worth, we will not be sure
with the obtained numerical values of the parameters, which mass-squared difference
and mixing angle are really handling the oscillation, while each experimental data
on the neutrino oscillation is usually provided in the form of the allowed region of the
parameters assuming a simplified two generation scheme with generic parameters
(0,Am 2 ).
Fortunately, the hierarchical structure of two mass-squared differences of neu-
trinos, suggested by the data of atmospheric and solar neutrinos experiments,

A m ^ < Am,!, (6.115)

enables us to reduce the problem of analyzing neutrino oscillations in the full three
generation scheme to that in the effective two generation scheme. Namely, "re-
duction formulae" are known to exist for each type of neutrino oscillations, which
obviously make the analysis of the allowed region of theory parameters quite similar
to those in the simplified two generation scheme, thus making the analysis quite
transparent and physically meaningful. In addition, in this way we can clearly see
which mixing angle and mass-squared difference out of the 6 parameters should
be identified with those appearing in the experimental data. As we will briefly
discuss in the next section, to explain the data on the atmospheric and solar
neutrino oscillation two different scales of mass-squared difference are inevitable:
Am2atm ~ 2 x 10~ 3 (eV2) and A m ^ l o r ~ 7 x 1CT5 (eV2). Thus the identification

Am231 = Am2atm ~ 2 x 1(T3 (eV2),

132 NEUTRINO MASSES AND NEUTRINO OSCILLATIONS

Am221 = Am2solar ~ 7 x 1(T 5 (eV2), (6.116)

should be made, which means the hierarchy (6.115) is a reasonably good assumption.
(In fact numerical calculations show that the reduction formulae given below provide
enough accurate oscillation probabilities.)
Now we will discuss these reduction formulae for each of vacuum and resonant
matter oscillations, successively below.

6.5.1 Vacuum Neutrino Oscillation in the Three Generation Scheme

Even under the hierarchical mass-squared difference (6.115), the reduction formulae
of the oscillation probabilities will take different formulae, depending on which of
A m ^ and Amfj is responsible for the oscillatory time-dependence of the oscillation
probability.

(A) Vacuum oscillation due to A m ^

We first consider the vacuum neutrino oscillation, where larger mass-squared
difference A m ^ handles the oscillation. This, in turn means that A m ^ is too
small to participate to the oscillation. It will be relevant for the experimental
situation satisfying,

Amf, _

Amf,

where L denotes the distance between the production and the detection points of
neutrinos. The oscillation of atmospheric neutrinos and terrestrial "long-baseline"
accelerator neutrino oscillation experiments may be classified into this category. It is
worth while noticing that the oscillation length of atmospheric neutrino oscillation,
implied by the zenith-angle dependence of its survival probability, is not the order
of the size of the earth, but is of the order of a few hundred kilometers, roughly the
distance from the Super Kamiokande detector till the top of the atmosphere at the
Zenith-angle = \. Strictly, for the up-going neutrinos in the Super-Kamiokande
detector, i.e. from the opposite side of the earth, the oscillation due to A m ^
may not be negligible, though we do not consider the effect in this textbook. The
wavelength is comparable to the baselines of K2K (KEK to Kamiokande) and other
proposed long baseline experiments.
Thus, ignoring the smaller mass-squared difference, A m ^ = 0, the general
oscillation formula (6.67) reduces into

P{ya -+ up) = \Sa0 - UP3U*a3(l - e - ^ ' ) ! 2 , (6.118)

 Neutrino Oscillation in the Three Generation Scheme 133

due to the unitarity J2i UpiUai ~ <W- Several survival and transition oscillation
probabilities, relevant, e.g., for the atmospheric neutrinos, now takes simple concrete
forms in terms of mixing angles

P(v» -» v„) = -4(l-|t/,3|2)|f/,3|2sin2(^)

Am:31
- 4(1 - sin 2 0 23 cos 2 0i 3 )sin 2 0 23 cos 2 0 13 sin 2 ( t)
AE
-sin22^3sin2(^-t), (6.119)
m
2MJT 2 l 2 „ : „ 2 ^ 3 1 ^
P{ye -> ve) = -4(l-|C/ e 3 | 2 )|[/e3| sin^(
AE
2
Am
sin2 20i3 sin 2 ( 31
t)~l, (6.120)
AE

P{»* Vr) = 4|C/ M 3 | 2 |C/ T 3| 2 sin 2 (^|li)

= sin 2 20 23 cos 4 0 13 s i n 2 ( ^ - i )
AE
J 22 , A m | l f ) )
~ sin 2023 sin ( (6.121)
AE
'31
Pfa-H'e) = 4|C/ e3 | 2 |t/ M 3| 2 sin 2 (- t)
AE
sin22013sin2023sin2(^-t)~O, (6.122)

where the approximate formulae are for small #i 3 , reported by e.g. CHOOZ exper-
iment,
sin 2 20 13 < 0.2. (6.123)

These formulae imply that, roughly speaking, ve does not oscillate and is decoupled
from other neutrino species, and only v^ O vT oscillation is possible. This is what
we expect for atmospheric neutrinos. We should also note that the formula for the
Vf! <-> vT oscillation (6.121) is of exactly the same form as the transition probabil-
ity in the simplified two generation scheme, i.e. as (6.69), though the replacement
0 -> ^23, Am 2 ! —> A m | j has been made. Namely, in this case the mixing handling
the oscillation is known to be 0 23 • Thus, the experimental data on the atmospheric
neutrinos should provide constraints on the set of parameters (0 2 3 ,Am 3 1 ), as we
will see in the next section. Concerning the possible CP violation in neutrino os-
cillations, the formulae given above are not suitable, since ignoring Am 2 1 and 0i 3
will erase the CP asymmetries, just as in the quark sector (see (9.57), (9.58)).

(B) Vacuum oscillation due to A m | j

Next, we consider the vacuum neutrino oscillation, where the smaller mass-
squared difference A m ^ handles the oscillation. This, in turn, means that Am| x is
134 NEUTRINO MASSES AND NEUTRINO OSCILLATIONS

so large that the oscillation length due to AmJi is much shorter than the baseline L,
i.e. the distance between the production and the detection points of the neutrinos.
Thus the oscillatory factor due to A m ^ should be time-averaged, i.e.
.Am?, . „ ,
C0S( } (6 124)
~ 2 # * ~* °' '
which means the interference of the matter waves of v3 with those of v\ and v2 can
be ignored. This treatment will be relevant for the experimental situation satisfying,
A 2 ( Am 2 i \ T

{
E (^f^) 100km>- '
\5MeV I
^ 3 1 L = 2 .0xl02. W ^ i Z ( _ £ _ ) » ! . (6.125)

This is exactly the case for the KamLAND experiment, recently started in Japan.
Ignoring the interference of the matter waves of v3 with those of v\ and v2, the
general oscillation formula (6.67) reduces into

P{ya -> vp) = \Sa0 - U03U*a3 - U02U*a2(l - e - ^ ^ 4 ) ! 2 + \U03\2\Ua3\2. (6.126)

In particular, the survival probabilities for a = ft are expressed by the following
reduction formula to the effective two generation scheme:

P{ua -+ va) = (1 - \Ua3\2)2 Peff(va ->• ua) + \Ua3\\ (6.127)

where Peff denotes the survival probability in the effective two generation scheme,
which is obtained by the replacing the factor sin20 in the formula for two generation
\u 2 12
scheme by a factor 1_iU ' p . Namely,

,„,,„. _ „., . , _ tMli-W-W-?) sin,(^kt). (6 , 28)

For instance, the concrete reduction formula in terms of mixing angle for the ue
survival probability, which we denote by 5, is given as
S = cos 4 6i 1 3-5e//(6ii2,Am^ 1 )+sin 4 (9 1 3,
~ cos 20i3-S e / / (0i2, A r r 4 ) (6.129)

where the survival probability of ve in the effective two generation system, denoted
by Seff is described by a set of parameters (#12, Am^i) and reads as

Seff(612, A m ^ ) = 1 - sin220i2 sin2(^^-t). (6.130)

The approximate formula in the second line of (6.129) is for small 9i3, (6.123). The
formula (6.129) with (6.130) will be also applicable to the survival probability of VI,
relevant for the KamLAND experiment, as well, as long as the small matter effect
 Neutrino Oscillation in the Three Generation Scheme 135

due to the neutrino interaction with the earth is negligible. Thus, in this case the
oscillation is known to be handled by the mixing angle #12-

(C) C P violation in neutrino oscillation

As we will discuss in some detail for the quark sector in Chapter 9, to get CP
violation we need the full interplay of three distinct generations of quark. In other
words, if there is mass degeneracy in up- or down-type quark sector, or if some of
generation mixing angles disappear, the CP violating observables will vanish (see
(9.57) and (9.58)). The same arguments hold for the lepton sector (at least when
neutrinos have Dirac masses). Thus the formulae given in (A) and (B) above, where
either A m ^ or A m ^ is responsible for the oscillations, are not applicable when we
consider the CP violation in neutrino oscillation, and we should go back to the
original formula (6.67).
The clear indication of CP violation in the leptonic sector will be the asymme-
try between the oscillation probabilities of neutrinos and the corresponding anti-
neutrinos defined by

A% = P(va-> vp) - P(9a-+up). (6.131)

The consequence of the CPT theorem shown in (6.72) (for the negligible matter
effect) implies that the CP asymmetry is identical to the T asymmetry defined by

Alp = P{va -> vp) - P{up -»• va). (6.132)

Namely, A^F = A^g. This can be confirmed explicitly in the formula (6.67).
Namely, the charge conjugation causes the replacement Uai —> U*i for every MNS
matrix elements, which is equivalent to the replacement a <->• /? in (6.67).
Then, a trivial relation A^p = —ATpa tells us

ACJ = -A%, (6.133)

which in particular means A%£ = 0, namely

P(ua -*• va) = P{va -» va). (6.134)

Thus the CP violating effects do not appear in the survival probabilities and appear
only in the transition probabilities. On the other hand, since the neutrino oscilla-
tion we are considering here does not break lepton number, the unitarity, or the
probability conservation read as

2 P(va-*vp)= J2 P^a-^vp) = l, (6.135)

P=e,ti.T 0=e,it.T

which means J2p Aap = °- Together with A%£ - 0 and A°% = -A^ we get a
relation for a = e, A^ = A^f. Similarly, for a = n we get A^J = Aff. We thus
136 NEUTRINO MASSES AND NEUTRINO OSCILLATIONS

obtain a remarkable relation as the specific feature of the three generation model,

A?f = A™ = A?ep = ACP, (6.136)

This is essentially the consequence of the fact that in the three generation model
the CP violating phase is unique, i.e. 8 alone.
Let us now derive the explicit form of AGP. From (6.67) we realize that for an
arbitrary choice of the pair (a,/3),

4
A% = P{»a -+ vp) - P(*a -> ftp) = ~ E MUaUtiUnUZj) •sin(^t).
(6.137)
Write TmiUdU^UpiU'j) = Jap,ij- By definition we readily know Jap,ij — ~Jp,a,ij
and Jap,ji = —Jap,ij- On the other hand, the orthogonality of the MNS matrix
implies £V Jap,ij = 0. We thus know for j = 2, e.g., J a/ 8,i 2 + Ja/3,22 + Ja/3,32 =
Jap,i2 + Ja0,32 = 0, Similarly, for j - 3 we get Jap,\z + Ja/3,23 = 0. By use of
the property Jap,ji = —Jap,ij, we thus conclude Jap,i2 = Ja0,23 = Ja0,3i- Similar
reasoning provides Jeii,ij = J\IT,%J = Jre,ij- In this way we have demonstrated that
actually Jap,ij is unique, in spite of the possible choices of the combinations of
a, /? and i,j. Hence we define J = J e /i,i2, which is the leptonic counterpart of the
Jarlskog parameter (9.58) in the quark sector. In terms of the parameterization in
(6.114),

J = Cl2Sl2C23S23Ci3Si3S<5, , (6.138)

where s$ = sin<$ with S being the CP violating phase. Now setting as a = e, ft = fi

in (6.137) and utilizing Je^,i2 = ^ , 2 3 = Jen,3i, we readily get

ACP = -4J { s i n ( ^ f ) + s i n ( ^ i ) + M ^ t ) } , (6.139)

where Amfj = rrif — m | . The appearance of the factor J is quite reasonable, as

it should be the unique re-phasing invariant measure of CP violation in the three
generation model, as we will see in section 9.4.
The above formula clearly shows that, as we anticipated from the introductory
argument, the CP asymmetry ACP vanishes when (at least) one of the following
situation is realized, though they may be unrealistic: (1) there is a degeneracy of
neutrino masses, i.e. Am^i = 0 or A m ^ = 0, (2) one of the mixing angles 0y or CP
violating phase S vanishes, (3) time average of the neutrino oscillation probabilities,
i.e. the average of the factor sin( 2E'^ 1S taken. Any experiment, which aims at
the observation of the CP asymmetry should carefully avoid the situations (1) and
(3). In particular, the experiment should be sensitive to the smaller mass-squared
difference A m ^ . It should also be mentioned that the presence of the matter effect
mimic the CP asymmetry, just because the matter effect for neutrinos and that for
 Neutrino Oscillation in the Three Generation Scheme 137

anti-neutrinos are not the same. Thus how to avoid or extract the matter effect is
another challenging issue.

6.5.2 Resonant Matter Oscillations of Neutrinos in the Three Gen-

eration Scheme
Once the matter effect due to the weak interaction of neutrinos with the matter, e.g.
inside the sun, is included, the time evolution equation is not analytically solvable,
and the analysis of the oscillation probabilities in the full three generation scheme
becomes quite cumbersome, in general. Still, however, some reduction formulae are
possible if the matter effect is much smaller than A m ^ , i.e.

^ p , V2GFNe « ^ , (6.140)

which is the case for solar neutrinos. (Concerning the neutrinos emitted from super-
novae, the matter effect can easily exceed Am^1/(2E) in the region near to the core,
and two-step level crossings become possible. Though there exists an extension of
the reduction formula, applicable for the case, it is out of scope of this textbook.)
Here let us focus on the ne survival probability S. We will now see that, even in
the presence of the matter effect, a similar reduction formula to the one given in
(6.129) holds;

S = cos 4 0 13 • Seff(612, Am2n;aeff) + sin 4 0i 3 , (6.141)

where aeff = cos 2 0 13 a(x) = cos 2 0 13 \J2GFN<,(X) is the "effective" matter effect,
reduced by the factor cos 2 #i3, in the effective two generation scheme (Lim, 1987;
Smirnov, 1992; Shi and Schramm, 1992).
Now let us derive this formula. We start with the time-evolution equation in
the three generation scheme, (6.91). What we attempt is to separate the 3 x
3 Hamiltonian matrix, though it cannot be exactly diagonalized because of the
presence of the matter effect, into those for one and two neutrino systems, i.e.
3 —» 1 + 2, by making some suitable approximation. Namely, we attempt to make
the Hamiltonian of the form of block-diagonal, relying on the hierarchy (6.140). It is
reasonable to expect that one mass eigenstate uz, having the dominant "energyh, i.e.
Am2
the diagonal matrix element 2£x, is decoupled form other two states, orthogonal
to 1/3, namely the linear combinations of v\ and v-i. Thus, in the base of mass-
eigenstates, the Hamiltonian is approximated by H ~ diag(0,0, -^gp-), roughly
speaking. Instead of mass-eigenstates i>i and i/2, however, it will be convenient to
take their linear combination defined by u'e = cosfl^i + sm6i2V2, ^ = — s i n f l ^ i +
cos#i2^2- The reason is, in this way, the 2 x 2 sub-matrix of the Hamiltonian in the
space of v'e and v' has a similar form to that in the two generation scheme, since
only v'e has the matter effect and the term of a(x) appears in the diagonal element
138 NEUTRINO MASSES AND NEUTRINO OSCILLATIONS

of the sub-matrix. Namely, we will work in the base

cos# 12 ^i + sin^12i/2
-sin^i 2 ^i + cos^i2^2 | • (6.142)
"3 / \ VZ

Recall that from (6.52), (6.114) the relation ipw = V ^ V i s V ^ m holds between the
column vectors of weak- and mass- eigenstates, \j)w = {veL,vllL,vTi,)i and ipm =
(viL,v2L,V3L)t- Thus, the relation (VJ3ipw) = Vi3(Vi2ipm) means £ = Vi2ipm =
V/3 (V23 ipw). More explicitly,

/ cos^i3i/e - smBlze~iSvT
( v'e \
^

1/3 /
= ^

V sin6>i3ei,5^e+cos^i3J>r
\

/
, (6-143)

where
sm
£M = COS#23^V ~ ^23^T - -7={Vfi ~ VT),

vT =
sin0 2 3^ + cos023^T - —j=(vn + vT), (6.144)
v2
where the approximate relations are for #23 — f, as suggested by the data on the
atmospheric neutrino oscillation. Actually, as far as the survival probability of ve
is concerned, we may set #23 = 0 without changing the answer, since both of initial
and final states have nothing to do with v^, uT and the matter effect is invariant
under the rotation of these states. As we expected, among v'e and 1/^, only v'e has
the ve component and, therefore, the matter effect.
Now the time-evolution equation in the base of £ is written as

S=\v
dt 12
l A12 W2 + ^
A 13,
' A i 2 5 ? 2 + ac? 3 Ai 2 si 2 c 1 2 0 \ / 0 0asi3ci3'
A12s12c12 A 12 c? 2 0 + 0 0 0 ) ^,
0 0 A13 + a s ? , / \as13C13O 0
(6.145)

Am|, A Am*
where A 1 2 = -fgf 1 ) ^13 — ~ ^ a n d c*j = cos dij, s^ - sin 6{j. The second matrix
in the right-hand-side of the above equation may be treated as a small perturbation
under the hierarchy (6.140), and may be ignored. One may wonder why the small
quantities in the 2 x 2 sub-matrix in the first matrix in the right-hand-side are kept.
This is because at the leading 0-th order of perturbation, i.e. eliminating these
small quantities as well, the Hamiltonian has degenerate eigenvalues 0 in the space
of sub-matrix. Thus the two eigenstates in the sub-space is not fixed at the 0-th
order, and the small perturbation should be kept only in the sub-space. We can
 Atmospheric and Solar Neutrino Oscillations 139

read off the factor cos2#i3 a(t) in the sub-matrix, which comes from the relation
v'e — cos#i3i/e - sin^ise - 1 " 5 ^. Namely, in this sub-space the matter effect is reduced
to

aeff = cos 2 0 13 a{t). (6.146)

Thus the time-evolution is described approximately by

dc ( Ai 2 s? 2 + acf3 Ai 2 s 12 ci2 0 \
if ~ A12s12c12 A 12 c? 2 0 £. (6.147)
V 0 0 Al3+as213J
Let us note that the separation 3 —> 1 + 2 has been realized, and the time-evolution
in the subspace of (u'e, u'^)1 is exactly the same as that of the two generation scheme,
except for the replacement of the matter effect a —> aeff.
Since ve state corresponds to (ci3,0, Si3et<5)', the survival probability S is given
as
2
S = l & ^ O O i / J + ^e"^"*''")'!
= cos40i3 • S e / / ( 0 i 2 , A m 2 1 ; a e / / ) + sin 4 0i 3
~ cos2913-Seff(e12,Aml1;aeff), (6.148)

which is nothing but the reduction formula (6.141). The approximate formula in the
third line is for small #13, (6.123). In the derivation we have ignored the interference
of 1/3 with other two states, as the oscillation due to A13 is very rapid: e~ t A l 3 ' —>• 0.
The effective survival probability 5 e //(0i 2 , Am 2 1 ; a e / / ) should be calculated by use
of the 2 x 2 sub-matrix in the Hamiltonian of (6.147). Let us note that if we neglect
the small #13, 5 = Seff and ve = v'e. Namely, in this case the two generation
treatment is exact and the solar neutrinos oscillate into v' = j>M;

Ve -»• -j=(Vlx-VT), (6.149)

and we expect roughly the same amount of v^ and vT in the solar neutrinos. The
error of the reduction formula (6.141) is anticipated to be handled by the relative
strength of the ignored off-diagonal matrix element to the dominant matrix element,
of the order a fin/13 < 10~ 2 , which has been confirmed by numerical calculation.
\ 2B I

6.6 Atmospheric and Solar Neutrino Oscillations

The smallness of neutrino masses and therefore small mass-squared differences

means that the oscillation lengths of possible neutrino oscillations are sizable, and
to test these oscillations are very challenging. Though a long distance L from the
source to the detector of the neutrinos is desirable to detect such oscillations with
long wave lengths, it in turn means the neutrino flux rapidly decreases as 1/L 2 .
140 NEUTRINO MASSES AND NEUTRINO OSCILLATIONS

Thus the neutrino source with large L and very intense flux should be ideal one to
test the oscillation. We can find such ideal neutrinos sources with astrophysical or
cosmological origin, i.e. atmospheric neutrinos and solar neutrinos.
As we now very briefly discuss, both of these neutrinos have experimental puz-
zles, called atmospheric and solar neutrino problems (anomalies): the observed
event rates of these two-types of neutrinos are significantly smaller than the theo-
retical predictions. We will see below that these reductions of the neutrino event
rates are naturally accounted for by virtue of the atmospheric and solar neutrino
oscillations.

6.6.1 Atmospheric Neutrino Oscillation

It has been known for long time that high energy particles (mainly protons) are
flowing toward the earth, almost isotropically. These particles are called (primary)
cosmic ray, and interacts with the atmosphere surrounding the earth via the strong
interaction, to produce pions. The decay of the pions due to the weak interaction
then copiously produces neutrinos, called "atmospheric neutrinos", through the
chain of decay processes:

7T+ -> n+ + i/„, n+ ->• e+ + ve + Dfj,. (6.150)

Let us note that the decay rate of n+ —> e + + ue is strongly suppressed compared
with that of n+ -> fi+ + v^ shown above roughly by a factor (me/m^)2 (see (2.49)).
Though the absolute values of vt and v^ fluxes have some uncertainties in their
calculations, in the ratio of these fluxes {yis,-\-vv)j{ytJrve) {v^ denoting the i/M flux,
etc.), such uncertainty is considerably reduced. The above decay chain suggests
that the ratio is roughly 2.
The atmospheric neutrino experiment by Super-Kamiokande collaboration claims
that the "double ratioh of the observed [y^ + PM)/(i/e + ve) to that of the predicted
value remarkably deviates from 1,

R= K + PMWfo+PeU ^Q6 ( 6 m )
\Vn + V^pred/yVe + Ue)pred

This discrepancy is so-called atmospheric neutrino problem (anomaly). To be more

precise, Super-Kamiokande data show that the v^ capture rate is considerably
smaller that the expectation, while that of ve is consistent with the prediction.
Thus, probably the most natural explanation of this problem is to invoke to the
atmospheric neutrino oscillation

i/M -> J/T, (6-152)

while atmospheric ue do not experience any oscillation. This observation is consis-

tent with the theoretical expectations of neutrino oscillation probabilities for small
#13 given in (6.119) - (6.122). Comparing the survival probability P(fM -> v^),
 Atmospheric and Solar Neutrino Oscillations 141

sub-GeV multi-GeV
250 76 r
e-like e-like
p < 0.4 GeV/c 200 . p>0.4GeV/c
200 -
150 150 • i

100
-frri j
1UU •\A±
50 50

0
200
H-like
160 p < 0.4 GeV/c
T 7 - r < 7 T 7 |\i 11 k i n v r r
120

80

40

0
"-1 -0.6 -0.2 0.2 0.6 •0.6 -0.2 0.2 0.6 1 -1 -0.6 -0.2 0.2 0.6 1 -0.6 -0.2 0.2 0.6 1
cos© cose cose

Fig. 6.8 The Zenith-angle (0) dependence of ;i-like and e-like neutrino events for the scattered
particles in the energy ranges of sub-GeV and multi-GeV. The hatched regions are expected events
for no neutrino oscillation, while the bold lines indicate the (best-fit) expected events assuming
the neutrino oscillation Vy. —• vT. From the result of Super-Kamiokande collaboration, (Super-
Kamiokande collaboration, 1998).

(6.119), with the data, Super-Kamiokande has derived the allowed parameter re-
gion

A"»3i = ( 1 . 3 - 3 . 0 ) x 10- J (eV2), (6.153)

sin 2
26*23 > 0.9. (6.154)

To be interesting, the best fit value of sin2 2#23 is 1. Namely the maximal mixing
has been realized in the lepton sector, in clear contrast to the case of quark sector !
To get the restrictive range of A m ^ , the data on the Zenith-angle dependence is
very helpful. Namely, the dependence of the atmospheric neutrino capture rate on
the Zenith-angle 9 at Kamioka, shown in Fig. 6.8, is probably the most convincing
direct evidence of neutrino oscillation.
The figures tell us that the muon-like event rates at cos 9 = —1, i.e. 9 = n,
corresponding to the atmospheric neutrinos from the opposite side of the earth,
indicates a clear deficit, while the event rate at 9 — 0 is quite consistent with the
prediction. This may be attributed to the oscillation of atmospheric neutrinos v^,
which traverse long way toward the Super-Kamiokande detector. The remarkable
decrease of the capture rate at 9 = | shows the wave length of the oscillation
is of the order a few hundred kilometers. This is why the terrestrial test of the
atmospheric neutrino oscillation at the long-baseline experiment, such as K2K, is
possible.
142 NEUTRINO MASSES AND NEUTRINO OSCILLATIONS

6.6.2 Solar Neutrino Oscillation

Another ideal source of neutrinos to perform the neutrino oscillation experiment
,sensitive to the very small neutrino mass-squared differences, is the Sun. As is
well-known, the energy emitted by the Sun is due to the nuclear fusion processes at
the center region of the Sun. A main series of reactions is called "p-p chainh:

p+p -> D + e+ + ue, < Ev > = 0.26(MeV),

e_+7Be -> 7
Li + i/e, Ev = 0.86(MeV),

8 8
B -> Be* + e+ + i/e, < Ev >= 7.2{MeV),
: , (6.155)
where three reactions, which mainly contribute to the event rates at the on-going
solar neutrino experiments, have been shown, together with the average energies <
E > (a fixed energy for the second reaction concerning Be) of the emitted neutrinos.
Actually, the second and the third reactions actually belong to different branches
of the chain reactions. The neutrinos emitted by these three processes are called
pp, Be and B neutrinos, respectively. The pp neutrino dominants the solar neutrino
flux, though its average energy is relatively low and its detection is not easy, while
Be and B neutrinos have higher energies and are detectable even in the experiments
with relatively higher threshold energies, such as Super-Kamiokande, Homestake
and SNO experiments, in the mines in Japan, U.S. and Canada, respectively .
The chain reactions are summarized in the following net reaction of the nuclear
fusion

2e~ + 4p -> 4 He + 2ue + 7 (26.73MeV). (6.156)

Though of course the nuclear fusion itself is due to the strong interaction, the
presence of the weak interaction causes the emission of the solar neutrinos ue.
The experiment for the detection of solar neutrinos, solar neutrino experiment,
started by the pioneering work by R. Davis and collaborators, at Homestake mine,
the "CI experiment" by use of C2CI4. Now the data on the solar neutrino capture
rates are available from four different types of solar neutrino experiments:

1. CI experiment:
i / e + 3 7 C l -> 37
Ar + e~, obs./exp. = 0.34 ± 0.03,
(mainly) sensitive to B, Be neutrinos (Eth = 0.81MeV)
2. Super-Kamiokande experiment:
v + e~ -» v + e~, obs./exp. = 0.47 ±0.02,
 Atmospheric and Solar Neutrino Oscillations 143

sensitive to B neutrinos (Eth = 5MeV)

3. Ga experiment (SAGE, GNO(GALLEX)) :
z / e + 7 1 G a -> 71
Ge + e", obs./exp. = 0.55 ± 0.05,
(mainly) sensitive to pp, Be, B neutrinos (Eth = 0.23MeV)
4. SNO experiment :
(charged current reaction)
ve+D -+e~+p + p, obs./exp. = 0.35 ± 0.02,
(neutral current reaction)
i/ + D -^v + n + p, obs./exp. = 1.01 ±0.13,
(sensitive to B neutrinos) {Eth = 6.8MeV), (6.157)

where CI and Ga experiments are radio-chemical experiments, while Super-

Kamiokande and SNO experiments utilize elastic scattering of neutrinos off
electron (Super-Kamiokande) and the charged current interaction of ve and the
neutral current interactions of all species of neutrinos, ve, v^, and vT, with the
deuteron D (SNO). In the above equations, "obs./exp." means the ratio of the ob-
served solar neutrino capture rate to the expected one, predicted by the standard
solar model (Bahcall, et al., 2001). Eth denotes the threshold energy of each exper-
iment. We realize that these experiments have all different Eth and are sensitive to
different energy ranges. Hence we can get independent complementary data from
these experiments, which enables us to restrict the allowed region of parameters of
the theory, as we will below. We know from the above "obs./exp." ratios that all
experiments, except SNO neutral current experiment, have shown clear deficits of
solar neutrinos. This anomaly is called "solar neutrino problem".
This solar neutrino problem find a natural solution in the neutrino oscilla-
tion of solar ve. As the CI and SNO charged current experiments claim that
obs./exp. ~ 1/3, it is difficult to explain the deficits in the framework of the vacuum
oscillation. Even though the (time-averaged) survival probability P(ue —» ve) = 1/3
is possible in he three scheme, it requires maximal mixings, i.e. the fine tuning of
mixing angles (see (6.80) and the argument below that equation). Thus we should
rely on the resonant matter oscillation of solar neutrinos in the three generation
scheme, discussed in subsection 6.5.2. We have seen there that, solar ve, roughly
speaking, oscillate into the even mixture of "active" (neutrino states with weak in-
teraction) i/M and vr (see (6.149)). Comparing the observed obs./exp. ratio with the
reduction formula (6.141) or (6.148) of the survival probability, we get (with a little
uncertainty of the small #13) a MSW triangle, Fig. 6.7, in the parameter space of
(log(sin22#i2/cos 2^12), log A m ^ ) for each of experiment. Each experiment, having
different deficit rates and energy sensitivities, will draw different MSW triangles.
By taking the overlap of these triangles, we finally get a allowed region of the pa-
rameters, called LMA (Large Mixing Angle)-MSW solution, as is seen in the right
144 NEUTRINO MASSES AND NEUTRINO OSCILLATIONS

10 • 1
sin2(26)

Fig. 6.9 The allowed area in the parameter space of (sin 2 (20i 2 ), A m f 2 ) , where 6\2 and Amf 2 are
simply written as 0 and A m 2 , imposed by the data of solar neutrino experiments. The hatched area
is the allowed area at 95% C.L. obtained by the combined analysis on the neutrino flux observed
in the Ga, CI and Super-Kamiokande experiments. The area of the triangular shape, roughly
speaking, is the exclusion area due to the day/night spectrum analysis by Super-Kamiokande.
From the result of (Super-)Kamiokande collaboration, (Super-Kamiokande collaboration, 2001).

upper area of Fig. 6.9.

The best fit values are:

Am22l = 7.0 x 1(T 5 (eV2), (6.158)

2
tan 0, 2 = 0.34. (6.159)

Let us note the SNO neutral current result obs./exp. = 1.01 ± 0.13 in (6.157)
is consistent with the neutrino oscillation (6.149), since the i/M and vT are active
states and contribute to the capture rates of solar neutrinos, just as ve does. Namely
only in the SNO neutral current experiment, the capture rate is expected to be the
same as the prediction of the standard solar model, which is exactly what has been
confirmed by the experiment.
When we discuss the deficit of the solar neutrinos, of course we have to know
the prediction of the standard solar model. There exist, however, a way to confirm
the neutrino oscillation, irrespectively of the solar model, by use of the combined
results of Super-Kamiokande (Super-Kamiokande collaboration, 2001) and SNO
 Atmospheric and Solar Neutrino Oscillations 145

charged current reaction (SNO collaboration, 2001)! This should be a direct and
convincing proof of the solar neutrino oscillation. The clue is in the fact that in
the elastic scattering of solar neutrinos off electron, not only ue but also v^ and
uT contribute to the event rate, though the ratio Ra of scattering cross section of
v^ and vT to that of ve is roughly Ra = 1/6 to 1/7, because of the presence of
the additional charged current process Fig. 6.4(b) in the case of the ve scattering.
On the other hand, in the SNO charged current reaction, only ve can contribute to
the process, as the charged current processes of other neutrinos require higher than
available neutrino energies. Thus if the B neutrino flux predicted by the standard
solar model is denoted by <£ggM, the "effective" fluxes at these experiments, denoted
by $gj{ and 0gj^Q (CC) should be given as

<^SK = <£ggMx{S + (l-S)i?CT} (6-160)

C C = x S
^SNO( ' ') ^SSM (6.161)

where S is the survival probability (6.141). (For illustrative purpose, we have ig-
nored the dependence of S and Ra on the solar neutrino energy.) It is quite impor-
tant to note that the observed value </>gj^ ~ 2.4 x 106 (/cm2/s) is certainly larger
than </>gjsfo(CC) — 1-8 x 106 (/cm2/s). This means S < 1 and that among the
solar neutrinos there surely exist the active neutrinos v^ and vT, which strongly
suggests the solar neutrino oscillation ve —> i/M, uT. (It also implies that the so-
lar neutrino oscillation into a sterile state (states without weak interaction) is not
favored.) Furthermore, equating (6.160) and (6.161) with the observed values, we
get 0 g g M = 5.4 x 106 (/cm2/a) and S = 0.33, for Ra = 1/6. The obtained <I>SSM
is quite consistent with the prediction of Bahcall et al., (5 ± 1) x 106 (/cm2/s).
The most recent remarkable experimental achievement is the confirmation of
the solar neutrino oscillation in quite different circumstance, i.e. the terrestrial
oscillation experiment of ue produced by nuclear reactors in Japan, i.e. KamLAND.
As we have seen in the subsection 6.5.1, the experimental setting is such that the
oscillation is due to the smaller mass-squared difference A m ^ and 9\2, just as
in the solar neutrino oscillation, though the matter effect may be safely ignored,
even though not negligible. Thus the formulae (6.129) and (6.130), similar to the
one in the solar neutrino oscillation (6.141), gives a prediction on the deficit of ve
at the detector. The predicted deficit rate for the MSW-LMA solution, (6.158)
and (6.159), is quite consistent with the observation, which started very recently
(KamLAND collaboration, 2003).

Problems
6.1 Show that the relation (6.17) holds, to verify that a free Weyl fermion and a
free Majorana fermion are equivalent, as far as they are massless.
146 NEUTRINO MASSES AND NEUTRINO OSCILLATIONS

6.2 Verify the relation (6.41) when (6.40) holds.

6.3 Show that the coefficient of the (irrelevant) operator (6.48) provides a Majorana
mass for VL when the Higgs doublet 4> develops its vacuum expectation value.

6.4 Verify the relation (6.66).

6.5 Verify that the adiabaticity condition (6.101) at the resonance point is equiva-
lent to the condition (6.103).

6.6 State the intuitive physical meaning of (6.108).

 Chapter 7
SUPERSYMMETRY

7.1 Supersymmetry and The Hierarchy Problem

Supersymmetry (SUSY for short) is a symmetry which connects particles and fields
with different spin-statistics, i.e. bosons and fermions. When a Lagrangian density
is invariant under a properly chosen transformation which connects particles whose
spins differ by 1/2, the theory is said to be supersymmetric. The supersymmetry
is independent of any internal symmetry such as gauge symmetry, and therefore
connects a pair of particles, "superpartners" of each other, with different spins
but the same quantum numbers such as electric charge, weak isospin, color etc.
As we will see later the generator of the supersymmetry transformation is known
to commute with the generator of space-time translation PM (4-momentum), and
therefore with P%. We thus learn that the superpartners have the same mass.
If we wish to apply the supersymmetry to the description of quarks and leptons,
the supersymmetry clearly should be broken spontaneously or explicitly, as we have
not seen any spin = 0 scalar particles with the same charge and the mass as those of
quarks and leptons. Though it is not hard to break supersymmetry spontaneously, it
easily happens that some of superpartners of light quarks or leptons are lighter than
these particles, thus contradicting with the reality. In the Lagrangian, to put by
hand the terms which break supersymmetry explicitly, such as the additional mass-
squared term for the scalar electron, is also possible and an easy way. However,
without any guiding principle, such arbitrarily introduced explicit SUSY breaking
terms easily lead to an unacceptable prediction, such as too large flavor changing
neutral currents. The detailed discussion on the mechanism of SUSY breaking is
beyond the scope of this textbook, though we briefly discuss the mechanism of
spontaneous SUSY breaking in a later section, and we just consider here how large
the SUSY breaking mass scale can be. The scale will be fixed so that the SUSY
works as a solution of the so-called hierarchy problem.
What is the hierarchy problem? Though it has been perfectly explaining all
phenomena (except the recently reported neutrino oscillation), the standard model
of elementary particles is widely believed not to be the final theory; in particular,

147
148 SUPERSYMMETRY

H
y ~ ^

H
* '
--^ H H (
r\ J H

(a) (b)

Fig. 7.1

it has too many free parameters and the gauge couplings are not really unified as
the gauge group is not a simple group. We thus expect that the standard model is
valid up to some energy (mass) scale A (a physical cutoff of the standard model),
and at A the theory is replaced by more fundamental new theory incorporating the
standard model, i.e. "beyond the standard model" or "new physics". The standard
model is thus regarded as the "low" energy effective theory with the cutoff A.
The cutoff A is MQUT if the new physics is Grand Unified Theories (GUT), with
e.g. MQUT ~ 1015GeV for SU(5) GUT, and Mpi ~ 1019GeV (Planck mass) if
the new physics is a unified theory with gravity. These mass scales, MGUT, MPI
are much higher than the weak scale Mw, the typical mass scale of the standard
model. The hierarchy problem is the problem of how to maintain the hierarchy
of these mass scales, i.e. Mw C MGUT,MPI. The problem is the most serious in
the sector of scalar particle such as the Higgs particle, as the Higgs mass-squared
get a huge quantum correction proportional to MGUT or Ml\. This large quantum
correction manifest itself as a "quadratic divergence" A2 if we calculate the quantum
correction to the Higgs mass-squared in the effective theory, i.e. in the standard
model: the problem of "quadratic divergence" (See Fig. 7.1(a)). Thus it is hard to
maintain the hierarchy naturally. Actually, it is possible to adjust the bare mass
of the Higgs, so that the renormalized Higgs mass remain to be the weak scale. It,
however, will require a fine tuning of the bare parameter at the precision roughly
of (MW/MGUT)2 ~ 10 - 2 6 , for instance, which we regard as unnatural.
Supersymmetry had been a subject of the theoretical research for rather long
time, but its application to the particle physics was strongly motivated by the desire
to solve the hierarchy problem. In fact, supersymmetric SU(5) GUT was proposed
by Sakai and Dimopoulos-Georgi (Sakai, 1981; Dimopoulos-Georgi, 1981) based on
the motivation. Let us note that in the supersymmetric theory coupling constants
of a particle and its superpartners are identical, and also that the Feynman rule
provides an additional negative sign for the diagram with a loop of fermion. Thus
the quadratic divergences from the two diagrams Fig. 7.1(a) and (b) are known
to cancel with each other as long as the supersymmetry is exact. In this way, the
 Two-component Notation 149

hierarchy problem of the quadratic divergence is solved. As the matter of fact,

the supersymmetry should be broken to the extent of "SUSY breaking mass scale"
MSUSY- We, however, are still free from the quadratic divergence, though we do
have an logarithmic divergence, ~ aMgUSY log (jf~)- We regard the logarithmic
correction as the order of aMgUSY • Then the SUSY breaking scale should not be too
high, since otherwise the hierarchy problem will arise again, i.e. aMgUSY < M^,.
This leads to a condition MSUSY < 1 (TeV), roughly speaking.
It should be also stressed that another very important motivation to study the
SUSY is that it almost automatically accommodates gravity. This is because the
commutation relation of two infinitesimal SUSY transformations is just space-time
translation. Thus, if SUSY is promoted to a local symmetry the resultant theory
should be invariant under a local transformations of space-time coordinates, i.e.
under the general coordinate transformations, thus leading to a supersymmetric
gravity theory, supergravity.
Discussions in this chapter is, in many respects, based upon the classic textbook
on the supersymmetry and supergravity by Wess and Bagger (Wess and Bagger,
1992).

7.2 Two-component Notation

Supersymmetry (SUSY) is in some sense a symmetry concerning (external) space-

time coordinates, instead of internal symmetries; the fact that the commutation
relation of SUSY transformations is space-time translation is the manifestation.
Actually, as we will see later, SUSY transformation can be understood as a trans-
lation in a Grassmannian space, "superspace". Just as the Lorentz transformation,
the SUSY transformation preserves (a little modified) chirality. Thus the most fun-
damental representation of SUSY has a definite chirality, and the usage of chiral
(Weyl) fermion, i.e. the usage of the two-component notation is useful. We thus first
summarize the notation and some useful relations in the two-component notation.
Weyl spinor, IPR,L = ^p^V") behave as irreducible representation of Lorentz
group, since [£^,75] = 0, with £M„ being the generator of the Lorentz group.
A Weyl spinor has a half degrees of freedom compared with a Dirac spinor and
is expected to be described by a two-component spinor, in a suitable base of 7-
matrices. To see this explicitly, we employ the Weyl base of 7-matrices, where 75
is a diagonal matrix, i.e. the same base as shown in (6.1)-(6.4);

T - U 0)•
ff" = (-/,0-i), (7.2)
d* = (-/,-o-i), (7.3)
150 SUPERSYMMETRY

75= i / y 7 V = ^ ^y ( 7.4)

where I is a 2 x 2 unit matrix and Oi (i = 1,2,3) are Pauli matrices. Then the Weyl
fermions are described by 2-component complex spinors as we expected:

V
</>* = ( ° ) , (7.5)

ri>L = I 1 J , ( a , d = l,2). (7.6)

The Lorentz generators have the form of block diagonal

r^S" cr"" = ^ C T ^ (78)

which clearly means that the chirality is preserved under the Lorentz transforma-
tion. It will be necessary to comment on the raising and lowering of the 2-component
spinor indices a or a and the meaning of the dot. We first note that the right-handed
Weyl spinor, t]a, is transformed by a Lorentz transformation as r)'a = Ma ^rip, where
the 2 x 2 matrix

M = expos'"'). (7-9)
with 6 transformation parameters eM„ (e„M = —e^). It should be noted that the M
is an element of SL(2,C). In fact, since the generators £M„ are traceless detM =
1 and also the 6 real degrees of freedom of eM„ just coincide with the degree of
SL(2,C), 2 x 2 2 — 2. We can easily show that in SL(2,C) an invariant inner-
product of two fundamental representations, i.e. doublets r\a and Xa, is given by
eal3T]aX0 (e a/3 : 2x2 Levi-Civita tensor, e12 = - e 2 1 = 1), by use of the property
detM = 1. Thus the invariant product can be written as

Vcc Xa, (7-10)

once the raising of the lower index is defined as

Xa = e Q / V (7.11)
In other words, the Levi-Civita tensor ea® behaves as the metric tensor in the 2-
component spinor space. Hereafter we will use the abbreviation, r\x = i]aX<x =
-VaXa = XaVa = Xn,WX = VaXa = Xn, for convenience.
Noting an identity W'"' = —ea^v*e, we find that the matrix of Lorentz transfor-
mation for a left-handed Weyl spinor should have the following indices,

-e*P (M*U %s (7.12)

 Two-component Notation 151

where the dot of the index denotes the complex conjugation of the quantity. To
be consistent, the left-handed 2-component spinor should have the following upper
dotted index

1pL=[ -a • (7.13)

We may also understand the origin of the dot as the result of charge-conjugation.
Namely, the charge-conjugation interchanges the chirality of Weyl spinors, e.g.
^-^•(IPR)0 = 0, while the charge conjugation always has the action of complex
conjugation. Thus, when the right-handed spinor has an undotted index the left-
handed spinor should have a dotted index. More explicitly, the charge-conjugation
of a Dirac-spinor ipr> = ,4>R + ipL is given as,

(i[>D)c = C^>D=i12xP*D=il

(7.14)

where rj^ = (r/a)* etc. This clearly shows that when £ = 77 the spinor

*M = ( 1% ) , (7-15)
V"
is self-conjugate under the charge-conjugation,

W>M) C =T/>M- (7.16)

This specific "real" spinor is called Majorana spinor, and the particle described by
the spinor is called Majorana particle.
Once the assignment of the spinor indices of Weyl spinors is fixed, the assignment
of the indices of various 2x2 matrices are determined as,
0
(*")««, ( r ' T V O a , ( ^ r ^etc. (7.17)
The Lagrangians of renormalizable theories have bi-linear terms of spinors. Weyl
spinors ipR and tpi behave as 2 and 2 representations of SL(2,C), or equivalently
( | , 0 ) and (0, | ) representations of the Lorentz group, where \ and 0 are the "spins"
of two 5J7(2)-like groups, contained in the Lorentz group. We then ask how bi-linear
forms of Weyl fermions behave under the Lorentz group. The bi-linear form of two
Weyl spinors with the same chirality decomposes into (1,0) + (1,0) or (0,1) + (1,0)
representations, as is known from the recombination of the spin: | x | = 0 + 1.
More concretely, the singlet (0,0) and the triplet (1,0) representations formed by
the product of two, e.g., right-handed spinors, 771 and 772, are written as

ea0mam0 = viart = -mm, K T ' 3 1 ^ ^ . (7.18)

152 SUPERSYMMETRY

The reader can easily check that {aliv)a^ is symmetric under the exchange of the
spinor indices a and /?, the property necessary for the triplet. This statement,
however, is a little puzzling, as the number of possible combinations of fi and v in
the a11" is 6, instead of 3. Actually, we find that not all of <r'"/ are independent; for
instance a01 and a23 are essentially identical matrices. (In fact, if our space-time
were Euclidean, this tensor satisfies the self-dual condition a^v = |C'" / K A (T K A , while
a^v satisfies the anti-self-dual condition.) Similarly, the product of two left-handed
spinors decomposes into (0,1) + (0,0) by the contractions with (^1/)&a and e d/ j. In
the four component-notation, where the right-handed spinors are written as ipi =
(f?ia,0)' and ip2 = (^2ai0)', the singlet and the triplet discussed above correspond
to the following bi-linear forms, ip\Cip2 a n d iplCE^^- Then what corresponds to
the Lorentz vector V ' I T ^ ^ ? (Note that V'IC'T/IV^ identically vanishes.) This time
the bi-linear form is made by the mixture of the dotted and undotted indices, i.e.
~Via {v*i)0"*rl2cc Namely, ( | , 0 ) x (0, | ) = ( | , | ) behaves as a Lorentz vector with
4 components.
We display in the Problem 7.1 some of formulae in the 2-component notation,
which are frequently used and useful in the discussion below.

7.3 SUSY Algebra and SUSY Group

As already stated, supersymmetry (SUSY) is the symmetry under the exchange of

bosons and fermions. Remarkably, it is not only a possible symmetry, but also is a
unique symmetry, consistent with relativistic quantum field theory, besides internal
symmetries such as gauge symmetry.
It had been known that generally the Lie algebra of possible symmetries of the S-
matrix is restricted to the one composed of FM, MM„, and Lorentz scalar generators
which are the generators of

space-time translation,
Lorentz transformation,
gauge transformation,

respectively. Supersymmetry, therefore, seems to be not allowed as a symmetry.

This no-go theorem (Coleman and Mandula, 1967), however, has been proved to be
evaded, once the Lie algebra is extended so that it may contain anti-commutation
relations as well as commutation relations, i.e. to an graded Lie algebra (Haag,
Lopuszanski and Sohnius, 1975): the algebra now takes a form

Q,Q' = X, [X,X'] = X", [Q,X] = Q",..„ (7.19)

where Q etc. and X etc. are, say, fermionic and bosonic generators. Since SUSY
transformation connects fields whose spins differ by 1/2, the generators of SUSY
 SUSY Algebra and SUSY Group 153

naturally carry spin 1/2, and therefore satisfy the anti-commutation relation due
to the Fermi-statistics, just as Q etc. do in (7.19).
Through such rigorous argument, it turns out that the SUSY algebra is given
as

{Qa, Qa) = - 2 K ) a d P^ (7-20)

{Qa, Qp} = {Qa, Qfi} = 0, (7.21)

[Plt,Qa] = [Plt,Q&]=0, (7-22)

[P»,P»] = 0, (7-23)
where Qa and Qa are complex conjugate, or hermitian conjugate in the sense of
operators, of each other, and are Grassmannian operators, as is seen typically in
(7.21). One may wonder why the anti-commutator in (7.21) should vanish, while
that in (7.20) is non-vanishing. Just as the product of two doublets \ x | decomposes
into a triplet and a singlet 1 + 0 in the calculus of spin, if {Qa, Qp} were non-
vanishing, it generally should have contained two parts C^a^ and C[a,0], i-e. the
parts symmetric and anti-symmetric under the exchange of a and /?, belonging
to (1,0) and (0,0) representations of the Lorentz group, respectively. The anti-
symmetric part, however, trivially vanishes, since {Qa, Qp} is symmetric under
the exchange. Let us note that this situation changes in the case of extended SUSY
with N (N > 1) kinds of SUSY transformations generated by Qla (and Q&ti),
where a "central charge" term of the form eap c ^ l , symmetric under the exchange
(a,i) «-» (/?, j), is allowed to exist. In the case of simple SUSY N = 1, howver,
the only possibility is the symmetric piece C{a<^, proportional to (<r'"/)a^ MM„.
A rigorous argument leads to the conclusion that this term is not allowed to exist,
being consistent with the SUSY algebra [PM,<5a] = 0.
We thus know that the commutator of SUSY transformations provides a space-
time translation, i.e.

[(eaQa), {e&Q")\ = [(eQ), (eQ)} = -2(£>")<>«€«) PM, (7.24)

where ea, ea are Grassmannian transformation parameters for SUSY and eM =
ea crM
( )ade a denotes the amount of the infinitesimal space-time translation. There-
fore, it is a rather trivial fact that once SUSY is promoted to a local symmetry the
resultant theory inevitably has a local translational invariance, i.e. the invariance
under the general coordinate transformations, thus leading to a SUSY theory in-
cluding gravity, i.e. supergravity. In this book, we will restrict our argument to the
global (rigid) supersymmetry where the transformation parameters are space-time
independent.
Having the SUSY algebra, we may define a SUSY group, which is a set of trans-
formations generated by the SUSY algebra, e*(£''p'J+<;<5+e<^, where the generators
154 SUPERSYMMETRY

Pft,Q and Q are the elements of SUSY algebra and ^,€,1 are the corresponding
transformation parameters. We need representations of SUSY algebra, on which
the transformations of SUSY group should act, so that we can assign our quarks
and leptons and gauge bosons to some of the representations. In the SU(2) gauge
theory, for instance, the representations are concretely expressed as multiplets, such
as doublet, triplet, etc., on which the group elements act in the form of matrices.
Similarly, the representations of SUSY group can be expressed as "SUSY multi-
plets". As the SUSY transformation connects fields whose spins differ by 1/2, it
is expected that each multiplet is made of "super-partners" of each other, such as
a (spin 0) scalar and a (spin 1/2) fermion. SUSY generators Q and Q commute
with p2. Thus all members of a SUSY multiplet on their mass-shell should have a
unique mass, and the SUSY multiplet can be classified by the mass M {p2 = M2).
Let us first consider massless (M = 0) SUSY multiplets. The massless multiplets
we discuss in this book are

"chiral multiplet" : (A, i(>L>R) (s = 0, s = 1/2) (7.25)

"vector multiplet" : (V^, A) (s = 1, s = 1/2), (7.26)

where s denotes the spin. A is a complex scalar field and A should be a Majorana
fermion, corresponding to the real vector field VM. The chirality of the Weyl fermion
of the chiral multiplet can be either L or R. We will be able to assign our quarks
and leptons and Higgs to the component fields of the chiral multiplets, together with
their super-partners. Gauge bosons, such as photon, are assigned to V^ together
with their superpartners, such as "photino". The important feature of the SUSY
multiplets is that the (real) degrees of freedom of boson and fermion fields just
coincide. This, however, seems to be not the case in the above multiplets; for
instance, in the chiral multiplet (A, IPL,R), the complex scalar clearly has 2 degrees,
while the Weyl fermion ipL seems to have 4 real degrees of freedom, at the first sight.
This superficial puzzle may be resolved if we realize that the on-shell fermionic state
has just the half degrees of freedom, due to the equation of motion. For instance a
massless Weyl fermion ipR = (r)a, 0)* with 4-momentum pM = (p, 0,0, p) satisfies a
Dirac (Weyl) equation in the momentum space

Vr»»-0»(-»l)(Z) = (l). (7,7,

which shows that the r]\ disappears and only 772 remains. This argument, in turn,
suggests that the SUSY algebra is closed only for on-shell states, or only by use of
the equations of motion. In order to make the SUSY algebra close even for off-shell
states, we therefore need to introduce additional bosonic states. Such additional
states, i.e. 1 complex scalar field in the case of chiral multiplet and 1 real scalar
field in the case of vector multiplet, should not have physical degrees of freedom,
since they disappear once the equations of motion are applied. Thus these fields
are called "auxiliary fields".
 Superfield Formulation 155

Concerning the SUSY multiplets of (on-shell) massive states with M ^ 0, the

situation is rather different from the massless case. A general argument (see for
instance (Wess and Bagger, 1992)) shows that the multiplets are generally "larger"
than those in the massless case. Let us take the example of a massive SUSY mul-
tiplet with the highest spin 1. This multiplet has one s = 1, two s = 1/2 and one
s = 0 component fields, though the SUSY transformation changes the spin of fields
only by 1/2. This may not be so surprising if we recall the Higgs mechanism in
ordinary gauge theories. For a gauge bosons to be massive, it needs to absorb a
scalar field. Thus for a vector SUSY multiplet to be massive, it needs to absorb
a chiral multiplet, which has inevitably a scalar component field. One remarkable
feature of the SUSY theory is that, at the same time as the gauge boson becomes
massive absorbing the scalar, gauge fermion (a superpartner of gauge boson, "gaug-
ino") absorbs the chiral fermion of chiral SUSY multiplet to form a massive Dirac
fermion (super-Higgs mechanism). For instance, suppose that the vector multiplet
(Vfj,, A) is made of the photon and "photino", and the scalar component A of the
chiral multiplet (A, IPL,R) is a charged scalar. Assuming that the real part of the
A develops the vacuum expectation value (VEV) by a suitable scalar potential, the
photon will absorb the imaginary part by the Higgs mechanism. Thus, we have a
massive photon with 3 degrees of freedom and a massive scalar of 1 degree, i.e. the
real part. As for the fermionic part, due to the super-Higgs mechanism, the photino
A and ip form a massive Dirac particle, which should be the super-partner of the
massive photon.

7.4 Superfield Formulation

We have formulated SUSY algebra and its representation, SUSY multiplets. The
next task is to find the rule of SUSY transformation among the component fields
of a SUSY multiplet, which is non-trivial. A powerful technique for such purpose
has been invented, which makes the formulation of SUSY transformation and the
construction of supersymmetric theories, i.e. the theories invariant under the SUSY
transformation, almost automatic.
The formulation is based on the notion of "superfields", which are the fields
defined on "superspace". Before going into the detail, let us recall that the gen-
erator of space-time translation, p M , acts on an ordinary field </>(xM) as a differ-
ential operator pM</>(a;M) = i-g^^x*1). Since (SUSY)2 ~ space-time translation,
roughly speaking, and a Lorentz vector can be expressed as a bi-linear form of a
spinor, eM = ea(a^)aaea as suggested by (7.24), it will be natural to expect that
the generators of SUSY transformations Q and Q act as a sort of translations
in a "fermionic space", with anti-commuting Grassmannian coordinates 9a, 6a
({6a,6p} = {0a,6a} = {8a,6a} = 0). The space with orninary space-time co-
ordinate and the Grassmannian coordinates, (x^, 0a, 6a), is called superspace, and
a field defined on the superspace is called superfield. Actually it turns out that the
156 SUPERSYMMETRY

SUSY generator is not a simple translation of 6 and 9 coordinates, but contains

a translation of xM, as well. Then a natural question is what is the differential
operator of SUSY generator, corresponding to p^ = igf^-?
Sometimes it is useful to regard a space (or space-time) as a "group manifold",
which is a space, whose point can be identified with an element of the group. For
instance, a point on the Earth, i.e. a 2-dimesional sphere 5 2 , may be obtained as
the result of the action of 3-dimensional rotation group 50(3) of the point on the
north-pole. Hence the point can be identified with the element of 50(3) group.
This identification is not unique, as the rotation about the z-axis, 50(2), does not
move the north pole and can be freely added; 5 2 may be regarded as a group man-
ifold of 5 0 ( 3 ) / 5 0 ( 2 ) , whose degree is 3 - 1 = 2, the same as that of 5 2 . More
explicitly, a point P(9, <fi) with polar coordinate (9, <j>) and Cartesian Coordinates
(sin#cos0, sin#sin0, cos#) on a unit 5 2 embedded in a 3-dimensional space is iden-
tified with an element of 50(3), G(6,4>,a) = R3(-(p)R2(-9)R3(a) with (see also
(4.179))

( Ci 0

0 1 0 ,
-si \

RaWi) =
( Ci

-St a 0
Si

,
0 \

(7.28)
Si 0 ct J \ 0 0 1 /
where Cj = cosOi and Sj = sin#;. In fact,
sinflcos./. \ / 0 \ / 0
sin0sin</> \ = R3(-<t>)R2(-9)R3(a) I 0 = R3(-(j>)R2(-9) I 0 | , (7.29)
cos0 / V1/ V1
where we find that the 50(2) rotation with angle a is irrelevant. Let Q(6', <j)') be a
point on the 5 2 , which is obtained by the action of a 50(3) rotation G(K, A, /?) on
the point P(9,cf>). Then how can we express the transformation of space coordinates
{6, <f>) -> (0',<j>') by use of K, A, /?? We note

sin#'cos</>' \ / 0 \ / sin^cos(/>
sinfl'sin^' = G(9',4>',j) 0 =G{K,\,0) sin^sin./.
cosfl' / \ 1 / V cos6>

= G(K,\,P){G(9,<p,a) | 0 \} = {G(K,\,/3)G(6,<l>,a)}l 0 |. (7.30)

This means the transformation of the space coordinates is completely determined

by the multiplication rule of group elements: G(9',$'',7)
= G{K,\,j3)G{6,<j),a). Such determined transformation automatically satisfies all
properties of 50(3) group, by construction.
Similarly, we can establish the transformation of superspace coordinates (x, 9,9).
Let G(x",9,9) = ei^p»+0Q+0Q) be an element of SUSY group and identify it with
 Superfield Formulation 157

a coordinate (a;",0,0). By use of the SUSY algebra (7.20)-(7.23) and the well-
known relation eAeB = eA+B+^[A,B\+--- w e g e t ( s i n c e higher commutators do not
contribute)
G(0,e,e) • G(x",9,6) = G(x^ + iOa^e - ie(7"9,9 + e,e + e). (7.31)

^From the identification we readily get a transformation property of the superspace

coordinates due to the action of G(0, e, e),
{x^,9,9) -> {xfM + i9afle-iea'i9,9 + e,6 + e), (7.32)

where 0<rMe = 9a(a^)a^e6', etc. Thus we can conclude that the action of the SUSY
transformation G(0, e, e) on a superfield is equivalent to the application of e^eC3+£<3'
with the following differential operators, corresponding to the (i times) SUSY gen-
erators,

iQa = 0^-i((^)aiS&dlt, (7.33)

iQa = —% + *'0°> M Wd M . (7.34)

89
The nice thing to use the group-manifold approach is that such constructed differ-
ential operators automatically satisfies the SUSY algebra, just because it is based
on the group element itself. For instance if G\Gi = G3 holds among the elements
of SUSY group, multiplying G(x M , 9,9) from the right of both sides and identifying
G(x>*,9,9) as a superfield <j> and G\,G2,G$ with the corresponding group elements
with differential operators, we immediately get G\G2<$> — Gz4>- In fact, we can
directly confirm that the differential operators (7.33) and (7.34) satisfy

{Qa,Q«} = -2i(a")a& aMI {Qa,Q0} = {Qd,Q^} = 0, (7.35)

to be consistent with (7.20) and (7.21).
We will now see how the component fields of SUSY multiplets are contained
in the superfields. Superfields containing the irreducible representation of SUSY
algebra, i.e. (massless) chiral multiplet and the (massless) vector' multiplet, are
called chiral superfield and vector superfield. Before going to the discussion of
each superfield, we first consider the most general complex superfield, <j)(x,9,9)
(the Lorentz and spinor indices of the coordinates have been suppressed). The
important feature of the superfied is that, when it is Taylor-expanded with respect
to Grassmannian coordinates 9, 9, it contains only a finite number of terms, whose
coefficients are functions solely of x, i.e. ordinary fields. This termination of the
expansion is due to the anti-commuting property of the Grassmannian coordinates;
e.g. 9a9p91 = 0 as the spinor indices a,/?,7 take only 1 or 2. Thus the general
superfield can be expanded as

(j)(x, 9,9) = C(x) + 9x{x) + 9Y(x) + 99M(x) + 99N(x)

158 SUPERSYMMETRY

+ Oa^ev^x) + eeex(x) + eeeip(x) + 6666D(X), (7.36)

a
where 66 = 9 8a etc.. The bosonic component fields, C, M, N, VM, £>, have in total
(o) ~*~ (2) + (4) = ^ complex degrees of freedom, while the fermionic component
fields, x, X'J ^> ip, have (*) + (3) = 8 degrees of freedom. Thus we have checked the
coincidence of bosonic and fermionic degrees of freedom, as expected from SUSY.
Altogether we have 8 + 8 = 16 degrees of freedom. This may be easily under-
stood, since each term of the Taylor expansion can be written, in general, in a form
Cijki(x) {6\)l{62)j(6i)k(6i)1- We have 2 4 = 16 choices of the coefficient functions
Cijki(x), as i,j,k,l take either 0 or 1.
The deviation of the superfield i5<j>(x) = <j)'{x) — (j>(x), under an infinitesimal
transformation, <f>'{x) = exp(i(eQ + tQ)) (p(x), with infinitesimal Grassmannian
parameters e and e can be easily calculated by use of the differential operators
(7.33) and (7.34) as

8<j> = i(eQ + eQ)cj> = { e Q ( ^ - i ^ J a d ^ f l , ) - *"(-— + tfaK)ad^)Ma;).

(7.37)
Writing 5(f>{x) — 5C{x) + 66x(x) + • • •, the infinitesimal deviation of some of com-
ponent fields can be read off as

SC = ex + ex, (7.38)
SXa = 2e a M + K ) a d e d ( y M + i a M C ) , (7.39)
Stf* = 2^N + ea{a^)a&{Vtl-idtlC), (7.40)
% a
SD = -{^)adl{e d,x - {d^rrn- (7.41)
We learn from the above exercise the following things, which play quite impor-
tant roles when we attempt to construct a supersymmetric theory, i.e. a theory
invariant under the SUSY transformation.

(a) The dimensionality of the component fields

As is suggested by (7.31), the bi-linear form of the Grassmannian coordinates
8a and 6& behaves as the ordinary space-time coordinate x^. Hence, the 6a and
da should have mass dimension d = —1/2. The generators of SUSY transformation
Qa and Q& therefore have mass dimension d = 1/2. This means by the SUSY
transformation a component field with mass dimension d is transformed into a field
with mass dimension d + \ or the space-time derivative of a field of lower mass
dimension d — | . We also learn that the component field accompanied by a higher
power of 6a and 6a has higher mass dimension.

(b) The property of D-term

The above argument suggests that the "highest component" D(x), or "D-term",
proportional to the highest power 6666, has the highest mass dimension. For in-
 Superfield Formulation 159

stance, if we assign the Lorentz vector field V^ the mass dimension d = 1, the D field
has d — 2. As the quadratic term of D(x) already has d = 4, we cannot include any
space-time derivative in the free Lagrangian of the D field, thus making the D an
auxiliary field, i.e. a field without dynamical degree of freedom and is describable
in terms of other physical fields, once the equation of motion for the field is used.
As the D field has the highest mass dimension, the infinitesimal deviation of
the D(x) under the SUSY transformation, is inevitably the space-time deriva-
tive of the field of lower mass dimension. In fact 6D in (7.41) can be rewritten
as a total derivative, noting that e and e are space-time independent constant:
SD = dll{%ea(atl)aa\a - %i>a(&^)aa£d'}- This means that if the Lagrangian of a
theory can be regarded as the D-trem of some superfield, the action is supersymmet-
ric, since a total derivative term of the Lagrangian is irrelevant. (If space-time has
some compact dimension, as usually assumed in higher dimensional theories, the
derivative term may not be ignored for non-trivial boundary conditions of fields.) It
is suggestive to note that a similar thing happens when a theory is invariant under
a space-time translation, i.e. also in this case the Lagrangian itself is not invariant
under the translation, but the deviation can be written as a total derivative, thus
making the action invariant.

(c) T h e p r o p e r t y of closure
To find out the multiplication rule of SUSY multiplets, namely a rule to re-
construct a new SUSY multiplet out of the product of two SUSY multiplets, is
a non-trivial task. In the language of the superfield, however, the multiplication
rule is quite simple, i.e. we can just multiply two superfields. If fa and fa are two
superfields, their product fa(x,8,8) = fa(x,8,8)-fa(x,8,8) can be Taylor expanded
again, and therefore behaves as a new superfield ("closure" property of superfields
under the multiplication). Each component field of fa is written as a bi-linear term
of the component fields of fa and fa, which provides us the multiplication rule.
This is essentially due to the fact that the SUSY generator is a linear differential
operator and the chain rule i(eQ + eQ)fa = {i(eQ + eQ)fa} -fa + fa • {i(eQ + eQ)fa}
holds.
By use of these properties, we can easily construct (some part of) the action of
a supersymmetric theory,

S = Jdixf(fa,fa,.--)\D, (7.42)

where f(fa,fa, • • -)\D denotes the D-term (the highest component) of the polyno-
mial function / of superfields, fa, fa, • • •, which is again a superfield. The infinites-
imal deviation of the D-term is a total derivative and the action is SUSY invariant.
Since SUSY transformation is a sort of translation in the superspace, it is naturally
expected that the integral of the Lagrangian over the entire superspace yields a
SUSY invariant action. In fact, we find that the action can be equivalently written
160 S UPERS YMMETR Y

as

S= fdixd28d28 /(</>!,&,•••). (7.43)

Let us note that the integration over Grassmannian variables is equivalent to the
differentiation in terms of these variables. Thus d26 d?8 is equivalent to extracting
the D-term.

7.5 Chiral superfleld

The general superfield turns out to be reducible representation of SUSY, as it con-

tains fields with spin s = 0,1/2,1 together, while SUSY transformation changes the
spin only by 1/2. It is thus important to identify the irreducible representations of
SUSY. There are two kinds of irreducible representation, i.e. chiral superfield and
vector superfield, which contain, as their component fields the chiral multiplet and
the vector multiplet, respectively.
In this section, we will discuss the chiral superfield, which contain component
fields with s = 0 and 1/2. Naive guess is that we can get it by adopting a "chiral"
superspace, whose Grassmannian coordinate is restricted to a Weyl fermion with
definite chirality, 8 or 8, but not the both, e.g. space with coordinates (x^,^ Q ).
Then the power series expansion with respect to 8a will terminate at the quadratic
term 88, and the appearance of the s = 1 vector field V^ will be evaded. Thus
we are tempted to impose a constraint in order to define, e.g., a "right-handed"
chiral superfield <fr, -^ <f> = 0, so that <j> does not depend on 8. This constraint
equation, however, does not work, since the condition is not compatible with SUSY,
i.e. {Qa, gfs-} 7^ 0. We then have to find out a suitable differential operator, which
(anti-)commutes with the SUSY generators. The desirable operators turn out to be

D
° = ^ + <KUfld5(.. (7-44)
a
D& = -^y-*0 (*")««» 0M- (7.45)

It is straightforward to check that these operators anti-commute with SUSY gener-

ator,

{Qa,D0} = {Q*,Da} = {Qa,D&} = {Q«,D0} = 0. (7.46)

Let us note that these "SUSY-covariant" derivatives Da and D& are quite similar
to iQa and iQa in (7.33) and (7.34), except the sign in front of the space-time
derivative. This is not an accident. Let us recall that the differential operator Q
and Q were read off from the relation, G(0,e,e) • G{x'i,d^) = G(x^ + ida^e -
iea^VjO + e,8 + e), where we identified the "left multiplication" of G(0,e,e) as
an action of the element of SUSY group. What happens if the multiplication of
 Chiral superfield 161

G(0, e, e) is from the right? The only change will be that the sign of the space-time
translation, proportional to e or e, is just opposite to the case of left-multiplication,
i.e. -iOa^e + iea^O. Thus we find that the operators D and D can be regarded
as the differential operators corresponding to the right-multiplication. It is now
trivial that they commute with the SUSY generators, just because the right- and
left-multiplications are mutually independent.
Now, for instance, a right-handed chiral superfield <> / can be defined as

A i <f> = 0. (7.47)

(The left-handed superfield is defined similarly by replacing D& by Da.) As ex-

pected, Do,0a = 0 means that the superfield <f> may depend on the right-handed
coordinate 0a. The space-time coordinate xM, however, is not a suitable coordinate
to describe cj>, as D ^ = —i 0a ((T^)aa does not vanish. We note that the bi-linear
form of 0 and 0 behaving as a Lorentz vector 6a^0 satisfies DaiOa^O) = ^ a (cr'') a ( i .
We thus find the suitable Lorentz vector to describe (j> is yM = x^ + i (0<rM0):
D& 2/M = 0. In this way, the chiral superfield should be a function of coordinates
(yM, 9a), and can be expanded in a power series of 6 as

0(y, 0) = A(y) + y/2 di/>(y) + eOF{y)

= eW^** (A(x) + V2 6iP{x) + 66F{x))
= A{x) + i (0ff"9) d„A{x) - J eeW DA(x) + V2 0r/f(x)

- -^=00 (d^^O + eOFix), (7.48)

where D = 9M9M and we have used the property that the power series expansion
of e'(e<r''9)di' ends up at {{0ali0)dll}2, because of the anti-commuting property of 8
and 0. In the derivation, some relations listed in Problem 7.1 may be useful. The
component field F has appeared in addition to the expected scalar and fermion
fields A and tp. The field F has mass dimension d = 2, and is anticipated to be an
auxiliary field, just as the D field in a general superfield. Let us note that in the
third line of the right hand side of (7.48), derivative terms have appeared though
there is no derivative of the chiral superfield at the beginning. This is due to the
"non-locality" of the coordinate yM, which has a deviation i(0a^0) from xM, and will
play an important role in the construction of the kinetic term for the component
fields.
In terms of the coordinates ?/M, 8a, the infinitesimal SUSY transformation reads
as

i{(eQ + eQ)} <£(</,0) = ( ^ ^ + 2i6a{a^)a^dll) <f>{y,9), (7.49)

where <9M = d/dy^. It is now easy to derive the transformation property of each
162 SUPERSYMMETRY

component field:

SA = V2eip, (7.50)
Stp = y/Hio^edpA + V2eF, (7.51)
SF = -V^iea^d^ip. (7.52)

We realize that again the deviation of the "highest" component field F can be
written as a total derivative (as e is ^-independent), which becomes important in
the construction of supersymmetric theory of chiral superfields.
In order to see the multiplication rule for chiral superfields, it is suggestive
to recall that multiplication of holomorphic functions f(z) and g(z) gives another
holomorphic function h(z) — f(z)g(z), while the multiplication of holomorphic and
anti-holomorphic functions is no longer holomorphic nor anti-holomorphic, but is a
general complex function. A complete analogy holds in the multiplication rule for
chiral and anti-chiral superfields, i.e.,

chiral x chiral = chiral,

anti-chiral x anti-chiral = anti-chiral, (7.53)
chiral x anti-chiral = general.

Thanks to the multiplication rule, especially the closure property among chiral
or anti-chiral superfields, it is quite easy to form a SUSY invariant monomials of
such superfields,

J diy{My,9)My,9)---<l>n(y,9)}\F~ J d4yd2e{4>1h---M, (7.54)

where F denotes the extraction of the F-term of the chiral superfield obtained by the
multiplication of the chiral superfields <j>\{y,6) etc., which is a total derivative with
respect to yM. This procedure will be used in the formation of the "superpotential"
which is discussed below. In (7.54), we may finally replace yM by the ordinary
space-time coordinate a;M.

7.6 Wess-Zumino Model

So called 4>4 theory is often discussed as a prototype theory of quantum field theory
for the purpose of getting some essential ingredients of the quantum field theory. In
this section, we discuss the supersymmetric extension of the </>4 theory, a supersym-
metric model of chiral multiplet (Ai, ipi) (i = 1,2, • • •), called Wess-Zumino model.
As the new feature of the Wess-Zumino model, we have not only self-interaction of
scalar fields but also Yukawa-type couplings between scalars and fermions. As the
consequence of the SUSY, the coupling constants of such interactions are identical,
and there should be mass degeneracy between the scalars and the fermions, unless
 Wess-Zumino Model 163

there is SUSY breaking.

• The kinetic term

To get kinetic terms for scalars and fermions, it is clear that we need quadratic
terms of chiral superfields. We learn from ordinary field theories that kinetic terms
are those which are hermitian by themselves. We thus expect the multiplication of a
chiral superfield and its hermitian conjugation, i.e. a anti-chiral superfield, provides
the desirable kinetic term. This turns out to be the case, since the difference between
yM and y^ yields the non-locality, and therefore the kinetic term. (As already has
been pointed out and we will see below more concretely, a simple square of a chiral
superfield just provides a quadratic term in "superpotential", but not a kinetic
term.) In fact, the multiplication yields for a given chiral superfield (j>

<Ky,0)1tKv,e) = ( e ^ " ^ 0 ( M ) ) t ( e ^ " ^ cf>(x,6))

= A(x)*A(x) + V26ip{x)A(x)* + VW^{x)A{x) + •••
+ eeM[-^-A*uA - ^{UA*)A + i(oMA*)(aMA)
l
+ -{d^)a^-l-^{d^)+F'F). (7.55)
According to the multiplication rule, this product behaves as a general superfield.
Thus taking the D-term we readily obtain the desirable SUSY invariant kinetic term
for the set of chiral superfields:

= (d^ADid^Ai) - i^^d^i + F*Fi + (total derivative). (7.56)

As we expected from dimensional counting, the field F has no derivative in this

"kinetic" term and therefore is an auxiliary field. Namely, F does not propagate by
itself, unless it couples with propagating fields through interactions. The remaining
physical fields, A and tj) have just ordinary forms of kinetic terms.

• The superpotential
The aforementioned kinetic term is a free Lagrangian and does not contain any
interactions. Let us now consider the self interaction of chiral superfields. Just
as the self-interaction of scalar fields Ai are described by scalar potential V(A),
supersymmetric "self-" interaction is generally describable in terms of a polynomial
of chiral superfield W((j>), called superpotential:

W{(f>) = Xi4>i + -mi^facpj + -gi,j,k<Pi<t>j<j>k- (7-57)

^From the closure property under the multiplication, the superpotential itself should
be a chiral superfield. Thus / d4x W((J))\F (after the replacement of j / M -> :rM) should
be SUSY invariant. Let the mass dimension of W{4>) be dw • Then its F-term should
164 SUPERSYMMETRY

have a mass dimension d + 1, as 06 carries d = 1. This is why the superpotential

ends up at cubic terms of chiral superfields. Namely if we introduce terms higher
than cubic, such as quartic terms, the mass dimension of the F-term should exceed
4, thus leading to a non-renormalizable theory. We will see below, more concretely,
that such cubic superpotential eventually yields quartic self-interaction of scalar
fields and Yukawa-type couplings.
Let us now calculate the F-term of the superpotential W. There is a sys-
tematic way to derive the F-term. We first calculate the F-term of a monomial
(0102 • • • 0 7 I ) | F - The F component is the sum of the terms, which are quadratic
in the Grassmannian coordinates 9. To get the quadratic terms there are only two
possibilities. One possibility is to pick up Fi component from some chiral superfield
(pi and take scalar components from all remaining chiral superfields:

9 ( A , A n )
A1 • • • An^Fn + A1---An^Fn^An + ••• = ' ^ Ft. (7.58)

The right hand side means that such term can be compactly expressed in terms of
the first derivative of the monomial, obtained by replacing all chiral superfields by
their scalar components. The next possibility is to take fermions fron two different
places of the monomial and take scalar fields from all remaining chiral superfields:

-A1---W-i1>na-A1---rft_2An-1il>na--- =-\d{AQ^A'.An)^^i- ( 7 - 59 )

As this compact way of writing is easily known to be valid for arbitrary monomials
and because differential operators are linear operators, this procedure is applicable
for an arbitrary superpotential. Hence, the F-term of the superpotential can be
written as

wm =
* -BAT Fi' 2dJM; * *'• (7 60)
-
The whole Lagrangian now reads

C = Ckin-{W{<t>)\F + h.C.}
= (dftADiWAi) - i ^ d ^ i + F*F
fdW(A) „ ld2W(A) , , , . ._„,

The auxiliary field does not possess an independent dynamical freedom by itself,
and is expected to be eliminated from the Lagrangian, if we wish. In fact, by use
of Euler-Lagrange equation for Fi, Fi can be expressed in terms of scalar fields,
as there is no derivative term of the auxiliary field in the Lagrangian. For such
purpose we first write down the terms containing Fi fields in £,
 Wess-Zumino Model 165

By use of the Euler-Lagrange equation, the quantities inside the bracket just Dis-
appear:

Hence, the (ordinary) scalar potential V(A) is given as

VIA) - I ^ f . (7.64)

In this way the Lagrangian of the Wess-Zumino model can be rewritten in terms
of only physical fields as

c = (aMA:)(aMi)-i^ff"aMVi
\1fd2W{A)
d2W{A) , , u .
2^8 Aid Aj
,dW{A).2
(7.65)
dAi
The first line at the right hand side is ordinary kinetic term and the second and the
third lines denote Yukawa coupling and scalar potential, respectively.

• Solving the hierarchy problem

To see how SUSY works as a mechanism to solve the hierarchy problem or
the problem of quadratic divergence, we will consider the Wess-Zumino model of a
single chiral multiplet, with superpotential W{(j>) = \m <j>2-\ g<f>3- The Lagrangian
is explicitly written as

£ = (dltA*)(&'A)-ili&tdllil>
+ [{^rn - gA) i/>2 + h.c]
- \mA-gA2\2. (7.66)

Thus we know that the complex scalar A has a mass m and the fermion ip has
the same Majorana mass m. Thus the degeneracy of the masses of boson and
fermion is realized, as an important consequence of SUSY. We further note that the
Yukawa coupling Aip2 and the scalar self-interaction (A* A)2 have identical coupling
constant g. This is another important consequence of SUSY. One remark here is
that the scalar self-coupling is proportional to g2, not g as in the Yukawa coupling,
and SUSY seems to be not manifest. One may convince oneself that SUSY is really
maintained, by noting that the self-interaction is originally due to the coupling
between scalar A and auxiliary field F, 2gFA2 + h.c, whose coupling constant is
proportional to g, thus preserving SUSY (see Fig. 7.2).
As was mentioned in the beginning of this Chapter, SUSY has a remarkable
feature to stabilize the scalar mass under the quantum correction, i.e. the problem of
166 SUPERSYMMETRY

A ' \

A -+~-4 2ig 2igV +~ - A => A --*--•*--- A

A -+-4-2ig -2igf-+~- A

Fig. 7.2

quadratic divergence is absent in SUSY theories. Let us now explicitly show that the
cancellation of the quadratic divergence does happen in the Wess-Zumino model.
What we calculate is the Feynman diagrams (a) and (b) shown in Fig. 7.2. In order
to calculate the diagram with V'-exchange, we need a propagator for the fermion ip.
Since we are familiar with the propagator for 4-componet spinor, and as ip has a
Majorana mass term ^mip2 + h.c., let us define a Majorana fermion tpM — ip + {ip)G,
with tp being understood as a 4-vector (if}a,oy, whose free Lagrangian is given by
\ V ' M ( * ^ 7 ' ' — m ) V'M- In terms oiipM, the relevant Yukawa coupling can be written
as —gAip2 + h.c. = gA^M^^-^M + h.c. According to the discussion in Chapter 6,
the propagator of Majorana particle ipM is just the same as the case of ordinary
Dirac fermion, except that not only IPM^PM DU^ a l s o V'MV'M tyPe propagator exist,
due to the fact (II>M)C = V'M (see (6.40) and (6.41)). Paying attention to this
specific feature of the Majorana particle, we may calculate Fig. 7.2(b) according to
the ordinary Feynman rule. We can directly check that there is a cancellation of the
quadratic divergence between the diagrams (a) and (b). Actually, we find that for
m = 0 the sum of these two diagrams identically vanish. We may also check that
when SUSY is broken (spontaneously or explicitly) and only A gets a mass MSUSY ,
denoting the extent of the SUSY breaking, there remains a logarithmic divergence
in the sum of these diagrams, i.e. (for m = 0) oc MgUSY log A. This is why MSUSY
should not exceed 1 (TeV) or so, since otherwise the hierarchy problem will arise
again. Let us note that another type of diagram coming from the contraction of
two identical vertices, e.g. of the type of AIJJM^[L'IPM, does not yield the quadratic
divergence, even for m ^ 0, and therefore has been ignored in the discussion above.
 Vector superfield 167

7.7 Vector superfield

Though the general superfield, discussed in section 7.4, is a complex field in general,
we may get an irreducible representation V of SUSY by imposing a hermiticity
condition

V f = V, (7.67)

since the infinitesimal transformation of SUSY is described by an operator, i(9Q +

8Q), which may be regarded to be hermitian as Q and Q are differential operators.
Then the component fields appearing as a coefficient of 9a^8 is a real vector field
VM, which may be identified as a gauge field in supersymmetric gauge theories. Here
we will retain in Abelian gauge theory. The generalization to non-Abelian case is
rather straightforward, once we realize what are the new features of SUSY gauge
theories from the simplified Abelian case, since SUSY and gauge symmetries are
mutually independent symmetries. The "real" superfield may be written as follows

l %
V = C + iex-iO~X+ -06(M + iN) - -M{M - iN)

- ea^v^ + ieee{\ + Va M x) - ieee{\ + Va M x)

+ leeee(D - \nc), (7.68)
where bosonic component fields C, M, N, VM and D are all real. In the above equa-
tion the lower component fields C, x, X have appeared also at the positions of higher
component fields with derivatives. To see the reason, let us consider a supersym-
metric extension of (Abelian) gauge transformation (see (7.93))

V-^V'=V + i(\-A.i), (7.69)

with the transformation parameter A being a chiral superfield. Expand the A in

terms of component fields as A = A(y) + \/26%j){y) + 69F(y). Then from the "non-
locality" of the coordinate j / M ,

i(A-At) = i(A-A*) + iV2(6il>-6lp) + i66F-i66F*

- OaVOdn(A + A*) - 4 = (PO0&,dlttl> - Mea^d^)

V2
l
- -eeMU{A-A*). (7.70)

It is now easy to read off the SUSY gauge transformation of each component field
C^C = C + i(A-A*), (7.71)

X -> X' = X + V2xP, (7.72)

168 SUPERSYMMETRY

M + iN -> M' + iN' = M + iN + 2F, (7.73)

V^V^V^ + d^A + A*), (7.74)

A -> A' = A, (7.75)

D ^ D' = D. (7.76)
We realize that, because of the specific way of writing in (7.68), the gauge fermion
A and the auxiliary field D are gauge invariant: (7.75) and (7.76). This property
is desirable, since the gauge fermion, e.g., belongs adjoint representation, Which
is gauge invariant in the Abelian gauge theory. On the other hand, by use of the
transformation property (7.71) - (7.73) , one immediately know that by suitable
choice of the parameters ImA, ip, F, the lower component fields C, x, M, N are
readily gauged away. Thus, without loss of generality, we may assume the following
form of the vector superfield, called Wess-Zumino gauge:

v = -e^ev^ + i{eee\ - m\) + heWD. (7.77)

In some sense, the above situation is similar to what we expect in ordinary gauge
theories, i.e. the longitudinal component of gauge field, which can be written as a
derivative of a scalar function may be gauged away. However, we should also note
that even after moving to the Wess-Zumino gauge, we still have a degree of gauge
transformation due to the real part of A(x),
T/M -> y ; = yM + 2dM(Re.4), (7.78)

which corresponds to the ordinary gauge transformation with transformation pa-

rameter Re A.
In this way, we have identified a SUSY multiplet containing vector field V^,
called vector multiplet,

(V^x), X(x), D{x)). (7.79)

The mass dimensions of these component fields are readily known to be (1, 3/2, 2).
Thus A should be a physical field called gauge fermion or gaugino, the SUSY partner
of gauge boson VM, while D is an auxiliary field, which should be written in terms
of physical scalar fields once Euler-Lagrange equation for D field is used.

• Field strength superfield

In order to get the kinetic term for the vector multiplet, we need to construct
some gauge invariant (covariant in non-Abelian case) superfield, whose product
provides the desirable kinetic term, out of gauge variant superfield V.
We have learned from the aforementioned gauge transformation property that
the gauge fermion A is gauge invariant, while VM transforms inhomogeneously. Thus
 Vector superfield 169

a reasonable way to get the gauge invariant superfield is to differentiate V, so that

it can extract the Aa field as the lowest component:

t°=£- V. (7.80)

This is not a correct way of doing, since the differential operators J= etc. do not
commute with SUSY algebra, and some modification is needed:

Wa = -^D~DDaV. (7.81)

Such defined Wa and Wa are called field strength superfield.

The field strength superfields turn out to be chiral superfields. In fact, we can
easily show, for instance,

Dp Wa = ~D0~DDDaV = 0, (7.82)

where we have used the property D„DD = 0 as the result of {Da, Dp} = 0. Another
important feature of Wa is that it is gauge invariant:

Wa-+W'a = Wa-l-D~DDa(A-A^ = Wa-l-D{D,Da} A

= Wa - l- (D,Da}DA = Wa, (7.83)

where we have invoked to the chiral nature of A and A, DaA = DaA^ = 0.

Let us power expand Wa over 6, in such a way where component fields are
functions of j / M . For such purpose we first rewrite V in terms of j / M = x^ + J(0CTM0):

v = -ea^evli(y) + i(e6ex(y)~ee9\(y))
+ ^eeM(D(y)+idflV'i(y)). (7.84)

We also rewrite "SUSY-covariant" derivatives (7.44), (7.45) in terms of y** and

9a,0&:

Da = • ^ + 2i((jnaaOAdfi, (7.85)

Then, according to the definition (7.81), Wa is obtained after some calculus to be

Wa = -iXa(y) + [6a0D(y)-l-(a'iana0(dliVv(y)-dvVli(y))}e0
+ ee^aad^iy). (7.87)
As we expected the lowest component is the gauge fermion A and all of A, D and
the field strength F^„ = d^Vv — dvV^ are gauge invariant.
170 S UPERS YMMETR Y

• Kinetic term for the gauge field

^,From the above expression for Wa, it is clear that W a W a |jr provides the kinetic
term —iXa^d^X, etc. for the members of gauge multiplet:

Cgauge-kin = ~ (WaWa\F + h.C.)

= -\F^Fflu-iXa(a")aadl,\& + ^D2. (7.88)

The kinetic term for gauge boson V^ is of ordinary type. The kinetic term for the
gauge fermion A may be understood as that for a Weyl fermion. Since the gauge
fields should be "real", it may be natural to expect that gauge fermion is described
by a Majorana fermion AM = (A a , A a )'. In fact, the kinetic term can be re-written
as | A M ^ A M , up to a total derivative. (A Weyl and a Majorana fermions are iden-
tical, as far as the fermions are massless.) As we expected the "kinetic term" for the
D field does not have any derivative and D should be understood as an auxiliary
field.

• Gauge invariant Lagrangian

Let us consider a SUSY U(l) gauge theory, whose typical example is SUSY
QED. Generalization to the non-Abelian case is rather straightforward.
Suppose we have chiral superfields fa, which have electric charges Qi. Under a
global U(l) gauge transformation the chiral superfield transform as

fa -> 4>'i = e~i2eQiA fa, (7.89)

where gauge transformation parameter A should be regarded as a chiral superfield,
not to spoil the chirality of <j> by the transformation. Just as in ordinary field theory,
the self interaction of chiral superfields, i.e. superpotential should be invariant under
the transformation (7.89):

W(fa) -> W(fai) = W(fa). (7.90)

When the U(l) transformation is made local, allowing A to depend on yM, the
superpotential clearly remains invariant. The kinetic term for fa, however, is no
longer invariant,

4fa ->• e-2ieQi^~h^ <f>lfa. (7.91)

In order to make it gauge invariant we now introduce the vector superfield V to com-
pensate the factor e-2teQi(A~A )5 j , e . the kinetic term is now replaced by "covariant
derivative" term,

4>U2eQiVfa, (7.92)
where V should transform under the local gauge transformation inhomogeneously,

V -* V' = V + *'(A-At). (7.93)

 SUSY QED 171

Actually the Taylor expansion of e2eQiV ends at V2: V2 = ^69eeV„V^, Vn = 0

for n > 3. By use of this fact we can check that the SUSY invariant (up to a total
derivative) D-term of (7.92) really contains the ordinary covariant derivatives for
scalars and fermions, though it also contain interaction terms of A, A, D, which are
characteristic to the SUSY gauge theory:

<j>\e2eQiV<j>i\D = (D^AiYiD^Ai) - i^^D^i + \Ft\2

- V2ieQi{Ai\xjji - A*\xjji) + eQtD\Ai\2, (7.94)

where _DM denotes the ordinary gauge-covariant derivatives

D^Ai = (<9M + ieQiV^Ai, D^t = (d„ + ieQiV^rpi. (7.95)

The terms in the second line of (7.94) are the new types of gauge interactions due
to the presence of gaugino A, A and auxiliary field D, whose coupling constants are
identical, i.e. e, because of the SUSY.

7.8 SUSY QED

As the immediate application of the SUSY U(l) gauge theory discussed above, let
us consider the supersymmetric extention of QED, namely SUSY QED.
The matter field, electron, should be identified with the fermionic components
of two (right-handed) chiral superfields,

4>- = {eR,e-R,FR), <P+ = (e*L,e+,F*L), (7.96)

where e^ and e^ = ( e ^ ) c are the right-handed components of electron and positron

in 2-component notation, and by ER^L we denote the SUSY partners of e^ L, "se-
lectron". The vector superfield V = (VJj,A, D) contains the photon field V^ and its
superpartner, "photino", A.
Since the electron has a bare mass term —mee in ordinary QED, we employ the
following superpotential

W = m<f>-cp+. (7.97)

Combining with the SUSY invariant and gauge invariant kinetic terms for vector
and chiral superfields, we get the full Lagrangian for the SUSY QED

£ = h,WaWa\F + h.c.)+4>le~2eVcj>_\D + 4,\e2'v4>+\D

- (m(j)^(f>+\F + h.c.). (7.98)

Writing explicitly in terms of the component fields, the Lagrangian reads as

1 2
L = • ^ " " F--F»»F^-iXa>
^-tACT^A d„X+-D
+i.
172 SUPERSYMMETRY

eR{k) -ie(A M + fc^) eR{k') eR \/2e±^£ eR eR —le eR

— -• •-
•T-
i
I
i
I
AM D

(b) (c)

Fig. 7.3

+ (D^eRr(D^eRH(D^eir(D^el)
- i^a»D^- ie+a'D^ + |FflJ^+ |F L | 2
+ V2ie(eRXeR - /i.e.) - \f2ie(e*j\eR~ - h.c.) - eZ?(|e fl | 2 \eL\2)
+ m ( e ^ e j + /i.e.) - m(eRF£ + e*LFR + h.c). (7.99)

So far spinors for electron and positron have been all written down by use of the
2-component notation. When we calculate Feynman diagrams, however, it will be
more convenient to utilize the familiar 4-component notation, so that we can use
the ordinary form for the spinor propagators. Thus we define 4-component spinors,
Dirac spinor e denoting the electron, and a Majorana spinor AM for the photino:

c
Ra (7.100)

AM = (7.101)
A*

In terms of these 4-component spinors, the Lagrangian is given as

+ I^Vfll 2 + \D^L\2 + ieQe + \FR\2 + \FL\2

+ V2ie{(eRJ^^^e + tj^^—^e) - h.c.} - eD(\eR\2 - |e L | 2 )
mee - m(e*RFL + e*LFR + h.c). (7.102)

As we clearly see in the Feynman rule shown in Fig. 7.3 the coupling constants of
the photon, photino and the auxiliary fields are the same, i.e. e.
After eliminating the auxiliary fields FRIL and D, the scalar potential gets con-
tributions from both of superpotential and gauge interaction. To see this explicitly,
let us write down the terms including the auxiliary fields in the Lagrangian

\FR\2 + \FL\2 - m(eRF*L + e*LFR + h.c.) + \D2 - eD(\eR\2 - \eL\2). (7.103)

 SUSY Yang-Mills theories 173

By use of equations of motion for the auxiliary fields

FR = meL, FL = meR,
2
D = e(\eR\ - \eL\2), (7.104)

we get the scalar potential

V = \FR\2 + \FL\2 + ±D2

= m 2 (|e f l | 2 + |e L | 2 ) + i e 2 ( | e f l | 2 - | e L | 2 ) 2 . (7.105)

We easily find that the vacuum state of this theory is achieved by vacuum expecta-
tion values (eR) = (e/,) = 0. The corresponding vacuum energy Ev — (0\H|0) also
vanishes, and it indicates that SUSY is not spontaneously broken, as we will see in
the following section. The term m2(\eR\2 + \&L\2) in the potential V thus provides
scalar mass-squared m2. This means the masses of electron and selectron are the
same as the inevitable consequence of SUSY.

7.9 S U S Y Yang-Mills theories

The argument for SUSY U{1) gauge theory is easily generalized to non-Abelian
gauge theory including the SUSY extension of the standard model, such as MSSM.
The transformation of a set of chiral superfields, denoted by a column vector <f>,
belonging to a representation of non-Abelian Yang-Mills gauge group, takes a form

4> _• 0 ' = e-i2gA (f>, (7.106)

where A is a matrix belonging to the Lie algebra

A = TaAa, (7.107)

where A a are transformation parameters and the generators Ta satisfy (in a suitable
normalization)

[Ta,Tb] = ifabcTc, Tr (TaTb) = ]- 6ab. (7.108)

Correspondingly the vector superfields Va belonging to the adjoint representation

of the group behave as
e29V _> e2gV = e - i 2 9 A * &2gVei29A (7 10Q)

where

V = TaVa, V =TaV'a, (7.110)

so that the term <fte29V<fi becomes gauge invariant.

174 SUPERSYMMETRY

A gauge covariant field strength superfield is obtained by a straightforward ex-

tension of the Abelian case (7.81),

Wa = - i 'DDe-29V Da e2sV. (7.111)

It is easy to show that such defined field strength has a desirable transformation
rule,

Wa -> W'a = e-i29AWaei2gA. (7.112)

Thus gauge invariant and SUSY invariant Yang-Mills theory with a gauge in-
variant superpotential W((p) is described by

C = ~Ti(WaWa\F + h.c.) + ^e23Vcl>\D

- (W(<j))\F + h.c.). (7.113)

The generalization to the case with several chiral superfields is quite straightforward.
Let us note that we still can impose a condition of Wess-Zumino gauge for each
of Va. The Feynman rule shown in Fig. 7.3 for SUSY QCD is now modified in a
trivial manner by a prescription,

-ie -> igTa, (7.114)

for the gauge interaction vertices of (V£, X%[, Da), with (e, e) being replaced by the
scalar and spinor components of tj>.

7.10 Minimal Supersymmetric Standard Model (MSSM)

The SUSY Yang-Mills theory of the most physical interest is the SUSY extension
of the standard model. Its minimal version, with minimal number of Higgs chiral
multiplets, is called Minimal Supersymmetric Standard Model (MSSM). In this sec-
tion, we discuss MSSM very briefly, focusing on the basic structure of the model.
For more detailed discussion on the MSSM and its various phenomenological im-
plications, we ask readers to refer to the nice reviews in the literature (Haber and
Kane, 1985; Weinberg, 2000).
Basically the construction of the MSSM is achieved just following the argument
of the previous section. In particular the Lagrangian is given by (7.113), though the
second term, the kinetic term (with covariant derivative) of chiral field <f>, should
be replaced by the sum of the kinetic term of each existing chiral field with distinct
gauge generator V = TaVa corresponding to its reperesentation of the gauge group
SU(3)C x SU(2)L x U(1)Y- The superpotential W of the chiral superfields may
take, in principle, the most general form compatible with gauge symmetry, as in
the standard model. We will argue, however, that some restriction is needed, since
 Minimal Super symmetric Standard Model (MSSM) 175

otherwise some serious problems, such as too rapid proton decay is induced. The
restriction is due to a global symmetry called i?-symmetry, as we will see below.
The gauge group of MSSM SU(S)C x SU(2)L x U(1)Y is just the same As that of
the standard model, and the gauge bosons and gauge fermions form vector mmmul-
tiplets, whose interactions are uniquely determined by gayge principle, as is seen in
(7.113). The matter fields, on the other hand, are of our choice. In the stabdard
model, the quarks and leptons to start with are Weyl fermions. Thus it is natural
to assign quarks and leptons as the members of chiral multiples. The remaining
matter fields, Higgs, being complex fields, should belong to a chiral multiplet.
Actually we realize that to construct a realistic theory, one SU{2)L doublet of
Higgs chiral superfields is not enough, and we should add another doublet. The
reason is two-fold.
(1) In the (non-SUSY) standard model the Yukawa couplings or masses of down-
type and up-type quarks are provided by <j> a n d <t> = *cr2</'*, respectively. One may
naively expect that the Yukawa couplings are just replaced by the corresponding
superpotential of the chiral superfields of quarks and the Higgs doublet. Unfortu-
nately, its does not work. This is because, 4> is obtained by the operation of complex
conjugation (charge conjugation), and therefore the chiralities of <f> and <j> are op-
posite, while superpotential should be the polynomial of superfields with the same
chirality. We thus need to introduce two independent Higgs doublets in order to
provide masses to down-type and up-type quarks, say HD and Hu, with the same
representations of the gauge group as those of (f> and </>, both being left-handed chiral
superfields.
(2) Suppose we introduce chiral superfields whose scalar components are the mem-
bers of one Higgs doublet. As the new feature of the SUSY theory, chiral superfields
also contain the fermionic superpartners of the Higgs scalars, i.e. Higgs fermions. As
the Higgs fermions are Weyl fermions, they yield triangle gauge anomalies. For ex-
ample "t/(l)y" anomaly, coming from the triangle diagram where all three vertices
are made of the currents coupled with the U(l)y gauge boson, arises by the presence
of the Higgs fermion. Ther also appear the (SU(2)L)2U(1)Y type anomaly as well.
Such anomalies will be vancelled by the introduction of another Higgs doublet,
just because they have just opposite quantum numbers, with weak-hypercharges
Y = ± 1 . For instance U(1)Y anomaly will vanish as (+1) 3 + ( - 1 ) 3 = 0.
Thus the matter fields of MSSM with 3 generations of quarks and leptons are
all given as left-handed chiral superfields

Qi = f d[ J , u», di; Li=l fl J , h (a = 1,2,3)

*° = ( t ) • H" - ( i ) •
where we have used the same letters to denote the chiral superfields as those of the
ordinary particles in the standard model. For instance for generic quark q and Higgs
176 SUPERSYMMETRY

H chiral superfields, their spin s = 0,1/2 components together with the auxiliary
fields Fq, FH are written as

q = (q,q,Fq), H=(H,H,FH), (7.116)

where q and H are the superpartners of ordinary particles, i.e. a squark and a Higgs
fermion.
Next thing to do for the construction of the MSSM is to choose the form of the
superpotential W. To provide the Yukawa couplings necessary to give quark and
lepton masses and the self-interaction of Higgs doublets, we employ the following
superpotential

W = fZZiQjHu + f'diQjHD + fljIiLjHi,

+ (j,HuHD, (7.117)
where i = 1,2,3 is the generation index, and the abbreviation QjHu etc. is used
for Qtj(ia2)Hu etc., to denote the SU(2) invariant products of two doublets.
Actually this superpotential is not the most general one compatible with gauge
symmetry. Let us note that in the standard model, the Higgs potential is the most
general one, so that it can prepare every counterterms for possible UV-divergences
at the quantum level. It is important to note that in the supersymmetric standard
model, the lepton doublet L and the Higgs doublet HD have exactly the same
quantum number and a priori there is no way to distinguish these two, though in
the non-SUSY standard model, they were clearly discriminated by the difference
of their spins. In fact, the above superpotential has not included, for instance, the
following type of possible gauge invariant operators

udd, dQL, IL2, (7.118)

These terms clearly break either of baryon or lepton number, B or L, and will easily
lead to too rapid B or L violating processes such as proton decay, or too frequent
FCNC processes such as /i —> cy, as the coefficients of these operators are generally
independent of the Yukawa couplings given in (7.117) and can be new sources of
flavor mixings. Thus it is desirable to devise a mechanism to guarantee the absence
of these undesirable operators.
Generally, to ignore terms, or operators, which is compatible with gauge sym-
metry is not allowed for the theory to be renormalizable, since such terms are
gauge invariant and they are generally induced at the quantum level with UV-
divergence and the counterterms are needed anyway. Thus the elimination of the
terms listed up in (7.118) does not seem to be justified. (One may expect that the
"non-renormalization theorem", characteristic to the SUSY theories, may work as
the justification. The SUSY, however, should be broken eventually for the theory
to be realistic.) The elimination is, at least, technically allowed if the elimination
is able to be regarded as the consequence of some global symmetry, independent of
the local gauge symmetry. It is not hard to imagine that the terms incompatible
 Minimal Supersymmetric Standard Model (MSSM) 177

with the global symmetry, imposed on the whole Lagrangian, will never appear at
any orders of quantum corrections. Let us note that while the tri-linear term in the
superpotential dQHo yields ordinary Yukawa coupling of ordinary particles dQHo,
undesirable term dQL will result in a tri-linear coupling with one superpartner and
two ordinary particles, dQL. Thus in order to distinguish these two types of terms,
it seems to be natural to devise a global symmetry, which assign different transfor-
mation properties for ordinary particles and their superpartners. We thus consider
i?-symmetry, which is a characteristic symmetry in SUSY theories.
What we consider is some global U(l) symmetry. It is trivial that to distinguish
ordinary particles and their superpartners, the global symmetry should assign dif-
ferent U(l) charges to the different members of the same chiral multiplet. Hence
a global symmetry which assign an overall phase for the all members of a chiral
multiplet is not enough for our purpose. .R-symmetry is the symmetry under the
chiral transformation of the Grassmannian coordinates of superspace,

6 -> ff = ei,fi9, 6^6' = e-iv§, (7.119)

or

e< (7 120)
I) - ™(*)- -
Accordingly a generic chiral superfield <j> transforms as

<f> = A + V26tp + OOF -><£' = eic,fi<f> = A' + \/26'ip' + 6'6'F', (7.121)

where c denotes the overall U(l) charge of the chiral multiplet. The each component
field therfore transforms as

A' = eic,fi, if>' = ei{c-l)vtl), F' = ei(c~2),fiF. (7.122)

The F-term of the superpotential W\F ~ p W should be invariant under the trans-
formation of .R-symmetry. Thus the transformation property of the superpotential
should be

W -*W' = e2ivW. (7.123)

For instance, for the superpotential W = <$>\4>i$z of some three chiral superfields
to be i?-symmetric, their U(l) charges should satisfy C\ + Ci + c$ = 2. Let us note
that, since the .R-symmetry is a chiral symmetry, other terms in the Lagrangian,
i.e. kinetic terms with covariant derivatives are automatically .R-symmetric, though
the gauge fermions transform non-trivially.
We will assign the U(\) charge c = 1 for all of quark and lepton chiral superfields,
<3, u, d, L, I, and c = 0 for Higgs chiral superfields HQ and H\j. By taking a specific
value ip — ir, we can define a discrete symmetry, ".R-parity". The (overall) parity
178 SUPERSYMMETRY

of quark and lepton superfields is -1, while that of Higgs superfields is 1. Now the
condition reduce, with <p = n, to

W' = W, J2Ci = ° ( m o d 2 )' ( 7 - 124 )

i

It is now easy to see that the all terms in (7.118) do not satisfy the above condition,
while the terms in (7.117) are all consitent with the condition.
For each component field of the chiral multiplets the assigned 1/(1) phases for
<p = 7r, i.e. .R-parity denoted by R, is given by a compact form,

R = (_i) 2 ->(-i) c = ( - l ) 2 J ( - l ) 3 B + L , (7.125)

where J is the spin of the field and B,L are baryon and lepton numbers, and
3B + L = 1 for quarks and leptons and 0 for Higgs doublets. We thus easily know
that all of the quarks, leptons and Higgs have R = 1 while their superpartnes all
have R = — 1. In this way we have suceeded to distinguish ordinary particles and
their superpartners. In addition, the absence of the undesirable terms in (7.118)
is justified as the consequence of the -R-parity symmetry. We further note that,
in general, hevier superpartners having odd .R-parity should decay into lighter su-
perpartners and other ordinary particles, but not into a final state composed only
of ordinary particles, which have even .R-parity. Namely the .R-parity invariance
enforces the lightest supersymmetric particle (LSP) to be absolutely stable, thus
being an interesting candidate of (cold) dark matter.
Though we have implicitly assumed that the SUSY is an exact symmetry, actu-
ally the SUSY should be broken spontaneously (see section 7.12) or explicitly (with
"soft" terms, whose coefficients have positive mass dimensions, in order not to spoil
the nice property of the quadratic divergence cancellation), since the superpartners
of ordinary quarks and leptons having the same masses as those of the ordinary
particles have not been discovered. The theory should also possess the spontaneous
gauge symmetry breaking. These two kinds of symmetry breaking cause the mixings
among various particles via their mass (-squared) terms, in various sectors, which
makes the complete analysis of MSSM (Haber and Kane, 1985; Weinberg, 2000)
rather cumbersome.
We finally briefly comment on the specific feature of MSSM in its Higgs sector.
After the elimination of the auxiliary fields FHUID, the quadratic terms of Higgs
bosons in the obtained scalar potential are \H\2(H'DHD + H^Hu)- This in turn
means that the spontaneous gauge symmetry breaking does not happen, as long as
SUSY is preserved. An intriguing mechanism to achieve the negative mass-squared
of Higgs boson, necessary for the symmetry breakdown, has been proposed (Inoue,
et al., 1982), where the negative mass-squared is realized by the SUSY breaking
and the quantum correction due to the loop of a heavy fermion with large Yukawa
coupling, i.e. t quark.
We also note that the (Higgs scalar) 4 term of the scalar potential is uniquely
 Some Phenomenological Predictions of SUSY Gauge Theories 179

provided by the D-term contribution, e.g. ^-(WDHD - HvHu)2 which is obtained

after the elimination of D-auxiliary field of U(l) vector multiplet. Thus the coeffi-
cient of the quartic term is completely determined by the gauge coupling constants
and is of the order of g2 or g'2. This means that the Higgs masses are comparable
to Mz or Mw, provided SUSY is an exact symmetry. Though actually the SUSY
breaking mass-squared terms for the Higgs bosons considerably modify the range
of the Higgs masses, there should remain at least one light physical Higgs boson,
whose mass < Mz- Intuitively, this is because at the limit of MSUSY -> oo, the
theory should reduce to the non-SUSY standard model, where one physical Higgs,
whose mass is independent of MSUSY, should exist.

7.11 Some Phenomenological Predictions of S U S Y Gauge Theories

In this section we will discuss very briefly how we can search for the trace of SUY
gauge theories for elementary particles and/or how we can test the characteristic
phenomenological predictions of SUSY gauge theories.
Very successful prediction of the SUSY version of the standard model, Minimal
Supersymmetric Standard Model (MSSM), is the gauge coupling unification. In
Grand Unified Theories (GUT), which unify all interactions (except gravity) of
elementary particles, three gauge couplings of SU(3)C, SU(2)L, U(1)Y, denoted
by <?3,<?2,<h for suitably normalized gauge generators, should be unified at higher
energies. Since these couplings are those for "strong", "weak" interactions etc.,
such couplings do not seem to be unified at least at low energy processes. Actually
the "asymptotic freedom" of non-Abelian gauge theories, and the "asymptotic non-
free" nature of Abelian gauge theory, briefly discussed in chapter 5, implied by the
renormalization group equations to discribe the energy evolution of gauge couplings,
make the grand unification possible. Since the dependence of the gauge couplings
on the energy is very mild, depending only logarithmically, the mass (energy) scale
of the grand unification MQUT is quite high. In the prototype (non-SUSY) SU(5)
GUT (Georgi and Glashow, 1974) MGUT = O(1015)(GeV). Such tremendously
high mass scale makes the life time of proton decay, typically predicted by GUT,
quite long, assuring the (approximate) stability of atoms.
The simplest SU(5) GUT model, however, seems to be facing the following diffi-
culties, (a) First, it predicts the proton life time TP ~ 3 x 1031 (years), much longer
the age of our universe. Nevertheless, the prediction has been ruled out by the
recent (Super-) Kamiokande result on the decay mode p -¥ ix° + e+. (b) Second,
the precision measurement of the gauge couplings g3, g2, g\ at E ~ 102 (GeV) at
LEP experiments (CERN) has revealed the fact that these couplings do not meet
with each other at an unique value of higher energy. It is quite impressive that
such difficulties can be evaded (though there still may remain some problem) in
SUSY SU(5) GUT (Sakai, 1981; Dimopoulos and Georgi, 1981), whose low-energy
180 SUPERSYMMETRY

effective theory is MSSM. Namely in the SUSY GUT, the evolution of gauge cou-
plings, or the ^-functions in the renormalization group equations are modified by
the presence of the newly introduced superpartners of ordinary particles (mainly
by the gauge fermions). If the SUSY breaking mass scale MSUSY is in a reasonable
range, MSUSY ~ l(TeV), all of such modified three "running" couplings meet at a
unique value of the energy ! Furthermore, the modification of the evolution raises
the unification scale a little bit, MQUT — 1016(GeV). This little change, however,
considerably raises the proton life time Tp (since Tp oc MQUT), thus evading the
lower bound on TP imposed by (Super-)Kamiokande experiment. It may be worth
noting that the most recent result on the lower bound of Tp from Super-Kamiokande
experiment is quite close to the prediction of the SUSY GUT, even though they
still seem to be mutually compatible.
Probably the most direct confirmation of SUSY will be the discovery of the char-
acteristic new heavy particles, i.e. the superpartners, in accelerator experiments,
and extensive efforts have been made in such direction. Such superpartners should
affect the low-energy physics as well, and their contributions may be (indirectly)
tested by use of various low energy observables, whose probabilities have been mea-
sured precisely. In particular the SUSY breaking masses of the order MSUSY of
these superpartners can be arbitrary a priori, as long as they are gauge invariant.
Therefore they may potentially spoil the successful predictions of the (non-SUSY)
standard model, unless there are some guiding principle to control them.
For illustrative purpose, here we will briefly discuss the contributions of super-
partners of quarks and leptons (squarks and sleptons) to the observables, which are
rendered to be small by global symmetries, i.e. (1) Ap and (2) Flavor Changing
Neutral Current (FCNC). These are quantities discussed in detail in chapters 8 and
9, respectively, and are handled by global symmetries, i.e. "custodial" and flavor
(or horizontal) symmetries. We will see in chapters 8 and 9, that because of these
global symmetries, these observables exactly vanish at the tree level. Though the
observables are induced at the quantum (loop) level, they are automatically finite
and are suppressed. Thus the smallness of these quantities are guaranteed without
any tuning of parameters of the theory, the property we call "natural". If SUSY is
exact, i.e. if there is no SUSY breaking, the situation essentially does not change,
and natural suppression of these observables are still operative. Once SUSY is
broken, which is necessary anyway for the models to be realistic, the situation re-
markably change, generally speaking. In particular, as we will argue below, FCNC
may get into trouble, since the SUSY breaking mass terms, being flavor or gen-
eration dependent, break the flavor symmetry. Thus unless there is some guiding
principle to control the mass terms, the natural suppression of FCNC is spoilt, and
it may cause a serious phenomenological problem of the model.
(l)Ap
The ^-parameter is defined in (4.105) and (8.5) as the ratio of charged and
neutral weak gauge boson mass-squared, p = 1 at the tree or classical level because
 Some Phenomenological Predictions of SUSY Gauge Theories 181

of the SU(2)v (or SU(2)L) global "custodial" symmetry which remain in the Higgs
sector even after the spontaneous symmetry breakdown (see 8.4). However, the
symmetry is broken in the entire Lagrangian by e.g. the mass splitting between
the members of SU{2)L doublet, such as {t,b)1, the parameter deviates from 1 ,
i.e. Ap = p — 1 ^ 0. The contribution of the heavy top quark to Ap turns out to
2
behave as -rri-. Then, one may naively expect that the superpartners, which are
m
w
supposed to be heavy, ~ l(TeV) or so, may strongly affect Ap and SUSY gauge
theories may contradict with the experimental upper bound on Ap. Fortunately the
SUSY breaking mass-squared given for the doublet of squarks, for instance (iijd)',
does not break the global symmetry. Namely the SU{2)i gauge invariance makes
the SUSY breaking mass-squared term for the squarks MgUSY(\u\2 + \d\2) invariant
under the transformation of the custodial SU(2) symmetry. Thus we expect the
contribution of the squark doublet is under control. In fact we will see in (8.18) that
the contribution is strongly suppressed by the factor {m\ — m2l)2/(MyVMgUSY) (the
mixing between the superpartners of left-handed and right-handed quarks has been
ignored for brevity). This kind of suppression is generally stated as the "decoupling"
phenomena of heavy particles, which we will systematically analyze in section 8.2.
Thus the newly added contributions, characteristic to SUSY gauge models, are not
troublesome concerning Ap, as long as MSUSY is not too small.
(2) F C N C
We will discuss in section 9.2 that to guarantee the natural suppression of FCNC
processes, whose rates are experimentally stringently bound and these processes are
often called rare processes, fermions of the same electric charge and chirality should
belong to the same representation of the gauge group (Glashow-Weinberg's condi-
tion) . Since the SUSY generators commute with those of gauge symmetry, these
two symmetries are mutually independent. Thus the superpartners should belong to
the same representations as those of ordinary quarks and leptons, and the Glashow-
Weinberg's condition is still satisfied in SUSY extension of the gauge theories sat-
isfying the condition. In fact, each of neutral currents coupled to Z weak gauge
boson in quark and squark sectors, for instance, is flavor-diagonal and does not have
FCNC. There appear, however, new type of neutral current in SUSY gauge theories,
namely the (fermionic) current coupled with gauge fermion, which is bi-linear form
of a fermion and its superpartner, of the type shown in Fig. 7.3(b). As we will see
below if SUSY breaking mass-squared for superpartners are flavor-dependent, there
arises a mismatching between the mass (-squared) matrices of fermions and super-
partners, and when we move to their mass-eigenstates the neutral current coupled
to the gauge fermion may have FCNC. This happens already at the SUSY QED
discussed in 7.8 once the model discussed there is generalized so that it includes
(three) generations of leptons. Thus leptonic FCNC process, for instance p, —> ej,
becomes possible via the exchange of the gauge fermion, photino AM , as shown in
Fig. 7.4.
We are now ready to confirm the above statement by a little explicit calculation
182 SUPERSYMMETRY

AM

Fig. 7.4

in the SUSY QED with three generations of leptons. The chiral multiples in (7.96)
is generalized to include three generations, a = 1,2,3, as

4>-a = (iaR,l-R,FaR), <t>+a = (raL,liR,KL) (a = 1,2,3), (7.126)

where la, la are current-eigenstates, the eigenstates of gauge interactions. The

Lagrangian is the same as (7.98) except that the chiral multiples become triplicate
and their superpotential is given by

W = mpa4>-a(f>+0, (7.127)

where mpa may be regarded as the (/?, a) elements of a mass matrix m. The mass
terms of leptons la and sleptons la are easily obtained as the generalization of the
mass and mass-squared terms for e and e given in (7.102) and (7.105):

Cm = -(hmlR + h.c.) - {lR{w)m)lR + l{(mrrJ)iL}, (7.128)

where the lepton and slepton states are denoted by column vectors IL<R =
(IIL,R^2L,R,13L,RY and IL,R = ('iL,fl^2L,fl,'3L,fl)'- It is obvious that the mass
matrix for lepton and mass-squared matrices for sleptons are diagonalized simulta-
neously by the same unitary matrices UL and UR:

UlmUR = mdiag, (7.129)

UR{w)m)UR = m|iag, Ul(mmi)UL = m£iag, (7.130)

and the vectors of current-eigenstates are related to the vectors of mass-eigenstates

lmL,R,lmL,R by h,R = UL.RIW,L,R, h,R = UL.RlmL,R- On the other hand the
interaction due to the neutral current coupled to the photino AM is given in the
base of current-eigenstates as

CNC = V2ie{lRJ^lR + VJ^IL - h.c.}. (7.131)

It is clear that even if we move to the mass-eigenstates lmL,R and lmL,R, the neutral
 Some Phenomenological Predictions of SUSY Gauge Theories 183

current is still flavor-diagonal and no FCNC is induced, i.e.

CNC = V2ie{VmRXMlmR + flmLXMlmL - h.c], (7.132)

since WL RUL,R = I (I is the 3 x 3 unit matrix). Namely all neutral currents have
no FCNC at tree level. More intuitively we may say that the superpotential (7.127)
itself can be diagonalized by UhyR without changing the kinetic terms for chiral
multiplets.
The situation changes once we introduce SUSY breaking mass-squared matrices
M\ R, only for sleptons (for brevity, possible mass-squared term to mix II and lR
is ignored),

CsusYbr = -{IR(MR)lR + ll(M2L)h}. (7.133)

Suppose the SUSY breaking is caused by some interactions, which are indepen-
dent of flavors, such as gauge interactions. Then the SUSY breaking mass-squared
matrices do not break the flavor symmetry and MRL should be invariant under
the unitary transformations by UL,R- This argument implies that MR L should be
proportional to the unit matrix,

M2L = M2SUSYtL-I, M2R = M2SUSY<R-I, (7.134)

where MSUSY,L and MSUSY,R denote the mass scales of the SUSY breaking. As
far as the SUSY breaking does not break the flavor (or horizontal) symmetry, we
expect that the same argument as that in the case of exact SUSY will holds. In
fact, we readily realize that the slepton mass-squared terms, including the SUSY
breaking terms, are simultaneously diagonalized as the diagonalization of the lepton
mass matrix,

UR{rrJm + M2SUSYRI)UR = m | i a g + M2SUSY^RI, (7.135)

Ulimrrf + MlUSYiLI)UR = mfa^ + M2SUSY<LI. (7.136)

Thus there does not appear FCNC in the neutral current coupled with the photino.
In the MSSM, the SUSY extension of the standard model, there appears the
flavor or generation mixings in the charged currents coupled with the superpartners
of W± as well. But such mixing is described by the same matrix as the KM matrix,
as far as the SUSY breaking mass-squared matrices are proportional to J. We
also learn from (7.135) and (7.136) that the mass-squared differences of squarks or
sleptons in MSSM, which play essential roles in FCNC processes, are the same as
those of quarks or leptons. For instance

m? - m? = m\ - m2u. (7.137)

Thus the rates of FCNC processes are suppressed by the small quark mass-squared
differences and/or small flavor (generation) mixings, just as in the case of the stan-
184 SUPERSYMMETRY

dard model. This mechanism may be called as "super-GIM" mechanism (Barbieri

and Gatto, 1982; Inami and Lim, 1982).
We, however, have to be aware of the fact that a priori there is no mean to
restrict the SUSY breaking mass-squared terms. In fact Ml R may be arbitrary,
without contradicting with the gauge symmetry. Hence, unless there is some good
reason to guarantee the flavor symmetry in the SUSY breaking masses, the model
will face a serious problem of too large event rates of FCNC processes, such as
p —> ej decay shown in Fig. 7.4, whose branching ratio should be extremely small,
< 4.9 x 1 0 - 1 1 even if the process ever exists. In such a sense, the study of FCNC
processes may be quite crucial in order to select the mechanism of SUSY breaking,
which strongly characterizes the SUSY theories.

7.12 Spontaneous S U S Y breaking

We have seen in the argument of SUSY QED that the masses of electron and se-
lectron should be the same provided the SUSY holds. Experimentally, however, no
selectron with the mass me has been observed. Thus SUSY should not be an exact
symmetry in the nature and should be broken spontaneously or explicitly. Unfortu-
nately, no successful spontaneous SUSY breaking in the sector containing ordinary
quarks and leptons is known because of some phenomenological difficulty, though
the spontaneous breaking is theoretically more appealing. Thus in the MSSM, for
instance, explicit SUSY breaking scalar mass-squared terms, for instance, are added
to the Lagrangian (by hand). We, however, discuss in this subsection the mecha-
nisms of spontaneous SUSY breaking, since it is not only theoretically appealing,
but also is expected to be the origin of the "explicit SUSY breaking". In fact, it
is argued that in the MSSM embedded into a supergravity theory, for instance, the
explicit SUSY breaking terms are brought into the "observed" sector of quarks and
leptons via supergravity interaction from some "Higgs sector", which is decoupled
from the observed sector once supergravity interaction is switched off and where
the SUSY is spontaneously broken.
First let us note that the vacuum energy, i.e. the vacuum expectation value
(VEV) of the Hamiltonian Ev — (0|-W|0) is the order parameter of the spontaneous
SUSY breaking. This is easily seen from the relation in SUSY algebra,

{Qa,<U = - 2 K ) a d J P M . (7.138)

As only a0 has non-zero trace, Trer0 = - 2 , we find

{QuQi} + {Q2,Qi} = -2Tra°P0 = 4H. (7.139)

As the each term in the left hand side is semi-positive definite operator, we find
Ev = (0\H\0) > 0. If Ev vanishes we conclude SUSY is not spontaneously broken
as Qa\0) — Qa\0) — 0. Thus to break SUSY spontaneously, EV > 0 is necessary.
 Spontaneous SUSY breaking 185

In fact if this is the case, under some SUSY transformation due to Qa or Q&, the
vacuum state is not invariant. Thus

Spontaneous SUSY breaking -> (0|tf|0) > 0. (7.140)

Actually it is quite easy to realize a SUSY model with Ev > 0. Let us imagine a
Wess-Zumino model with a chiral superfield <> / and a simple superpotential W(<f>) =
c 4> (c : constant). The resultant scalar potential is positive, V = | g^ '\2 = |c| 2 >
0, and SUSY is expected to be broken, formally speaking. It, however, does not lead
to any mass splitting between the scalar A and the spinor %p components of <f>. In
fact, W\F = c F (F : auxiliary field), which has no self-interaction, nor mass terms
for A and ip. Thus Ev > 0 is necessary but not sufficient condition for "physical"
SUSY breaking. (This is why the arrow in (7.140) has only one direction.) In the
following we discuss possible mechanisms to get physical SUSY breaking.
In general, scalar potential is the sum of the squared absolute values of auxiliary
fields,

V = \F\2 + \{Da)\ (7.141)

where the auxiliary fields are written in terms of scalar fields Aj as implied by the
equations of motion,

*i=^p-> D<> = 9 A*(Ta)iAj, (7.142)

where g is the gauge coupling and Ta are the gauge generators acting on A{. Thus
to achieve spontaneous SUSY breaking with constant scalar vacuum expectation
values, necessary for Lotentz invariance, Ev = (0|V|0) = |(JFi)|2 + \(Da)2 > 0 is
needed ((F?) = (0\F?\0) = a ^ ^ > > ) , etc.). We therefore think of three possibilities
to realize the spontaneous SUSY breaking:
(a) (Fi) i£ 0, irrespectively of (Da)
(b) (Da) ^ 0, irrespectively of (Fi) (7.143)
(c) (Ft) = 0 and (Da) = 0 do not hold simultaneously,
where in the former two cases we do not have to care whether other types of auxil-
iary fields develop vanishing VEV's or not. The spontaneous SUSY breaking of type
(a) and (c) are well-known and were proposed by O'Raifeartaigh (O'Raifeartaigh,
1975) and Fayet-Iliopoulos (Fayet and Iliopoulos, 1974). The third type (b) (Fayet,
1976; Inami, Lim and Sakai, 1983) is less-known, but will be discussed here, as it
is one logical possibility and has a nice feature that SUSY is spontaneously broken
purely due to the gauge interaction.

(a) (Fi) ^ 0 (O'Raifeartaigh mechanism)

We seek a model where (Ft) = 0 are not satisfied for all i, i.e. at least (Fi) ^ 0
for some i, no matter (Da) ^ 0 or not for some a. Hence we consider Wess-Zumino
186 SUPERSYMMETRY

model with several chiral superfields, without considering their gauge interaction.
As we have pointed out above, (F) ^ 0 itself is easily realized only with 1 chiral
superfield, though it does not lead to any physical SUSY breaking. O'Raifeartaigh
found that to realize physically meaningful SUSY breaking at least three chiral
superfields are necessary. Namely he found that a superpotential W for three chiral
superfields fa (i = 1,2,3),

W(0) = A 0 i ( ^ - M 2 ) + /x^203, (7.144)

yields the SUSY breaking, since not all of

„ - 2 dA^ = ^ - 0 . (7,45)
2

F3* = ^ ^ - = 2XA1A3 + f,A2 = 0,

are satisfied for any VEV's of ^1,2,3. Assume, for simplicity, that M, A, fi are
all real and /i 2 > 2A 2 M 2 , then the minimum of the scalar potential V = X2\Al -
M 2 | 2 + A*2|A3|2 + |2AA!A3 + fiA2\2 is known to be achieved by (Ax) = (A2) =
(A3) = 0. Thus (V) = A 2 M 4 > 0, as expected. Furthermore, we can check
that there appear mass splitting among the masses of scalars and fermions. Since
all VEV's of scalars are 0, the masses of fermions and scalars are easily read off
from the quadratic terms in W and V, respectively: fj.tp2ip3 for the fermion mass,
-2A 2 M 2 Re(A|) + ii2(\A2\2 + \A3\2) for the scalar mass-squared. Namely the Weyl
fermions ip2, if>3 are combined to form a Dirac fermion $ = (V,3iV'2)t with a mass

my = n (7.146)

while the scalar masses are given as

m2Al = 0, m\ = fS, m2ReA3 = fj2 - 2A 2 M 2 , m2ImM = »2 + 2A 2 M 2 . (7.147)

Thus the masses of fermions and scalars are not identical any more, as the con-
sequence of SUSY breaking. We, however, also note that, even though the mass
splitting does occur, the average mass-squared of fermions and scalars are still
identical. To be more strict, so-called super-trace of mass-squared matrix, i.e.
StrM 2 = ^2j(-l)2J(2J + l)m2j vanishes when SUSY is spontaneously broken.
Here mj is the mass of the field with spin J field, where each of Majorana fermion
or real scalar with mass m contributes to the sum as —2m2 and m2. For instance,
in the O'Raifeartaigh model we can explicitly confirm that StrM 2 = 2 x (0 + (j?) +
(fi2 - 2A 2 M 2 ) + (/i2 + 2A 2 M 2 ) - 2 x (2 x \ + l ) ^ 2 = 0, where the multiplication
factor 2 denotes that the corresponding masses are for complex scalars and a Dirac
fermion.
 Spontaneous SUSY breaking 187

It is exactly this fact of vanishing super-trace that causes the phenomenological

difficulty when we attempt to construct a SUSY model where SUSY is spontaneously
broken in the sector containing ordinary quarks and leptons, since it means that
some of scalar partners of quarks or leptons should be lighter than corresponding
quarks or leptons, which is not acceptable as we have not observed any of such
superpartners.
In the O'Raifeartaigh model, once gauge interaction is included, all of the su-
perfields $1,2,3 should be singlets under gauge transformation, for W to be gauge
invariant, which may be not appealing if SUSY is to be relevant for high energy
theory. Thus we next consider the case (c), which realizes the spontaneous SUSY
breaking in a SUSY gauge theory.

(c) (Fi) = 0 and (Da) = 0 do not hold simultaneously (Fayet-Iliopoulos

mechanism)
Fayet and Iliopoulos proposed a model where (Fi) = 0 and (Da) = 0 are incom-
patible, and therefore SUSY is spontaneously broken. Thus, in this model, SUSY
is broken as the result of the interplay between the superpotential, i.e the self-
interaction of chiral fields, and gauge interaction, and the mechanism is different
from (b), discussed below, where SUSY is broken purely due to the gauge interaction.
They essentially considered a SUSY QED discussed in section 7.8. They, how-
ever, put an additional SUSY and gauge invariant term with a real parameter £,

-£V\D = -£D, (7.148)

which is called Fayet-Iliopoulos D-term. (Of course, in no-Abelian gauge theory

this term is no longer gauge invariant and is forbidden.)
The scalar potential for the selectrons en,L is now modified from (7.105) into

V = m2(\eR\2 + \eL\2) + i ( e | e f l | 2 - e\eL\2 + £) 2 . (7.149)

We immediately realize that if m = 0 or f = 0, SUSY is not broken, since the

vacuum states with Ev = 0 are realized by the VEV's |(efl)| 2 = |(ez,)|2 — ^ or
=
\{&R)\ \(&L)\ = 0, respectively. On the other hand if both of m and £ are non-
vanishing, SUSY is spontaneously broken, since the two conditions

|e f l | 2 + | e t | 2 = 0 , e | e f l | 2 - e | e L | 2 + £ = 0, (7.150)

are not satisfied simultaneously.

Depending on the sign of m 2 — e£ (assuming £ > 0), the minimum of the potential
V is given by the following VEV's

(SR) = (eL) = 0, for m 2 - e£ > 0,

(eR) = 0, (eL) = ^ e t - m \ for m

2
- e£ < 0. (7.151)
188 S UPERS YMMETR Y

*-eL

(a) (b)

Fig. 7.5

In the former case, though SUSY is broken, the U{\) gauge symmetry is not spon-
taneously broken with vanishing VEV's, while in the latter case both symmetries
are spontaneously broken (see Fig. 7.5).
In the case m 2 — e£ > 0, for instance, the masses of electron and selectrons are
given as

me = m, m\ = m2 + e£, ms — m — e£. (7.152)

Thus, again we get the remarkable relation

StrM 2 = 2 x (m 2 + e£) + 2 x (m 2 - e£) - 2 x (2 x - + l)m 2 = 0. (7.153)

(b) (Da) ^ 0 (The third possibility)

As the third possibility we consider a case where (Da) = 0 are not simultaneously
satisfied by any choice of VEV's of scalar fields, leading to Ev > 0. Namely, in
this case SUSY is spontaneously broken purely due to gauge interaction. For the
transparency of the argument, here we neglect superpotential, since the presence is
irrelevant for the mechanism to work.
Ignoring the superpotential, the scalar potential, the values of VEV's and the
breaking of SUSY is completely determined by the choice of gauge group and the
group representations of matter chiral superfields. One restriction is that the gauge
group should have at least one 17(1) factor, since otherwise Fayet-Iliopoulos D-term
is not allowed and supersymmetric vacuum state is easily realized for all vanishing
VEV's of scalar fields.
The simplest model to realize the SUSY breaking turns out to be 5(7(2) x [/(l)
SUSY gauge theory (Fayet, 1976; Inami, Lim and Sakai, 1983) with Fayet-Iliopoulos
D-term -g£D (we denote gauge couplings of SU(2) and [/(!) by g and g) and two
 Spontaneous SUSY breaking 189

chiral superfields

<P (Q=\), <t>' (Q = -\), (Q •• U(l) charge), (7.154)

both belonging to the doublet representation of SU(2). Let us note the assignment
of the U(l) charges Q = 1/2, —1/2 is such that the triangle gauge anomaly disap-
pears. By use of the equations of motion for D auxiliary fields, the auxiliary fields
are written in terms of scalar fields, which we denote by the same letters cf> and </>',
as

Da = g(^Tacj) + <j>'iTa<P') (Ta = y ) for SU(2)

D =~g(\<f>U~ | * ' V + 0 fortf(l). (7.155)

By use of the SU(2) x U(l) symmetry of the theory, we can assume the following
form of the VEV's of scalar fields, without loss of generality,

where x,y,z and 6 are all real.

Then the conditions (Da) = 0 and (D) = 0 read

yz = 0 for (D1) = (D2) = 0,

x2 - y2 + z2 = 0 for (D 3 ) = 0, (7.157)
2 2 2
x - y - z + 2£ = 0 for (D) = 0.

It is easy to check that for £ ^ 0 these three conditions are not met simultaneously
By any choice of x, y and z. Thus, we conclude that Ev = (V) = \(Da)2+ \{D)2 > 0
and SUSY is spontaneously broken, irrespectively of the choice of the superpotential.
This argument can be generalized to SU(n) x U(l) SUSY gauge theories. It
turns out that SUSY is spontaneously broken provided the number of pairs of
chiral superfields belonging to n and n representations of SU(n) with opposite
U(l) charges is less than n (Inami, Lim and Sakai, 1983). The SU(2) x [7(1) theory
with a single pair of chiral superfields just discussed above is the simplest example.

Problems
7.1 Prove the following relations concerning the 2-component spinors (anti-commuting
Grassman numbers) 6 and 6 and the 2 x 2 matrices a1* and CTM. In 96, 6a§ etc.,
spinor indices have been suppressed:

a. 0°gt> = - I e ^ 0 0 , 6a90 = ±ea066,

190 SUPERSYMMETRY

m& = l_e«0§§} S J . = -.±e&.§§ (7.158)

b. {6a^e){dau6) = 50"" (00) (00) (7.159)

c. (0^i) (0^,) = -i(V'iV' 2 )(00) (7.160)

d. det(<T"pM) = det(CT^pM) = p M p" (7.161)
e. (a")*« = e ^ c ^ K ) ^ (7.162)
2 ,/ /3
f- {a' a + ( T ' ' ^ } a ^ = 2p'"'<5 a (7.163)
v A &
9- {^(T" + a a"} ^ = 2g^S $ (7.164)
/
h. Tria^W) = 2 5 '" (7.165)
i. K ) . ^ , ) " = 2<5a ^ d * (7.166)
_£0123
3- O»OV0X - < f V a " = - 2 ^ * * 3 * (e0123 = = 1}
(7.167)
w x x v AK
k. a*t o - a a a» = 2ie^ <7K (7.168)
liX v
1. a ' W * + ffVff* = -2{g a
v x
- gfty - <r°%
- (7.169)
m. (j»o o + ( r V / = -2(3" V - 0"V -- 5 " " f f A ) . (7.170)
7.2 Verify the relation in (7.31),
G(0, e, e) • G(x", 0,0) = G^ + iO^e - iea»6,0 + e, 0 + e). (7.171)
7.3 Find the propagator of the auxiliary field F of a chiral superfield (consulting,
for instance, with the Lagrangian given in (7.61)). Apply the propagator to the
calculation of the diagram in the left of Fig. 7.2(a), and show that the coupling
—4ig2 should be assigned for the vertex of the diagram in the right of Fig. 7.2(a).

7.4 Show that in the simplified Wess-Zumino model with only one chiral superfield
(/>, whose Lagrangian is given in (7.66), there is an exact cancellation of the quadratic
divergence between the diagrams Fig. 7.2(a) and (b). Show also that the another
type of diagrams with the loop of fermionic field, coming form the contraction of
two identical vertices, e.g. AipM^-^ipM, does not yield the quadratic divergence,
even for m / 0 .

7.5 Verify (7.87) and (7.88).

 Chapter 8
PRECISION TEST OF ELECTROWEAK
RADIATIVE CORRECTIONS AND
N E W PHYSICS

8.1 The Meaning of Precision Test of Electroweak Radiative Cor-

rections

The standard model of elementary particles, especially its electroweak sector, was
devised as a renormalizable gauge theory with massive vector bosons such as Z°
and W±. Already at the classical (tree) level, without quantum corrections, the
standard model has its characteristic properties and such properties have been ex-
tensively tested experimentally. We may pick up some typical examples of such
tests: discovery of the predicted neutral current processes, confirmation of the ex-
istence of Z° and W± with their predicted masses, forward-backward asymmetries
AFB in the scattering e+e~ -> ff (f : fi,etc.).
The genuine features of the theory as the renormalizable theory, however, should
be studied by comparing the calculated (finite but small) quantum effects of the
theory on the physical observables, i.e. radiative corrections, with the data obtained
in the precision experiments, such as LEP, CDF, etc.
Let us recall how the renormalization procedure goes, namely how the finite
(calculable) radiative corrections are obtainable in the renormalizable theory. To
avoid unnecessary complication, let us focus on the gauge sector, namely the part
of the Lagrangian, which contains gauge bosons. Then we have 3 bare parameters:
2 gauge couplings g, g' of SU(2)L and U(1)Y and the vacuum expectation value
(VEV) of the neutral Higgs boson, v. What we should do first is to calculate
observables, which have been excellently well-measured experimentally, i.e.

Mz = 91.150(30) GeV (from LEP, SLC), (8.1)

5 -2
GF = 1.16637(2) x 10~ (GeV) (from fi -> eu^), (8.2)
_1
a = 137.0359895(61) (from g- 2 of electron), (8.3)

including quantum effects in terms of the 3 bare parameters. Namely, we calculate

these quantities as the functions of 3 bare parameters and other parameters, such

191
192 PRECISION TEST OF ELECTROWEAK RADIATIVE CORRECTIONS ...

as mt, TUH'-

Mz(g,g',v;mt,m.H,- • •), GF{9,9'',v;mt,mH,- • •), a(g,g',v;mt,mH,-••), (8.4)

which at the first sight have UV (ultra-violet) divergences, but actually are finite
"renormalized" quantities. (Or we may say that the bare parameters themselves
are UV divergent.) Equating these 3 functions with the measured values listed
above provides formulae, which enables us to express the bare parameters in terms
of observables Mz etc.: g = (j>(Mz,GF,a;mt,mH,- • •)> etc. Then, we may cal-
culate any other observables in terms of bare parameters, / ( # , <?',u;m t ,m#, • • •),
which are superficially divergent. Once the functions <f>(Mz,---) are substituted
for the bare parameters g etc., they become the functions of well-measured quan-
tities: f(g,g',v;Tnt,mH,---) = f(Mz,GF,a;mt,mH,- • •)• The obtained quantities
/ should be all finite, as long as the theory is renormalizable. These are real pre-
dictions of the renormalizable theory.
The above argument on the quantum effects is easily understood, relying on
(local gauge invariant) operators. As is well-known in 4 space-time dimension, a
renormalizable theory should contain only marginal or relevant operators with mass
dimension d = 4 or d < 4 to start with. After including quantum effects, there also
appear irrelevant operators with d > 4, together with the marginal and relevant
operators. The marginal and relevant operators get ultra-violet (UV) divergent cor-
rections on the coefficients of their operators, such as A2 or In ( £ ) , with A and
p, being momentum cutoff and the renormalization scale. (The momentum cutoff
potentially break local gauge invariance and usually alternative regularization meth-
ods such as dimensional regularization may be used.) The quantum corrections to
g and g' and the corrections to gauge boson mass-squared M§, M^,, keeping the
relation of bare masses,

belong to this category. On the other hand the coefficients of irrelevant operators,
induced by the quantum effects, should be automatically finite; otherwise the UV-
divergences cannot be removed, since there are no irrelevant operators nor countert-
erms in the original Lagrangian. The physical observable Ap = p — 1 or equivalently
Mf cos2# - My/i caused, e.g., by a mass splitting between the members of S U ( 2 ) L
doublet, is a typical example. Apparently, it seems that the radiatively induced Ap
is UV-divergent, as the gauge boson mass-squared seems to be the coefficient of a
d = 2 relevant operator. This apparent contradiction is resolved once we invoke to
the language of (local) gauge invariant operators with Higgs field included. As far
as the theory to start with is gauge invariant, every operators induced or corrected
by quantum or radiative effects should be all gauge invariant. This is, say, a picture
in the "symmetric phase" of the theory. Of course the gauge boson mass-squared
stems from the spontaneous symmetry breaking (SSB), i.e. in the "broken phase"
 The Meaning of Precision Test of Electroweak Radiative Corrections 193

of the theory. Thus the mass-squared (difference) relevant for Ap should be under-
stood as the result of the replacement of the (neutral) Higgs field by its VEV in
an irrelevant operator composed of Higgs field, as well as gauge bosons. In fact we
will see in the later section that Ap can be regarded as the coefficient of a d = 6
operator. This is why the radiatively induced Ap is finite.
The precision tests of such finite radiative "corrections" (quantum effects) to
electroweak parameters, such as gauge couplings and gauge boson masses, are quite
important to check the validity of the standard model, since these finite effects are
genuine predictions of the model. For instance such tests are performed by the
precision measurements of various parity violating asymmetries such as forward-
backward asymmetry or LR asymmetry in e+e~ scattering (LEP etc.). The heavy
particles of the SM, like t quark and Higgs, and still unknown heavy particles
predicted by theories beyond the standard model ("New Physics") do not directly
appear in the external lines of Feynman diagrams of lower energy processes. They
appear only or mainly through their contributions to gauge boson self-energies,
i.e. through indirect "oblique corrections". Strictly speaking, e.g. the t quark
may contribute to bbZ vertex, and these extra contributions have to be estimated
depending on each case. It, however, is still true that the oblique corrections exist
universally when we consider the radiative corrections of heavy particles and are
worth general investigations.
As the coefficients of irrelevant operators have negative mass dimensions (d <
0), we may expect that the finite quantum effects due to heavy particles may be
suppressed by the inverse powers of the heavy particle masses ("decoupling" of
heavy particles in low energy processes). But it turns out that this is not always
true, and some radiative corrections are not suppressed by the inverse powers. Such
"non-decoupling" effects in oblique corrections potentially appear only in restricted
number of "oblique parameters", S, T and U, which parameterize electro-weak
radiative corrections. The precision tests of the 3 parameters S, T, U are quite
important not only to get useful information of the heavy particles in the standard
model, m t and mu, but also to test various theories of New Physics. In fact,
(original version of) "technicolor" theory was ruled out as the result of the precision
test of the 5-parameter (Peskin and Takeuchi, 1990).
The purpose of this chapter is to discuss the precision test of the radiative
corrections to the electro-weak parameters in some detail, not only due to the heavy
(comparable or greater than Mw) t quark or Higgs in the standard model, but also
due to the heavy particles predicted by various New Physics. We, however, assume
that the gauge group we take is the same as the standard model, i.e. SU(2)L X
l/(l)y.
194 PRECISION TEST OF ELECTROWEAK RADIATIVE CORRECTIONS ...

8.2 Decoupling and Non-decoupling

As we discussed above, precision test of the radiative corrections due to heavy

particles provides useful information on the properties of heavy particles such as
their masses, or even rules out their existence. The prediction of c quark mass by
Gaillard and Lee (Gaillard and Lee, 1974) before its discovery, utilizing the data
on the flavor changing neutral current processes of neutral kaons K°, K°, is the
remarkable example.
Heavy particles are heavy in the sense that the processes we treat are those
for lighter particles at lower energies. That is why to get information on their
masses indirectly through radiative corrections is meaningful. Heavy particles, even
though they cannot appear as the real states in the lower energy processes, may
participate in the processes as the virtual or intermediate states. We, however,
naively anticipate that the effects are suppressed by the inverse powers of the heavy
particle masses, which we generically denote by M, since the propagators of heavy
particles behave as (for scalars)

fc2-M2~ M2' K
'
2 2
in lower (\k \ -C M ) energies. At the tree level, this replacement is justified. In
the loop diagrams, however, the replacement is no longer justified, since the 4-
momentum k^ varies as the loop momentum in the integration. Thus whether the
heavy particle contributions are really suppressed by the inverse powers of their
masses or not is a non-trivial question to be addressed.
In gauge theories without SSB, having only parity preserving vector-like (non-
chiral) interactions, such as QED or QCD, so-called "decoupling theorem" holds
(Appelquist and Carrazone, 1975): the contributions of heavy particles with mass
M to physical observables are suppressed by the powers of 1/M. Let us take the case
of QED as an example. Suppose we have a heavy charged lepton E~ with the mass
M, say a partner of ordinary electron belonging to hypothetical heavy generation.
Let us consider its contributions as virtual states in the processes described by
Feynman diagrams with n external photon lines (see Fig. 8.1).
The quantum effects provide effective Lagrangian for external photon fields

where Oi denote local gauge invariant operators with n photon fields A^ and Cj are
2

their coefficient functions, which depend M in addition to a = | ^ as the result of

loop integration of internal lines. Note that the mass dimension of the operators
Oi are greater than n, di > n, in general. This is because to make gauge invariant
operators, we need field strength F^ rather than A^ itself, which has derivatives
of A^, thus making the dimension of the operators higher than n. Thus, as is clear
from dimensional analysis, for n > 4 the coefficient functions are suppressed by
 Decoupling and Non-decoupling 195

Fig. 8.1

inverse powers of M, 1/M di ~ 4 (di > n > 4). What all of these are saying is that
the coefficients of irrelevant operators are suppressed by the inverse powers of M.
This kind of dimensional analysis no longer holds for the case of marginal op-
erator, which appears in the case of n = 2, i.e. self-energy diagram. This just
corresponds to the correction to Ffil/F,iV. (We also get operators with higher
derivatives, which we ignore as they are contained in the category of irrelevant
operators.) The coefficient is dimensionless and behave as J0 dt In ( M2 _w 1 _ t N g2 ))
with A, t being momentum cutoff and Feynman parameter. Obviously, this co-
efficient is not suppressed by large M. This correction, however, is not a pre-
diction of the theory as it is UV-divergent. After imposing a renormilization
condition that the correction disappear at —q2 = /z2, which is met by adding a
counterterm made from bare parameters, the correction is rendered to be finite:
Jo dt In (
Mi-t{i-t)qi) ~ ( _ 9 2 - • /^2)> which may be approximated at low energies
H , \q \ < M 2 to be, - £ dt t(l - t) ^f-
2 2
= -\*~$r- « 1. Hence, again the net
effect of the heavy particle is suppressed by an inverse power of M, though it does
contribute to the renormalization of the bare parameters. These are contents of the
"decoupling theorem".
This situation is psychologically good, since when we calculate some physical
quantities just relying on the standard model we do not have to worry about what
really is the theory in higher energies where unknown heavy particles are supposed
to exist: these heavy particles may give some extra contributions but they are
safely small. However, it also says that we cannot have significant information con-
cerning the properties, such as masses, of the heavy particles or concerning New
Physics. Fortunately, in chiral theories with SSB like the standard model, some
"non-decoupling" effects are known to exist. As the above argument of decoupling
theorem is a convincing one, one may wonder what is new in such theories. The
key ingredient, which leads to the non-decoupling, is the fact that now (at least
some of) the coupling constants are proportional to masses of the relevant par-
ticles. More precisely, for instance in the standard model, all massive particles
(including Higgs itself) get their masses only through spontaneous gauge symmetry
196 PRECISION TEST OF ELECTROWEAK RADIATIVE CORRECTIONS ...

\ / N /
\ H / N /
) ( s /
/ s
/ \
/ \ / X
V V V V
(a) (b)

Fig. 8.2

breaking (SSB), i.e. through their couplings with the Higgs field, which develops
VEV. Thus when some particle is heavy that means its coupling with the Higgs,
such as Yukawa couplings and Higgs self-coupling, are strong. (Hereafter we will
assume that the couplings are not extremely strong, not to spoil the perturbative
expansion.) For instance the Yukawa coupling of the top quark / ( i s proportional to
mt, ft — mt/(v/V2) = T/5i§^7- Such couplings appear in the vertices of Feynman
diagrams and therefore in the numerators of transition amplitudes of the processes
we consider, and the above argument on the decoupling theorem, solely relying
on dimensional analysis, is no longer valid. We thus anticipate that in some suit-
ably chosen processes, heavy particle contributions may be enhanced by (positive)
powers of M.
We will list up some known examples of such non-decoupling effects of heavy
particles in the following.

(a) Non-linear sigma model

The non-decoupling effect is obtainable already at tree level. This seems to
contradict with our intuition that the propagator of heavy particles are suppressed
by 1/M 2 as we have discussed above. Concerning the Higgs self-interaction in the
standard model, however, this naive argument no longer holds. Let us consider
4-point scattering amplitude of charged Higgs ip+ ("would-be Goldstone" particle)
as shown in Fig. 8.2:
In Fig. 8.2(a), the exchange of the "physical" neutral Higgs H yields a propa-
gator i/(q2 - Mjj) ~ -i(llMjj)-i{q2IMfj), which is suppressed by 1/M^. On the
other hand the vertices are both enhanced by an factor proportional to Mfj, i.e. by
—ig2MH • Thus this Higgs-exchange diagram provides an effective Lagrangian (in
momentum space)

1 i / . M% , , . , ., o2 . 1 o2 , Mt , , .,
( -) + ] {ipip ? (8 8)
-* 2^MF ^ W ^ ^¥ ^ M M ~- '
2
The first term without q implies we have effectively got a contact interaction en-
hanced as Mfj. Actually this term is exactly cancelled by the original quartic
 Decoupling and Non-decoupling 197

interaction, shown in the second diagram Fig. 8.2(b), - gMaH (ip+ip~)2. Thus in
the large MH or equivalently low energy {\q2\ C Mjy) limit, we have only derivative
interaction —5^5- (tp+<p~) D(<p+ip~~). This term is independent of Mjj and indi-
cates a non-decoupling effect of heavy Higgs. We may attach the lines of <p+, <p~ to
the internal line of H as many times as we want to get higher dimensional operators
of the fields tp. Such operators turn out to be all derivative interactions and to be
summarized in a compact form (u/\/2 being the VEV of the Higgs)

~ Tr {dJJ d"U), U = eiS^, (G+ = - ^ ^ = - i y + , G 3 = v^Imy, 0 )-

4 y2
(8.9)
This Lagrangian contain an infinite number of non-linear terms, and corresponds to
a description of the Higgs doublet (f> in terms of "polar coordinates system" instead
of ordinary Cartesian coordinates:

$ = (#) = v-±^- U. (8.10)

This non-linear realization of the Higgs field is the counterpart of non-linear sigma
model in QCD, where <pl should be understood as 3 pion fields -K% (i = 1,2,3), which
are only particles appearing in the (very) low energy regime of QCD. It is now clear
why the contact interaction term disappeared: in the polar coordinate system the
Higgs potential is independent of <p%, as U^U = / (/ : unit matrix).

(b) Ap
We have defined the p-parameter as the ratio of charged gauge boson mass-
squared to that of the neutral gauge boson (with an extra factor cos2 0w), which is
exactly 1 at the tree or classical level. The parameter, however, deviates from 1, i.e.
Ap = p — 1 ^ 0 , once the quantum correction is taken into account. Let us consider
the quantum correction due to the SU(2)L doublet (t, b)1. If there ever exists
extra generation of quarks and leptons, the doublet may be replaced by the doublet
of fourth generation quarks (£', b')* as well. Since W£ = cos6\y ZM -I- sin Owl11
and the photon field 7M never gets mass correction due to the unbroken U(l)em
symmetry, the quantum correction to cos26w M\ is equivalent to the correction
to the operator W^^W^, with 3 denoting the third component of SU(2) adjoint
representation. So at 1-loop level, namely as far as 0(a) corrections (denoted by
the quantities accompanied by 6) are concerned,

Ap 6{ X
- M*W3>- M>W + 6M*W3 -M%,{ 2 SM
n>> (8
-H)

where we have replaced the mass of the charged gauge boson by Mwt 2 :
M ^ + W + W - " = M*,+ w^w?+w^w3 . (8.H) is graphically expressed in Fig. 8.3,
where n a b(0) (a, b = 1 — 3) denote the scalar part of the vacuum polarization tensors
198 PRECISION TEST OF ELECTROWEAK RADIATIVE CORRECTIONS ...

t,b t,b t,b

W2\ -W3
^ b,t b,t ' b,t

~ 2 ( n u ( o ) + n 22 (o)) -n 3 3 (o)

Fig. 8.3

of Wa^W^ 2 point functions (after factoring out g^u) at q2 = 0. Let us note that
the scalar parts at 0 momentum correspond to the quantum corrections to gauge
boson mass-squared.
The contributions of the (t, b)1 doublet is readily obtained by modifying the
original calculation by Veltman (Veltman, 1977) for the contribution of heavy lepton
doublet to be

A
P = ,„ • 2 „ T?T (mt + mb o \ In - 4 ) , (8.12)
167rsin2 ^ ^ ml-m2b m\n v ;

where the factor 3 comes from the colors of quarks. One may check that this
expression vanishes in the limit of "degenerate doublet" mt = m^. (The factor
—5^—j In ^ is not singular in this limit, but just yields a derivative of the log-
arithmic function, 1/m2. If we take the limit of mj C mi, as is realized in the
nature, we get
. 3a m?
l07TSin C'lV Mff

which clearly shows the enhancement of the quantum effect of heavy top quark,
proportional to m2, i.e. a non-decoupling phenomenon.

(c) Flavor Changing Neutral Current (FCNC) Processes

As we have already mentioned, Gaillard and Lee could predict the c quark
mass before its discovery (Gaillard and Lee, 1974), studying the quantum effects
of the c quark in low energy FCNC processes of neutral kaons K°, K°, such as
K° o K° mixing. As the neutral kaons are the bound states of d and s quarks,
K° ~ s^d, K° ~ J75S, this mixing is caused by an elementary process sd «-» sd.
In this transition, flavor quantum number "strangeness" S changes by two units,
|A5| = 2, while there is no change of the electric charge. This is why such processes
are called Flavor Changing Neutral Current (FCNC) processes. We will discuss
these FCNC processes in some detail in the next chapter. As we discuss there, since
there is no FCNC interaction at the tree level of the standard model, the sd +•> sd
transition is induced by the loop diagram, called "box diagram" (Fig. 8.4).
When Gaillard and Lee discussed the process the third generation had not been
confirmed. But now its existence is known and it is an important question to ask if
the heavy t quark gives a non-decoupling quantum effect to this process, thus making
 Decoupling and Non-decoupling 199

U,C,t
——*—

W+,<p+ SW~'V~
u,c,t
»

Fig. 8.4

it meaningful to estimate mj or its flavor mixings with lower generation quarks by

studying this K° f+ K° mixing, just following the philosophy of Gaillard and Lee.
For such purpose it is necessary, since mt > Mw, to calculate the box diagram
exactly without assuming the internal quark mass is much smaller than Mw, which
was adopted by Gaillard-Lee's calculation in the view of the fact mc -C Mw- The
result of such exact calculation (Inami and Lim, 1981) is written in the form of
4-Fermi effective Lagrangian (see (9.11))

4 / f = 2 = , J*°F2 a £ (V^V^iV^Vja) Eix^xj) (sllxLd)(srLd), (8.14)

4\/27rsin Vw itj=Ctt

where L = ^ p 2 - , Vjs etc. are KM matrix elements and the coefficient function
E(xi,Xj) denotes the contributions of internal up-type quarks with masses m*, m,-
2
(xi = -Tji-). The (pure) t quark contribution reads as,
m
w

EM , E (x„ Ie ) . Aj^L-f ln „ . [i _ | _ L _ _ ? _ ! _ ] x, (8.15)

For the imaginative limit of xt > 1, i.e. m% » Mw (though it is not so bad

approximation), the function behaves as

<8 16)
**>*-H~\jk- -
which behaves as mf and is another example of the non-decoupling effect of the
heavy fermion. It is worth noticing that for c quark, with xc <C 1, E(xc) ~ —xc,
which just recovers the result of Gaillard-Lee. The calculation (Inami and Lim,
1981) also shows that similar enhancement proportional to m\ appears in the in-
duced FCNC vertex sdZ ("Z-penguin"), but not in sdj vertex (ordinary "penguin"),
where we have only logarithmic enhancement ln -n!j- due to the property of the
CVC. As far as we know the heavy fermion contributions to Ap and these FCNC
processes are only known examples where we have an quadratically growing non-
decoupling effects. Interestingly, it turns out that (at least at 1-loop level) the
contribution of a heavy Higgs to Ap does not grow as M\, but grows only as
ln
m ^M
a --
w
200 PRECISION TEST OF ELECTROWEAK RADIATIVE CORRECTIONS ...

We have seen that the interesting non-decoupling effects arise in chiral theories
with SSB. Even in such chiral theories, however, heavy particles effects may be
decoupling. It is the case when heavy particle mass is not provided by SSB, i.e.
not through large couplings with Higgs, but by a new large mass scale, independent
of Mw, characterizing some New Physics. Such new mass scales are inevitably
invariant concerning the gauge symmetry of the standard model, since otherwise
it will break the gauge symmetry at much higher energies than Mw, which con-
tradicts with reality. In such a sense, the situation is similar to the case of the
decoupling phenomena we discussed above about a heavy lepton E~, whose large
mass was also gauge invariant and came to the denominator of the coefficients of the
effective operators. We will list up some typical examples of such decoupling effects.

(d) Seesaw mechanism

In Chapter 6, we discussed the seesaw mechanism. The seesaw mechanism is
known to be a scenario which naturally lead to light Majorana neutrinos. As we
2

have already seen in 6.2, the small Majorana mass ^f-, with mr>,m,R being a
Dirac mass and the Majorana mass for VR, is induced through a tree level dia-
gram in Fig. 6.1. The fact that the mass is suppressed by 1/TUR can be under-
stood as to indicate the decoupling effect of the intermediate state VR. Let us
note that m f i ( » Mw) is SU{2)L X U(1)Y invariant mass, which may be related
with the new mass scale of, say, left-right symmetric model of electro-weak interac-
tion SU{2)L x SU(2)R x U{\)B-L (Mohapatra and Senjanovic, 1980) (B, L denote
baryon and lepton numbers), where SU(2)R symmetry and therefore parity is spon-
taneously broken at the scale 0(rriR).

(e) The contribution of super-partners to Ap

As was discussed in Chapter 7, supersymmetric theories predict the presence of
super-partners of ordinary particles. Here let us consider the contribution of the
SU(2)L doublet (u, d)1, the superpartners of light generation quarks (u, d)1. As the
partners have not been observed in, e.g. e + e~ collider experiments, these partners
should be heavy: m\ = m2u + MgUSY, m2- = m2d + MSUSY, where MSUSY is
"SUSY breaking mass scale", denoting the extent in which SUSY is broken. (We
have ignored the effect of possible mixing between the super-partners of left- and
right-handed quarks etc., to avoid unnecessary complication.) These partners are
heavy not because the Yukawa couplings are large (mUid <£ Mw), but because the
gauge invariant MSUSY is large. We thus anticipate the decoupling of these partners
in their quantum effects. Their contribution to Ap is given as (Alvarez, Gaume,
Polchinski and Wise, 1983; Barbieri and Maiani, 1983; Lim, Inami and Sakai, 1984)

Ap = ^ h r - TRT K + m\ - ^ % In % . (8.17)
H
167rsin2 6W Mw y u d
ml-ml m\> y
'
Superficially, the form of this equation is the same as that in (8.12) for (t, b)( doublet
 Oblique Corrections and S, T, U Parameters 201

/'

V/QAAA/

/'

Z „ 7 7 „ Z
z _ z

Fig. 8.5

contribution, and the decoupling of super-partners does not seem to be the case.
As the matter of fact, the decoupling holds: Since mu>d <£ MSUSY, we can expand
m2
the formula for Ap in terms of powers of m,/ri"'d to get
SUSY

AP- ^ — A ( f :fd)2 . (8-18)

167rsin2 dw M$y M2SUSY
which shows the suppression by -j-^— — ~\— and therefore the decoupling of
M
SUSY a,J
super-partners. The factor (m u — m2d)2 indicates that Ap is described by an ir-
2

relevant operator which is quartic in the Higgs field, as we will see later in this
chapter.

8.3 Oblique Corrections a n d S, T, U P a r a m e t e r s

As has been already stated in 6.1, heavy particles of the standard model, t quark
and Higgs, and heavy particles predicted by New Physics do not directly appear
in the external lines of Feynman diagrams for the lower energy processes of light
particles. They appear only or at least mainly through their contributions to gauge
boson self-energies, so called "oblique corrections". Up to the order of 0(a2), or up
to the 1-loop order, the oblique corrections to the process / / —> / ' / ' are shown in
Fig. 7.5, where the blob denote the radiative corrections to gauge boson self-energies
due to the heavy particles.
Such oblique corrections are known to be conveniently incorporated by "star
prescription" (Kennedy and Lynn, 1989), namely by replacing the bare quantities
at tree level amplitude e, s 2 (= sin29w), etc. by corresponding "star" quantities
e*(q2), s2(q2), etc.. which incorporate quantum corrections as well and depend on
q2, with q^ being the momentum of intermediate gauge bosons. More explicitly,
202 PRECISION TEST OF ELECTROWEAK RADIATIVE CORRECTIONS ...

the effective 4-Fermi Lagrangian for the process (in momentum space) is written as

Ceff = elQQ' ( / 7 „ / ) 1 (f'-ff)

+ ^(H[hL-slQ)f)-^—{j'^%L-slQ']n (8-19)

where Q, Q' and h,^ are electric charges (in the unit of e) and the third components
of weak isospin of fermions / and / ' , respectively.
Let us demonstrate this and find the expression of the star quantities in terms
of the vacuum polarization functions n(<72). We start from writing the radiative
corrections to gauge boson self-energies denoted by blobs in Fig. 8.5, in the form
of induced quadratic terms for gauge bosons in the effective Lagrangian, in terms
of the vacuum polarization functions II(g 2 ):

| U^g^A^A" + | Uzzg^Z^Z" + I ^ p ^ Z M " + Uwwgtll/W+'iW-v, (8.20)

where we have included the correction to charged gauge boson self-energy Uww, for
later use. We have ignored the part proportional to q^,qv in the vacuum polarization
tensors, since in the scattering amplitude this part provides relatively negligible
terms which are at most proportional to external light fermion masses after the
usage of equations of motion, such as q^f^y^f = 2m/ / 7 s / . At the order of 0(a2)
or at the 1-loop level, the effective 4-Fermi Lagrangian relevant for the scattering
is given as

Ceff = (eQftrf, f-sH(hL-s2Q)f )

2 1
9 -n 77 Z7 -n y ( eQ'f'rf
-riz 7 q2 - M2 - nzz ) V tJ'-f^L ~ *2 W
* ( eQf7llf , %H(I3L-s2Q)f )

g '(i-n!, T ) ? S ^ \ f eQ'f'-ff
J%, „ ^ - J V fJ'l^L-s2Q')f

+ -^(hAhL - (s2 - csUZy)Q]f)

c s
1
(f'^m - («2 - csUZj)Q'}f'), (8.21)
q2 - M2 - Uzz

where I I 7 7 = q2 II 7 7 , Uzz = q2 U'zz, and Ml = f ^ i s t n e b a r e z boson mass-

squared. In the last step, we have used the fact that the off-diagonal element of the
n' n'
matrix for propagators, n2 fa, may be replaced by n2_fazlu— a t t n e 1-loop level,
 Oblique Corrections and S, T, U Parameters 203

Fig. 8.6

and can be combined with the contribution of diagonal element accompanied by

Hzz- The key ingredient here is that not only II 7 7 but also Hz-y is proportional to
q2 because of the CVC of QED. Namely in the diagram with 7 — Z mixing in Fig.
8.6, the factor of photon propagator 1/q2 is cancelled by the factor q2 in Uz-y, thus
making the photon propagator shrunk, yielding a amplitude exactly of the same
form as that of (the vector coupling part of) pure Z-exchange diagram (see Fig.
8.6).
The factor of Z-propagator 2 _ M a _ n — should have a pole at q2 = Mz, with
Mz being the well-determined physical Z boson mass: Mz ~ ^0 ~ Tlzz(Mz) = 0>
which may be used to eliminate the bare mass MQ from our final result. Namely,
the Taylor expansion of Hzz(q2) around M§ gives an expression in terms of the
physical mass: q2 - M2 - UZZ(q2) = q2 - M2 - (q2 - M2z)^\q2=M2 - UreS cz
(! - ^\q>=Ml)(q2 -M2Z- n r es), where n r e s denotes the part 0((q2 - Mf) 2 ):
Tires = Tlzz(q2)-Uzz(M2)-(q2-M2)^\q2=M2. The factor ( 1 - ^ | , 2 = M | )
is the residue at the pole of the propagator.
Now the expression (8.21) takes the same form as (8.19), and we can easily read
off the star-quantities:

~ — {i-[n77(<z2)-n77(0)]}, (8.22)
Ana
2 _ „2
3^ = 3 cs IL'Z„ (8.23)

dUzz. , c2 - s2
l + ^!^=Mi-n;7-^—-U' Z y } (8.24)
_lg2=M|-n77-—-
M2 = Mlz +
t- U
Ilres
= M2+Uzz(q2)-Uzz(M2)-(q2-M2z)^\q2=M2, (8.25)

where ^ ^ = ^ [1 —II77(0)] has been used. Let us note that generally when the vac-
uum polarization functions U(q2) are Taylor expanded in the powers of q2, only 11(0)
and the first derivative ^\q2=0 are possibly UV-divergent, since they correspond
204 PRECISION TEST OF ELECTROWEAK RADIATIVE CORRECTIONS ...

to the radiative corrections to gauge boson mass-squared and kinetic terms. This
statement is true for the Taylor expansion around arbitrary point, such as q2 = M§.
Thus n ; 7 ( g 2 ) - n ; 7 ( 0 ) and II r es = V.zz{q2) -UZZ(M2) - (q2 - M 2 ) ^ | , 2 = M | ,
e.g., are finite, since they contain only the terms with 2nd or higher derivatives, and
we have got formulae in which the finiteness of physical quantities, like e* and M*
are manifest . The finiteness of Z* is less trivial (See Problem 8.4). The remaining
formula for s* contains bare quantity s2 and is not manifestly finite. So we attempt
to rewrite s2 by a quantity made from well-measured observables and e* at the Z
pole, sin.20w\z = (TTOGM 5 ") 1 ^ 2 - ^rom t n e relations

47TQ = e 2 (l + n; 7 (0)) (8.26)

M
z = -^+^zz(M2z) (8.27)

~j5-w{1--#Mr]' (8 28)
-
We get a formula for Ow\z
2cV
sm26w\z = s2 + S (s 2 ) = s 2 + 2csS6w = s2 + -r ~ 6(ln(sm26w))

c2s2 ,6a _ J O F _ SMj

a GF M2Z

^ ^ < ^+ ^ - S ^ ) . <-)
Thus substituting s2 obtained from the above relation, we finally get

A - ™2M* - j g s <n;, + ^ - M l , _ cs m 7 y , . (8.30,

The part described by II 's turns out to be finite from a similar reasoning to the
case of Z* and from the property that IIww(O) — c 2 IIzz(0) is finite, as it should
be. All of these finite radiative corrections are the genuine predictions of the theory
we consider and can be precisely tested by various experiments. For instance the
corrected s2 may be tested by studying parity violating asymmetries. The asym-
metry of the scattering cross-sections of polarized electrons at the Z boson pole is
an example, which is very sensitive to the precise value of the Weinberg angle, since
the vector coupling of the Z boson to the charged lepton like electron is close to
zero, — \ + sin2 9\y — 0:

A - <e~Le+ ~» Z) ' <e~Re+ -* Z) - 8(|-sin2flw) 1 2

ALR
~ a(eie+->Z)+a(e-Re+^Z) ~ 1 + (1 - 4sin2 6W)2 " 8(
°wh
4 Sm

(8.31)
The vacuum polarization functions II 7 7 , Tlz-y, Rzz can be written, factoring
the gauge coupling constants out and indicating the indices of SU(2)L and U(l)em
 Oblique Corrections and S, T, U Parameters 205

adjoint representations by 1,2,3, and Q, as

n77 = e 2 IlQQ, (8.32)

UZy = ^-s(n3Q-s2UQQ), (8.33)

nzz = ~ (n33 - 2 S 2 n 3Q + s 4 n QQ ), (8.34)

2
e
Uww = -o I I n , (8.35)

where Tl22 = IIn due to the unbroken t / ( l ) e m symmetry, and it may also be written
as n "+ n 22 as well. We can understand the form of IIz 7 , Tizz easily if we recall
that neutral current coupled with Z takes a form —^ /7 M (J 3 L — s2Q)f.
As an example we calculate the contributions to these polarization functions
from a pair of quarks (t, b)1, which may be regarded as the quarks of another gen-
eration including 4-th generation as well, if it ever exists. All vacuum polarization
functions are reduced to just 2 independent functions which correspond to the vac-
uum polarization amplitudes induced by two electro-weak currents of quarks with
the same and opposite chiralities, respectively:

ULLim^m^q2) = II f l f l (m 2 ,m 2 ,g 2 )
dtln ( 22 jV2 ., , 1
= ~jBr [ 2
- *" "~2)'[t(1"t)q2 ~ ^M% (8-36)
'M -t{l-t)q2
^LR{m\,ml,q
i\,m\,q') = T\RL{m\,ml,q2)
1
12 f A2 1
dUn{
-~(4TT)2 J0 M*-t(l-t)qz>'2

where M 2 = tm\ + (1 — t)m2. Writing polarization function induced by vector

currents as

nvv = nL+R,L+R = nLL + uRR + uLR + uRL = 2(uLL + uLR), (8.38)

the polarization functions of our interest are now written in terms of these two
functions as

IW<72) = Q2tUvv(m2,mlq2) + Q2bUvv(m2b,mlq2), (8.39)

n3Q(g 2 ) = -QtTlLv{mt,m2,q2) - -QbULV(m2b,m2b,q2),

= -[QtUvv{m2,m2,q2)-QbUvv(mlm2b,q2)}, (8.40)

n33(<72) = \[ULL(mlm2,q2)+ULL(m2b,m2b,q2)}, (8.41)

IMg2) = \uLL{m2,mW). (8.42)

206 PRECISION TEST OF ELECTROWEAK RADIATIVE CORRECTIONS ...

It is worth noting that the "chiral amplitude" has non-decoupling effect for mt »
mb,
6 A2
2 2 2
ULL(m t,mlq ) ~ —— m \n{-^), (8.43)
2
3 ,A , 1,
nLL(m2,m2,g2) ~ 7 --T^?
7/•r m
?2 m
22 n
t £ (In
( l nv((—
—)
29 +-), (8.44)
(47r) m
which contribute to Ap as

n n ( 0 ) - n 3 3 ( 0 ) ~ ±HLL(m2,m2,0) - ilILi(m2,m2,0) ~ J L m2. (g.45)

On the other hand, when IILL and ULR are summed up to get TLvv, relevant for
the vector-like theories such as QED or QCD, the quadratic term m 2 are cancelled
out and I I w is proportional to q2:
24 f1 A2
2 2 2 2
nvv{m t,m ,q ) = q {-J-^ jQ t(l-t)ln(m2t_tii_t)q2)dt}. (8.46)

This proportionality to q2 is the result of CVC valid in the vector-like theories, and
is an indication that in such theories the decoupling theorem holds.
The star quantities are functions of q2 and each of them has an infinite number
of observables as the functions of heavy particle masses appearing as the coeffi-
cients of the Taylor expansion in terms of q2. We, however, can argue that pos-
sible non-decoupling radiative corrections, oblique corrections, of heavy particles
are concentrated only in three parameters, called 5 , T, and [/-parameters (Peskin
and Takeuchi, 1990), which therefore make the analysis of New Physics simple and
transparent.
To see this, suppose we have a vacuum polarization function Uij(q2) where the
indices (i, j) take either SU(2) adjoint indices 1,2,3 or Q corresponding to unbroken
U ( l ) e m . At least as far as we retain in the 1-loop level, vertices of the diagrams
of vacuum polarizations are all proportional to dimensionless gauge couplings (at
higher loop levels we may have mass dependent couplings due to Higgs exchanges),
and we may rely on a naive dimensional analysis. Namely when we Taylor expand
it as the power series in q2 or equivalently in - ^ (M: a generic mass of intermediate
heavy particles),

Uijiq2) = n 4 j (0) + - j f2| , * = o Q2 + • • •, (8-47)

dq

we readily know that the order of magnitudes of Hij(0) and -j4 L |,2 = 0 are 0(M2)
and 0{q2), respectively, and the residual terms are at most of 0(*jfi-), which give
only decoupling effects. Thus as far as our focus is on the possible non-decoupling
effects we may treat only the first two terms of the expansion. Thus, at the first
sight, there seem to exist 8 (= 2 x 4) parameters, corresponding to the 4 choices
of the gauge indices (i,j): (1,1) = (2,2), (3,3), (3,<2), (Q,Q), or equivalently the
 Oblique Corrections and S, T, U Parameters 207

choices of gauge bosons in the external lines of the vacuum polarization diagrams
(W+,W~), (Z, Z), (7,7), (Z, 7). Actually, not all of these parameters exist. In
fact, I1QQ(0) = n 3 Q(0) = 0 as the result of CVC of U(l)em: we never have mass
renormalization to the photon as we wish the photon to travel at the speed of light.
We thus have 8 — 2 = 6 parameters remain. We also should aware of the fact that
some of the remaining 6 parameters should be understood as to be used for the
renormaliation of three bare quantities, g,g',v describing the gauge sector. This
is because the first two terms of the Taylor expansion correspond to the radiative
corrections to the operators with mass dimensions d = 2 and d = 4 (in "broken
phase" of the theory). Thus among 6 parameters, three parameters are used for
renormalization, i.e. in the process to fix the bare parameters in terms of a, GF, MZ
etc.. In other words, these are inputs rather than the prediction of the theory. In
this way, the outputs of the theory, i.e. the genuine predictions of the theory, are in
the remaining 6 — 3 = 3 parameters, which are nothing but the S,T, {/-parameters
we are interested in.
As is clear from the above argument, such three parameters S,T,U should be
automatically finite quantities, as they correspond to operators, which do not exist
in the original Lagrangian: they need not to be renormalized. Actually, such finite
combinations of poralization functions have already appeared in the discussion of
star-quantities, i.e. in the attempt to show the finiteness of the star-quantities.
Namely the parameters are defined as follows:

aS = 4e2[n^3(0)-n^(0)], (8.48)

aT [ r i l l ( 0 ) n 3 3 ( 0 ) ] (8 49)
= c^| - ' "
2
aU = 4e [n' u (0)-1133(0)], (8.50)

where we have used a notation IIJ(O) to denote -^-j*-|92=0, which are nearly equal
to Uij(q2)/q2 for the case of IIjj(O) = 0, which were originally denoted as IIJ -, once
we ignore 0{^fi-). We can confirm that S, T, U really do not suffer from the
UV-divergences. The finiteness of S-parameter is guaranteed by the fact that there
is no mixing term of SU(2)L and U(l)y field strengths, Fl^v B^v in the original La-
grangian. T and U should not get UV-divergence since whose existence contradicts
with the "custodial symmetry" of the terms £ L i F^ F^v and ^ - £ 3 = i A^A^
in the original Lagrangian: no couterterm can be prepared for the quantities which
violate the symmetry of the original Lagrangian.
We can easily check that, in the approximation that C( ^ 7 2 ) is negligible, the
star quantities are now describable in terms of 5, T, [/-parameters:
J_ 1_
(8.51)
e2 \-KOL ~~

sl-sm28w\z ~ (8.52)
cr — S J 4
208 PRECISION TEST OF ELECTROWEAK RADIATIVE CORRECTIONS ...

z
*~1 - 4hs> < 8 - 53)
Ml-Ml ~ 0, (8.54)

where e* and M* do not obtain quantum effects in this approximation, i.e. for
q2 <g. M2, since these quantum effects stem from 2nd or higher derivatives of Ily.
In addition, we can list up a related star-quantity p*(0) which is defined as the ratio
of the 4-Fermi coupling of low energy (q2 = 0) neutral current process to that of
charged current process:

Qff' = - ^ P*(0)(f^[hL - slQW-y^L - slQW). (8.55)

After some arithmetic we find

p*(0) - 1 ~ a T. (8.56)

Thus we know that the T-parameter (times a) is equivalent to Ap we had been

discussing.
As the example we consider the contribution of a pair of quarks (t, b) (or equiv-
alent^ that of (t',b'), the hypothetical quarks of 4th generation) to the S, T pa-
rameters. By use of (8.48), (8.49), (8.39)-( 8.42) and (8.36)-(8.38), we easily find
their contributions to be:
s
=^ - l O - <8-57)
Z7T 6 mi,
m 3 1 1 r 2 2 2m?m? , m?. .„ „„.
T [m? +m r ln L (8 58)
= i^^A4 ^T^f 4 -
where the formula for T is just A p / a , as we expected. We learn that in the limit
of global SU(2) "custodial" (or weak isospin) symmetry, i.e. mt = mj, T vanishes,
just as Ap does, but (the first term of) S remains: S = ^ .
This result is relevant especially for the technifermion contribution, since in
technicolor theory (Susskind, 1979; Weinberg, 1979), just as in QCD, the tech-
nifermion condensation, corresponding to the VEV of Higgs, preserves the custo-
dial symmetry: < T\jTy > — < TDTD >, which means that the resultant dynamical
masses due to technicolor interaction are degenerated, and each technifermion dou-
blet contributes to S-parameter additively. Thus the precision test on 5 parameter
is expected to yield a meaningful constraint on the theory. To be precise the above
calculation cannot be directly applied to the contributions of bound states such
as techni-hadrons, since technifermions are treated as free fermions in the above
analysis. Thus more sophisticated approach is desirable, though the above results
are enough for rough estimation. A sophisticated analysis, relying on the dispersion
relations of vacuum polarization functions of vector- and axial-vector types, whose
imaginary parts are measurable i?-ratios, provides
S = 0.4 (for 1 technifermion doublet)
 Global Symmetries 209

= 2.1 (for 1 "generation" techinicolor model), (8.59)

assuming that the number of the technicolor NTC = 4. Comparing with the upper
bound on S obtained from the precision measurements at LEP etc., Peskin and
Takeuchi were able to rule out the 1 "generation" technicolor model (Peskin and
Takeuchi, 1990).
The constraints on S, T, {/-parameters imposed by the precision experiments at
LEP etc. are not only used to get information on the new physics, but also quite
useful to constrain the masses of heavy particles in the standard model, i.e. mt and
TTIH • Especially, since the T-parameter (Ap) is quite sensitive to mf, it was possible
to predict rather precisely the value of mt as somewhere around 175(GeV), before
the direct discovery of the t quark at Fermi Lab.

8.4 Global Symmetries

We have seen that all of S, T, U parameters are all "calculable" quantities, free from
UV-divergence without any need of renormalization. ^From a bit different point of
view this is the consequences of global symmetries, which remain in the gauge sector
of the theory even after the SSB. To see this, let us take a matrix form of the Higgs
fields and introduce a global SU(2)n symmetry.
We start by writing the Yukawa coupling in the following form with the matrix
form $ of the Higgs field (ignoring its VEV for a while),

^•Yukawa = fu ( UL, dL ) f . « i ! + / i ! ( U£, dL ) f 0* ) d,R + fl.C.

= (*• * ) * ( o l){2)+k-c- <8 60)

-
where

which just corresponds to the linear-u model discussed in QCD to describe the low
energy effective theory, where H and Gl are replaced by a scalar and pseudo-scalar
bound states of quarks, a and 7rl (i = 1,2,3). The technicolor theory takes this
analogy seriously and regard the Higgs fields as the bound states of hypothetical
fermions with quantum numbers "technicolors", instead of ordinary colors, called
technifermions T. In particular H ~ fvTu + TDTD (for 1 doublet model of techni-
color), though the Arc, the mass scale where the technicolor interaction becomes
strong, is scaled up to 0(lTeV).
If the degeneracy mu = ma is realized, the theory is invariant under a global
210 PRECISION TEST OF ELECTROWEAK RADIATIVE CORRECTIONS ...

chiral symmetry SU(2)L x SU(2)R:

I) - (%)="{%)•&)<%)-«{%
$ -> $ ' = 9L * 9*R.
( 9L, 9R- elements of SU(2) i and SU(2) fl ) (8.62)

Let us note the squared absolute-value of the Higgs doublet is given in this matrix
form as Tr($t $) = H2+Gi2, which is clearly invariant under the SU(2)LxSU(2)R:
Tr($'t $') = Tr($+ $) : H'2 + G'i2 = H2 + Gi2. This means the real fields
H, G% behave as the fundamental representation of 5 0 ( 4 ) , which is equivalent to
SU(2) x SU(2), as is well-known in the group theory. Just as in QCD, after Higgs
develops its VEV v (H —> v + H), corresponding to non-vanishing < uu > = < dd >
in QCD, the chiral symmetry is spontaneously broken, but leaving its vector-like
subgroup SU(2)v generated by gi, = gR = g unbroken:

< 8 6 3 )
' " ( * * ) • '

This remaining vector-like symmetry is called "custodial symmetry" (Sikivie et al.,

1980), or we may just call it as (global) weak isospin symmetry. Let us note that
such custodial symmetry is valid only when the degeneracy of the quark masses
mu = rrid is realized, since

unless fu = fd- This is intuitively trivial as mu ^ rrid should break the weak isospin
symmetry.
The SU{2)L gauge bosons A1^ (i = 1,2,3) and U(l)y gauge boson B^ transform
under the chiral symmetry SU(2)i x SU(2)R and the custodial symmetry as

(4,4,4): (3,1) and 3 (8.65)

B": (1,3)+ (1,1) and 3 + 1 , (8.66)

where the representation of the U(1)Y gauge boson breaks into the two pieces
(1,3) and (1,1), since the weak-hypercharge may be written as y = I^R + ^y^S
where I$R is the eigenvalue of the 3rd component of "right-handed weak-isospin"
SU(2)R and B, L denote baryon and lepton numbers respectively. (Or we may
rewrite the Nakano-Nishijima-Gell-Mann relation as Q = I^i + I3R + ^ = ^ in
a left-right symmetric form, which really holds in the left-right symmetric gauge
theory SU(2)L X SU(2)R X U(1)B-L, though in the gauge group of the standard
model, we are taking now, the SU(2)R is not gauged.)
Paying attention to the fact that II33 — U'3Q = \TVZY denotes the mixing of
SU(2)L and U(l)y gauge bosons in the kinetic term, and that IIn — II33 = ifrin +
 Operator Analysis 211

II22] — II33 we realize the representation of the oblique parameters under the chiral
and custodial symmetries:

S: ( 3 , 3 ) + (3,1) 1+ 3 + 5 (8.67)
T,U: (5,1) 5. (8.68)

where in the representation of S, 1 + 5 comes from (3.3) while 3 comes from (3.1).
Let us recall the simple algebra in SU(2) group theory, 3 x 3 = 1 + 3 + 5, and
note that only 1 and 5 are relevant for us, since vacuum polarization functions
should be symmetric under the exchange of gauge indices. The linear combination
5 [II11 + II22] — II33 can be expressed by a traceless 3 x 3 matrix and should belong
to 5 representation of SU(2), while traceful combination I l n + II22 + II33 behave
as the singlet. It is now easy to understand that T, [/-parameters should vanish in
the limit of custodial symmetry: T, U behave as a non-singlet of SU(2)v, namely
as 5, and should vanish in the limit of exact SU(2)v symmetry, since in the all
orders of quantum corrections, quantities contradicting with the symmetry never
appear. In such sense, the symmetry handling the T, U or Ap is the custodial
symmetry. Strictly speaking, it may be more suitable to say that actually what
handle these parameters is the global SU(2)L rather than the custodial symmetry
(Inami-Lim-Yamada, 1992) simply because these parameters are concerned only
with SU(2)L- (In fact to get 5 representation we need quartic term of quark masses,
or equivalently an operator quartic in the Higgs field, as we will see in the following
subsection, while the quark mass matrix Mq = -j?di&g(fu, fd) has a piece of 3
under SU(2)v and quadratic term of quark mass is enough to get 5, if we invoke
to the custodial symmetry. This becomes clear in the operator analysis extended
in the next section.) Then, what symmetry handles the S-parameter ? This time
obviously custodial symmetry is not the one, since the part (3,3) of S contains a
singlet 1 under the SU(2)y, the constant term ^ in (8.57), which survives even
in the limit of degenerate doublet (mt = m^). We may say that the relevant
symmetry for the parameter S is just the chiral symmetry SU(2)L X SU(2)R, as
the S-parameter behaves as (3,3), therefore not as an invariant under the symmetry
(Inami, Lim and Yamada, 1992). In this way we can readily understand why each of
the technifermion doublet gives additive contribution to S: each of the condensation
< TjyTfy > = < TDTD > break the chiral symmetry spontaneously, thus contributing
to 5.

8.5 Operator Analysis

We have seen T, for instance, is calculable finite quantity, although the gauge bo-
son mass-squared seems to be the coefficient of relevant operators with d = 2,
W+W~ etc. As was discussed in section 8.1, in the renormalizable theory a calcu-
lable quantity should be a coefficient of some irrelevant operator. It is known that
212 PRECISION TEST OF ELECTROWEAK RADIATIVE CORRECTIONS ...

the T-parameter is described by the coefficient of the following irrelevant operator

(Problem 8.6)

0T = ( ^ DM <t>){<t>t D»cj>) - | (0t £>„£>" 0 ) ( ^ ) . (8.69)

Similarly, 5 parameter is proportional to the coefficient of another d = 6 operator,

Os = [tf {F;y) cfr] B»". (8.70)

It is easy to see that when the replacement § —> (0, -T=)' is made this operator
causes the mixing between the field strengths of SU(2)L and U(l)y, i.e. F^„ and

It is now obvious why the S, T, [/-parameters are all calculable finite quantities:
they are all described by the coefficients of higher dimensional d > 4 irrelevant
operators, such as those listed above, which do not exist in the original Lagrangian.
Thus the coefficients should be all finite as long as the theory is renormalizable. Now
the difference of decoupling and non-decoupling effects of heavy particles is also easy
to understand. The examples we discussed above tell us that when heavy particles'
contributions to physical observables are decoupling ones, there is a characteristic
feature that heavy particles get their masses through some new mass scales, such as
MSUSY, which are inevitably gauge invariant. In that case, the coefficients of the
irrelevant operators induced by a heavy particle will be suppressed by the inverse
powers of M, with M denoting a generic new large mass scale. For instance the
effective Lagrangian for the 5-parameter behave as

~ j ^ W (W>'M B»\ (8.71)

as is clear from dimensional analysis. This means the heavy particle contribution
is suppressed by the factor 1/M 2 . On the other hand, we have learned that when
heavy particles' contributions are non-decoupling ones, heavy particles get their
masses through "strong" couplings with Higgs, the origin of the masses being the
VEV of the Higgs (SSB), without demanding any new large mass scales. In this
case Mw is a unique mass scale and the coefficients of irrelevant operators are not
suppressed, and even enhanced by the "strong" couplings in some suitably chosen
cases. This is why such contributions are non-decoupling.
One defect of the operator analysis extended above is that we may put arbitrary
powers of a gauge invariant factor $$ to a given irrelevant operator, all of them
giving the same operator in the broken phase. When we are considering some
decoupling effects, it will not be a real problem, since the insertion of $<f> will make
the coefficients of the operators more suppressed by higher inverse powers of M,
such as 1/M4,1/M6, which are less important. In the case of non-decoupling effects,
however, such further suppression is not the (tf<f>)/v2 -» 0(1) after the
replacement of the Higgs by its VEV. This problem is evaded once we move to the
 Operator Analysis 213

non-linear realization of the Higgs field, as discussed in section 8.2:

U = eiS^, (8.72)
where the physical Higgs field H has been ignored, since it does not appear in the
processes we are interested in (at 1-loop level). In this non-linear representation we
do not have the problem of getting infinite number of operators for one observable,
since WU = / is a c-number. Analysis based on the non-linear realization shows
that potential non-decoupling effects in gauge boson 2 and 3 point functions are
given by 7 independent operators. The reason why we can treat both of the 2 and
3 point functions on an equal footing is that in non-Abelian gauge theory they are
mutually related. For instance, Os not only has gauge boson 2-point function, but
also 3-point function as well, since Fl has both linear and quadratic terms of gauge
bosons.

Problems
8.1 Calculate the diagrams in Fig.8.3 to get the formula for Ap in (8.12). Also
show that (8.12) vanishes for nit — nn,.

8.2 Show that (8.17) is approximated by (8.18) when m„ d <C MgUSY {m\ —
m
l + MsusY,m2j = ™2d + M2SUSY).

8.3 Verify the relation in (8.21) at the 1-loop level.

8.4 Prove that Z, defined in (8.25) is finite.

8.5 Verify the formula (8.57) by use of (8.40), (8.41) and (8.48).

8.6 Show that the T-parameter is given as the coefficient of the operator OT in
(8.69).
This page is intentionally left blank
 Chapter 9
FLAVOR PHYSICS AND C P
VIOLATION

9.1 The Interest in Flavor Physics and C P Violation

The gauge sector of the standard model (SM) is theoretically uniquely determined
by gauge principle. The Higgs sector, on the other hand, has many arbitrary pa-
rameters, whose values cannot be theoretically predicted. Especially so far we have
no definite idea about the origin of quark and lepton masses and generation or fla-
vor mixings, i.e. the origin of the Yukawa couplings of Higgs field. Issues related
with quark and lepton flavors, "flavor physics", is therefore very important clue, not
only to the the confirmation of the standard model, but also to the search for some
theories beyond the standard model, "New Physics", where the origin of flavors is
expected to be understood more deeply.
Among such flavor physics, Flavor Changing Neutral Current (FCNC) processes
are of special interest. They do not exist at tree or classical level in the standard
model, and are induced only at loop or quantum level. Therefore the observation of
such FCNC processes will provide us valuable information on the contributions of
all particles as the intermediate states. They are, therefore, very suitable, not only
to get information about the flavor mixings of relatively heavy t quark with lower
generations, but also to search for possible heavy particle effects of New Physics.
Namely the studies into the FCNC processes is important for the progress of particle
physics. The rates of such FCNC processes are known to be suppressed by both of
the higher order of the perturbation theory and the small mass difference or small
flavor mixings. Thus, the FCNC processes are often called rare processes.
In fact, historically the rare processes have played very important role in the
foundation of the particle physics. We may list up some of the most important
events, some of them have already been discussed in 4.6.2: the introduction of
c quark, in order to naturally suppress the rare processes of neutral kaon such as
K° <-> K° mixing, by GIM (Glashow, Iliopoulos and Maiani, 1970), the prediction of
the mass of such predicted c quark by Gaillard and Lee (Gaillard and Lee, 1974), the
introduction of the third generation to implement the observed CP violation in the
neutral kaon system by Kobayashi and Maskawa (Kobayashi and Maskawa, 1973),

215
216 FLAVOR PHYSICS AND CP VIOLATION

which embodies the standard model nowadays, the lower bound on mt imposed by
the data on B° <-> 5 ° mixings, obtained before the direct discovery of t quark, etc.
The CP-violating processes are also known to happen rather rarely in neutral
kaon system. In such sense, the physics of CP violation also provides valuable
information of the heavy t quark and/or heavy unknown particles of New Physics.
We should further note that the Kobayashi-Maskawa (KM) model of 3 generations
was originally devised as the theory to accommodate CP violation. In spite of the
remarkable success of the KM model, such as N„ = 3 (N„: the number of light
neutrinos) at LEP, yet there has been no final conclusive argument on the origin
of CP violation. So, the confirmation of the prediction on CP violating observables
of the model, such as CP asymmetries in B decays, is an urgent necessity to really
establish the standard model. We may even find some evidence of some New Physics
in the course of the investigation.
The FCNC processes and CP violation is, in principle, mutually independent. In
fact, in the model of spontaneous (or soft) CP violation (Lee, 1974; Weinberg, 1976),
the CP violation is attributed to the complex phase of the vacuum expectation value
(VEV) of the Higgs created by the suitably chosen scalar potential, and has no direct
relation with the flavor mixing. On the other hand, in the mechanism of KM, both
of quark mass matrix causing the flavor mixing and CP violation come from the
same Yukawa couplings. Therefore they are mutually very closely related. In fact
in KM model we will see below that the CP violation needs mass differences and
flavor mixing, (of course) in addition to the complex phase in the KM matrix. CP
violation, therefore, is observed in FCNC processes, such as K° -H- K° and B° «-> B°
mixings.
The purpose of this chapter is to discuss the rare processes due to FCNC and
the CP violation. We assume the theory we work on is the standard model, since to
provide the results in SM is quite useful even if we further investigate New Physics.
Furthermore, some of the formulae we derive are readily applicable for the class
of New Physics, where the Yukawa coupling of the Higgs has the same structure
as that of the standard model, e.g. the model with four or more generations,
minimal supersymmetric stadard model. In this chapter we will discuss FCNC rare
processes only in the quark sector. Once neutrinos become massive, the lepton flavor
violation, i.e. FCNC processes in lepton sector, also becomes physically meaningful.
We, however, will not deal with the lepton flavor violation here, since the neutrino
oscillation as the typical example of lepton flavor violation has been extensively
discussed in chapter 6.

9.2 Flavor Symmetry and F C N C Rare Processes

As was discussed in 4.6.2, GIM escaped FCNC processes at tree level, by introduc-
ing c quark. What GIM has proposed is that by introducing c quark, the weak
isospin of s quark became the same as that of d quark and even after the uni-
 Flavor Symmetry and FCNC Rare Processes 217

tary transformation to the mass eigenstates the neutral current is kept flavor- or
generation-diagonal. We may say that the essence of their idea is to make the
gauge interaction of each sector of fermions with definite electric charge and chiral-
ity universal, i.e. to make it invariant under global SU(ng) transformation, with ng
being the number of the generations (ng = 3 in KM model). If we call the gauge
interaction to connect the elements of doublet "vertical symmetry", this flavor (or
generation) symmetry may be called as "horizontal symmetry". By imposing such
global flavor symmetry, the FCNC, such as sj^d coupled with Z boson or photon are
strictly forbidden at the tree level. We call such flavor conservation in the neutral
currents as "natural flavor conservation". The concept of natural is used to denote
some property which is naturally guaranteed by some symmetry of the theory irre-
spectively of the tuning of the parameters of the theory. From such point of view,
for instance, small Higgs mass in the standard model is not natural (the hierarchy
problem), but it can be naturally small in the SUSY standard model, etc. Glashow
and Weinberg have summarized the conditions for the natural flavor conservation
in neutral currents coupled with gauge or Higgs bosons (Glashow-Weinberg, 1977):
1. Fermions with the same electric charge and chirality should belong to the same
representation of the gauge group.
2. Fermions with definite electric charge should couple with only one Higgs doublet.
The first condition is the condition for the neutral current gauge interactions do
not have FCNC. If the condition is met, a neutral current J^ ' generally has a form
in the base of weak-eigenstates

(WZ,U2L,- • • ,UlR,U-2R,.. .,diL,d-2L,' • -,dlR,d2R,. ..)

x diag(a, a,...,b,b,...,c,c,...,d,d,...)
x (UXL,U2L,- •• ,UIR,U2R,.. .,diL,d2L,-- -,diR,d2R,.. Y, (9.1)

and this form is invariant under the unitary transformations to the mass eigenstates,
since the transformation is among the fields with the same electric charge and chi-
rality, such as UIL, U2L, •. -, and therefore the unitary matrix for the transformation
takes the form of block-diagonal:

[/£diag(a, a,.. .)UL = diag(a, a , . . . ) , (9.2)

etc. The second condition is necessary for the Yukawa coupling to naturally con-
serve flavors. If, for instance, up-type quarks are allowed to couple with two Higgs
doublets <f>i,<f>2, their neutral Higgs fields generally have the following Yukawa cou-
plings,

^ L [ / § V i + /j?V°] «>fl + h.c (9.3)

One linear combination of the Higgs fields, ,\ 2 [vi y>\ + v2 tp®], ( < <p\ 2 >—
7/f)) has a flavor-diagonal Yukawa coupling with the mass eigenstates u, c , . . . in
accordance with the diagonalization of the mass matrix for up-type quarks, while
218 FLAVOR PHYSICS AND CP VIOLATION

w+

\ u,c,t I
£
s — • — - » -^ » -2—» d

Fig. 9.1

the Yukawa coupling of the orthogonal linear combination ,} [v2 <£? — ^i ip®]
V"l+,;2
causes the FCNC at the tree level. Let us note that in SUSY standard model,
because of the chirality (holomorphic property) of the super-potential, even though
we have two Higgs doublets, there exists a selection rule to forbid the simultaneous
Yukawa couplings of the two doublet Higgs.
Since the standard model satisfies the Glashow-Weinberg's condition, the FCNC
is forbidden at the tree level. The flavor symmetry SU(ng), however, is broken by
the quark mass differences in each of up- and down-type quark sectors, i.e. by
mu 5^ m c ^ mt etc. We may still have symmetries of each generation number, i.e.
a sub-symmetry U(l)ng, if there is no generation mixings, but we know that there
exist such mixings and the flavor symmetry is thus completely broken. We thus
expect that at the quantum or loop level of Feynman diagram the FCNC processes
are induced by charged current interactions, which have generation mixings via
KM matrix U. In Fig. 9.1, we have a diagram with W + -exchange, which induces
effective dsZ and ds-y vertices.
The amplitude of Fig. 9.1 is expected to vanish for an exact SU(3) symmetry,
rnu = mc — mt- In fact, we easily find this is the case, since in this case the
orthogonality of the KM-matrix J2% U*dUis = 0 works. Thus, to get FCNC we
need the violation of flavor symmetry due to both of non-degenerate quark masses
at up- (and down-) type quark sector, and flavor mixings. More intuitively we
may say that when the degeneracy mu = mc = mt holds there is no mean to
distinguish the (u,c,t) quarks. Thus we may treat (u',c',t') obtained by arbitrary
unitary transformation of (u, c, t) as mass eigenstates with an equal right. Thus it
is always possible to choose the unitary transformation of the left-handed up-type
quarks, so that it makes the KM-matrix unit matrix, thus making the generation
mixing meaningless. Suppose we have only two generations with a mixing angle 6C.
2 2 2_ 2
Then the above amplitude is proportional to U*dUcs " 1 ^ f i m " = cos#c sin#c m'Milv —
6 x 1 0 - 5 . We learn from this simple exercise that the amplitudes of FCNC processes
are suppressed by small mass-squared differences (the GIM mechanism). If we
include the third generation the relevant mass differences ml — m\ c are not small
 Rare Processes in Kaon System 219

compared with M ^ . But, on the other hand, the mixings of t with 1st or 2nd
generations are small compared with sin#c, thus making the amplitude rather small.
Thus, the rates of FCNC processes are tiny and these processes are often called "rare
processes". It is worth noticing that the FCNC processes in quark sector and the
neutrino oscillation in lepton sector are both induced by the mass differences of
different generations and generation mixing. This is not surprising, as the neutrino
oscillation is one of FCNC processes.

9.3 Rare Processes in Kaon System

As we have seen in the introductory argument in section 9.1, the rare processes in
neutral kaon system of {K°, K°) have played very important roles in the establish-
ment of the standard model. (K°, K°) are the bound states of s and d quarks and
their anti-quarks, K° ~ §75d, K° ~ J75S. There is a freedom of relative phase
of these two states, but here we adopt a convention CP \K°) = —\K°). We may
define a quantum number (a U{1) charge) "strangeness" S carried only by s and s
quarks, S = - 1 for s, S = 1 for s, S = 0 otherwise. The neutral kaons, K°, K°,
thus have S = —1, 1, respectively. In the FCNC processes of kaons, the strangeness
changes, and we may classify the rare processes by the extent of the change of the
strangeness:
(a) | A 5 | = 2: K ° H K°,
(b) | A 5 | = 1: KL -> up, K+ -»• TT+I/P,
where in the decay K+ —• -K+VV, the contributions of neutrinos of all generations
should be summed up, as long as they are light. We discuss these two types of
FCNC processes separately below.

(a) | A 5 | = 2 process
The CPT theorem tells us that the masses and the life time (total decay width)
of particle and its anti-particle should be exactly the same. Thus if only strong
interaction QCD is switched on, K° and K°, being anti-particles of one another, are
two degenerate states without any interaction (as the strong interaction preserves
the strangeness S). The situation is somehow similar to the case of double-well
potential with very high wall separating two degenerate vacuum states |1), |2) in
the quantum mechanics (see Fig. 9.2).
In the case of double-well potential, as the height of the barrier wall gets reduced
the tunneling effects connecting two states |1), |2) starts to operative. Then the
Hamiltonian in the base of these states takes a form

(9 4)
" = ( S %)• -
220 FLAVOR PHYSICS AND CP VIOLATION

V(x)

Fig. 9.2

Even if the tunneling effect Et is small, the energy eigenstates get large modification,

^|[|1>-|2>], ^ [ | 1 ) + |2>], (9.5)

with energy eigenvalues E0 — Et and E0 + Et.

Similarly, when weak interaction is switched on to the neutral kaon system,
the strangeness is no longer preserved and FCNC process K° «-> K° arises. The
Hamiltonian in the base of (K°, K°) at the rest frame now reads as

M M12
H = (9.6)
12 M
where M is the common mass implied by the CPT theorem and Mu is due to the
K° «-» K° mixing. For simplicity, here we have ignored absorptive part, denoted by
T, coming from the on-shell intermediate states, which correspond to the fact that
the neutral kaons are unstable states decaying into irn etc. Thus the Hamiltonian is
hermitian, though we will treat non-hermitian Hamiltonian including the absorptive
part later in this section. We further simplify by ignoring a small CP violating
effect in the element M\i due to the CP phase of KM matrix for a while, assuming
M,12 Mi2. Then just as in the case of quantum mechanics, we find two mass
(energy) eigenstates

\Ki) = ^= [\K°) ~ \K°)], l*2> = ^ = [\K°) + 1*°)], (9.7)

with the mass eigenvalues

mi = M - M12, m 2 = M + M i 2 . (9.8)

As we have ignored the possible CP violating effect, these states have definite CP
eigenvalues:

CP \K,) = \Ki), CP \K2) = -\K2). (9.9)

 Rare Processes in Kaon System 221

U, C, t u,c,t
S «—5 « T—•— d S «—5 « 1 « d

\w+ V
u,c,t < u,c,t
d —*—£ •*- <
~~*— s •

(a) (b)

u, c, t
s —«- -*— d •— d

<pi 'W
u,c,t
d -*- • —

(c) (d)

Fig. 9.3

Actually when we include the CP violating effect and the absorptive parts, real
eigenstates of the Hamiltonian are Kg and KL states with shorter and longer life
times, which are nearly equal to K\ and K2, respectively. Their mass difference

Am*: = mKL - mKs - 2Mi 2 , (9.10)

is an observable, and is known to be extremely small: ATUK = (3.552 ± 0.016) x
10~ 12 MeV.
The elementary process to cause the mixing at the quark level is the 1-loop
diagram, called "box diagram" (See Fig. 9.3).
In Fig. 9.3, tp^- are the would-be Nambu-Goldstone boson, which has a propaga-
tor in i?£ gauge (discussed in the gauge theories with spontaneous gauge symmetry
breaking), M 5 As we see below the effective Lagrangian due to the top
-

quark exchange in this box diagram has a term proportional to m\, which comes
from the diagram with </?+ and (p~ exchanges (at least for £ 7^ 0), which has 4
Yukawa couplings yielding a factor (jm ( /Miv) 4 . The factor is then multiplied by
1/wt 2 (for mt > Mw) coming from dimensional analysis, thus producing a net
effect ( p 4 m t ) / M ^ , a non-decoupling effect of heavy t quark. Note that this box
diagram is only possible quark diagram at 1-loop level; the diagrams with 1-loop
induced FCNC Z- or 7 (photon)-vertices, which are relevant for | A 5 | = 1 processes,
such as KL —> up,, changes the strangeness only by 1 unit and do no contribute to
the | A 5 | = 2 process K° *•> K°. In the pioneering calculation by Gaillard-Lee,
an assumption that the masses of intermediate up-type quarks are much smaller
than Mw was made, which is a good approximation in the 2 generation (4 quark)
model, m c <S Mw, which was the standard picture of that time. The calculation,
however, needs to be revised, since we now know that there are 3 generations of
quarks and mt is greater than the weak-scale Mw- The calculation which is valid
222 FLAVOR PHYSICS AND CP VIOLATION

for the arbitrary masses of intermediate up-type quarks was performed (Inami and
Lim, 1981) and the result can be represented in a form of effective 4-Fermi | AS\ = 2
operator (see also (8.14)):

4/?"2 = - f 47rsin2 g E (KVid)(V*sVjd) E{Xi,Xj) (s^Ld^Ld),

i,j=c,t
(9.H)
where the contributions of two internal quarks Ui, Uj (i,j = u,c,t) are contained
2
in the coefficient function (with x% = -rA-)
M
w

E(Xi,Xj) = -xixj{^\{\-\^-\ J ^-^)\nxi-{ X i ^xi)]

3 1
h (9-12)
4 (a* - 1 ) ^ - 1)
and L = ^-j^, Vis etc. are K-M matrix elements. We have replaced the original
coefficient function E(xi,Xj) by E{xi,Xj) = E(xi,Xj) — E(xu,Xj) — E(xi,xu) +
E(xu,xu), setting xu = -rrf = 0. The additional three terms exactly vanish by
m
w
the orthogonality of K-M matrix J2i V*sVid = 0, £ \ V£ V^ = 0. This is why the
summation of i, j is only over c and t. When both intermediate quarks coincide we
get
3
EM ss E(Xi,«) = - ^ ) In *, - «[\ - f ^ - I j ^ l (9.13)
Though the above result is valid for an arbitrary intermediate quark masses, the
external quark masses md, ms and their 4-momenta have been ignored there, which
is a good approximation. To get M 12 , we need to evaluate the matrix element of
the 4-Fermi effective operator with respect to kaon states,

(K^s^Ld^s^Ld^KO) = | ?Km2KB, (9.14)

where /K, mx are decay constant and the mass of if-meson (TTIK — M in (9.6)), and
the B is the "bag-parameter", which parameterizes the ambiguity due to the non-
perturbative QCD effects to form the bound states K° and K°, and the reference
value B = 1 corresponds to the vacuum saturation, (K°\(sjliLd)\0)(0\(s'y,iLd)\Ko).
Thus writing the off-diagonal element of the neutral kaon mass-squared matrix as

6m2 = -(K°\c[Aff=2\K0)
= ~%R \ f2Km2KB [(V*Vid)(V;sVjd) Eix^xj)}, (9.15)
V2 67rsin 9w
the mass-squared matrix reads as

{6m2 M2)> (9 16)

-
 Rare Processes in Kaon System 223

and taking the square-root we obtain the Hamiltonian at the rest frame, with an
approximation 8m2 <C M 2 ,
M
H^['^ Sm'2 J £ ), (9-17)
2M
s 2
which means M 12 ~ j ^ - . We thus obtain
„_ , , Re<5m2
ArriK - 2ReMi 2 —
M
f
- ~%K \ KmKB M(V*V )(V*sVjd))
id E{Xi,Xi)
V2 67rsin 8w
~ - 3 . 8 x 1 0 - 2 m K B M(V*sVid)(V*sVjd)} E{XUXJ). (9.18)

In the case of restricted 2 generation model, noting E(xc) ~ — xc for xc <C 1, we

get

AmK s °G*&mK (sin flccos fle)2 mg

6V27rsin2 6W M2,
which reduces to the result obtained by Gaillard and Lee. Comparing this with the
observed Amx they succeeded to predict the mass of c quark to be m c ~ 1.3(GeV),
before the discovery of the c quark. Unfortunately, in the full 3 generation model,
the generation mixings of t quark with lower generations have not been determined
precisely, thus making the prediction of mt not straightforward. In addition, though
the above result is suitable for perturbative contributions of heavier quarks c, t, there
may exist non-perturbative effects of intermediate u quark, which make our predic-
tion not conclusive. The CP violation observable e discussed later, however, being
proportional to Im(V*sVidV*sVjd), does not suffer from such problem, since the inter-
mediate u quark does not contribute, and the above formula is directly applicable.
Thus by use of e we can impose a useful constraint on the generation mixings and
CP phase, namely on the unitarity triangle discussed later in this chapter.

(b) | A 5 | = 1 processes
The typical | A 5 | = 1 processes are KL —> Up- and K+ —> TX+VV. (KL —> 7r°e + e _
is the signature of CP violation and is another interesting process to study. But
the derivation of its decay rate is obtained in a similar manner with the argument
below, and will not be given.)
Now the diagrams with 1-loop induced FCNC Z- or 7 (photon)-vertices also
contribute to the processes, and the quark Feynman diagrams, relevant for the
+
KL —• nfi and K -> TT+UU, are shown in Fig. 9.4, where ui denotes all possible
light neutrinos (I = e,/u,r for the standard model). Note that the 7-exchange
diagram obviously does not contribute to the decay K+ —> TT+UU. The 1-loop
induced FCNC Z- or 7 (photon)-vertices, denoted by blobs in Fig. 9.4, are the sum
of the sub-diagrams shown in Fig. 9.5.
224 FLAVOR PHYSICS AND CP VIOLATION

W+,ip+ W+,<p+
[ W W W r rvww\>—•- vi

U,C,t u,c,t •>r

w-,<p- w-,<p-
A* vi
(a)
•wvwvi
(b)

H,Vl

t*,Vl

Fig. 9.4

One may wonder why we should include the 1-particle reducible diagrams with
FCNC self-energies such as Fig. 9.5(a), and wonder if such FCNC self-energy should
be eliminated by adding a suitable counterterm for the external quarks d, s to have
physical meaning as the eigenstates of propagation. Such procedure is also pos-
sible, but it turns out that the contributions of all counterterms just cancel with
each other in the FCNC processes (Botella and Lim, 1986), intuitively because the
counterterms for the FCNC quark self energy and for the FCNC vertices of Z or 7
gauge bosons are not independent.
We first display the results for the 1-loop induced effective FCNC Z- and 7-
vertices (Inami and Lim, 1981):

£>sdZ 2
J2 (V*Vid)rz(xi)(s-L^dL) Z\ (9.20)
47TSin #iy COS0W ^

47rsin 2 0 w 2M 2 , £ (v*vid)
i=c,t
s[Fi(xi)(q2^ - q^i)L + F2{xi)atlviqv{msL + mdR)]d A", (9.21)

where the effective sd-y vertex is given in the momentum space (q^ = p£ — p^). The
coefficient functions are given in terms of the interemediate up-type quark masses
x, = ml/Myy (i = u, c, t) as

rl
r*(*i) = [\ In Xi + 7 ( 1 1 , O i (9.22)
4 8x~^llXi+8(xi-l)2
1 1 13 1 1 1
Fi(Xi) = Q{[ ] Xi
12 n - 1 + 12 (Xi - l ) 2
2 (xi - l ) 3
1 1 1 1 1
+ ri
3xi-l 3(xi-l)2 6 ( z i - l ) 3 + 2(x,
—] xMxi - ^H~)}
 Rare Processes in Kaon System 225

7 1 13 1 1 1
[ +
3xi-l 12(34-1)2 2{Xi-\f]Xi
1 1 35 1 5 1 1 1
l
6 Xi - 1 12 (a* - l ) 2 6 (Xi - l ) 3 +
2 (a* - l ) 4 J Xi
'
- 27(^,0, (9-23)
rw x ^rr 1 1 3 1 3 1 . 3 xl . ,
*i(*i) = - Q { [ - i - ^ + i I - _ ^ + -I-_^]a;i--I-_TFIna;i}

r1 1 9 1 3 1 . 3 a ; ? . ,,._,.

where the term ln(^-) in F\ (x^ has an infra-red singularity at the limit xu, mu —> 0,
and xu dependence has been kept only there. This term just corresponds to the
contribution of "penguin-diagram" with gluon-exchange (see Fig. 9.6) discussed
below (we may call the logarithmic contribution as the contribution of "electro-
weak penguin" diagram). In Fifi{xi), Q denotes the electric charge of the internal
up-type quarks, i.e. Q = §. We have left the factor Q free, so that we can clearly
distinguish the contributions of the different types of diagrams where the photon is
attached to the internal or external quarks, respectively. This makes, for instance,
the calculation of sdg (g : gluon) vertex, namely the calculation of the gluon-
penguin-diagram, really straightforward (Problem 9.2). The appearance of the £-
dependent term
x
f ,\ 1 ,3 1 1 1 , ,

X
4. r (5£ + l ) z , 2x-jx\

implies that the effective vertices are not gauge invariant, though the Pauli-term
accompanied by F2 is gauge invariant, since only this term contributes to the decay
s —> dj with the on-shell condition q2 = 0. The remaining gauge dependence is
cancelled when the contributions from box diagrams are also summed up, as we will
see below.
Let us note that the sdj vertex has two pieces, the terms with the currents
[s(<Z 2 7 M _ 9 M ^)^ a n d {sGnvd)q", called "charge radius" and "Pauli" terms, respec-
tively. These currents automatically vanish (without the use of equations of motion),
when multiplied by <?M, thus satisfying the current conservation of electromagnetic
current (CVC). Note that in the FCNC vertices the current conservation is not
satisfied by use of the equations of motion, since s and d have different masses, but
satisfied by these specific forms of the currents including the external momentum
q^. Thus the sdj vertex vanish in the limit of gM = 0 , and we need to calculate the
diagram keeping the 0(q2) terms, in contrast to the case of calculations of other
FCNC effective Lagrangians. This proportionality to q2 makes the non-decoupling
effect of heavy t quark rather mild in the sdj vertex, i.e. In (mt/Mw), while in
226 FLAVOR PHYSICS AND CP VIOLATION

w+
ipT

s S, u,c,t ^d d s d

Z,7 z, 7
(a) (b)

W+
w+
S "" ~" N
s s ,y(^d s s '' ^ d s d s

z,i Zn z,i Z,7

(c) (d) (e) (0
s d s d s d s d
+
\ w+ / Xw+<p /'
\ ^
>
/
V >
Z,J £,7 Zn
(g) (h) (i) (j)

Fig. 9.5

sdZ vertex we have an non-decoupling effect proportional to m2, just as in the box
diagram (see (9.22)-(9.24)). The reason is rather easy to see; in the sdZ vertex
the (p+-exchange gives the Yukawa coupling factor mt/Mw twice, thus giving the
non-decoupling effect proportional to m2, while in the sdj vertex, the additional
q2 factor enforces the remaining piece to have mass-dimension d = — 2, behaving as
1/mf, thus cancelling the m2 from the Yukawa couplings. In this way, when the
FCNC 7 vertex or similarly FCNC gluon vertex is relevant, the contributions from
the lighter u.c quarks become significant. Especially the penguin-diagram shown
in Fig. 9.6 with FCNC gluon vertex is expected to play an important role in the
explanation of "A7 = | rule" in K —> TTTT decays.
Write the effective 4-Fermi Lagrangian relevant for the decays of our interest as

4/f=1 = ^ 2f 47rsin
, . ° L E (v;avid)[C{xi)(sZ'rltdL)(Tii'f^)
v °w it?,t
3
- ££(xi,^)(^7MdL)(I^7'Vii)]) (9.26)

where yj = jjf- with m;;. denoting the charged lepton mass of the j-th generation.
 Rare Processes in Kaon System 227

w+

Fig. 9.6

The coefficient functions are the sum of the contributions from the box diagrams
and the Z-exchange diagrams in Fig. 9.4:

C(xi) = Ca(xi)+Cz(xi), D(xi,yj) = Dn(xi,yj) + Dz(xi). (9.27)

We readily know that Cz(xi) = Dz(xi) = Tz(xi). The contributions from the box
diagrams are known to be
d X%
Cn(xi) = lnxj — l(x,Q (9.28)
8 (xt - l ) 2 > Xi — 1

Dn(xi,yj) - fVi
8 Vj - Xi JJi
yj - 1
J.• r r Xi
Xi „fXjX{ **. \0
^2 , *^i
•"% ' •l -i i
x
+ 8\yj - Xi Xi - 1 ' Xi-V \xi\nxi
9 1
-•y(x,£). (9.29)
8 yj - 1 Xi - 1
Combining these two types of contributions we finally get gauge invariant coefficient
functions
1 3 Xi
C(xi) = -( ) lnx, + -xt - (9.30)
4 Xi - 1 ; X' 4 ^ - 1'

^ ^ = -8^^(^ri)ln^
1, • Xj
*fc j l l l j / j
81?/.,- -XiXi - V (xi-1)2
1 3. „ 1 . xt
+ -:Xi--(l +3 -) -. (9.31)
4 8 yj - 1 Xi - 1
Note that the piece with vector current fij^n does not contributes to the KL —>
^/i decay. This is because when gM coming from the hadronic matrix element
(0|si7pdL|.Kz,) = — !(0|s7M75(i|.ft'L) is multiplied to the leptonic current, the CVC
implies qli(p,'y^n) = 0. Thus the photon-exchange diagram in Fig. 9.4 does not
contribute to the process KL —» fiji, as the matter of fact.
228 FLAVOR PHYSICS AND CP VIOLATION

Fig. 9.7

We thus obtain formulae for the ratios of the branching ratios of two processes to
those of reference decay processes, in terms of the coefficient functions given above,
Br(KL -> w2) sd T(KL) a 2 [ R e ( £ i = C | t VSVidCfo))] 2 . .
Br(K+ - £„„) ~ T(K+) Uirsm2ew> V*s > ^ )
Br{K+->it+v9) a 2 ^ | £<=c,t V S ^ O * ^ ) ! 2
Br{K+ _> noe+„e) ^4,mbx29w> 2- (Ks)2 - &•*>)

where the subscript "sd" denotes the short distance (or perturbative) contribution
to the process, without including the two-photon process discussed below, and in
the estimation of KL —> fip, the difference of phase spaces of KL and K+ decays
have been ignored. In the case of KL -> up, there is an additional contribution from
FCNC two photon vertex, as shown in Fig. 9.7, whose presence makes the precise
prediction of the branching ratio difficult, though including the two photon process
the above formula gives a reasonable prediction, compatible with the observed value
Br(KL -» tip,) = (7.25 ± 0.16) x lCT 9 .
On the other hand, the branching ratio Br(K+ —> •K+VV) can be reliably esti-
mated, as the photon process does not contribute to this decay. When the values of
quark masses and generation mixing angles (or KM matrix elements) are substituted
and an approximation yj = 0 is used, i.e.,

D(Xi) = D(Xi,0) = I JO^-Xilnxi +X-l+ ! - ^ - , (9.34)

4 (Xi - iy 4 4 Xi — 1

the above formula gives Br(K+ ->• -K+VV) of the order of 0(1O - 1 0 ), which seems
to be quite consistent with the recent result of BNL experiment, which has claimed
the observation of the rare decay.
We learn from (9.7) that K° or K° is the mixture of Ki and K2, or more
precisely Ks and KL, with mass difference A m # due to the K° «-» K° mixing.
Hence, if we can prepare an initial state of K° or K°, it's existence probability
will oscillate in time due to the interference effects of matter waves of Ks and KL,
just as the neutrino oscillation discussed in chapter 6. A feature not shared by the
neutrino oscillation is that neutral kaons decay into TTTT etc., when they propagate.
Thus the time evolution Hamiltonian have (effective) imaginary or non-hermitian
part to effectively represent the decay. In order to accommodate the absorptive
 flare Processes in Kaon System 229

part and possible CP violating effects we correct (9.6) as

/ M-fr M12-ir12\
(9 35)
Ur2-§r12 M-fr J' -
where M, V, corresponding to the mass and the decay width of neutral mesons in the
absence of the mixing, are real, while M\2 and Ti2 may have imaginary parts due
to CP violation. Though this Hamiltonian is applicable to both of neutral kaon and
neutral B-meson system (Bd ~ b^d,Bd ~ d^b;Bs ~ b^s,Bs ~ S756), here we
will discuss the neutral B meson oscillation. (Here we discuss only (Bd, Bj) system,
and will simply use the abbreviation B° for Bd, unless otherwise mentioned.) One
reason for this is that in the neutral kaon system the lifetimes of KL and Ks are
too different for the interference to be operative, while in the B meson system the
difference of lifetimes due to Ti2 is very small.
The eigenstates of the Hamiltonian

|Bl) s
vw^{PlB0)~q{*0))' (9-36)

l*> = ^ T W ( P | B ° ) + 9|^0>)' (9 37)

-
where

1
\JM{2 - f r; 2
(9.38)
p
y/M12 - f r 12

are known to have eigenvalues

Ai,2 = mi,j - %- r i , 2 , (9.39)

AA = A2 - Ai = Am - %- A r
(Am = m 2 - m i , Ar = r 2 - r 1 ) , (9.40)

with

l
Am - i Ar = 2 ^ M 1 2 - ^r 1 2 yJMt2 - -T*12. (9.41)

The time evolution of a state ip(t) = c(t) \B°)+c(t) \B°) is governed by a Schrodinger
equation
230 FLAVOR PHYSICS AND CP VIOLATION

where the Hamiltonian H has been given in (9.35). The diagonalization of the
Hamiltonian

V\p\2 + \q\2 (q -P\ „ i ( P P W A I O\

2 2
2M V? P ) V\P\ + \Q\ \-i i) v° W'
enables us to solve the differential equation as
'<t)\ ( p p\fe-^ 0 \(± -£Wc(0)
iAat
w V-<? <?A o «- A£ £ Ae(o)
ff+(t) -|j.(t)Uc(0)
(9.44)
-*M*) ff+W J Ve(0)
where

g±(t) = |[e-^*±e-^«]

= ie-^'e-iTOl' [l±e-^*e-'Ami]. (9.45)

Let |2?(i)) and |B(t)) be the states with the initial conditions |2?(0)) = |B°),
|B(0)) = |B°). As |B°), e.g. corresponds to c(0) = 1,5(0) = 0 we easily find

|B(t)> = p+(t)|B0>-^_(t)|B°>, (9-46)

\B(t)) = g+(t)\B°)-Zg-(t)\B°). (9.47)

The "survival" and transition probabilities for a state starting as B° at time 0 to

be found as B° and B° at time t are easily found to be

P(B°->B°) = \(B°m))\2 = \9+(jt)\2^e-rtcos2{^t), (9.48)

P(B°->B°) = \(B°\B(t))\2 = | V W I 2 - | ^ | 2 e - r t s i n 2 ( ^ i ) , (9.49)

where the specific property of neutral B meson system A r <§; r ^ , Ti ~ T2 ~ T
has been used. Note that in neutral kaon system A r ~ T, which causes very
different life-times for the two eigenstates, KL, K$. Why does such difference in
two systems appear? In the B meson system, the generation mixings of t quark
with lower generations are not suppressed compared with the mixing with the 3rd
generation, |V^6V^d| ~ |V^,VCd| ~ IV^Vtdl, as we will see below. On the other hand
the, Am, A r , produced by the dispersive and absorptive parts of box diagrams,
M\2, r ^ , as shown in Fig. 9.8, get contributions from intermediate t and u,c
quarks, respectively.
In the case of absorptive part only u, c intermediate states are kinematically
allowed to be on-shell. In the case of dispersive part all of u, c, t contribute, in
principle. We, however, know that the coefficient function E(xi) of (9.13) grows
2
up as — | xt — —\jji- (non-decoupling effect) for Xi > 1 and the t quark gives the
 Rare Processes in Kaon System 231

U,C |

u,c I
b
(a) (b)

Fig. 9.8

dominant contribution. Thus the absorptive part is suppressed compared with the
dispersive part by 0(m\lm\) -C 1. This is the reason why A r is negligible in the B
system. The formulae (9.48), (9.49) remind us of those in neutrino oscillation (with
maximal mixing). One remarkable difference is that in the B° <-> B° oscillation,
the sum of the probabilities decrease as the time goes on, which is the consequence
of the loss of the unitarity due to the non-hermitian Hamiltonian. So, when we
study observables sensitive to the oscillation, such as CP asymmetries in B°, B°
decays, Am x r = ^- = x needs to be sizable, which is satisfied in B°, B\ system,
where x is known to be ~ 0.7. (For the B°, B° system, x is much larger, relatively
enhanced by \Vts/Vtd\2-)
We may define the ratio of the time-integrated transition and survival probabil-
ities
J~ P(B° ^ BQ)dt x-
~ /0°° P(J3° -> B°)dt ~ 2 + x 2 ' {
'
where a formula

f
Jo
e-rtcos(Ami)dt=^^-^ (a; = - p ) , (9.51)

has been used. By a reason stated below we have also set | 2 | = 1.

collider experiments, the ratio r can be measured by observing a ratio

— (~ 2r for small r and assuming CP invariance), (9.52)

Ni+i-
namely the ratio of the event rates of the same sign leptons to that of the opposite
sign leptons. In the collision B°, B° are pair created, but some of them may
experience oscillation, so that there appear a possibility that both of mesons become
identical, producing charged leptons as their decay products with the same sign, via
processes b -+ c +1~ + i>i or b —> c + l+ + i>i. Historically such experiments first gave
an useful information on Am, which in turn was used to impose the lower bound
on mj.
This mixing due to Am or the box diagram is a FCNC process of |A£?| = 2
with B denoting "bottom number". We may think of a CP violating or T violating
232 FLAVOR PHYSICS AND CP VIOLATION

observable in this | A J 5 | = 2 mixing process in mass matrix, sometimes called as

"indirect CP violation" (in contrast to "direct CP violation" in the decay amplitude
of B meson itself). Namely the event rate difference of the same sign di-lepton events
Wm+ ~ N,-i-1 will signal a difference P(B° -> B°) - P{B° -> B°), which is a clear
signal of CP or T violation. (9.46), (9.47) tells us that the difference is possible
only when

If J2 ^ If I" » # # ! • (9-53)
Unfortunately, this deviation of |*| from 1 is quite small, and to see the CP violating
effect is not plausible. The reason is the following. The key ingredient to understand
the situation is in (9.38). There we see that if the M i 2 and Ti2 have the same phase
factor, | | | = 1, even though £ itself has a large phase factor (, which produces
large CP asymmetries in B decays as we will see later). As we see in Fig. 9.8, the
absorptive part Ti2 is due to the contributions of u, c quarks. In the hypothetical
limit m u = m c , the orthogonality of the K-M matrix V*bVuli + V*bVcd = —V*bVtd
implies that M12, handled by t quark, and Ti2, handles by u,c quarks have the
same phase factor, leading to |£| = 1. We thus learn the deviation of | a | from 1
is strongly suppressed by a factor (m2 - m2u)lm2. The difference m2 — m2u was
compared with m2, since in the limit of Ti2/Mi2 —> 0, again |^| = 1 is realized.

9.4 C P Violation in Kobayashi-Maskawa Model

The 3 generation Kobayashi-Maskawa model was proposed to account for the ob-
served CP violation. As we have briefly discussed in the section 9.1, in this model
CP violation is closely related with FCNC processes. In other words, the CP vio-
lation needs the breaking of the global "horizontal" or generation symmetry ST/(3)
(ng = 3), which is also needed in FCNC. To see this, suppose mu = mc. Then,
51/(2) sub-group of SU(S) in the up-type quark sector, i.e. the symmetry between
u and c quarks becomes an exact symmetry. By use of this SU(2) symmetry, it
is always possible to perform a suitable unitary transformation, belonging to the
SU(2), so that Vud vanishes, while we can always make the 1st column and the 1st
row of the KM matrix real by suitable re-phasing of quark fields q —> el<i> q. Then,
it is easy to show (see Problem 9.4) that we can perform a re-phasing such that the
all matrix elements become real. We learn from this example that if there is a mass
degeneracy in either of up- or down-type quark sector, CP violation never happens
(except for the possible strong CP violation, which we ignore).
Thus there arises a natural question: what is the (necessary and sufficient)
condition for the CP violation in the Kobayashi-Maskawa model? In particular we
ask what is the re-phasing invariant quantity which characterizes the CP violation.
The importance of the re-phasing invariance should be stressed here. In (4.181), the
CP violating phase 6 appears in the matrix element of, e.g. Va,, while in (4.183) the
 CP Violation in Kobayashi-Maskawa Model 233

U | U

W+ <ww\x(^J d
W w+

G G*

Fig. 9.10

matrix element is real. This is a little puzzling, since if we adopt (4.181) it seems
a process with t -> b transition has CP violating effect, while if we adopt (4.183)
there seems to be no CP violation in the process. Actually the CP violating phase
of (4.181) can be eliminated by a suitable re-phasing of t,b. As the physics should
be the same, irrespectively of the re-phasing, this means it is meaningless to talk
about such re-phasing variant quantity Vtb to judge whether the CP symmetry is
violated or not, and we should seek a re-phasing invariant quantity, which is due
to the interference of different KM matrix elements. Let us now come back to the
condition for the CP violation. We first note that any observables related to quarks
are represented by the cuts of Feynman diagrams with closed fermion loops. In fact
if a transition amplitude A is represented by a diagram G, its complex conjugation
A* is represented by a diagram G* where the directions of the arrows of all fermion
lines are reversed, as the complex conjugation implies the charge conjugation. Thus
\A\2 = AA* is represented by combining G and G*, namely by the cut of closed
fermion loop (see Fig. 9.9 for the example of W+ —> ud decay).
In the language of weak-eigenstates, the origin of the CP violation is attributed
to the complex Yukawa couplings or mass matrices of up- and down-type quarks,
Mu, Md, inserted into the fermion line of the closed loop (see Fig. 9.10).
234 FLAVOR PHYSICS AND CP VIOLATION

Actually Mu, Md appear in the loop diagram in the combination of hermitian

matrices Hu = MUM*, Hd = MdMd, as the right handed quarks have no charged
weak current interaction. Thus if

Im TI[Pl{Hd)P[{Hu)P2{Hd)P!l{Hu) • ••P'n{Hu))} ± 0, (9.54)

for some monomials of Hu,d, P\{Hd), P[(HU) etc., it is the indication of CP
violation. In the case of n = 1, we easily find that lmTi[Pi(Hd)P{(Hu)} =
1
0, as {TrlP^H^PKHu)}}* = T r ^ i W ^ ) ] = Tr[P{(Hu)P1(Hd)} =
Tr[Pi(Hd)P[(Hu)]. The first non-zero imaginary part appear at n = 2:

Im Tv{HdHuH2dH2u\. (9.55)
It can be shown that there appear no more independent imaginary part (Gronau,
Kfir and Loewy, 1986). On the other hand some arithmetic shows that
Im Tr[HdHuHJH2} ^ 0 <-> det [Hu,Hd] ^ 0. Thus we find the condition for
the CP violation is
det [MuMl MdMd) ± 0. (9.56)

By use of the fact Hu = U^L dia(m£,m;?,mf) UUL etc. {UUL is the unitary matrix
used to diagonalize the up-type quark mass matrix), we can show that the above
condition is equivalent to (see Problem 9.5)
{ml - m\){m\ - m2t){m2c - m2)(m2d - m2s)(m2d - m2b)(m2s - m2b) x J ^ 0, (9.57)

where J is the so-called Jarlskog parameter (Jarlskog, 1985), defined by

J = |Im {yiaVjaVjpV*0)\ — sin20isin#2sin03COS0icos02cos03sin(5

~ A2X5T) = O(10-6)(i,j=u,c,t;a,l3 = d,s,b;i^j,a^p), (9.58)

where A, A, r) are parameters in Wolfenstein's parameterization of KM matrix (see

(4.184)). As we will see below, even though there seem to be 9 choices of the
combination of i,j, a, P, the imaginary part is unique, up to its sign, corresponding
to the fact that in 3 generation model there remains only one CP-violating phase.
We take J to be a positive value. As we anticipated, (9.57) clearly shows that if
there is a mass degeneracy in either of up- and down-type quark sector, such as
mu = mc or md = ms, CP violation never happens.
Let us also note that J is invariant under re-phasing of quark fields, Ui ->•
e^ Ui (ui = u,c,t), da -¥ e**" da (da = d,s,b). In fact under the re-phasing,
Via -> e^-^+^Via and J -> e^-^+^e^-^e^-^+Me^'-M J = J.
The invariance comes from the fact that we have been considering the closed loop
of fermion, where the creation and the annihilation of all quarks happen in pair.
The uniqueness of the J parameter is easily understood noting the orthogonality of
the unitary matrix V. For instance

Im (VidV*dVjsV*) = - I m [VidV*dVjdVid + VidV;dVjbV*b] = - I m (VidV*dVjbV*b).

(9.59)
 CP Violation in Kobayashi-Maskawa Model 235

W w+ Im

(a) (b)

Fig. 9.11

The origin of the J-parameter can be understood diagrammatically. Let us work

in mass-eigenstates of quarks. Then the origin of the CP violation is attributed to
the CP phase of KM matrix appearing in the vertices of charged currents. We may
first think of a diagram with two vertices of charged current where W+ and W~~
are attached, Fig. 9.11(a).
This diagram does not yield an imaginary part, since Im {ViaV*a) = 0.
Next candidate is the diagram given in Fig. 9.11(b), yielding a non-vanishing
imaginary part Im(ViQV^QVj/3VJ), which is nothing but the J parameter, up
to the possible sign difference. We can confirm that there appears no more
independent imaginary part, even if additional pairs of W+ and W~ are at-
tached; for instance, I m ^ ^ V t / k , ^ ) = -\Via\2Im(V*aVj0Vk*0Vka) -
\vk0\2im(viav*avj0v*0) (i^j^k^i, a^(3^1^a).
Suppose the i-th element of a-th and /?-th column vectors (a ^ /3) have relative
phase factor Vi a /|Vj a | = ±e** (V~i0/\Vi0\) or arg(Vi a VJ) = 4>, irrespectively of the
choice of i. In this case by a suitable re-phasing, e.g. da —> e - 1 ^ da, we can make
all of ViaV*p real, eliminating the relative phase. If suitable re-phasing for each of
Ui is further performed, both of a-th and /?-th column vectors can be made real,
leading to the reality of the third column vector (by a re-phasing) as well, implied
by the orthogonality of the column vectors. We thus conclude that, in this case,
there should not be any CP violating phenomena. Let us note that if these three
complex numbers ViaV*0 (i = 1,2,3) are drawn as three vectors in a complex plane,
every vectors are parallel or anti-parallel to each other. On the other hand the
unitarily of the K-M matrix J2i ViaV*0 = 0 means that the three vectors form a
closed triangle, though the two inner angles vanish, or the triangle is a closed one.
This argument suggests that the CP violation is closely related with the inner
angles of such triangles, called "unitarity triangles". Namely, three independent
orthogonality conditions 5 ^ ViaV*/j = 0 (a / ^) yield three independent unitarity
triangles, depending on the choice of the pair (a,/3), as shown in Fig. 9.12.
As we will see below, CP asymmetries, such as those discussed in the neutral
B meson system, correspond to sin 20, with 0 being a relevant inner angle of an
unitarity triangle. The reason of the necessity of the factor 2 in the argument 26
236 FLAVOR PHYSICS AND CP VIOLATION

vcavc}

Fig. 9.12

vcdv;b (~ A 3 ) vcdv;s (~ A)
(a) (b)

Fig. 9.13

is easily understood; when the triangle is closed the inner angles are either of 0,
j or 7r, which means sin 2fl = 0 for all inner angles. Let us note that now the
re-phasing invariance of CP violating observables are manifest; though the each
side of the unitarity triangle rotates under the re-phasing, the shape of the triangle
is manifestly invariant, so are the inner angles. The shape itself clearly changes
depending on the choice of (a,/3), while we know every CP violating observable is
described by an unique parameter, the Jarlskog parameter J. What happens is,
although we have several different triangles depending on (a, /?), their area S is
known to be unique and just the half of J:

J = |Im (ViaV*aVj0V;0)\ = \ViaVCp\\VjaV;0\ sin fly = 25, (9.60)

where fly is the angle between ViaV*0 and VjaV^ (fly < n). As long as the area S
is unique, the CP symmetry will get large in the system whose triangle has three
sizes with comparable lengths, as in the case of B meson system (a = b, (3 = d) we
will see later, while if only one of the sizes is very small the inner angles will be near
to 0 or | , and the CP asymmetry will get small, as is the case of K meson system
(a = s,P = d). See Fig. 9.13.
So far our argument has been focused on the CP phase in the KM matrix and the
resultant Jarlskog parameter. We have also seen that for CP to be really broken the
non-degeneracy of quarks masses in both of the up-type and down type sectors is
needed. But, actually we still need an additional condition for the CP violation: the
 CP Violation in the Neutral K System 237

presence of the absorptive part in the relevant transition amplitude (the cut in the
Feynman diagram). In fact, in the end of the section 9.3, we have already seen that
T 12 = 0 leads to a vanishing CP violating effect in B meson system. The necessity
of the absorptive part is seen from the following illustrative argument. Suppose the
transition amplitude of a process A gets contributions form two different Feynman
diagrams A = A\ + A2. We will decompose the each amplitude to the part which
possibly contains the CP phase 8 due to the weak interaction, and a phase factor
indicating the possible presence of the absorptive part due to the on-shell property
of the intermediate state: A1<2 = Awl>2 ei5l~2, A = Awl eiSl + Aw2 ei62. The
transition amplitude of the process between the CP-conjugated states is written as
A(cp) = A*, etSl + A^2 e'*2. Let us note the phases 8i>2 do not change their sings,
since they have dynamical origin and have nothing to do with the KM CP phase.
The CP asymmetry proportional to the difference of squared absolute values of A
and A(cp} now reads as
| A | 2 _ ^ ( C P ) |2 = 4 g i n {h _ h ) lm{A*wiAw2y (g.ei)

Here we clearly see that to get CP violation as a physical observable, in addition

to the CP phase contributing to \m.(A*wlAw2), the presence of the absorptive parts
denoted by 8it2 is needed. We also see the interference of multiple diagrams, such
as A*wlAw2, is inevitable to get the CP violating observable, as suggested by the
re-phasing invariance: the CP phase in a single diagram can be always rotated
away by a suitable re-phasing of an external quark field. Thus a tree diagram
alone never produces any CP violation. We will see below that the possible large
CP asymmetries in B system of 0(10%) is due to the interference of the decay
amplitudes of the B meson and the amplitude for B +* B mixing, namely the
interference between AB = 1 and AB = 2 amplitudes.

9.5 C P Violation in the Neutral K System

The CP violation had been observed only in the system of neutral kaon system
(K°, K°), although the B-factory experiments in KEK and SLAC, recently started,
have begun to release the data on the CP asymmetries in the decay processes of
neutral B meson system (B°,B°). In fact, historically the first confirmation of the
CP violation was made by the detection of the long-lived neutral kaon KL —> TTTT.
In this section we focus on the CP violation in the kaon system, as it is still playing
an important and complimentary roles in the study of CP violation.
As has been already mentioned above, the CP asymmetry in the neutral kaon
system turns out to be small. This comes from the fact |Vt«V^| <C l^usKTdl ~
|V-CSVC^| (\VtsVt*d\ = 0(A2X5), \VusV:d\ = 0(A)). The CP asymmetry in the kaon
system is thus expected to be 0(A2Xi) = 0(1O - 3 ), which has been confirmed
experimentally.
Our main concern is the aforementioned KL -> TTTT decay. In the kaon system,
238 FLAVOR PHYSICS AND CP VIOLATION

in clear contrast to the B meson system, the life times of KL and Ks are quite dif-
ferent, as suggested by the subscripts L and S, thus making to isolate the KL state
possible. The decay of the KL state is a clear signature of the CP violation. The
state KL is determined as an eigenstate of the 2 x 2 Hamiltonian of the (K°, K°)
system. If there is no CP violating effect in the Hamiltonian, the eigenstates should
be the eigenstates of CP transformation, i.e. KL = K2 (see (9.7), (9.9)). On
the other hand, the K2 state is odd under the CP transformation, while the final
state of the decay, the 7T7T system, is even under the transformation. This simple
argument shows the decay KL —> W7r never happens, provided the theory is CP
invariant. Therefore, its detection should be a clear signature of the CP violation
in the theory. This argument also suggests that actually such CP violation decay
may have two distinct sources of CP violation, i.e. "indirect CP violation" in the
Hamiltonian of the (K°,K°), and the "direct CP violation" in the dacay amplitude
K-i -> 7T7T. We will discuss these issues successively below.

"Indirect CP violation"
The indirect CP violation in the Hamiltonian of the (K°, K°) is due to the imag-
inary parts of the amplitude of | A 5 | = 2 mixing process K° «-> K°, denoted by Mi 2
and Ti2. As the expected CP asymmetry in the kaon system is very small, (9(10 - 3 ),
we expect ImMi2 <C ReMi2 etc., and the deviation of the ratio of coefficients of two
neutral kaon states qa/pK (see (9.36) - (9.38) from unity is small. Thus, writing
PK/QK — (1 — e)/(l + e), the two eigenstates of the Hamiltonian are re-written as

w
- 75™ l ( 1 + E ) l *° ) + < 1 - € ) l * 0 > 1
- 7mm[m+llKl)]
1
(9 62)
'
\KS) = . [(1 + e)\K°) - (1 - e)\K°)}

1
[\K!)+^K2)]. (9.63)
VW+W)
We learn that non-vanishing e indicates KL, KS are not pure eigenstates of CP, im-
plying CP symmetry is violated in the Hamiltonian. As lm(VcsV*d) = -Im(VtsV*d)
(see (4.183) and (4.184)) and the contributions of the intermediate quarks in the
box daigram behave as E(xc)/E(xt) ~ rnl/rrfi < 1 (see (9.13)), we conclude
Imri2 <S ImMi 2 . The experimental values tell us A r ^ = TL — Ts(— —Ts) =
2 r i 2 ~ -2AmK = -4Mi2, i.e. r i 2 ~ - 2 M i 2 . We thus get a relation from (9.38)

_ e t i ImMi2. ,„„,..
e ~ —= (— ). (9.64)

Let us note that e cannot be physical observable, since it is re-phasing variant:

under the U{\) global transformation concerning strangeness S, s -> e**s or K° ->
 CP Violation in the Neutral K System 239

W+

Fig. 9.14

e-i<t>Ko # o _> ei<t>K0 <ZJL(= I=f\ _> e-m ajc Or, for infinitesimal transformation
v
' ' PK 1+ C ' PK _
\(j>\ <S 1, e —> e — i</>. Thus, though Ree ~ | ( 1 — lyxfl) *s re-phasing invariant, Ime
is not re-phasing invariant.
The re-phasing invariant Ree has been measured by the observation of the
"charge asymmetry"
g = r(KL ->• it-l+vi) - T(KL -» TT+1-UI) ^ \pK\2 - \qK?
L
~ TiKL^Tt-l+vO + TiKL^ir+l-vt) \pK\ + \qK\2
2

~ 1 - 1 — | ~ 2Ree = (3.30 ± 0.12) x 1 0 - 3 , (9.65)

PK

which clearly vanishes if CP is conserved, as TT~1+V[ and n+l~i'i are CP conju-

gate states of one another. As we expected, the observed CP asymmetry is of
0(1O - 3 ). This charge asymmetry is equivalent to the non-orthogonality of KL,KS:
6L = (KL\KS).

Including "direct CP violation"

CP violating effect may "directly" appear in the amplitude of K°,K° —> mr
themselves. As a pion has an isospin 1 = 1, the final state TTTT has either / = 0 or 2
(1=1 contradicts with Bose statistics of scalar fields). Write the amplitudes of the
decay to these two possible isospin states as

A(K° -t 7T7r(J)) = eiSl Aj (I = 0,2), (9.66)

where Ai are due to the weak interaction, including possible CP phase, while 5i
denote the absorptive parts coming from the strong (final state) interactions. The
corresponding amplitudes for K° decays are obtained by replacing Ai by —A},
without changing the signs of 5/. The AQ gets a remarkable contribution from the
"penguin" contribution with gluon-exchange (see Fig. 9.6), which is expected to
be responsible for the relative enhancement of the 1 = 0 amplitude compared with
1 = 2 amplitude in the K -»• 7T7T decays, p | ^ n ~ 22, so called AI = | rule, while
the Ai gets the main contribution from the ordinary charged current process shown
in Fig. 9.14, though its CP violating imaginary part may be provided by "electro-
weak penguin" diagram, which is obtained by replacing the gluon by 7 or Z in the
penguin diagram.
The final state TTTT is either ir+n~ or 7r°7r°. Write the corresponding measures of
240 FLAVOR PHYSICS AND CP VIOLATION

CP violation as
A(KL ->• TT+TT~) .
v+- = -rnr z—k = e + e 9 67
-
A(KL -> 7r°7r°) „. ,

in terms of two parameters e and e', which denote the part common for two kinds
of decay and the difference of the amplitudes of the decays, respectively. By use of
a relation

l ? r +
0 = \jI \iric(I = 0)) + y ^ |7T7r(J = 2)) (9.69)

k°*r°> = _ y i | 7 r 7 r ( / = 0)) + y||7r7r(7 = 2)), (9.70)

and A(KL ->• irn)/A(Ks -> 7T7T) ~ (eA(Ki ->• 7T7T) + A(J£T2 -> 7r7r))/A(i<r1 -»• 7T7T) =
e + {A{K2 -> TTTT)/A(KI —> 7T7r)), we easily get the following approximate relation

, K ^ . n * , ^ - . . ^ , , ^ . ^ , , (,72)
where we have used the facts A2 is sufficiently smaller than Ao in both in their real
part ( A / = 1/2 rule) and in their imaginary part (electromagnetic penguin contri-
bution is suppressed compared to ordinary penguin contribution by the difference
of gauge couplings). As KL and IT are physical states, the decay amplitudes, and
therefore e, e', should be re-phasing invariant. In fact, neglecting A2,
A*
A(KL^nn) l - f f ^ _ 1, QK A*0
A(Ks-+irit) i + 2^45. ~ 2L p * V {
'
PK AO

As ^ —> e - 2 1 ^ 2iL
, and ilo —• e~%^Ao under the strangeness 1/(1) transformation,
A*
the combination SM--^- is re-phasing invariant. The re-phasing invariance of e' is
trivial.
In the KM model, T" 1 ^ 0 is less important compared with e, and e ~ e. Thus in
terms of the coefficient function E(x) obtained from the box diagram,

^-^O^l T^-WCTO^} .E(^), (9.74)

247rsin 6w ArriK Mfo
though additional factor coming from perturbative QCD correction of 0(1) should
also be multiplied. The experimental value (Particle Data Group, 2002)

|e| = (2.282 ± 0.017) x H T 3 , (9.75)

 CP Violation in the Neutral B System 241

is useful to restrict the shape of the unitarity triangle, or on (1 - p)r) (in terms
of Wolfenstein's parameterization) from the factor Im{(V^*Vtd)2}, since we now
know mt ^ 175GeV. The prediction of another CP violating variable e' is less
reliable, since we have two competing terms p " ^ 2 and pjffi- Though |ImA 2 | <£
|ImA 0 |, the AI = 1/2 rule relatively enhances the importance of ImA 2 , which gets
a contribution from the electro-weak penguin diagram. As the coefficient function
of the FCNC sdZ vertex, Tz(jk-) in (9.20) grows roughly in proportion to m% for
larger mt, the non-decoupling contribution of the intermediate heavy t quark tends
to cancel the effect of {"M0 coming from ordinary penguin diagram, thus making the
prediction of e' rather smaller than naively expected, i.e. |^-| = 0(1O - 4 ) to 0(1O - 3 ),
while the present experimental data (Particle Data Group, 2002),

Re ( - j « - = (1.8 ± 0.4) x 10~ 3 , (9.76)

seem to support this expectation. In accordance with our general argument, the CP
violating observables e, e' result from the interference effects of different amplitudes,
e.g. the interference of AQ and A^ in the case of e'. We may summarize this section
by classifying the CP violating observables depending on where they have their
origin, i.e. in the AS = 2 "indirect" CP violation in the mass matrix, or in the
AS = 1 "direct" CP violation in the decay amplitudes or in the interference of these
two types:
• pure AS = 2 CP violation: Re e (6L)
• interference between AS — 2 and AS = 1: Im e
• pure AS — 1 CP violation: e'.

9.6 CP Violation in the Neutral B System

In the KM model, as long as there are three independent unitarity triangles de-
pending on the choice of (a,/?) with a,/3 = d,s,b, there should be CP violating
phenomena, not only in the K system (a = d, /3 = s), but also in the neutral B
meson systems, Bd (a = d,/3 = b) and Bs (a = s,/3 = b). Hence, the observa-
tions of the CP violating events, in the rates predicted by the KM model will be
very useful for the final confirmation of the model, as the model for explaining CP
violation, or by chance we may even be able to find some deviation from the pre-
diction, which then will be an indication of some "new physics". In addition, as
was discussed in section 9.4, we expect large CP asymmetries, of 0(10%), in the
Bd system, stemming from the fact that the three sizes of the unitarity triangle for
(a = d, (3 = b) are all of comparable lengths, 0(AX3). These are main motivations
for currently on-going B-factory experiments at KEK and SLAC. In this section we
mainly discuss the Bd system, and we will simply use the abbreviation B° for Bd,
unless otherwise mentioned.
242 FLAVOR PHYSICS AND CP VIOLATION

Although the origin of the CP violation in the B system in KM model is the

same as in the K system, i.e. the violation of flavor symmetry and the CP violating
J-parameter, there are a few characteristic features of the B system listed below,
which requires some sophisticated observational method of the CP asymmetries in
the system, a bit different from that in the K system:
a. In the B system, since A r <C Am ~ Y ( r i 2 / M i 2 ~ ml/ml), the two eigenstates
of the Hamiltonin, B\, Bi defined in (9.36), (9.37), have almost the same life time.
Thus, in contrast to the case of kaon system, we cannot extract pure Bi state,
corresponding to K^: the observation of CP violation by use of the decays of B\
or £?2 is practically impossible, unless one of these states has been prepared as an
initial state.
b. In the collision experiments such as those in the B-factories, the initial states
should be either 5 ° or B°, the (coherent) admixtures of B\, B2 states, and we
expect the flavor oscillation B° ++ B°, as was discussed in 9.3, which is similar to
the neutrino oscillations, except the B mesons decay as the time goes by with the
almost equal life time r = 1/r. If the oscillation in the duration of the lifetime is
too rapid, i.e. if x = Am/T = Am r ^> 1, we in practice cannot distinguish B° and
B° states, and the CP asymmetries, sensitive to the difference of these two states,
will be washed out, no matter which initial state we prepare. Fortunately, the ob-
tained value of x from the observation of the same sign leptons in e+e~ collisions
(see (9.50)) is in a suitable range in order to avoid such washing out, x ~ 0.7.
c. Although we generally expect the CP asymmetry in the B system is quite large,
the pure AB = 2 CP asymmetry, oc P(B° -> B°) - P(B° -> B°), is much smaller
2

than the naive expectation, being suppressed by |^| — 1 ~ ^ (see the argument
below (9.53)). We also note the rates of "direct" CP violation in the pure AB = 1
decay amplitudes cannot be predicted with accuracy: the (partial) decay rate asym-
metries oc T(B° -> / ) — F(B° -> / ) (/, / : distinct final states) need the interfer-
ence among different amplitudes, which have absorptive parts with different phases,
whose estimation is not easy especially for those caused by strong interactions.
d. Thus the remaining candidate to see the sizable CP asymmetries with reli-
able predictions, is the CP asymmetries caused by the interference effect between
AB — 2 and AB = 1 amplitudes, corresponding to Im e in the K system. In fact,
we will see below the CP asymmetry is due to
qMB^Jl
Im{ (977)
pA(B0^f)^
which corresponds to — 21m e in K system (see (9.73)). Such type of CP asymmetries
is known to be of 0(10%), as was first pointed out by Sanda and collaborators
(Bigi and Sanda, 2000). Another nice thing of this type of CP asymmetry is the
possible phase due to strong interaction exactly cancelled out, as long as we choose
a final state / such that the decay amplitude A is dominated by one Feynman
diagram, as in the case of / = ipKs- In this type of CP violation B° o B°
 CP Violation in the Neutral B System 243

oscillation obviously plays a central role. In fact, the CP asymmetry is caused by the
interference of two distinct amplitudes, B°(B°) -» f and B°(B°) -> B°(B°) -> / ,
realized by the oscillation. (We may rewrite the CP violating factor Im ( | J g o Z ^ j )
as
u(Bo_»f)|2 Im ( | A(B° ->• / ) A(B° - • /)*).) Thus we need a non-vanishing Am
for the oscillation, and the observed value of x ~ 0.7 is in an ideal range,
e. Although the CP asymmetry of this type is expected to be quite large, this
in turn means the lengths of the three sizes of the relevant unitarity triangle are
tiny, 0(AX3). Namely the branching ratios of the decay processes to observe the
CP asymmetries are rather small, e.g. Br(B ->• ipKs) = 0(1O - 4 ). Thus, even
though the CP asymmetries are sizable, we still need high luminosity for the collision
experiments. Let us note that in the K system, Ks -> TTTT was the main decay mode.
Of course, the large CP asymmetries will help a lot in the statistical significance of
the signal: needed luminosity is proportional to 1/A2, instead of I/A, with A being
a generic CP asymmetry.
We are now ready to discuss the sizable CP asymmetry caused by the B° f+ B°
oscillation. For instance in e + e~ collision experiments, e+e~ -> B°B°, B+B~,
we a priori do not know which one of two neutral mesons is B°, which makes the
observation of CP asymmetry difficult. We start with the idealized situation, where
we know that at time 0 B° or B° was produced, and denote their state at time t
as B°(t) or B°(t) with B°(0) = B° or B°(0) = B°, which are linear combinations
of B° and B°, as shown in (9.46) and (9.47). Let us consider a CP asymmetry,
which is the difference of the probabilities, or event rates, for the state starting as
B° or B° to decay into the final state / or / at time t, divided by the sum of the
probabilities:

= P(fl°(t)->/)-P(fl»(t)->/)
AW j
~ P(B0(t) -+ f) + P{B°(t) -> / ) ' ^

where

P(B°(t)->/) oc \g+(t)A(B°^f)-^g-(t)A(B^^f)\2, (9.79)

P(B°(t)->/) <x \g+(t)A{B°^f)-2g-(t)A{B°^f)\2. (9.80)

As we have mentioned above, we can safely assume |^| = 1. We also take a CP

self-conjugate state as the final state, / = ± / , which has a merit to allow the both
processes B° —• / and B° —> / possible, and to eliminate the uncertainty in the
phases due to the strong interaction, "strong phases", in the ratio A\Bo^fl- As
the strong interaction preserves CP, generally the strong phases of A(B° -> / ) and
A(B° —> / ) are exactly the same, which means for the case of / = / the ratio is
free from the strong phase. The typical examples are / = ipKs, 7T7T. Under these
244 FLAVOR PHYSICS AND CP VIOLATION

assumption the time-dependent asymmetry A(t) may be written as

cos(Amt)[r(i? 0 -» / ) - T(BQ -> /)] + 2sin(Amf)r(g° -» /)Im (^gfejj)

(r(s° -> /) + r(s° -> /))
(9.81)
where, the partial decay widths are proportional to the squared-transition- ampli-
tudes:
T(B° -> / ) oc \A(B° -»• / ) | 2 , T(B° -> / ) oc |A(£° -> / ) | 2 . (9.82)

In the expression for the time-dependent CP asymmetry A(t), the term accompanied
by cos(Amt) does not go away at t = 0, and is the indication of the direct CP
violation in the AB = 1 decay amplitude, which does not need the help of the
flavor oscillation, while the term accompanied by sin(Ami) and Im ( £ ^ | B ° ~ > ^ )
vanishes at t — 0, and indicates the CP violation due to the interference between
A B = 2 and AB = 1 amplitudes, with large CP asymmetry, we are seeking. As
we have discussed above, we are interested in the process where the decay into /
is dominated by one Feynman diagram. In that case the direct CP violation is
negligible, i.e. T(B° - • / ) ~ T(B° - • / ) . In this way, the formula for the CP
asymmetry greatly simplified into

A[t) = sin(Ami) • Im ( ? ^ j | ^ ± i j ) . (9 .83)

It seems the time-integrated or time-averaged CP asymmetry vanishes, as the time-

average of sin(Amt) just goes away. This is the case provided T = 0, as in the
CP asymmetries in neutrino oscillations, The finite lifetime of B mesons, however,
makes the smearing of the CP asymmetry due to the oscillation incomplete. From
a relation,
/•oo /»oo
( / e'Tt sin (Ami) dt)/( / e'rt dt) = (9.84)
Jo Jo l + x2
we get the time-integrated asymmetry,
/0°° dt [P(B°(t) -> / ) - P(B°(t) -> /)]
A =
/0°° dt [P(B°(t) ->f) + P(B°(t) -»• /)]

= T T ^ I m ( M ( f l o _,/))• <9-85)
where the smearing factor j ^ — 0.5 is fortunately not small. Write the time-
independent CP asymmetry factor as a(f) = Im ( £ ^ f ° ^ ^ ) - As the three sizes
of the unitarity triangle are of comparable lengths, and the coefficient function
representing the contributions of itj = u,c,t to AB - 2 box diagram grows roughly
as ~ ml/Mw the Mi 2 , Ti 2 stemming form the box diagram are dominated by the
t quark contributions. Thus 1 ~ J jfc is written in terms of KM matrix elements
 CP Violation in the Neutral B System 245

related to t quark, (VtaVt*d)2. Thus for the most promising channel to observe the
CP asymmetry in the B-factory experiments,

a(7TKs) n I m c K f f a ) = - ^ ^ i , (9.86)
V
" td\

where ly'ffii = e i(</,1_,r) was used (see Fig. 9.13(a)). As we expected, the CP
asymmetry is described by the sin of an inner angle of the unitarity triangle in
(b,d) sector. In terms of Wolfenstein's parameters p and 77 of O(l) (see (4.184)),

and the \O,(T;KS)\ can be of order 10%. This large CP asymmetry has already been
observed at B-factories: Belle and BaBar experiments have reported \a(nKs)\ ~
70%.
One may wonder why the expression above is not re-phasing invariant: the phase
of {VtsVt*d)2 changes by the re-phasing. Actually when the factor J g o I ^ K s ) ~
v v" « — i/'tv \2wVv n is taken into account, we realize that the whole ex-
pression is re-phasing invariant, as it should be. Let us note that VcjVcs comes from
the decay b -» ccs due to the ordinary charged current process, while PK/QR —
(V*sVcd)/(VC8V*d) reflects that K° and K° are contained in Ks by the fractions px
and qK- As the matter of fact, however, V^Vcs is almost real and the CP violating
phase appearing in p*K/qK ~ 1 + 2e* is suppressed by |e| = 0(1O - 3 ). That is why
the CP asymmetry is dominated by the phase of \ydyi\-
So far we have assumed that we can prepare a neutral B meson with definite
flavor, i.e. B° or B°, at t = 0. Such identification is crucial when we discuss CP
asymmetry, since the interchange B° +-> B° at t = 0 will change the sign of CP
asymmetry. Unfortunately such identification of flavor is not possible in a realistic
situation of e+e~~ experiment, such as the B-factory experiments Belle and BaBar,
since B° and B° are produced in a pair. In addition the presence of B° -H B°
oscillation makes the identification difficult. One intriguing idea is to tag the flavor
of one of the mesons to be, e.g. B°, by the observation of its decay products
at some time t. As B mesons are produced at the collider experiment through
e+e~ —> T(45) -> B°B°, the relative orbital angular momentum I of two mesons
is 1, i.e. p-wave, whose wave function is antisymmetric under the exchange of two
mesons. Thus the remaining meson at time t should have opposite flavor, e.g. B°,
even though two mesons may have the same flavor at different times as the result
of the oscillation. If we regard the time of the tagging t as a sort of initial time we
are able to repeat the argument extended above for the idealized situation.
Let us illustrate the idea more concretely by explicit calculations. At the initial
time when B mesons are produced via the decay of T, the state of two mesons with
246 FLAVOR PHYSICS AND CP VIOLATION

momenta k and k' is described by

-j=[\B°(k))\B°(k')) + c\B°(k'))\S°(k))], (9.88)

where c = (—1)'. Though in the on-going B-factory experiment 1 = 1, c = - 1 as

explained, c = 1 is possible, e.g. in the process T(55) -> B°B°* -> B°B°j, where
I = 0 is possible. Let /„ a final state, which both of B° and B° can decay into,
i.e. fa = fa, and /& a final state, which only B° can decay into. Namely fi, can
be utilized for the tagging of B°. Let P(fa,ff,t,t')cdtdt' be the probability for the
mesons with momenta k and k' to decay into fa and / t during the time intervals
(t,t + dt) and (t',t' + dt'), respectively. Then,

P(fa,fb\t,t%
<x |-^=[< fa\B°(t) X fb\B°(t') > +c < fb\B°(t') >< fb\B°(t) >]| 2

oc T(B° -> / a ) r ( 5 ° -> / 6 ) e - r t +

{1 Im(
' ~ pl(g°^/J ) s i n ( A m ^ ) } ' (9 89)
-
where t± = i ± f (t+ > 0, - i + < t_ < t+). Similarly,

P(fa,h;t,t%
oc r ( B ° -> / o ) r ( B 0 -* / 6 ) e - r t +
{1 + I m (
' pl(B°^/I) ) s i n ( A m ^ ) } ' (9 90)
-
where a good approximations r ( B ° -» / a ) ~ r(B° ->• / a ) , and r(f?° -> /ft) ~
r ( B ° —> fb) have been used, which are valid for decay processes dominated by tree
level diagrams. The subscript =F correspond to c = =F1. At the time £' of tagging of
the meson with momentum k', another meson is known to have different flavor, and
if we regard i_ as the time t in the ideal situation it is expected that we have a CP
asymmetry with the same time dependence as what we got above for the idealizes
case. In fact, for the case of our main interest c = —1, after the integration over t+

j 0 0 P(fa,fb;t,t')-. dt+ « e - r l * - l { l - I m (gAA{,l°0^ffah sin(Ami_)}, (9.91)

J\t-\ pA(±S" -> Ja)
I" P(fa,ff,t,t% dt+ oc e- r '«-l{l + Im ( g ^ ° ~*fa)) sin(Ami-)},. (9.92)
J\t-\ pA{t5v -»• Ja)
As we expected if we replace t- by t we recover the previous results. One remarkable
difference from the previous case is that now £_ is not positive-definite, but may be
negative (t < t'). Thus if we further integrate the probabilities over i_, the term
proportional to the CP asymmetry Im (f JgoZw"!) w u l disappear. Thus we need
to identify the decay times t and t1. This is why the on-going B-factory experiments
 CP Violation in the Neutral B System 247

are performed by asymmetric collider where the energies of e + and e~ are different
and therefore the center of mass system is moving. Thus the difference of t and t1 is
measurable by identifying the corresponding decay points. If we denote the number
of events for the two mesons to decay into fa and fb at time t and t' with t > t',
etc. by N(fa,ft)t>t' etc., the time-integrated CP asymmetry A = J — J o-(fa) is
given as
A _ N(fa,fb)t>t' -N(fa,fb)t>t' - N(fa,fb)t<tl + N(fa, fo)t<t' ,~ „„\
[
N(fa, f b ) t > v + N(fa, f b ) t > t , + N(fa, }b)t<v + N(fa, f b ) t < t , " - '

Problems
9.1 Consider the Yukawa couplings of two Higgs doublets <f>i, fc, given in (9.3).
Show that a linear combination of their neutral components
—j===^{vi<p\ -\-v-i^) («i2 = (v?,2)) n a s flavor-diagonal Yukawa coupling (no
FCNC).

9.2 Utilize the effective FCNC 7-vertex given in (9.21) and (9.23), (9.24) to get
the effective FCNC gluon-vertex, i.e. the contribution of "penguin" diagram. The
comment below (9.24), concerning the Q-dependence of the effective 7-vertex may
be helpful.

9.3 Solve the time-evolution equation (9.42) to get the solution (9.44), (9.45).

9.4 Prove that when mu = mc, for instance, all the elements of the KM (Kobayashi-
Maskawa) matrix can be made real, by use of suitable re-phasing of quark fields.

9.5 Verify that the condition (9.56) is equivalent to (9.57).

This page is intentionally left blank
 Appendix A

Notation and Useful Relations

A.l Four vectors

Throughout this book, we use the natural unit h{=. /i/27r) = c = 1, where h is the
Planck constant and c is the light velocity. We follow the convention of Bjorken
and Drell (Bjorken and Drell, 1964).
Space-time coordinates x = (i, x, y, z) = (t, x) are described by the contravariant
4-vector

{x°,x ,x ,x ) = (t,x,y,z) (A.l)

The metric tensor is denned as

/ i 0
\
-l
g"" = 9^ (A.2)
-l
\o •i /

and produces the covariant vector

Xn = 9»vx" = (XQ,X\,X2,X3) = (t,-x,-y,-z) (A.3)

The repeated indices are summed unless otherwise specified. Then the scalar prod-
uct of space-time points is given as
2 z2
1^ = 9!" Qliv*& •£ — Q XpXy — Z X (A.4)

The four-momenta are similarly given as

P" = (P0,P\P2,P3) = (E,px,py,pz) . (A.5)

The four-gradient is defined as 9M = d/dx1* or 9M = d/dx^. Then the momentaum

operator in the coordinate representation is given as

^ = 0«=(if,-*v) (A.6)

249
250 Notation and Useful Relations

Thus,

^pM = _ 5 ^ M = - g + V2 = - D , (A.7)

where V 2 and • are called the Laplacian and the d'Alembertian operator, respec-
tively.

A.2 7 matrices

The Dirac 4 x 4 7-matrices satisfy the following anticommutation relations,

{^,Y} = 2g^I (A.8)

where / is the 4 x 4 unit matrix. (A.8) leads to the following relations:

7 M 7M = 4, (A.9)

7"7^ = -27* , (A.10)

Ylvlpln = ±9vP , (A.ll)
M
7 7^7P7<T7M = -llalplv • (A.12)

Define the chirality operator 75 as

5
7 = 75 = t7°7 1 7 2 7 3 - (A.13)
which satisfies

ll = 1 , (A.14)
and
{75,7"} = 0 . (A.15)

By using the Levi-Civita tensor which is totally antisymmetric, one can also write
75 as
i
(A.16)
4!'"'"
where
+1 for even permutations of 0,1,2,3,
pCT
W o - = -e"" = { - 1 for odd permutations of 0,1,2,3, (A.17)
0 otherwise.
The Levi-Civita tensor satisfies the useful relations:

-6${SW - 5$5»p) , (A.18)

a v
e ^ ea0pa = - 2 ( < ^ - 8*%), (A.19)
 7 matrices 251

ea^eaMp = -6^ (A.20)

a0l5
e ea^s = -24, (A.21)

where 8% is the Kronecker's delta,

The scalar product of a 7-matrix and a 4-vector is given by

7MaM =i = 7 0 a 0 - 7 • a (A.23)

and similarly

7MP" = h ^ =ty = ij°dt + ij • V . (A.24)

There are two kinds of useful representation of 7-matrices.

(1) 7°-diagonal representation:
o o
'-(!- ,)-'-(- ,;)—oo- <*-»
(2) 75-diagonal representation:

'-(Vo')-'-(^)-*-(S- 0 0- ^
where I is the 2 x 2 unit matrix and a is the 2 x 2 Pauli matrix

0 1 \ / 0 -» \ / 1 0
". = 1 n • "ff. == " n • CTz
" . == n " • (A-2?)
1 0 J' » I i 0 J' I 0 -1
The spin tensor is defined as

^ = ^(7 M 7" - W O = ^ [ 7 " , 7 l , (A.28)

which, in the 7°-digonal representation, leads to

(i,j,fc = 1,2,3)

where e'J'* is the totally antisymmetric tensor with e 123 = 1.

252 Notation and Useful Relations

A.3 Trace theorems

In calculating the transition matrix elements, the following trace theorems are use-
ful:

TrJ = 4, (A.30)
Tr 7 5 = 0, (A.31)
Tr(odd number of 7's) = 0, (A.32)
T±(W») = 4$„„ , (A.33)
Tr(7^7„7p7<r) = 4:{g„ugp<r - g„pgva + g^gvp) , (A.34)
Tt(757M) = 0, (A.35)
TV( 7 5 7M7^) = 0, (A.36)
Tf(7S7/.7„7„) = 0, (A.37)

Tr(757M7v7p7CT) = 4ieM„po. , (A.38)

The following relations are also useful:

TrWh'ffo) = 4 [pfrf + fi}# - g^(Pl • p,)] (A.39)

Tr [ y (1 - 75)^17"(1 - 76)jfe] = 2Ti{^i>ll"h) - Sie^^jn^p (A.40)

MYhYh^ii^hiuh) = 32 [fa • P 3 ) ( P 2 • P 4 ) + (PI • P4)(P2 •P3)] (A.4i)

Tr(7"A7"75^2)Tr(7 M ^37 v 75^4) = 32 [(Pl • p3)(p2 • p 4 ) - (pi • p 4 )(p 2 • p 3 )] (A.42)

Tr [7M(1 - 76)j*i7"(l - 7s)jfe] Tr [ 7 M (1 - 7 s ) M * ( l - TlOjM = 256( Pl • p3)(p2 • p 4 ),

(A.43)

A.4 Right- and left-handed Dirac spinors

The Dirac equation for a free fermion with mass m is given by

(i'fd,, - m)tp{x) = 0 , (A.44)

where ip(x) represents the 4 component spinor. It is convenient to define chiral

spinors (left- and right-handed spinors) as

IPR = ^ - ^ = P*V (P* = i ^ ) , (A.45)

 Dirac equation in the ^-diagonal representation: 253

where Pi and PR are projection operators of left-handed and right-handed spinors,

respectively, and satisfy

PL + PR = 1, PI = PL, PR = PR,
PI = PL, PR = PR, PLPR = PRPL =0. (A.46)

Note that
an
IhipR = V>fl d TsV'i = ~i>L • (A-47)

Thus, tpL and ipR are eigenvectors of the chirality operator 75. Define the Dirac
adjoint spinor i/> = ^ 7 ° . Then we have the following important identities:

i>PR = $L , i>PL = 4>R , (A.48)

1
$ 7 ^ = tpL^lpL + •4>Rl' 1pR , (A.49)
4>l"l5ip = ^Rj^ipR - rpLl^rpL , (A.50)
"0-0 = V'flV'L + V'LV'fl • (A.51)

A.5 Dirac e q u a t i o n in t h e 75-diagonal r e p r e s e n t a t i o n :

When we write the 4-component Dirac spinor in terms of 2-component Weyl spinors
X and 77 as

we have the relations in the 75-diagonal representation

PRIP = i>R = ( * ) , P L V = i k = ( ° ) , (A.53)

where x and 17 are unrelated 2-component spinors. In the 75-diagonal representation,

the Dirac equation is written as

U-V*"*v")(:)-(:)-. <-»
which results in the following simultaneous equation

(-i-Ki - i<? • V)x - rrvq = 0 , (A.55a)

(-i^-+ia-V)r]-mx = 0. (A.55b)
254 Notation and Useful Relations

If m = 0, we have two independent Weyl equations,

.8
-dtX = -icr-Vx, (A.56a)

i-^-r] — ia-Vr). (A.56b)

Assume the plane-wave,

X(x) = xip)e~ipx , (A.57a)

ipx
r)(x) = fj(p)e- . (A.57b)

Then we have

EX = +V-PX- (A.58a)
Erj = -a-pq , (A.58b)

For positive energy solutions with E = \p\ > 0, we have

a •p
X = +x, (A.59a)
\P\
aV
' f] - -rj . (A.59b)

Defining the helicity operator by

l
h='-4=
1^1
M-
2 |p|
(A.60)
Then we can write

% = +2X, (A.61a)

/IT? = -J17 , (A.61b)

2
an<
that is, helicity of x i ^ is + § and — | , respectively.
If we use the 4-component notation for spin as s = I ' _ . 1, we have
V 0 0-/27

hip = s-pip= I Q ' >'

(7 • ft/2 J \ 77

2 I n
0 -11 )MU„ )/ == 2'oTsV' , (A.62)

where p = i§r. Therefore, helicity /i is the same as chirality 75 for massless fermions
with positive energy. For negative-energy fermions(which correspond to antiparti-
cles), helicity h has the opposite sign to chirality 75. If a fermion is massive (m ^ 0),
 Dirac equation in the ^-diagonal representation: 255

X and r] are still eigenstates of chirality, i.e.

but they are no longer eigenstates of helicity h because of the non-zero mass term
in the Dirac equation.

a • px = +Ex + mfj . (A.64a)

a • pf\ = -Ef\ — mx , (A.64b)
But for high energy particles with E ^> m, we can approximate ^ ~ Tgf, and
then we have

X + (A 65a)
hx = + 2 ^ ' -
T 1 m , . „_, .
hn = ~2T,~2EX- (A.65b)
By using the eigenstates oih, x' and r/, as defined as hx' = + | x ' and hr)' = —1»/,
respectively, one can write them as
771
X « x' + ^ V , (A.66a)
777

i; « r/--X'. (A.66b)
That is, x and 77 are given by the mixture of both helicity states where contribution
of wrong sign helicity is order of m/E.
This page is intentionally left blank
 Appendix B
Cross sections and Feynman rule

B.l Cross sections

The scattering cross section or the transition probability for particle reactions is
described by the S matrix as

S = l+iT (B.l)

where T, being called the transition matrix, denotes the transition between initial
i and final / states.
First, consider a 2-body collision, a(pa) + b(pt) —• c(pc) + d{pd), as an example,
where the momentum of each particle is given in parentheses. The transition matrix
element for this process can be written as

Tfi = (27r)4S4(Pa +Pb-Pc- Pd)NaNbNcNdMfi. (B.2)

For the time being, we consider the case of spinless particles. Extention to more
general case including particles with spins is straightforward as described later. In
(B.2), all dynamics for this transition are included in M;i which is Lorentz invariant
and called the invariant amplitude. N is a normalization factor which depends on
the convention of the wave function normalization. Here we take the convention
for normalizing 2E particles in a volume V, i.e. N = -k= by defining the free field
wave function as

ftx) = ^ = 6 - * * . (B.3)

which leads to / pdV = 2E with p = i ((f>*d0(t> - d0<t>*4>) = ^-. Substituting N =

-j= for all particles involved into (B.2), we can write

Tfi = (2n)i6\pa +Pb -Pc -Pd)±Mfi, (B.4)

The transition probability per unit time and unit volume for this collision is

257
258 Cross sections and Feynman rule

given by

IT/-12
Wfi = LM_, (B.5)

where T is the time interval of the interaction. By using the relation

[8\Q)]2 -+ 6\Q)j~jdixei^(Q = 0) = <J4(Q)j^j, (B.6)

Wfi can be written as

Wfi = ^-64(Pa+pb-pc-pd)\Mfi\2. (B.7)

From (B.7), we can obtain the cross section a by multipling it by the number of
available final states and dividing it by the incident flux F. As is well known, the
number of states in which a particle in a volume V has its momentum between
P = {Px,Py,Pz) and p + dp= {px + dpx,py + dpy,pz + dpz) is ^ f . Since we are
normalizing IE particles in V, the number of states per one particle is i2ix\32E •
Then, the number of available final states for the process a + b -> c + d is i2„)32B '
(2TT)*2E • ^he m c m e n t flux F can be calculated as follows; if we consider the process
in the laboratory frame, the number of the incident particle a passing through unit
area per unit time in this collision is ^p^-l^ml) where Vin is the velocity of the
incident particle a, while the number of the target particle b per unit volume is ^ - .
Thus, the incident flux for the collision of a and b is F = \v%n\^h^v^- Then we can
obtain the cross section for this process as follows;

1 ,4,
da(a + b -> c + d) = T ^ v ^ (Pa + Pb - Pc - Pd)
d3pc d3pd
\M(a + b^c + d)\\ (B.8)
2EC 2Ed

The normalization volume V disappeared from (B.8). This is natural because V

is not a physical parameter. Therefore, hereafter we set V = 1 which corresponds
to take the wave function normalization as N = 1. This is the reason why we can
assign 1 to the external line of the spinless particle in Feynman diagrams as shown
below.
In the general collinear collision of a and b, Vin (incoming velocity of a) in the
incident flux F = \vin\4EaEb in the laboratory frame is replaced by t; re i(relative
velocity of a and b). F can be rewritten into the Lorentz invariant form as

F = \vrei\4EaEb = 4\va - vb\EaEb

= mpa\Ea + \pb\Eb) = ^(pa-Pb? - mlml,. (B.9)
 Decay width 259

Now, the extension of this formula to the n particle production processes, a+b -+
1 + 2H + n, is straightforward,
da(a + b -+ 1 + 2 + • • • + n)
1 1 ,4

One can easily generalize these formulas to processes including particles with spin
by replacing |M/i| 2 in (B.8) and (B.10) by X^ gpin l-^/*l2> which means that the spin
degree of freedom must be summed for the final states and averaged for the initial
states if the spin of initial particles is not polarized. In addition, for spin | fermions
the normalization convention for the wave function,

u ( s ) (p)u ( s ) (p) = 2m, © w (p)w w (p) = - 2 m , (B.ll)

and also the energy projection operator,

A+(p) = 5>«(p)uW(p)=^ + m>

s

A_(p) = -5^« (,) (p)«W(p) = -ji + m, (B.12)

8

must be used in these cross section formulas.

When we write the cross section formula (B.10) as

da(a + b-> 1 + 2 +•••+n) = l^L-dflt"), (B.13)

F
dRW is called the Lorentz invariant phase space which is given by

dH<»> = (2TT)464 (Pa +Pb- Pl -pj pn)

d 3 pi d3p2 d3Pn (B.14)
3
(2TT) 3 2£ 1 (27r) 2£ 2 (2n)32En'

B.2 Decay width

The decay width of a particle can be calculated similarly; the decay width formula
for a particle with mass M decaying into n particles in its rest frame is given by

dT{M -> 1 + 2 + • • • + n)
= -i
2M I T
(27T)(3"-4) \Mfi\264(Pa +Pb-Pl~P2 Pn)
2M (2TT)( 3 "- 4 ) '
d 3 Pi d 3 p 2 d?pn (B.15)
2E\ 2E2 2En
260 Cross sections and Feynman rule

B.3 Feynman rule

The invariant amplitude Mfi for an specific process is given by the sum of Feynman
diagrams corresponding to its process. They are usually composed of three kinds
of factors, i.e. external lines, propagators (internal lines) and vertices, which are
described below.

Feynman rule for QED

(l)External lines
• spin=0 bosons: For both incoming and outgoing bosons, assign 1.

• s p i n = | fermions: For an incoming fermion or anti-fermion, assign u^(p) or

v^(p), respectively. For an outgoing fermion or anti-fermion, assign u^ )(p') or
v(" )(p'), respectively.

• spin=l photon: For an incoming photon, assign ej, (k). For an outgoing photon,
assign e£ (&')

(2)Propagators
• Boson with spin=0:

_ i
p2 — m2

• Fermion with spin=^:

_ i _ . ^+m
$—m p2 — m2

• Photon with spin=l:

where £ is the gauge parameter. In the Feynman gauge, f = 1.

(3)Vertices
• spin=0 boson-photon vertex:
 Feynman rule

-ieip + p')^
(for charge +e)

2ie2g„,v

s p i n = | fermion-photon vertex:

(for charge +e)

Feynman rule for QCD

(l)External lines
- spin— 2- quarks: For an incoming quark or anti-quark, assign u^(p) or v^(p),
respectively. For an outgoing quark or anti-quark, assign u^s \p') or v^s ^(p1), re
spectively. Though quarks have the flavor and color degree of freedom, it is not
written explicitly in these spinors.

• spin=l gluon: For an incoming gluon, assign e), ( & ) T - ^ o r a n o u t g ° m g glu°n>

assign e'J- ' (&') ^r- Here 4f- (i = 1,2, • • •, 8) represent the color degree of freedom.

(2)Propagators
• Quark with spin=^:

. p" + m
rfi -
m m"

• Gluon with spin=l:

262 Cross sections and Feynman rule

^TBTnrBTnjTnmrc^ -£(-.- +a-o*>

where £ is the gauge parameter. In the Feynman gauge, £ = 1.

(3)Vertices
• Quark-gluon vertex:

-*s«y7^

• Gluon-gluon vertex:

fi, fci, i ,A, /C3, h

-9sfijk[g^(ki ~ k2)x + 9v\{k2 - k3)p

+9\n(k3 - kx)u]

[i, ki, i

-i9s[fijmfktm(9»\9vp ~ 9p.p9v\)
+fitmfjkm(9nv9\p ~ 9p.\9up)
v,k2,j X,k3,k +fikmfljm{9np9v\ — 9p.v9\p))

Feynman rule for the G W S model of electroweak interactions

(l)External lines
• spin=0 Higgs bosons: For both incoming and outgoing Higgs bosons, assign 1.

• s p i n = | leptons and quarks: For an incoming fermion or anti-fermion, assign

u^(p) or v^(p), respectively. For an outgoing fermion or anti-fermion, assign
v,(s \p') or v^" '(p')> respectively. Electroweak interactions are not affected by color
charges.
 Feynman rule 263

• spin=l gauge bosons: For an incoming gauge boson, assign ej, '(k). For an out-
going gauge boson, assign e*f} (&')

(2) Propagators
• Higgs boson with spin=0:

p2_M2

• Lepton and quark with spin=|:

i _ . ft + m
•p — m p2 — m2

Gauge boson with spin=l:

V-Ml\9 +
Ml )

(3) Vertices
• Charged current vertex:

._9_ 1-75
V2^ 2

• 9 a 1-75
 264 Cross sections and Feynman rule

U,C

qFi
^sin6,c7"~T^
— is for s-u-W vertex
+ is for d-c-W vertex
W,n

• Neutral current vertex:

—le
sin 6w cos 6w M ('i1-- 7 5 , j 1 + 75

Z°,fi

where

a{ = — - + sin 2 #w, a^ = sin2^vv! for/ = e",/i",T"

1 2 2
a{ = - - - s i n 2 ^ , a£ = - - s i n 2 0 w , f o r / = u,c,t

a{ = - - + - s i n 2 ^ w , a£ = -sin 2 0M/, for f = d,s,b.

and for massless neutrinos, a£ = | , and a^ = 0.

• Gauge boson vertex:

1. Three point interaction vertex:
7 W + W ' - vertex:

W+,ku^ W-,k2,\

ie[gv\{k\ - k2)^ + g\li(k2 - fc7)I/

7 , Ky, fJ,

^ory+xy- vertex:
 Feynman rule 265

W+,kuv. w-,k2,\

iecotOw[gvx{h - fc2)M + g\li{k2 - fo)v

+9»v(k3 ~h)\]

2. Four point interaction vertex:

l,a.

-ie2(2gapgllv - ga^g0V - gau9pti)

W+,fi W~,v

7>«.

-ie2 cot 6w(2ga09iiu - 9a^9&v - 9av9$^

W+,M W~,v

Z , a.

-ie2 cot 2 6w(2ga0glil/ - gaiigpv - gav9i3ij.)

W+,ii W~sv

W+,a w-,P
26
. 2 a {I9na9v0 ~ 9»&9av ~ 9iiv9ap)
sin Vw
W+,n W~,v

• Higgs boson (H) vertex:

1. HW+W- vertex:
266 Cross sections and Feynman rule
+
W ,kuv. W+,k2,X

ie
-Mwgv)i
sin# w

2. HZ°Z° vertex:

Z°,kuu Z°,k2,X
^ r

lie
Mz9v\
sin 20w

3. Fermion-Higgs Yukawa coupling vertex:

ie m,f
2 sin 6w Mw

4. Higgs boson self-coupling vertex:

H^ M

3H2e
2Mw cos 9w

5. Four point interaction vertex including Higgs boson:

HHW+W- coupling:

ff. M
ie2
9lLV
2 sin 8\v

W+,fi W,v
 Feynman rule 267

HHZ°Z° coupling:

^ ^ . ' ' lie2

2 Q»v
sin 26w

HHHH coupling:

H^ JI
^^ ,-''' 3i/x2e2

It'' "^H
This page is intentionally left blank
 Appendix C
Basics of the group theory

The Lagrangian is the fundamental object describing the dynamics of the physical
system and is to be constructed so that the interactions are invariant under certain
symmetry transformations. The requirement of invariance under symmetry trans-
formations is a guiding principle for the construction of interacting field theories.
Here we give a little mathematical basics of the symmetry, i.e. the group theory, in
particular, an SU(n) group and its algebra.

C.l Group and representation

A group G is defined as a set of elements {a, b,c,- • •} with the follwing conditions,

(1) If a and b are in G, a • b is also in G.

(2) There is an associativity law, a • (b • c) = (a • b) • c, for any elements in G.
(3) There exists a unit element e in G, which satisfies a • e = e • a = a for any
a in G.
(4) There exists an inverse element a - 1 in G for any a in G, such that a-a-1 =
a - 1 • a = e.

Among many examples of groups, the n dimensional orthogonal group (0(n)) and
special unitary group (SU(n)) are interesting in particle physics. 0(n) is the one in
which the length r 2 = x\ + x\ H 1- x\ in the n dimensional real space is invariant
under any rotation in this real space. U{n) is the one in which the length defined
by s 2 = |a;i| 2 + \x%\2 + 1- \xn\2 in the n dimensional complex space is invariant
under any rotation in this complex space and in addition, when the determinant of
the transformation matrix U is 1, i.e. detU = 1, it is called the SU(n) group.
In case of a • b = b • a for any elements a, b in G, the group is called the Abelian
or commutative group, while in case of a • b ^ b • a, it is called the non-Abelian or
non-commutative group. One-dimensional translation, two-dimensional rotation
(0(2)) or phase transformation (U(l)) etc. belong to the Abelian group, while
three-dimensional rotation (0(3)) or special unitary transformation (SU(n) with

269
270 Basics of the group theory

n>2) etc. belong to the non-Abelian group. Specially interesting group in particle
physics is the Lie group in which the group element is the analytic function of the
continuous parameter.
Let us assume that there is a matrix or operator M{a) for any elements a in G.
Then, if M(a) satisfys the conditions, M(a)M(b) — M(ab), M(a~l) — M'1(a) and
M ( l ) = 1, then M(a) is called representation of the group G.

C.2 SU(n) group and Lie algebra

We are interested in the SU(n) group defined in the n dimensional complex space,
which is expressed by an n x n unitary matrix with its determinant being equal to
1

Ui = U~1, dettf = 1. (C.l)

Then, how many parametrs are there in U1 As is well known, an n x n complex ma-
trix has 2n 2 real parameters. But there are n 2 constraints among these parameters
because of the unitarity relation WU = 1 and moreover, there is one additional
constraint due to the condition of detC/ = 1. Therefore, the number of the real
parameters of the matrix U is 2n 2 — (n 2 + 1) = n2 — 1. For example, the unitary
matrix for SU(2) and SU(3) has 3 and 8 parameters, respectively. Using these
parameters 6l (i = 1,2, • • •,n 2 — 1), the nxn unitary matrix U for SU(n) is written
by

U{9) = e-ieiLi = e~ilL (C.2)

where Ll are the nxn matrices called the generators for the SU(n) group. Here
the summation over i is implied. Since U is unitary, L1 are hermitian matrices. In
addition, L% is traceless due to detC/ = 1. Explicit expressions of Ll, for example,
for SU(2) are given by the 2 x 2 Pauli spin matrices ^- (i = 1,2,3) or, equivalently,
2 x 2 isospin matrices y (i = 1,2,3) with

- = ( : ; ) . ---(!-.')• M 5 - 0 ' <c-3»

and for SU(3), they are given by the 3 x 3 Gell-Mann matrices ^ (i = 1,2, • • •, 8)
with
/ 0 1 0 \ / 0 -t 0 \ / 1 0 0 \
1 2
A = 1 0 0 , A = I i 0 0 , A3 = 0 -1 0 ,
\ 0 0 0 / \ 0 0 0 / \ 0 0 0 /
/ 0 0 1 \ / 0 0 -t \ / 0 0 0 \
A4 = 0 0 0 I , A5 = I 0 0 0 , A 6
= 0 0 1 ,
\ l 0 0 / \ * 0 0 / \ 0 1 0 /
 SU(n) group and Lie algebra 271

A7 = 0 0 -i , A8 = -== 0 1 0 . (C.4)

As explicitly shown above, the SU{2) group has one diagonal matrix, ^-, and
the SU(3) group has two diagonal matrices, ^- and ^-. The number of diagonal
matrices is equal to the rank of the group. The SU(ri) group has n — 1 diagonal
generators and is a group of rank n — 1.
From the group property of U, we can see that these generators satisfy the
following algebra which is called the Lie algebra

[Li,V]=ifijkLk, (z,j,fc = l , 2 , - - - , n 2 - l ) (C.5)

where fijk, being antisymmetric constants for exchange of any pairs of i, j and
k, are called the structure constants of the group. Here the summation over k is
implied. This statement can be proved as follows; let us start with the foil wing
relation for non-commuting operators A and B,
A . eB = eA+B+i[A,B]+ ( C 6 )
e

Now let A and B to be given by A = ialL' = ia-L and B = ifilLl = i/3-L with n 2 - 1
parameters a1 and p*, respectively, where the summation over i is implied. Then,
since U(a) = eA and £/(/?) = eB are the elements of SU(n), U(a) • U(0) = eA • eB
is also the element of SU(n) because of the group property. Hence, eA • eB must
be also written by ec with C = ij • L. Therefore, the commutator [Ll, V\ must be
linearly related to the sum of the generators as described in (C.5). The structure
constants faf. are determined from the associativity law of the group.
The Jacobi identity

[V, [Lj,Lk]} + [V, [LkM]} + [L\ [L\V]] = 0, (C.7)

with the Lie algebra (C.5) leads to the relation

fjktfitm + fkitfjlm + fijtfktm = 0. (C8)

Then, if we identify the matrices T' as (Tl) k = -ifijk, (C.8) results in

[Ti,Ti]=ifijkTk, (C.9)

which is just the same algebra as (C.5). That is to say, the generators being just
equal to the structure constants of the group satisfy the Lie algebra. In other
words, the structure constants themselves generate a representation of the Lie alge-
bra. The representation generated by the structure constants is called the "adjoint
representation". The dimention of the adjoint representation is just the number of
generators, which is n 2 — 1 for SU(n).
272 Basics of the group theory

C.3 Representation in SU(n) group

Let us consider a field <j> composed of n complex fields ipa (a— 1,2, • • •, n) and write
it by a column vector,

f2

<t> = (CIO)

\<Pn J
If <>
/ changes into <p' under the SU(n) transformation as

<j> -> <f>' = Ucf>, or <pa-*<Pa = Uba<pb = Uabipb, (C.ll)

ifa is said to be in the "fundamental representation" of SU(n) and denoted by n.
Now, let us consider an infinitesimal transformation
U{6) „-iSL 1-iO-L, for 0 < 1. (C.12)
Then, we can write the field transformation as

fa - Va. ~ Stpa, (C.13)

with

6ipa = i(d- L)ba(pb - itbaipb. (C.14)

Next, let us introduce the conjugate fundamental representation n*. If we take

the complex conjugate of (C.ll) and use the relation (U*)ba = (U*)ab = {U*T)ba =
(W)ba = (U^)l, where W is the hermite conjugate of U, then one can obtain the
transformation of the congujate states y>* as

<pl-*rt = {u*)baV>i = <pUut)i. (C.15)

Now, it is convenient to introduce the upper indices to these conjugate fields as
ifa = <p*a. Then we can rewrite (C.15) as

V" V'° = <Pb(U^ab, (C.16)

The representation for those fields ipa is called the "conjugate fundamental repre-
sentation" and denoted by n*. Since the infinitesimal transformation associated
with U*(0) is 17*(0) ~ 1 + ifl'L" = 1 - iO^-L"), the conjugate representation
<p = y?* are generated by — IS*. Therefore, all eigenvectors for generators Ll of
a

diagonal matrices are still eigenvectors of — L%* but of opposite sign. Regarding this
property, the SU(2) group is specially interesting because it has only one diagonal
matrix Ij. Then for SU(2), we can expect that the 2 and 2* representation is
equivalent. In fact, this is the case as shown in the following. Let us look for a
 Representation in SU(n) group 273

matrix A so that A<j> transforms, like <j> in (C.ll), as A<j>* -¥ (A(f>*)' = U{A(f>*)
or 4>* -» 0'* = (A-lUA)4>* = £/*<£*. Thus, we must have A~XUA = U*, that
is, A~X^A = -^- (i = 1,2,3) since U ~ 1 - iQ% for SU(2). This can be
possible, when we choose A = ir2 = I J. Actually, we can see that

**• - (-°i I) ( $ ) = ( v ) tanst°™ "ke ( £ ) • •* r = 2 -

Now, let us consider the direct product of n and n* representations.

<Pa ® x" = (VaX6 - £ * W ) + ^JVcX6- (C17)

By taking the trace of (C.17), we find Ti(ipa ® X6) = ¥>aX°- Hence the 1st term of
the right-hand side of (C.17) is the traceless matrix with n 2 - 1 components,

r
« = V-* 6 - i * W , (C18)
n
and denoted by n 2 - 1. The 2nd term is just the trace term having only one
component,

Sba = -ShcX0, (C19)

n
and is denoted by 1. From (C.ll) and (C.16), we see that the transformation law
of T and S under the SU(n) transformation is given as follows;

Tb -+ Tb = U^mXnUlb - ~6baU^mXnUnc, (C.20)

and

Sba -+ S'ab = UbaU?VmXnUt<. (C.21)

Using the unitarity condition U™U^ — S™ where the summation over c is implied,
these transformation law can be written in the matrix form as

T - • I " = UTU\ (C.22)

and

S->S' = S. (C.23)

As described above, the matrix Tb is an n x n traceless matrix with n2—1 components

and thus it can be expanded by the n2—1 traceless generators Ll (i — 1,2, • • •, n2—1)
of the SU(n) group as

I* = At{Li)ta, (i = l , 2 , - - - , n 2 - l ) (C.24)
274 Basics of the group theory

where A1 are the parameters of the SU(n) group. Here the summation over i
is implied. Now, let us consider the transformation law of A1 under the SU(n)
transformation. Under an infinitesimal transformation

<Pa^<Pa = <Pa-i£<Pb, if" ~+<p'a = ¥>" - i<p" 6%, (C.25)

the transformation law (C.22) leads to

T -> T' = T - i[e,T], {e = ajLj, T = AkLk) (C.26)

which can be written as
AiLi -> AHU = A{V - iaj[Lj, Lk]Ak
= ^^-ioPiifjuL^A11. (C.27)
That is,
A'i = Ai + fijkajAk, (C.28)
1
or by identifying —ifjik = (T^)ik, which is the adjoint representation, A transform
as
8A* = fijkaiAk = -i(-ifjik)a^Ak
= -i(Ti)ikajAk = -i(f-a)ikAk. (C.29)
In these equations the summation over the same subscripts is implied.
In summary, the T* which transform according to (C.22) or (C.26) and are given
by (C.24), have n 2 - 1 components and are called the "adjoint representation" of
the SU(n) group denoted by n 2 - 1, while 5* is invariant under the SU(n) trans-
formation, that is, it transforms as a singlet and denoted by 1. We can decompose
the product of n and n* representation into n 2 — 1 and 1,

n ® n* = n 2 - 1 © 1. (C.30)
 Appendix D
C, P and T transformation

In addition to the continuous transformations, there are several important discrete

symmetries in physics. Here we concentrate our attention on space inversion (P),
time inversion (T) and charge conjugation (C). In view of the field-theoretical model
building, it is important to study whether or not the physical system is invariant
under these transformations, and if not, how and in what degree these symmetries
are violated.

D.l Parity or space inversion P

A parity inversion is the reflection of a spatial plane. This is equivalent to a space

refelection followed by a rotation with an angle n about an axis perpendicular to
that plane after the space reflection. Thus a parity operation is equivalent to a
space inversion. A parity operation is defined as

P : x -> x1 = -x, t->t' = t. (D.l)

Invariance of a physical system under parity transformation means that any inter-
actions in the system should be left-right symmetric.
Here we just list up the transformation rule of fields under parity operation. For
simplicity, we fix the phase factor to be r\p — 1.

Scalar field : <j>(x,t) -> (/>p(f ,t') = <j>{-x,t), (D.2)

p
Pseudoscalar field : r](x,t) -> T? (f ,t') = -r](—x,t), (D-3)

Dirac field • { ^ ^ "* ^ ^ ' ^ = W ( - * . * ) , m 4^

Dn-ac field . \ ^ ( f > t ) _, ^ ( ^ ^ = ^ ( _ f | t ) l b > (D-4)
Vector field : V{x,t) -»• V""(x1 ,t') = V M (-f,«). (D.5)
Furthermore, the bilinear covariants (S(scalar), P(pseudoscalar), V(vector),
A(axial-vector), T(tensor)) composed of Dirac fields transform as

S : i>a{x,t)il)b(x,t) ->• tpa(-x,t)xpb(-x,t), (D.6)

275
276 C, P and T transformation

P M2,t)rM2>t) -> -M-2,tft*ih(-3,t), (D.7)

V 4>a{x,t)-fipb(2, t) - • 4>a(-x,t)^rpb(-x,t), (D.8)
A & ( f , t ) j 5 ^ ^ b ( x , t) -> -$a{-x,t)j5y„xpb(-x,t), (D.9)
l ,
T ipa(x,t)a " ipb(x,t) - • ^a(-x,t)a^tpb(-x,t). (D.10)

An important example of parity transformation is the one for the weak left-handed
current denned by J^ ' = $aL(z',t)jli.il)bL{x,t), where the left- and right-handed
fermion fields transform under parity inversion as

1pL,R.{x,t) -> {ljJ_P)L,R{x,t) = 70ll>R,L(-X,t),

V>Z,,flO?,t) -> {lp^P)L,R{x,t)
. _ , * * -=. * _1pR,L(-X,t)j
. , _ * * wQ- (D"n)

Then, the weak left-handed current J^ ' transforms as

J^ = ,ipaL(x,t)'yliijjbL(x,t) -> 4>aR{-x,t)j0'yli-yoil>bR(-x,t)

= TpaR(-X,t)j»lPbR{-X,t), (D.12)

where we used the relation, 707^70 = 7^ = 7M-

D.2 Time inversion T

A time inversion is a reflection of time as defined as

T: x ->• f = f, t^t' = -t. (D.13)

Here we just list up the transformation rule of fields under time inversion. For
simplicity, we fix the phase to be T\T — 1.

Scalar field : <£(£, t) -> (/>T(f, *') = <£(£, - * ) , (D.14)

T
Pseudoscalar field : 77(2, £) -> ri {x',t') = -t]{x, -t), (D.15)

{DA6)
Dxrac field . { ^ t ) ^frpj^^fr-tW,
Vector field : V{x,t) -» V " T ( f ,t') = VM(f, - t ) . (D.17)

One should not be confused at the superscript T to Dirac fields in (D.16); the
superscript T on the left-hand side means the time-reversed field, while the one on
the right-hand side does the transposed field.
Furthermore, the bilinear covariants composed of Dirac fields transform as

S 1pa(x,t)lpb(x,t) ->• 1pa(x,-t)i>b(x,-t), (D.18)

5 5
P i>a(x,t)j Mx,t) -)• - & ( * , -t)j Mx, -t), (D.19)
V ^a{x,t)-f^b{x,t) -> $a(x, -t)-y^b{x, -t), (D.20)
A 5
& ( £ , i ) 7 V ^ ( £ , *) -> & ( * , -t)7 7MV'6(f, - t ) , (D.21)
 Charge conjugation C 277

•ipa{x,t)afiui()b{x,t) -+ -xl>a{x,-t)<Tltvipb(x,-t). (D.22)

D.3 Charge conjugation C

In addition to the space-time symmetries discussed above, there are other symme-
tries called internal symmetries. Among many such internal symmetries, we are
here interested in the charge conjugation. This is a discrete transformation of par-
ticles into antiparticles, in which the electric charge and other quantum numbers of
particles are reversed without changing any kinematic attributes.
Here we just list up the transformation rule of fields under charge conjugation.
For simplicity, we fix the phase to be rjc = 1.
Scalar field <f>(x,t) -+<j>c(x',t') = tf(x,t), (D.23)
c
Pseudoscalar field 71(x,t)^rl (x',t')=rf(x,t), (D.24)
f i>(x,t)->i>c(x',t')=C$T(x,t),
Dirac field (D.25)
\ $(3,t) -> H>c{x',t') = -$T(x,t)C-1,
Vector field : V"(2, t) -4 V^ ( f , t') = -V^(x, t), (D.26)
where the charge conjugation operator C in (D.25) is explicitly written by C = Z727°
and satisfies the relations, C'^C = - ^ and C = - C _ 1 = - C * = -CT.
Furthermore, the bilinear covariants composed of Dirac fields transform as

S $a(x,t)i>b(x,t) -> $b(Z,t)ll>a{x,t), (D.27)

5 5
P $a(£,i)7 V>&(£,*) -*• ipb{x,t)^ rpa(x,t), (D.28)
V $a(x,t)'fll>b(x',t) -> -ll)b{x,t)Y^a{x,t), (D.29)
A ^ a (£,*hVV> 6 (£,<) -> ^6(f,t)7 8 7 M V'a(2,t). (D.30)
T tiaix^^Mx,*) -• -rkfrWI)*(x,t). (D.31)

Now, a comment on the transformation of the left- and right-fields under charge
conjugation is useful. Applying the transformation rule (D.25) to the left-handed
Dirac spinor, we can obtain the following result,
T
G
(VL) = ai>i=c (- ^ V O V
2
J*y> 1-75
= c p Ci° P
1 + 75 .0,/,* 1 + 75
C7°V - C-ip1
2
1 + 75 c
i> = (i>c)R = tfg. (D.32)

Similarly we obtain the result, (IPR)C = (II>C)L = V'L J f° r the right-handed spinor.
Furthermore, we can also obtain the similar result even for these adjoint spinors.
278 C, P and T transformation

We summarized them as

4>L,R "»• {i>°)L,R = {i>R,L)C = Ci>lL,

{ $L,R "• $C)L,R = $R,L)C = VR LC ^ ^

D.4 C P transformation

We are here interested in the CP transformation properties of Dirac spinors and

their bilinear covariants. A CP inversion is the direct product of parity operation
and charge conjugation. Then, from the transformation rules described above, we
obtain the following rules for Dirac spinors,

if>(2,t) -> V C P = WPc(-x,t) = j0CiPT(-x,t),

CP c
4>{x,t) -»• 4> {x,t) = i> (-x,t)j0 = ^T(-x,t)Cj0

and furthermore, combining (D.ll) and (D.33), similar rules are derived for the left-
and right-handed spinors as

•HA*'*) -»• WCP)L,R = r/o{rl)c)_n,L(-2,t) = j0CipliR(-x,t),

CP ,C
^ L , B ( ^ , * ) -»• (ip )L,R(x,t) = (V )fl,L(-^*)7o = tplR(-x,t)C-y0.

By using these transformation rules, we can obtain the CP transformation of the

weak left-handed current as

J^ab) = •<PaL(x,t)'ylll/)bL(x',t) -r 1plLC-)olv.loCi>lL{-X,t)

= -^bL{-X,t)^i)aL{-X,t)

where the minus sign in the last two equations originates from the interchange of
fermion fields in taking the transposition.
Let us consider the transformation of the charged current weak interaction under
CP inversion. The Lagrangian is given in (4.164) as

4 ° c = ^f'L^VD'LW+ + ^D'LV^U'LW-, (D.37)

where V is the unitary matrix, being the Cabibbo-Kobayashi-Maskawa(CKM)

matrix for the case of 3 generations, and W^ is the charged weak boson fields
defined by (4.55). If we write the charged currents as JM~ = U'L^VD'L and
JV+ = D'LJ^VW'L, we obtain the CP transformed currents as

r {jncp = -D'Ll,VTU'L,
+ CP (D 38)
I (J» ) = -U>Lj»V*D<L. -
 CP transformation 279

Furthermore, the application of C and P inversions to the vecor field Wjjf yields
w
jt -> W$P± = -W*. (D.39)
Then, we get the following transformation,

j dix{U'LYVD'LW+ + D'L>fVlU'LW-)

-»• J ^x{p'LltlVT\J'LW^- + U'L-yllV*D'LW',+). (D.40)

Therefore, if V = V*, i.e. V is real, the action is invariant under CP inversion,

that is, CP is conserved. On the contrary, if V ^ V*, i.e. V is not real, then CP is
violated.
This page is intentionally left blank
 Appendix E
The quark model

In modern particle physics, the symmetry and the constituent of matter having its
symmetry are powerful tools to understand the structure of matter and its physics.
One such example is the proton and neutron which are the constituents of all nuclei
and possess the SU(2) symmetry of isospin. Various properties of nuclei are well
understood by the dynamics of protons and neutrons having the SU(2) symmetry
of isospin. Now, it is known that there are more than 300 hadrons, including the
proton and the neutron. Then, it is natural to expect that there must be new
fundamental constituents at a deeper level in Nature which build those hadrons
systematically based on new underlying symmetries. In 1964, Gell-Mann and Zweig
(Gell-Mann, 1964; Zweig, 1964) proposed such a new constituent model of hadrons
called the quark model, in which the hadrons are beautifully classified with the
SU(3) symmetry. The idea of quarks as the fundamental constituents of hadrons
has been developed with a lot of theoretical and experimental efforts in 1960s and
1970s, and the quark model was established as an important component of the
standard model of modern particle physics. Here we would like to discuss the quark
model.

E.l Isospin symmetry

Before discussing the quark model, let us begin by briefly reviewing the isospin
symmetry. Nuclei are bound states of protons and neutrons and their nuclear force
is issued from pion exchange. Interestingly the nuclear force does not distinguish
the proton and the neutron. The proton mass mp = 938.27200 ± 0.00004 MeV is
very close to the neutron mass m„ = 939.56533 ± 0.00004 MeV and their small
mass difference is considered to be due to the electromagnetic interaction; a proton
has a positive charge, while a neutron is neutral. This situation naturally suggests
that the proton and the neutron are two manifestations of the same particle, called
the nucleon, and one can write it as a doublet 2 = I ) of the fundamental

281
282 The quark model

representation of the SU(2) group, that is, the proton and neutron are represented
in the isospin space as

W = ( J ) , <!•-+!>; W - ( ; ) , C = -i> (E..)

which are just the same representation as the spin-up and spin-down states of spin
| particles, respectively. Similarly, the masses of TT+, n° and w~ are very close and
they are considered to be three manifestations of the same entity, the pion. This
suggests that these particles are degenerate without electromagnetic interactions
and the isospin symmetry is a good symmetry under strong interactions. In fact,
nuclear interactions between nucleons and pions have an SU(2) isospin symmetry
which works very well. The isospin symmetry is very effective to understand the
structure of nuclei and nuclear interaction dynamics.
Mathematically the isospin symmetry is just the copy of the spin symmetry; the
generators P — y (i = 1,2,3) satisfy the same SU(2) algebra as the ones of the
spin

[l\P]=ieijkIk. (E.2)

where e^ is the totally antisymmetric Levi-Civita symbol with £123 = 1 and the
summation over k is implied. r% have the same expression as the Pauli matrices
a1 (i = 1,2,3) and their explicit expression are presented in (C.3).
For the two-riucleon system, analogously to the spin composition of two spin | ,
one can construct the J = 1 (triplet) and / = 0 (singlet) states of the nucleon-nucleon
system,

r \i = i,i3 = i)=pp,
I = l(triplet) I \I = 1, J 3 = 0) = ^{jm + np), (E.3)
[ \I = l,I3 = -l) = nn,

I = O(singlet) \I = 0, h = 0) = -^{pn - np). (E.4)

v2
As can be seen here, / = 1 (triplet) and I = 0(singlet) states are symmetric and
antisymmetric, respectively, for interchange of the 1st and 2nd nucleon in the two-
nucleon system. One can illustrate this situation in terms of the so-called Young
tableaux, in which a doublet state 2, i.e. the fundamental representation of SU{2)
is represented as • = 2.

D®D = g©m, (E.5)

2®2 = 1©3

where [j = 1 (singlet) and CD = 3(triplet) represent the antisymmetric and sym-

metric states, respectively. The number below the Young tableaux of (E.5) denotes
the dimension of each irreducible representation.
 Quark model and SU(3) symmetry 283

One should be careful in constructing multiplets of nucleon-antinucleon systems.

To see this, consider a rotation with n around the 2nd axis in an isospin frame, as
an example of isospin transformation of a nucleon doublet,

Now, let us define the charge conjugation operator C which transforms particles
into antiparticles,

C\p) = \p) and C\n)=n), (E.7)

where the charge of p and n is - 1 and 0, respectively. Applying C to (E.6) leads to

(i)-(J ?)(:)•
However, the sign of the eigenvalue of I3 for antiparticles is opposite to the one for
particles and we want to arrange h to the upper component (I 3 = + | ) and p to
the lower component (I 3 = — | ) in the antiparticle doublet, as 2* = I J. Then,
if we introduce a minus sign to n, we can see that it transforms in exactly the same
way as the the nucleon doublet I 1 under the SU{2) isospin transformation, as
shown

( ? ) - ( ; ?)(?)•
(E.9) is certainly the same transformation as (E.6). Namely, 2* and 2 transform
in the same way under the SU(2) transformation, i.e. 2* ~ 2. (Here don't worry
about the nasty minus sign to n.) This can be done only for the SU(2) symmetry
as proved in Appendix C.
Now, we can construct the isospin states of the nucleon-antinucleon system as
follows;

\I = 1J3 = 1) =-pn,
J = 1 (triplet) < \I=l,I3=0) = j-(pp-nn), (E.10)
\I - 1,I3 = - 1 ) =np,

I = O(singlet) \I = 0,1 3 = 0) = -^=(pp + nn). (E.ll)

v2

E.2 Q u a r k m o d e l a n d SU(3) symmetry

In 1950s, new hadrons were discovered. Those new hadrons were surprisingly long-
lived compared to the strong interaction scale. For example, the A0 and K° are
284 The quark model

easily produced in high energy n p scattering but those particles decay into light
hadrons with very long lifetime,

7T- + p^A° + K°, (E.12)

0 +
A -> p + TT~, K°->n n-. (E.13)

To explain the fact that while the production of A0 and K° occurs with strong inter-
action scale, the decay of those particles does with weak interaction scale, Nakano
and Nishijima and, independently, Gell-Mann introduced a new additive quantum
number called "strangeness" (S). They assigned S as S = 0 for p, TT~ , S = + 1
for K° and S = — 1 for A0, and considered that while the strong interaction con-
serves the quantum number S, the weak interaction does not. In the production
process (E.12), S is conserved but in the decay processes (E.13), it is not. Soon
later, the idea of Nakano, Nishijima and Gell-Mann was confirmed from observed
properties of many strange particles discovered those days. The conservation of the
strangeness S is similar to the one of the charge Q due to the (7(1) electromag-
netic symmetry. This suggests the existence of a new {7(1) symmetry. Actually
one can introduce the symmetry called the U(l) hypercharge symmetry, where the
new quantum number Y, called hypercharge, is defined by the sum of the baryon
number B and the strangeness S,Y = B + S. Because of this new (7(1) symmetry,
the strong interaction conserves the hypercharge Y and hence, the strangeness S
is also conserved in strong interactions, because the baryon number B is a good
quantum number for the strong interaction. Then, we can see that the following
relation, being called the Nakano-Nishijima-Gell-Mann (NNG) relation (Nakano
and Nishijima, 1953; Gell-Mann, 1953),

Q = I3 + ^, (E.14)

works well for all hadrons discovered those days, where Q and I3 are the charge
and the 3rd component of the isospin of the hadron, respectively.
In 1964, Gell-Mann and Zweig introduced the quarks as physical substances
to realize the relation (E.14). In the quark model, all hadrons are made of a few
quarks; while all baryons are made of 3 quarks q, all mesons are made of a quark q
and an antiquark q, where all quantum numbers of q is opposite to those of q. Since
the quark model should make even strange hadrons like A0 and K°, we need a new
quark, i.e. the s(strange) quark in addition to the u(up) and d(down) quarks which
nicely build the non-strange hadrons like p, n, ir, etc. Thus, in the original quark
model the u, d and s quarks were considered to be the fundamental constituents of
hadrons and to have the SU(3) symmetry. Note that though the 5(7(2) symmetry of
isospin is a rather good symmetry due to almost equal masses of p and n, the 5(7(3)
symmetry is not such a good symmetry because mass differences of strange hadrons
and non-strange hadrons are rather big. Later, the existence of more heavier quarks
c(charm), 6(bottom) and t(top) quarks were also established. Now, we have 6
 Quark model and SU(3) symmetry 285

I I3 s B Y Q
1 I 1
u 2 +1 0 3 3 ^3
1 1 1 1 1
d 2 2 0 3 3 3
1 2 1
s 0 0 -1 3 3 3

Table E.l Quantum numbers of u,d, s quarks.

different kinds of quarks qi (i = u,d,s, c, b, t) and these degree of freedom is called

"flavor", which is unrelated to another degree of freedom "color", being the strong
interaction charge which plays an important role in quantum chromodynamics, the
field theory of strong interactions, being discussed in Chapter 5. The quantum
numbers of u, d, s quarks are given in Table E.l. By taking account of these
quantum numbers of quarks, we can make many hadrons from those quarks, like
p = (uud), n = (udd), A0 = (uds), n+ = (ud), -K~ = (ud), K° = (ds), etc.
In the SU(3) symmetric world,* the fundamental representation of the quark is
given by the triplet (=3)

( u\
Q = d (E.15)

Vs J
Now, let us consider the 517(3) transformation of this triplet,

qa->q'a = U(6)baqb, (a,b = u,d,s) (E.16)

-i8*
U(9) = e

where 6% (i = 1,2, • • •, 8) are parameters of this group and ^- (i = 1,2, • • •, 8) are

the generators of the SU(3) group. Here the summation over i is implied. Explicit
expression of A1 is given by (C.4). These generators satisfy the Lie algebra;
A*
— Ijijk (E.17)
2 ' 2
where fak is called the structure constant and anti-symmetric under any exchange
of i, j and k. Here the summation over k is implied. The group SU(3) includes
the SU(2) of isospin and the U(l) of hypercharge as subgroups and hence it is the
group of rank 2. It has 2 diagonal matrices, A3 and A8. It is convenient to define
the isospin and hypercharge operators as I3 = ^- and Y = A=, respectively. Then,

"Here we consider an idealized world of equal quark masses of u, d and s, though they are, in fact,
different. Therefore, the same mathematics can be applied for the color SU(3), which is an exact
symmetry.
286 The quark model

* / 3

(a) q = 3 (b)g = 3*

Fig. E.l Weight diagram for q = 3 and q = 3*.

we can plot the members of the quark triplet with their quantum numbers of I3
and Y in (I3, Y) space, as shown in Fig. E.l (a). The members of the antiquark
triplet (=3*) are also ploted in Fig. E.l(b). (Fig. E.l is called the weight diagram.)
Introducing I3 and Y into (E.14), we obtain the charge operator,

3 0 0 \
A8
Q = i3 + ^ = 0 1 0 (E.18)
+ 2v/3 ~ 3
1
lo 0 3 /

The diagonal elements of Q are certainly equal to the charge of the u, d and s
quarks given in Table E.l. Then we see that the NNG relation works well even for
quarks.

E.3 Representations of mesons and baryons

In the quark model, mesons are composed of qq, while baryons are of qqq. The
fundamental representation q = 3 transforms as (E.16) under U, while the conjugate
representation q = 3* transforms as

b a
9" -» «" = q (U^) b (E.19)

under W, where U is unitary, W = U~l. (see (C.16)) Therefore, it is interesting

to see the product of representations to discuss the spectroscopy of mesons and
baryons. In general, the product of representations is not irreducible. If some rere-
sentations of the group can be decomposed into a direct sum of other independent
representations, they are called "reducible" and if not, they are called "irreducible".
 Representations of mesons and baryons 287

First consider the product qaqb — 3 <g> 3*, which is reducible. When we decom-
pose it as

qaq" = (qaqb ~ \sbaqcqc) + \sbaqcqc, (E.20)

we see that the 1st term is traceless because of Tv(qaqb) = qaqa and has 8 compo-
nents. The 2nd term is just the trace term and of only one component. The 1st
and 2nd terms do not mix each other and hence they are irreducible. Then we can
say that 3 ® 3* can be resolved into the octet 8 and singlet 1 as

3®3*=8©1. (E.21)

Next, consider the direct product qaqb = 3 <g> 3. It is also reducible. To resolve
this reducible representation into two irreducible representations, let us separate it
into the symmetric and antisymmetric parts as

qaqb = -j[qaqb + < M 0 ) + ~(qaqb - <?&<?«), (E.22)

where the 1st and 2nd terms are symmetric and antisymmetric, respectively, for
interchange of qa and qt- The antisymmetric part has 3 components and identical
to the set of objects ea6cQ69ci where eabc is the totally antisymmetric Levi-Civita
symbol with £123 = 1- Then if we define eabcqbqc = 9 a , w e c a n see that qaqa is
invariant under the transformation of U as follows;

q'aq'a = £abcq'aqWc = £abcU^u^u^qa-qb'qc'

= £a' b' c' det Uqai qv qc' ~ £a'b' & qa1 <7f>' q&
= qa-qa'=qaqa, (E.23)

where det U = 1. Since U is unitary, from the transformation rule of (E.16) for qa
and the result of (E.23), we can easily find that qa, i.e. the antisymmetric part of
(E.22), transforms as the conjugate representation 3*. The symmetric part has 6
components and thus, we can obtain the decomposition rule,

3 ® 3 = 3* © 6. (E.24)

3* and 6 do not mix under the 51/(3) transformation and each one cannot be
decomposed any more. They are irreducible.
Finally, what happens when we multiply one more quark state to (E.24)? In this
case, we have two products 3* ® 3 and 6 ® 3. The product 3* ® 3 has 9 components
and is decomposed into the octet 8 and singlet 1 as presented in (E.21). But here
we should be careful about that in 1 all quarks are totally antisymmetric (as can be
seen from qaqa in (E.23)) and in 8 the first 2 quarks are antisymmetric. The product
6 ® 3 has 18 components and is decomposed into the decuplet 10 and octet 8. In
10 all quarks are totally symmetric, while in 8 only first 2 quarks are symmetric.
288 The quark model

Y
1 I

ds 1 US

/-> "\
/ \
/ \
/ 1 \
/ \
' dj/ 3 \
i ^•v \

duL/\^\ 1 I ' O • J3
1u \ ' 1 \ v ' »% / I ' ,
2
» \ ' 2 ' 1
\ \ / '
\ \ '/ 2 ''
^s \
si
\\ \ 3 '
<s~y .
^s~ / /
\r*i-^^ . r~s\

SU -1 sd

(a) qq = 3 X 3*

»js I3

(b) 0 _ octet (8) (c) 0~ singlet (1)

Fig. E.2 Mesons in the quark model.

Thus, we can summarize the result of the decomposition of qqq states as

3®3®3 = (3* © 6)<g>3 = (3* <g> 3) © (6 ® 3)

= 1©8A©8S©10, (E.25)

where the first 2 quarks are antisymmetric and symmetric in 8A and 85, respectively.
Likewise in the SU(2) case, we can illustrate the quark triplet (E.15) as • = 3.
 Representations of mesons and baryons 289

In the Young tableaux, [j represents the antisymmetric combination of 2 parti-

cles. Since the antisymmetric part of (E.22) being identical to the set of objects
£abcQbqc = Qa has 3 components, we illustrate it as j j = 3*, which is the conjugate
representation corresponding to the antiquark q. Moreover, CD represents the sym-
metric combination of 2 particles and thus, the symmetric part of (E.22) having 6
components are illustrated as CD = 6.

• Mesons

In the quark model, mesons are made up from a quark q and an antiquark q. Here
we consider an example of pseudoscalar mesons with Jp = 0~.
qq = 3 <g> 3* = 1 © 8. (E.26)
Or using Young tableaux, one can write this as

^ = • 0 0 = §©EB (E.27)
3 <g> 3* = 1©8
The number below the Young tableaux denotes the dimension of each irreducible
representation. Therefore, mesons are represented by a sum of the octet 8 and the
singlet 1. By combining the weight diagrams of q (Fig. E.l(a)) and q (Fig. E.l(b)),
we can show this in a different way as shown in Fig. E.2. The same configuration
is obtained even for vector mesons with Jp = 1~.

• Baryons

Baryons are composed of three quarks qqq and its product of representations are
given in (E.25). The Young tableaux for this configuration is given as

qqq = D®D®D= (El®™) ®a= =

®EF@EP®rTJ1' (E-28)
3<g>3®3 = (3*©6) <g>3 = l © 8 © 8 © 10
Similarly to mesons, by combining the weight diagram of Fig. E.l(a), we can
represent the baryons as in Fig. E.3.
 The quark model

(a) Jp = \ baryon octet (8)

Y
-,
A- A0 A+ A++

E*-\ E*° /£*+

I3

(b) Jp = | +
baryon decuplet (10)

Fig. E.3 Baryons in the quark model.

 Bibliography

Abers, E. S. and Lee, B. W.(1973), Phys. Rep. C9 1

Adler, S. L. (1969), Phys. Rev. 177, 2426
Aitchison, I. J. R. anf Hey, A. J. G. (1989), "Gauge Theories in Particle Physics", (2nd
ed.), Adam Hilger Ltd., Bristol
Akhmedov, E. Kh. (1988), Phys. Lett. B213, 64
Alvarez-Gaume, L., Polchinski, J. and Wise, M. (1983), Nucl. Phys. B 2 2 1 , 495
Appelquist, T. and Carazzone, J. (1975), Phys. Rev. D l l , 2856
Bahcall, J. N., Pinsonneault, M. H. and Basu, S. (2001), Astorphys. Journal 555, 990
Barbieri, R. and Gatto, R. (1982), Phys. Lett. B110, 211
Barbieri, R. and Maiani, L. (1983), Nucl. Phys. B224, 32
Bardeen, W. A. (1969), Phys. Rev. 184, 1848
Beg, M. A. and Sirlin, A. (1982), Phys. Rep. 88, 1
Bell,J. S. and Jackiw, R. (1969), Nuovo Cim. A60, 427
Bigi, I. I. and Sanda, A. I. (2000), "CP Violation", Cambridge Univ. Press
Bjorken, J. D. and Drell, S. D. (1964), "Relativistic Quantum Mechanics", McGraw-Hill,
New York
Bludman, S. and Klein A. (1962), Phys. Rev. 131, 2363
Botella, F. J. and Lim, C. S. (1986), Phys. Rev. D 3 4 , 301
Bouchiat, C, Iliopoulos, J. and Meyer Ph. (1972), Phys. Lett. B38, 519
Bourquin, M. et al. (1985), Z. Phys. C21, 27
Broun, H. et al. (1975), Nucl. Phys. B89, 210
Cabibbo, N. (1963), Phys. Rev. Lett. 10, 531
Cheng, T. P. and Li, L. F. (1984), "Gauge Theory of Elementary Particle Physics", Oxford
Univ. Press, New York
Coleman, S. (1975), 'Secret Symmetry', in "Laws of Hadronic Matter", (Erice School
Lecture, 1973), ed. by Zichichi, A., Academic press, NewYork and London
Coleman, S. and Mandula, J. (1967), Phys. Rev. 159, 1251
Dimopoulos, S. and Georgi, H. (1981), Nucl. Phys. B193, 150
Englert, F. and Brout, R. (1964), Phys. Rev. Lett. 13, 321
Fayet, P. (1976), Phys. Lett. B64, 156
Fayet P. and Iliopoulos, J. (1974), Phys. Lett. B 5 1 , 461
Fermi, E. (1933), La Ricerca Scientifica 4, 491
Fermi, E. (1934), Z. Phys. 88, 161
Feynman, R. and Gell-mann, M. (1958), 109, 193
Frauenfelder, H. et al. (1957), Phys. Rev. 106, 386

291
292 Bibliography

Fukugita, M. and Yanagida, T. (2003), "Physics of Neutrinos and Applications to Astro-

physics", Springer, Berlin, Germany
Gaillard, M. K. and Lee, B. W. (1974), Phys. Rev. D10, 897
Gell-Mann, M. (1953), Phys. Rev. 92, 833
Gell-Mann, M. (1964), Phys. Lett. 8, 214
Gell-Mann, M. and Neeman, Y. (1964), "The Eightfold Way", Benjamin, New York
Gell-Mann, M., Rarnond, P. and Slansky, R. (1979), in "Supergravity", ed. by P. van
Nieuwenhuizen and D.Z. Preedman, North Holland, New York
Georgi, H. and Glashow, S. L. (1972), Phys. Rev. D6, 429
Georgi, H. and Glashow, S. L. (1974), Phys. Rev. Lett. 32, 438
Glashow, S. L. (1961), Nucl. Phys. 22, 579
Glashow, S. L., Iliopolulos, J. and Maiani, L. (1970), Phys. Rev. D 2 , 1285
Glashow, S. L. and Weinberg, S. (1977), Phys. Rev. D 1 5 , 1958
Goldhaber, M. et al., (1958), Phys. Rev. 109, 1015
Goldstone, J. (1961), Nuovo Cim. 19, 154
Goldstone, J., Salam, A. and Weinberg, S. (1962), Phys. Rev. 127, 965
Gronau, M., Kiir, A. and Loewy, R. (1986), Phys. Rev. Lett. 56, 1538
Gross, D. and Jackiw, R. (1972), Phys. Rev. D6, 477
Guralnik, G. S., Hagen, C. R. and Kibble, T. W. B. (1965), Phys. Rev. Lett. 13, 585
Haag, R., Lopuszanski, J. and Sohnius, M. (1975) Nucl. Phys. B 8 8 , 257
Haber, H. E. and Kane, G. L. (1985), Phys. Rep. 117, 75
Halzen, F. and Martin, A. D. (1984), "Quarks and Leptons: An Introductory Course in
Modern Particle Physics", John Willey &: Sons, New York
Herrmannsfedt, W. B. et al. (1957), Phys. Rev. 107, 641
Higgs, P. W. (1964), Phys. Lett. 12, 132
Inami, T. and Lira, C. S. (1981), Progr. Theor. Phys. 65, 297; (1981), ibid. 65, 1772
(Erratum)
Inami, T. and Lim, C. S. (1982), Nucl. Phys. B207, 533
Inami, T., Lim, C. S. and Sakai, N. (1983) Phys. Lett. B123, 311
Inami, T., Lim, C. S. and Yamada, A. (1982), Mod. Phys. Lett. A 7 , 2789
Inazawa, H. and Morii, T. (1988), Phys. Lett. B203, 279; ibid. B207, 520 (Erratum)
Inoue, K., Kakuto, A., Komatsu. H. and Takeshita, S. (1982), Progr. Theor. Phys. 68,
927
Jarlskog, C. (1985), Phys. Rev. Lett. 55, 1039; (1985), Z. Phys., C29, 491
Jauch, J. M. and Rohrlich, F. (1976), "Theory of Photons and Electrons", Springer-Verlag,
Berlin
Juaneau, L. (1997), "Historical Introduction from f) Radioactivity to The V — A Hypoth-
esis", pp.1-27 in "Weak Interactions", ed. Gaillard, M. K. and Nikolic, M., Institut
National de Physique Nucleaire et de Physique des Particules, Paris
Kam-LAND collaboration, (2003), Phys. Rev. Lett. 90, 021802
Kennedy, D. C. and Lynn, B. W. (1989), Nucl. Phys. B 3 2 2 , 1
Kibble, T. W. (1967), Phys. Rev. 155, 1554
Kistner, O. C. et al. (1956), Phys. Rev. 104, 154
Kobayashi, M. and Lim, C. S. (2001), Phys. Rev. D 6 4 , 013003
Kobayashi, M. and Maskawa, T. (1973), Prog. Theor. Phys. 49, 652
Landau, L. (1932), Phys. Zeits. D. Sowjetunion, 2, 46
Langacker, P. (1982), Phys. Rep. 72, 185
Lee, T. D. (1974), Phys. Rep. 9 C , 143
Lee, T. D. and Yang, C. N. (1956), Phys. Rev. 104, 254
Lim, C. S. (1987), in Proceedings of BNL Neutrino Workshop, Upton, NY, USA, ed. by
 Bibliography 293

M. J. Murtagh, BNL-52079,C87/02/05
Lim, C. S., Inami, T. and Sakai, N. (1984), Phys. Rev. D29, 1488
Lim, C. S. and Marciano, W. J. (1988), Phys. Rev. D37, 1368
Maki, Z., Nakagawa, M. and Sakata, S. (1962), Progr. Theor. Phys. 28, 870
Marshak, R. E. and Sudarshan, E. C. G. (1958), Phys. Rev. 109, 1860
Mikheyev, S. P. and Smirnov, A. Yu. (1985), Yad. Fiz. 42, 1441 [Sov. J. Nucl. Phys. 42,
913 (1985)]
Mohapatra, R. N. and Senjanovic, G. (1980), Phys. Rev. Lett. 44, 912
Muta, T. (1998), "Foundations of Quantum Chromodynamics-An Introduction to Pertur-
bative Methods in Gauge Theories" (2nd Edition), World Scientific Pub.
Nagashima, Y. (1999), "Standard Theory of Elementary Particles and Their Experimental
Ground", (in Japanese), Asakura Pub, Tokyo
Nakanao, T. and Nishijima, K. (1953), Prog. Theor. Phys. 10, 581
Noether, E. (1918), Nachr. Kgl. Ges. Wiss. Gottingen, p.235
O'Raifeartaigh, L. (1975), Nucl. Phys. B96, 331
Particle Data Group, Hagiwara, K. et al. (2002), Phys. Rev. D66, 010001
Peccei, R. D. (1989), "The Strong CP Problem", pp. 503-552 in "CP Violation", ed. by
Jarlskog, C., World Scientific Pub.
Peskin, M. E. and Schroeder, D. V. (1995), "An Introduction to Quantum Field Theory",
Addison-Wesley Pub.
Peskin, M. and Takeuchi, T. (1990), Phys. Rev. Lett. 65, 964
Pokorski, S. (2000), "Gauge Field Theories", 2nd edition, Cambridge Univ. Press, UK
Quigg, C. (1984), "Gauge Theories of The Strong, Weak, and Electromagnetic Interac-
tions" , The Benjamin/Cummings Pub.
Roos, M. and Sirlin, A. (1971), Nucl. Phys. B29, 296
Ross, G. G. (1985), "Grand Unified Thoeries", Benjamin/Cummings Pub.
Rudaz, S. (1986), Lecture at 1986 CERN School of Physics, "Introduction to Symmetries
and Gauge Theories", Proceedings, CERN-87-02 (1987)
Sakai, N. (1981), Z. Phys. C l l , 153
Salam, A. (1968), "Elementary Particle Theory: Relativistic Group and Analyticity" (No-
bel Symposium No.8), ed. by N. Svartholm, Almqvist and Wiksell, Stockholm,
p.367
Susskind, L. (1979), Phys. Rev. D20, 2619
Shi, X. and Schramm, D. N. (1992), Phys. Lett. B283, 305
Shi, X., Schramm, D. N. and Bahcall, J. N. (1992), Phys. Rev. Lett. 69, 717
Sikivie, P., Susskind, L., Voloshin, M. and Zakharov, V. I. (1980), Nucl. Phys. B173, 189
Smirnov, A. Yu. (1992) in Proceedings of the International Symposium on Neutrino As-
trophysics, Takayama, Japan: Frontiers of Neutrino Physics, ed. by Y. Suzuki and
K. Nakamura
SNO collaboration, (2001), Phys. Rev. Lett., 87, 071301
Strassler M. J. and Peskin, M. E. (1991), Phys. Rev. D 4 3 , 1500
Super-Kamiokande collaboration, (1998), Phys. Rev. Lett. 81, 1562
Super-Kamiokande collaboration, (2001), Phys. Rev. Lett. 86, 5656
Takahashi, Y. (1957), Nuovo Cim. 6, 370
Veltman, M. (1977), Nucl. Phys. B123, 89
Voloshin, M. et al. (1988), J E T P 64, 446
Ward, J. C. (1950), Phys. Rev. 78, 1824
Weiberg, S. (1967), Phys. Rev. Lett. 19, 1264
Weinberg, S. (1976) Phys. Rev. Lett. 37, 657
Weinberg, S. (1979), Phys. Rev. D19, 1277
294 Bibliography

Weinberg, S. (2000), "The Quantum Theory of Fields", Vol. Ill, Cambridge University
Press
Wess, J. and Bagger, J. (1992), "Supersymmetry and Supergravity", Princeton Univ.
Press
Wolfenstein, L. (1978), Phys. Rev. D17, 2369
Wolfenstein, L. (1981), Nucl. Phys. B186, 147
Wolfenstein, L. (1983), Phys. Rev. Lett. 51, 1945
Wu, C. S. et al. (1957), Phys. Rev. 105, 1413
Yanagida, T. (1979), Proceedings of "Workshop on Unified Theory and Baryon Number
in the Universe", ed. by Sawada, O. and Sugamoto, A. KEK, Japan
Yang, C. N. and Mills, R. L. (1954), Phys. Rev. 96, 191
Yndurain, F. J. (1992), "The Theory of Quark and Gluon Interactions", Springer-Verlag
Zener, C. (1932), Proc. Roy. Soc. London, A137, 696
Zweig, G. (1964), CERN Report 8182/TH.401; 8419/TH.412
 Index

Abelian group, 269 charged current interaction, 66

Abelian local U(l) gauge theory, 37, 51 charge screening, 96
adiabaticity, 126 charm quark, 80
adiabaticity condition, 127 chirality, 11, 254
adjoint representation, 271, 274 chirality operator, 12
anomaly, triangle anomaly, chiral multiplet, 154
Adler-Bell-Jackiw (ABJ) anomaly, 75 chiral superfield, 160
or axial anomaly, 84 closure property, 159
anomaly free condition, 86 CI experiment, 142
atmospheric neutrino, 139 color, 2, 5
atmospheric neutrino oscillation, 140 color charge, 89
atmospheric neutrino problem, 140 color interaction, 93
asymmetric collider, 247 confinement, 4, 103
asymtotic free, 5, 103 conjugate fundamental representation, 272
auxiliary field, 154 contribution of super-partners to Ap, 200
covariant derivative, 38, 41, 59
/3-decay, 7 CP asymmetry, 237, 243
beyond the standard model, new physics, CPT theorem, 219
115, 148, 193, 215 CP violation, 215
B-factory experiments, 237, 241, 245 CP violation in Kobayashi-Maskawa
bi-unitary transformation, 71, 76 model, 232
box diagram, 221 CP violation in neutrino oscillation, 133,
B° o B° oscillation, 231, 242 135
CP violation in the neutral B system, 241
Cabibbo angle, 23, 79 CP violation in the neutral K system, 237
Cabibbo currents, 22 CP violating phase <5, 232
Cabibbo-favored decays, 23 cross section, 257
Cabibbo-Kobayashi-Maskawa current conservation law, 35
matrix(CKM matrix), 25, 62 current (weak) eigenstate, 72, 77
Cabibbo mixing, 24 custodial symmetry, 207, 210
Cabibbo-suppressed decays, 23
canonical cummutation relations, 34 decay width, 259
central charge, 153 decoupling, 193, 194
charge, 35 decoupling theorem, 194
charge conjugation C, 277 A7 = § rule, 226, 239

295
296 Index

AmK, 221 Glashow-Weinberg-Salam (GWS) model,

Ap, 180, 197 55
|AS| = 1 process, 223 global SU(2) invariant Lagrangian, 48
| A 5 | = 2 process, 219 global symmetry, 33
Dirac equation in the 75-diagonal global transformation, 34
representation, 253 global £/(l) invariant Lagrangian, 45
direct CP violation, 239 gluons, 2
D-term, 158, 163 Goldstone bosons, 44, 47, 50, 61
Goldstone theorem, 44, 47
elastic scattering of neutrino, 143 graded Lie algebra, 152
electromagnetic current, 10 grand unified theories (GUT), 148
electromagnetic field-strength tensor, 39 Grassmannian coodinates, 155
electromagnetic interaction, 4, 39 gravitational interaction, 4
electron number, 18 group, 269
electro-weak penguin, 225 group manifold, 156
energy projection operator, 259
e, e', 240 helicity, 11, 254
Euler-Lagrange equation, 34 hidden symmetry, 47
explicit SUSY breaking, 184 hierarchy problem, 147, 165
extended SUSY, 153 Higgs boson, 52, 61
Higgs mechanism, 51, 52, 62
Fayet-Iliopoulos mechanism, 187 Homestake, 142
FCNC, 181 horizontal symmetry, 217
FCNC rare processes, 216 hypercharge operator, 285
Fermi constant, 17, 20
Fermi theory, 7 indirect CP violation, 238
Fermi transition, 14, 32 intermediate weak boson model, 28
Feynman rule, 260 irreducible, 286
field strength superfield, 168 irrelevant operators, 192, 212
fine-structure constant, 96 isospin symmetry, 281
flavor, 2, 285
flavor changing neutral current (FCNC), Jarlskog parameter, 234
198, 215
flavor mixing, 78, 115, 215 KamLAND experiment, 134
flavor physics, 215 KL -> MM> 223
F-term, 162, 164 Kobayashi-Maskawa model, 216
fundamental represenation, 272 K+ -> TT+I/P, 223
K° <-» K° mixing, 220
7 matrix, 250
Ga experiment, 143 Landau-Zener formula, 129
Gamow-Teller transition, 14, 32 large CP asymmetry, 241
gauge fields, 36 left-handed fermion, 12
gaugino, 155 left-right symmetric model, 200, 210
gauge fermoin, 168 leptons, 2
gaugino, 168 lepton current universality, 17
Gell-Mann matrix, 89, 92, 270 level crossing, 129
general superfield, 157 Levi-Civita symbol, 41, 57, 240
generator, 35, 36, 270 Lie algebra, 35, 36, 270
GIM mechanism, 79, 81 local gauge symmetry, 36
 Index 297

long-baseline accelerator neutrino penguin-diagram, 225, 226

oscillation experiments, 132 photino, 154, 155, 171
Lorentz invariant phase space, 259 pion decays, 20
low energy effective theory, 148 pion decay constant / „ , 21
precision test of electroweak radiative
Majorana fermion, 107, 166, 170 corrections, 191
Majorana neutrino, 109 problem of quadratic divergence, 148
Majorana particle, 151 projection operator, 12
Majorana-phase, 120 pseudo-Dirac neutrino, 111
Majorana spinor, 107, 172
Majorana type mass term, 107 quadratic divergence, 148
Maki-Nakagawa-Sakata (MNS) matrix, quantum chromodynamics, 89
116 quantum electrodynamics, U(l) model, 37
marginal or relevant operators, 192 quark, 2
mass eigenstate, 72, 77 quark mass matrix, 76
matter effect, 125 quark model, 1, 89, 281
maximal mixing, 123
minimal coupling, 38 radiative corrections, 191, 193
minimal model, 60 radio-chemical experiment, 143
Minimal Supersymmetric Standard Model rare processes, 215
(MSSM), 174 reducible, 286
MSW effect, 128 reduction formula, 131, 137, 143
MSW triangle, 130, 143 renormalizable theory, 192
muon number, 18 renormalized charge, renormalized
coupling, 100
Nakano-Nishijima-Gell-Mann (NNG) re-phasing invariant quanitity, 232
relation, 284 representation, 269
Nambu-Goldstone phase, 46 resonant neutrino oscillation, 123
natural flavor conservation, 217 p parameter, 197
natural unit, 249 right-handed fermion, 12
neutrino oscillation, 114 R-parity, 177
new physics, 148 R symmetry, 175, 177
Noether current, 35 running coupling constant, 101
Noether's theorem, 33, 35
no-go theorem, 152 seesaw mechanism, 114, 200
non-Abelian gauge theory, 40 selectron, 171
non-Abelian group, 270 SNO, 142
non-Abelian 5172) model, 52 SNO experiment, 143
non-decoupling, 193, 194, 199 solar neutrino, 139
non-linear realization of Higgs field, 212 solar neutrino experiment, 123, 142
non-linear sigma model, 196 solar neutrino oscillation, 142
non-renormalizability, 28 solar neutrino problem, 143
spontaneous (or soft) CP violation, 216
oblique corrections, 193, 201 spontaneous SUSY breaking, 184
oblique parameters, 193 spontaneous symmetry breaking (SSB),
operator analysis, 211 44, 47, 199
O'Raifeartaigh mechanism, 185 standard model, 4, 55
standard parametrization, 83
parity or space inversion P, 275 star prescription, 201
parity violation in weak interaction, 8 strangeness conserving weak decays, 22
298 Index

strangeness non-conserving weak decays, W * boson, 64

22 weak bosons, 2
strong interaction, 4 weak coupling constant, 10
structure constant, 36, 41 weak hyperchatge, 58
S, T, [/-parameters, 193, 201, 207, 212 weak interaction, 4
SU(n), 34 weak isospin, 56
SU(n) group, 270 weak isospin current, 57
superfield, 155 weak lepton current, 10
superfield formulation, 155 weak mixing angle, Weinberg angle, 63
supergravity, 149, 153 weak nucleon current, 10
super-Higgs mechanism, 155 weak scale, 67
Super-Kamiokande, 142 Wess-Zumino gauge, 168
Super-Kamiokande collaboration, 140 Wess-Zumino model, 162
Super-Kamiokande experiment, 142 weight diagram, 286, 289
super-partner, 154 Wigner phase, 45
superpotential, 163 Wolfenstein parameterization of KM
superspace,149, 155 matrix, 83, 234
supersymmetry, 147 would-be Nambu-Goldstone bodson, 221
survival probability of i/e, 121
SUSY algebra, 152, 153 Yang-Mills theory, SU(2) model, 40
SUSY breaking, 162 Young tableaux, 282, 289
SUSY breaking mass scale, 148 Yukawa interaction, Yukawa coupling, 44,
SUSY-covariant derivative, 169 60, 75
SUSY group, 152, 153
SUSY multiplet, 154, 155 Z boson, 64
SUSY QCD, 171 Zenith-angle dependence, 141
SUSY 1/(1) gauge theory, 170
SUSY Yang-Mills theory, 173
SU(3) symmetry, 273

tau number, 18
T — 6 puzzle, 8
technicolor theory, 208
time inversion T, 276
transition probability of the process
ve -*• Vp, 121
two-component notation, 149

unitarity triangle, 235

unitarity violation, 26
unitary gauge, 51, 62
universality of weak interactions, 18
C/(l) model, 47, 51

vacuum expectation value (VEV), 45, 61

vacuum polarization, 96
vector multiplet, 154, 168
vector superfield, 160, 167
V — A (current-current) interaction, 15,
17
 This book provides a unified description
of elementary particle interactions and the
underlying theories, namely the Standard
Model and beyond. The authors have
aimed at a concise presentation but have
taken care that all the basic concepts are
clearly described. Written primarily for
graduate students in theoretical and
experimental particle physics, The Physics
of the Standard Model and Beyond
conveys the excitement of particle physics,
centering upon experimental observations
(new and old) and a variety of ideas for
their interpretation.

World Scientific
www.worldscientific.com
4655 he