100% found this document useful (2 votes)
1K views151 pages

The Origin of Mass - Iliopoulos

Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
100% found this document useful (2 votes)
1K views151 pages

The Origin of Mass - Iliopoulos

Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 151

THE ORIGIN OF MASS

THE ORIGIN
OF MASS
Elementary Particles and Fundamental
Symmetries

John Iliopoulos

3
3
Great Clarendon Street, Oxford, OX2 6DP,
United Kingdom
Oxford University Press is a department of the University of Oxford.
It furthers the University’s objective of excellence in research, scholarship,
and education by publishing worldwide. Oxford is a registered trade mark of
Oxford University Press in the UK and in certain other countries
© John Iliopoulos 2017
The moral rights of the author have been asserted
Translation from the French language edition of: Aux origins de la masse
by Jean Iliopoulos © 2015 Editions EDP Sciences, Paris, France
First Edition published in 2017
Impression: 1
All rights reserved. No part of this publication may be reproduced, stored in
a retrieval system, or transmitted, in any form or by any means, without the
prior permission in writing of Oxford University Press, or as expressly permitted
by law, by licence or under terms agreed with the appropriate reprographics
rights organization. Enquiries concerning reproduction outside the scope of the
above should be sent to the Rights Department, Oxford University Press, at the
address above
You must not circulate this work in any other form
and you must impose this same condition on any acquirer
Published in the United States of America by Oxford University Press
198 Madison Avenue, New York, NY 10016, United States of America
British Library Cataloguing in Publication Data
Data available
Library of Congress Control Number: 2017932404
ISBN 978–0–19–880517–5
Printed and bound by
CPI Group (UK) Ltd, Croydon, CR0 4YY
Foreword

This book by John Iliopoulos begins with a question. Given that


a large number of elementary particles had already been dis-
covered, why did the announcement by CERN on 4 July 2012 of
the discovery of a new particle become such an extraordinary
worldwide media event, a situation highly unusual in element-
ary particle physics? The author tells us that this was not just
a matter of finding a new particle, but, most likely, ‘a window
into one of the most extraordinary phenomena in the history of
the Universe.’ This extraordinary phenomenon is the mechanism
which allows elementary particles to acquire a mass.
It is this phenomenon, which in the terminology of physi-
cists is referred to as ‘spontaneous symmetry breaking in the
presence of local internal symmetries’, that the author explains
to us in simple but correct terms. He shows us how this phe-
nomenon fits into our current understanding of elementary
particle physics summarised in the Standard Model, to which
he has himself made important contributions. In reading this
book, we also come to understand that, after the CERN dis-
covery, the Standard Model marks the boundary between the
known and the unknown, and we learn why our current vision
of this unknown sees knowledge of the ‘infinitesimally small’ of
elementary particles as potentially containing knowledge of the
‘infinitely large’ of the observable Universe.
      
The story told in this book forms a part of the history of the search
for a rational understanding of the world. Let me briefly review
this history.
Physics, as it is understood today, is an attempt to interpret a
wide diversity of phenomena as particular manifestations of gen-
eral laws which can be verified experimentally. This concept of a
world governed by verifiable general laws is a relatively recent idea
vi Foreword

in human history. It originated in Europe during the Renaissance,


and then underwent an extraordinarily rapid development. Its
success is largely due to the universality of the revolutionary
vision of Galileo (1564–1642). He was the first to introduce the
principle of inertia, which states the impossibility of detecting a
uniform rectilinear motion of a physical system, whether anim-
ate or inanimate, by an experiment performed within the system
itself. This principle suggests that motion in a straight line at
constant velocity need not be the result of any cause.
In accordance with the principle of inertia, in the late sev-
enteenth century Newton formulated the celebrated law of the
universal attraction of masses. Newton viewed the world as com-
posed of tiny entities interacting with each other by means of
forces which cause changes of velocity. These tiny ‘point’ entit-
ies have become what today we refer to as elementary particles.
In the nineteenth century Maxwell introduced the concept of a
field, which, in contrast to the tiny entities of Newton, fills an
entire region of space. He formulated in these new terms the
general laws of electromagnetism which govern all phenomena
associated with electricity, magnetism, and light. The concepts
of field and particle were unified during the first decades of the
twentieth century by quantum mechanics, with particles becom-
ing the ‘quantum’ constituents of fluctuating fields. In addition,
at the beginning of the twentieth century, Einstein extended the
principle of Galilean inertia to electromagnetism by developing
the theory of special relativity, which modifies our concepts of
time and space. He then generalised Newton’s law of the grav-
itational attraction of masses to create the theory of general
relativity, which opens up a path to the scientific study of the
cosmological expansion of the Universe.
Therefore, owing in particular to the impressive progress made
during the first half of the twentieth century, we arrived at a
picture of the world in which all phenomena, from the atomic
level to the limits of the observable Universe, appeared to be gov-
erned uniquely by two fundamental and known laws: the general
Foreword vii

relativity of Einstein and quantum electrodynamics, which is


the transcription of Maxwell’s theory of electromagnetism into
quantum mechanics.
The gravitational and electromagnetic interactions are long-
range interactions, that is, they act on objects separated by any
distance. However, the discovery of subatomic structures indic-
ated that additional fundamental interactions of short range exist
whose effect is negligible at our scale. At the beginning of the 1960s
their theoretical interpretation seemed to pose insurmount-
able problems. This is where the narrative of John Iliopoulos
begins.
      
In 1960 Nambu introduced into elementary particle theory the
concept of ‘spontaneous symmetry breaking’. Generalised in
1964 to ‘local internal symmetries’ by Brout and Englert, and
independently by Higgs, this idea permitted these authors to
construct a theoretical ‘mechanism’ which induces a transmu-
tation of long-range interactions into short-range interactions by
endowing the particles transmitting the force with a mass. More
generally, this Brout–Englert–Higgs (BEH) mechanism allowed
understanding of the origin of the elementary particle masses.
What do these concepts mean and how does this mechanism
work?
John Iliopoulos explains these concepts and then shows how
the Standard Model of elementary particles is constructed start-
ing from the BEH mechanism. He recounts the experimental
verification and validation of the mechanism itself by the detec-
tion at CERN of the particle which is its essential element. His
narrative leads us onward to theoretical speculations which, per-
taining to as yet unexplored energies, go beyond the Standard
Model.
From this perspective, the author shows how we are led to
the necessity of fusing the ‘infinitely large’ of cosmology with
the ‘infinitesimally small’ of elementary particles. The discovery
in 1965 of the cosmic microwave background radiation indicated
viii Foreword

the existence of a hot primordial universe whose structure is


becoming more and more accessible to us. On the one hand,
improvements in the techniques for observing this primordial
cosmic radiation, coupled with theoretical developments in cos-
mology, allow us to look back in time to the origin of the
quantum fluctuations of this radiation at an epoch close to the
birth of our Universe. On the other hand, theoretical and exper-
imental results in elementary particle physics are now leading us
to the analysis of the physics at energies corresponding to tem-
peratures of around a million billions of degrees, and thus to the
study of the primordial structure of the Universe at such temper-
atures, and then onward to the formulation of conjectures for
even higher temperatures.
The data on the ‘infinitely large’ and the ‘infinitesimally small’
are coming to overlap each other more and more and are mak-
ing the fusion of the theory in these two domains inevitable. Will
this lead us all the way to an understanding of the birth of the
Universe from quantum fluctuations of gravity? Unfortunately,
the quantum extension of general relativity which might describe
this era is, at best, embryonic, and it is too early to tell if we will
be able to obtain a rational understanding of the birth of the
Universe itself.
      
The popularisation of science for a general audience is a dif-
ficult art. Explaining the achievements of theoretical physics,
particularly those of contemporary physics, to an uninformed
public faces a double barrier. The language of mathematics
which furnishes the physicist with the mental shortcuts needed
to express the concepts involved is inaccessible to the general
public. Moreover, the concepts themselves are so remote from
familiar ideas that they have no evident analogue. How does
one explain the meaning of ‘spontaneous symmetry breaking’
or ‘local internal symmetries’, concepts which are essential for
understanding modern elementary particle physics? While the
Foreword ix

mathematical description of these concepts is unambiguous, their


approximate translation into everyday language requires going
beyond habitual ways of thinking.
The difficulty is the greater the further away from the sub-
ject is the everyday experience of the reader. A casual perusal of
this book, whose text is studded with a few equations, suggests
that it is written mainly for students who are beginning work
on scientific subjects. Undoubtedly it will be quite useful to this
audience, but it can also be read profitably by a much broader pub-
lic. Anyone interested in the fundamental questions which arise
in the search for a rational understanding of the world will find
some enlightenment here. The reader who is repelled by the pres-
ence of equations can simply ignore them and attentively read the
accompanying explanations. The latter are sufficiently explicit to
give a qualitative understanding of the subject, which in fact is the
goal of scientific popularisation.
The author’s description of the mathematical analysis in
ordinary language without any loss of rigour in explaining the
physical concepts is exemplary. Some effort on the part of the
reader is, of course, required; however, the reader’s task is aided
by the fact that the author, in limiting himself to the essential,
has brilliantly succeeded in summarising in a hundred or so
pages a series of topics unfamiliar to a non-scientist. The thread
of the discussion can be followed throughout the reading of
the text without interruption by an accumulation of secondary
details. The success of this little book at explaining ideas and facts
so foreign to everyday experience in an attractive and accessible
manner owing to its focus on the essential makes this a book of
rare quality.

François Englert
Professor Emeritus,
Université Libre de Bruxelles,
Nobel Prize 2013
Acknowledgment
An earlier version of this book appeared in French in the «Editions
de Physique». The author wants to thank Professor Michel Le
Bellac and Ms Nicole Ribet for their friendly help and advice.
Contents

1 Introduction 1
2 A Brief History of Cosmology 6
3 Symmetries 22
3.1 Space symmetries 24
3.1.1 Translation symmetry 26
3.1.2 Rotation symmetry 27
3.1.3 Space inversion 27
3.2 Time symmetries 29
3.3 Internal symmetries 30
3.4 Local or gauge symmetries 33
3.4.1 Local translations 34
3.4.2 Internal gauge symmetries 36
4 A Problem of Mass 40
4.1 Mass and the range of interactions 40
4.2 Gauge interactions 43
4.3 The masses of matter constituents: quarks and leptons 45
4.3.1 Chirality 45
4.3.2 Chirality and weak interactions 47
5 Spontaneously Broken Symmetries 50
5.1 Curie’s theorem 50
5.2 Spontaneous symmetry breaking in classical physics 52
5.3 Spontaneous symmetry breaking
in quantum physics 58
5.4 Goldstone theorem 62
5.5 Spontaneous symmetry breaking
in the presence of gauge interactions 65
6 The Standard Theory 71
6.1 Introduction 71
6.2 The electromagnetic and weak interactions 72
xii Contents

6.3 The strong interactions 77


6.4 The Standard Theory and experiment 85
7 Epilogue 91
Appendix 1 The Elementary Particles 97
A1.1 Introduction 97
A1.2 The four interactions 102
A1.3 Some basic notions 105
A1.4 The neutrino saga 110
A1.5 The table of elementary particles 117
A1.5.1 The elementary particles in 1932:
the world is simple 117
A1.5.2 The elementary particles today 119
The disorder 119
The quarks 120
The table today 122
Appendix 2 From Sophus Lie to Élie Cartan 125
Index 133
1
Introduction

On 4 July 2012 the European Organisation for Nuclear Research


(CERN) announced a discovery which made headlines in
the world media. Through a system of teleconferences, the
announcement was made simultaneously at the Large Lecture
Hall at CERN and in Australia, where the International
Conference for High Energy Physics was taking place, but also
directly on-line for the attention of universities and research
centres world wide as well as the general public (Figure 1.1). It
was the discovery of a new particle that completed a rather rich
collection which physicists have accumulated over the years. In
Appendix 1 we present a table which summarises our knowledge
on what we call elementary particles. We find quite a large number
of entries with exotic names, such as quarks, gluons, intermediate vector
bosons, etc. Therefore, why all the fuss about the last one? (see
Figure 1.2). Answering this question is the subject of this book.
We want to show that it is not just a new particle but, probably,
the trace of one of the strangest events that occurred in the early
Universe: the phenomenon which allowed most elementary
particles to acquire a mass.
In the course of presenting this extraordinary phenomenon,
we will touch upon many subjects in physics and mathematics.
The physics of elementary particles, naturally, but also that of
cosmology and, to a certain extent, the physics of phase trans-
itions. To prevent any misunderstandings, we want to stress here
what this book is not supposed to be: it is not a book on element-
ary particles; it is not a book on cosmology; and it is not a book on
phase transitions. It is not even a book on the discovery of this new
particle. The discovery itself, which could make for a fascinating
2 The Origin of Mass

Figure 1.1 The announcement of the discovery at CERN. In the centre


on the podium we see Fabiola Gianotti, spokesperson of the ATLAS col-
laboration (also at the top right), Rolf-Dieter Heuer, CERN’s Director
General, and Joe Incandela, spokesman of the CMS collaboration (also
at the top left). Bottom right : François Englert and Peter Higgs.
Dr A. Hoecker, CERN.

story, will not be described. It is a book on the physical significance


of the discovery. Out of all these subjects we will present only
the elements necessary to understand the phenomenon of mass
generation in the early Universe. We shall describe an important
chapter of the physics of elementary particles without present-
ing the full theory of particles and their interactions. The reader
who has already some familiarity with the field will find it easier
to follow the arguments. To help them, but also all those who are
unfamiliar with the subject, we include a rather long Appendix 1
which summarises everything one is supposed to know on the
ultimate constituents of matter. The reader who already possesses
this knowledge may safely ignore it, but the novice is strongly
advised to read it before reading the main chapters of the book.
Introduction 3

Figure 1.2 A collection of international press cuttings announcing


CERN’s discovery of the new boson. Dr A. Hoecker, CERN.

A second appendix attempts to explain very briefly some math-


ematical concepts to which we refer occasionally in the main text.
They are some notions of group theory applied to the symmetry
properties of a physical system. They have been included for com-
pleteness, but are not absolutely necessary for understanding the
main subject.
The structure of the book is as follows. In the first chapter
we present a brief history of cosmology: its origins and its evolu-
tion, together with our present ideas (some would say prejudices)
regarding the history of the Universe. This choice requires an
explanation. Indeed, it is not obvious guessing the relevance of
cosmology in the microscopic world in general, and the recent
CERN discovery in particular. Naturally, our first goal is to show
that this choice is justified, but in fact we want to go further and
show a profound connection between the infinitely large and the
infinitely small. This connection will be developed throughout
this book. We want to convince the reader that the structure of
4 The Origin of Mass

the Universe in all its immensity is due to the laws of microscopic


physics, and that the Universe is the best laboratory for applying
our ideas on the structure of matter.
The subsequent chapters develop the main subject. The dir-
ecting principle is that of symmetry, a concept which has been
fundamental in all the progress made during the past few dec-
ades in our understanding of the world. We first show the most
intuitive aspects of this concept, namely the symmetries of our
familiar space. We shall see in these examples that this concept
brings us close to geometry in its simplest form. Gradually the
symmetry concept will become more abstract together with the
geometrical ideas. Geometry, in the mathematical sense, would
have been the natural language for this book, but we have decided
to stay away from mathematics. Therefore, our main task will be
to translate mathematics into plain English. We shall pursue this
more or less successfully; less being the more probable. Following
a sequence of increasing abstractions we shall guide the reader
towards a complex edifice called the Standard Model. This epitom-
ises all our knowledge of the sub-atomic world and it is inside
this framework that the significance of the last discovery will be
revealed to us.
A remark, which is also a warning: the phenomena we are
going to present in this book are those we study in high energy
physics experiments, or observations in cosmology. They are phe-
nomena which are governed by the laws of quantum mechanics.
We shall try to describe them using analogies from classical phys-
ics which are closer to our everyday experience. These analogies
are of limited validity. At best, they can capture some aspects of
the phenomenon, never the real thing. The reader should not
take them too literally.
A last remark concerns the system of units we are going to
use. In classical physics we often choose the metre as a unit of
length (m), the kilogram as unit of mass (kg), and the second
as unit of time (s). This system is ill-adapted to the phenomena
of relativistic quantum physics, which are characterised by two
Introduction 5

physical constants: the speed of light in vacuum c and Planck’s


constant h. The numerical values of these constants in our MKS
(metre-kilogram-second) system are c = 299792458 ms–1 and h̄ =
h/2π = 1.054571726(47) × 10–34 Js,1 not very convenient num-
bers to carry around. We see that c has the dimensions of a
velocity, [c] = [distance][time]–1 and h̄ those of an action, namely
[h̄] = [energy][time]. Therefore, we are going to choose a system
of units in which these two constants are represented by dimen-
sionless numbers equal to 1. This means that we shall measure
all speeds as fractions of the speed of light and all actions as mul-
tiplets of h̄. In this system the dimensions of all physical quantities
are related and we find, for example,

[distance] = [time]
[mass] = [energy] = [distance]–1

With this choice there remains only one unit to complete our
system and we choose the electronvolt, written eV, as the unit of
energy.2 We shall mainly use its multiplets, the megaelectronvolt,
1 MeV = 106 eV and the gigaelectronvolt, 1 Gev = 109 eV. The pre-
vious relations show that time and distance will be measured in
inverse eV and, using the numerical values of c and h̄, we find the
approximate relation:

10–15 m = [200 MeV]–1 (1.1)

We often call the distance 10–15 m one fermi, denoted by 1 f.

1 The units are joule × second. 1 J = 1 kg m2 s–2 . The number in parentheses


in the value of h̄ represents the experimental uncertainty. There is nothing sim-
ilar in the value of c because, since 1983, this value is part of the definition of the
international system of units.
2 It is defined as the energy of an electron accelerated by a potential differ-

ence of 1 volt. 1 eV = 1.602176565 10–19 J.


2
A Brief History of Cosmology

In contradistinction with astronomy, whose origins go back


to the dawn of humanity, cosmology, as a natural science, is
relatively young. As late as the beginning of the last century
astronomers believed that, at large scale, the Universe was static.1
No evolution was perceptible. This belief was based on the fact
that the optical instruments of the time limited the observations
to the immediate vicinity of our galaxy. The same way that our
ancestors, who could not see much further than their horizon,
thought that the Earth was flat, Einstein’s contemporaries were
convinced that the Universe had no history. Religious beliefs
notwithstanding, the Universe appeared to have no evolution,
neither beginning, nor end.
We credit Edwin Powell Hubble2 as the first to shatter this world
of eternal stillness; but the real history is more complex. It is true
that when Albert Einstein, in around 1915, was looking for solu-
tions to the equations of general relativity, following his time’s

1 This notion of ‘large scale’ is not well defined. Firstly, because it evolves
with time: what was ‘large scale’ for early twentieth century astronomers is no
longer the case. Secondly, because it depends on the object we are studying.
Obviously, scientists did not wait until last century to understand that the solar
system is not static. Already in 1755 the philosopher Immanuel Kant had pub-
lished a study arguing that the solar system could have been formed out of a
gas condensed under the influence of gravity. But this idea of creation follow-
ing the laws of physics had not been applied to large structures, let alone to the
entire Universe.
2 Hubble is a rather strange character for a scientist of such reputation. He

was born in 1889 at Marshfield, a small town in the state of Missouri in the
United States. Apparently, he had been passionate about astronomy since he
A Brief History of Cosmology 7

prejudices, he was looking for a static solution, i.e. a solution


independent of time. Such a solution would describe a Universe
without evolution. In fact, in 1917 he introduced the concept of
a cosmological constant following this motivation. An example of a
discovery made for the wrong reasons.
The first voices to challenge this vision of immobility arose
around 1920 on two fronts: firstly from the side of mathematics
and then from that of physics.
Already in 1917 the Dutch mathematician Willem de Sitter
(1872–1934) had shown that the equations of general relativity
in the presence of a cosmological constant, contrary to what
Einstein thought, admitted a solution describing a Universe in
very fast expansion. At the time this solution was viewed as
a mathematical curiosity, but today we have good reasons to
believe that it describes the present state of our Universe. A little
later, in 1922, another mathematician, from the Soviet Union this
time, Alexander Alexandrovich Friedmann (1888–1925), studied
Einstein’s equations in the presence of a homogeneous and iso-
tropic mass distribution. He also obtained non-static solutions of
an expanding universe. Einstein thought at the beginning that
the solution was wrong, but he was soon convinced of its validity.
The existence of solutions is one thing, but the physical reality
could be different. Is the world we observe static, or not? In 1927
the Belgian physicist Georges Lemaître, a catholic priest but also

was a child, but during his early studies at the University of Chicago, he was
mainly distinguished for his performances in sports. His records would have
allowed him today to become a professional. He was more or less forced by his
father to enter the Law School, a subject which did not interest him at all. It
was only after the death of his father that he finally followed his own path and
obtained his PhD in astronomy in 1917. He had to interrupt his studies for two
years because of the First World War and it was in 1919 that he obtained his
first position at the Mount Wilson Observatory, in California. The Director of
the Observatory was the famous astronomer George Ellery Hale (1868–1938),
who started the construction of the largest optical telescope of the time, with
a diameter equal to 100 in. It was again Hale who later built the telescope of
Mount Palomar (diameter, 200 in). As a result Hubble was able to work all his
life with the world’s best instruments.
8 The Origin of Mass

an expert in general relativity, wrote the first formulation of the


theory which today we call the theory of an expanding Universe.3
Lemaître’s work was based on observational data. This was its
great merit. Since the late nineteenth century, many astronomers
had tried to measure the relative velocities of celestial bodies.
The best data came from the American Vesto Melvin Slipher
(1875–1969), of the Lowell Observatory in Arizona. How can we
measure the speed of an object several light-years away from us,
at a distance often poorly known? Slipher applied a spectroscopic
method which has since become classic. Using fine spectroscop-
ical analysis of the light received from a given source he observed
that the frequencies corresponding to known atomic transitions
were redshifted with respect to those measured on earth. This
phenomenon is due to the relative speed with which the source
moves away from us; the analogue of the frequency change of
the sound of a police siren when the car is moving. It is the fam-
ous Doppler–Fizeau effect.4 The result of these measurements
was a kind of ‘map’ containing many luminous sources, not with

3 If we consider atypical figures in science, the abbot, and later bishop,


Lemaître is certainly one. He followed parallel studies in mathematical phys-
ics and theology. He too had to interrupt his studies because of the First World
War, during which he served as an artillery officer. He entered the theology
seminary in 1920 and was ordained as a priest in 1923. As a scientist he travelled
all over the world and visited the best universities, Cambridge, Harvard, MIT,
CalTech, where he met the most famous physicists, such as Einstein, Eddington,
and Hubble. He studied with Eddington and obtained his PhD from MIT. All his
life he insisted on keeping distance between his two specialities, theology and
science and constantly opposed any attempt to connect them. After he intro-
duced the theory known as ‘the Big Bang’, he wrote: ‘As far as I can see, such a
theory remains entirely outside any metaphysical or religious question’.
4 The first to try this method was Christian Andreas Doppler (1803–53) him-

self in his 1842 article. Doppler was not an astronomer and his ideas on this
subject were rather primitive. His article is an extraordinary mixture of an
ingenious idea with totally wrong assumptions. The person who understood
the power of the method was Armand Hippolyte Louis Fizeau (1819–96). In a
conference in la Société Philomatique de Paris in 1848 he presented the method and
insisted on the importance of the frequency shift in the spectral lines. From
an observational point of view, the first to obtain reliable results was Hermann
A Brief History of Cosmology 9

respect to their positions in the sky, but to their velocities relat-


ive to the earth. The exact positions of each of these sources were
little known and, even worse, the very existence of celestial bodies
outside our own galaxy was not generally admitted.
Hubble’s first great contribution was, precisely, to give a con-
clusive answer to this question. Using the excellent resolution of
his Mount Wilson telescope, he could compare the luminosities
of a large number of sources and establish unambiguously that
many of them lay well outside our galaxy, and, even more so,
some were forming entire galaxies. Although this ground break-
ing result was not immediately accepted by all astronomers, it
is generally considered that 1925 marks the birth of extragalactic
astronomy.
Lemaître had access to dual information: the distance of many
luminous sources as well as their relative speeds with respect to
the earth. In 1927 he established an astonishing correlation: the
galaxies run away from each other with speeds which appear to be
proportional to their respective distances.5 The further away they
are, the faster they appear to move. Lemaître had proof that the
Universe was following the laws of the general theory of relativ-
ity. In 1930 he published the work for which he is best known,
containing a model for the origin of the Universe, which he called
The primitive atom. It was the first formulation of the theory which
remained in the literature as The Big Bang. Figure 2.1 shows the
founding fathers of modern cosmology.
The purpose of this book is not to present the evolution of
our ideas on the Cosmos. We just want to show that, among the
great scientific revolutions of the twentieth century, there was
Carl Vogel (1841–1907), Director of the Potsdam Observatory, who, in 1892,
published a catalogue with the radial speeds of 51 sources.
5 Lemaître first published this work under the title ‘A homogeneous

Universe with constant mass and increasing radius’ (Un univers homogène de masse
constante et de rayon croissant), in an obscure journal, The Annals of Brussel’s Scientific
Society. This is probably the reason why this law of speed–distance proportion-
ality does not bear his name, but it is known as Hubble’s law, who established it
independently with better precision two years later, in 1929.
10 The Origin of Mass

Figure 2.1 The ‘fathers’ of cosmology: Albert Einstein (1879–1955,


Nobel Prize 1921); Edwin Powell Hubble (1889–1953); Georges Lemaître
(1894–1966). Lemaître : Univ. C. Louvain , G. Lemaître Archives.

one which gave birth to a new science, that of cosmology.6 Since


the Universe evolves in time, it is natural to study this evolution,
describing the history of the Universe from the most remote past,
to its eventual fate in the most distant future.
The reader will certainly not be surprised to learn that this
young science reached its age of maturity together with the devel-
opment of high performance technical means of observation.
Today these include: ground based stations (very large optical
telescopes with sophisticated adaptive optics, radio-telescopes,
cosmic ray observatories, . . . ), high altitude balloons, and, over
recent years, a large arsenal of satellite and space stations.
Something we must keep in mind is that, in the Cosmos, ‘far’
means ‘old’. When we receive a letter from a friend who lives in
a faraway country (this example refers to the times when people
were still exchanging letters!), we do not learn his news of the day,
but of the day the letter was posted. Similarly, when we observe
6 People attribute to Merleau-Ponty the following poetic formula: ‘. . . a genius

of physics (Einstein) and a gigantic telescope (Mount Wilson) used by an astronomer of com-
parable status (Hubble) brought to natural philosophy, the first an idea (general relativity),
the second a vision of the Universe (the expansion), and we do not know which one was more
surprising and more thrilling.’ We have just seen that, in spite of its poetic value, the
formula is a historical shortcut.
A Brief History of Cosmology 11

a distant galaxy we do not see its ‘present’ state, we see the state
in which it was when the light started its long journey to reach
the earth. This way, by looking further and further away, we
obtain an increasingly older picture of the Universe; we follow
with our very eyes the birth and evolution of the celestial bodies,
the history of the Cosmos.
In the 1920s cosmology ceased to be a subject of philosophy and
became part of the natural sciences; but with a fundamental dif-
ference. A traditional scientific method proceeds through three
distinct steps: observation, elaboration of theoretical models and,
finally, their experimental testing. Experiment is the repetition
of observation under controlled conditions. It is only through
experiment that a theoretical model is validated and it is only
through such validation that it is promoted to a physical the-
ory. But in cosmology this last step is missing. The experiment
occurred once and we can only observe the results. Any men-
tal representation we can build on the evolution of the Universe,
even if it is corroborated by an impressive number of high preci-
sion observations, remains, from an epistemological point of view,
as a purely theoretical hypothesis.
Figure 2.2 shows an artist’s view of our present ideas on the
evolution of the Universe. It is the theory of the Big Bang.7 We
often find in the popular literature rather imaginative descrip-
tions of this theory. We say that ‘big bang’ means ‘great explosion’,
a singular event which caused the creation of space and time. It
was, to a certain extent, Lemaître’s point of view: at the beginning
there was a point; it was the only point that existed and there was
nothing else outside that point. Warning! You should not imagine
the point inside an ambient space. The point was all the space. And then

7 It seems that this rather pejorative name was given to the theory by one of

its opponents, the famous British astronomer and cosmologist Sir Fred Hoyle
(1915–2001). Hoyle had his own theory, known as the steady state theory, which
assumed a stationary Universe. This theory has been abandoned today because
it does not explain in a natural way recent high precision measurements of the
microwave background radiation (see footnote 11).
12 The Origin of Mass

Figure 2.2 The evolution of the Universe according to the Big Bang.
The axis represents the age of the Universe but also the ambient temper-
ature expressed in eV (1 eV ∼ 12,000 kelvin). The numerical values are
approximate and are shown only for indicative purposes. LBL, Berkeley.

the point exploded. This cosmic explosion produced an infinite


amount of energy which gave rise to the entire Universe.
Although this image offers a certain approximation to this
unique event, it is based simply on an extrapolation of our present
knowledge, without any serious justification. We do not really
know whether such an explosion did take place. Looking at
Figure 2.2 we see that as we go closer to the origin, we encounter
conditions characterised by extreme values of temperature (T)
and matter density (D). The figure stops at a time around 10–43 s.
Our present knowledge does not allow us to imagine the state
of matter prior to this time. We think that the values of temper-
ature and density were such that any description would require
the combined effects of gravity and quantum mechanics. But,
for the moment, we do not have the right equations for such a
A Brief History of Cosmology 13

description. The extrapolation to earlier times is not justified and


the ‘Big Bang’ is the expression of our ignorance of the state of
matter under these extreme conditions.
Following Figure 2.2 we see that, as time passes, the Universe
expands and cools down. We arrive soon at values of T and D
accessible to our experiments. We cross ‘the wall of ignorance’ and
enter familiar grounds.
At present this wall of ignorance stands at around 10–12 seconds.
In the figure it is marked ‘LHC’. What happened at earlier times
we can only speculate. But after this time our equations are
solidly anchored to experimental results. The evolution of the
Universe follows precisely the laws of physics as established in our
laboratories.
It is now the right moment to talk about this surprising con-
nection between the infinitely large and the infinitesimally small
we alluded to in the introduction. The study of the first is made
through telescopes, that of the second through microscopes. A
priori one wouldn’t expect to find any connection between the
two. Nevertheless, such a connection is not difficult to under-
stand: the collisions between the particles we study in our accel-
erators reproduce, to a certain extent, a microscopic image of
the conditions which prevailed in the early Universe. This is the
reason why we put along the time axis in Figure 2.2 the values of
the energy which are equivalent to each temperature. This way
we can see how the ‘wall of ignorance’ moves to earlier times fol-
lowing the increase in the energy of our accelerators. At present
the most powerful accelerator is the LHC (large hadron collider) of
the European Organisation for Nuclear Research (CERN) at the
border of France and Switzerland near Geneva. It is the most
powerful microscope man has ever made. Its resolution power
goes down to 10–19 m and its energy, which reaches 1013 eV, cor-
responds to a time t = 10–12 s in Figure 2.2. From this time on, the
evolution of the Universe follows our familiar laws of physics.8
8 As shown in the figure, the very high energy cosmic rays go above the LHC
energy, but the flux which reaches the earth is very small.
14 The Origin of Mass

At t < 10–12 s the values of temperature and density are very


high, above anything we have studied in our laboratories. We
see in the figure that, at very early times the expansion of the
Universe is very fast. We call this the inflation era. Because of it, the
entire Universe which is visible to us today, comes from a tiny
region in these very early times (10–34 s). This is shown in the fig-
ure. Although we have no direct observation of this epoch, we
have convincing indirect evidence of its occurrence. Between this
time and the ‘wall of knowledge’ of 10–12 s, the Universe consists
of a very hot gas of elementary particles. Among them there are
certainly those we know today and whose properties are presen-
ted in Appendix 1. However, we have good reasons to believe that
this primordial gas contains in addition particles which have not
so far been identified. We hope that the LHC will discover them.
In Chapter 4 we show that most of these particles in the prim-
ordial gas had a mass equal to zero, including those we know
today, such as the electron, which appears in our experiments to
be massive.
The Universe continues to expand. The temperature drops. We
see in Figure 2.2 that at a temperature of the order of 300 GeV
(1 GeV = 109 eV), we find a strange phenomenon marked as ‘EW-
SSB’, which stands for electroweak spontaneous symmetry breaking. What is
this? In everyday life we are used to the phenomenon of phase
transitions. If we cool water we observe a sudden change: at
T = 0 ◦ C the water turns to ice. The water molecules are still the
same, but the macroscopic properties of the system are radically
different. We call this phenomenon a phase transition and we know
many other examples of this kind. We believe that such a phase
transition occurred for the entire Universe at t ∼10–11 s after the
Big Bang, i.e. after a very short time. We are going to justify the
words ‘electroweak’ and ‘spontaneous symmetry breaking’ later,
but for the moment, we want to point out that during this phase
transition a fraction of the energy produced by the explosion was
transformed into mass, according to the well-known Einstein
formula E = mc2 , where E is the energy, m the mass, and c the
A Brief History of Cosmology 15

Figure 2.3 The mass generation mechanism: Robert Brout (1928–


2011). Brout died in 2011 and did not experience the triumph of
the theory to the elaboration of which he had contributed; François
Englert (1932–, Nobel Prize 2013); Peter Higgs (1929–, Nobel Prize 2013);
Brout : F. Englert archives; Englert : F. Englert archives; Higgs : Nobel
Foundation archives.

speed of light in vacuum. Most particles became massive, with the


exception of the photon, which remained massless.
Fifty years ago, in 1964, three theoretical physicists, the Belgian
François Englert, the American Robert Brout,9 and Peter Higgs
from Britain (Figure 2.3) proposed a mechanism which could
explain this phase transition. This mechanism implied the exist-
ence of a new particle, commonly named the Higgs particle,10 which
has all the properties of the newly discovered particle at CERN.
The Higgshunting lasted half a century, but it has been successful.
This explains the great emotion which accompanied the discov-
ery: it was not just a new particle, it was the element which could
shed light on the mystery of mass generation in the Universe.
It is the subject of this book and in the following chapters we
shall guide the reader through this extraordinary adventure (see
Figure 2.3).
But let us come back to Figure 2.2 and the evolution of the
Universe. We see that the expansion and the resulting cooling of
the Universe continues. The quarks bind among themselves and
9 Brout spent most of his professional life in Brussels and obtained Belgian

nationality.
10 In this book we shall often call it the Brout–Englert–Higgs or BEH particle,

which is historically more accurate.


16 The Origin of Mass

form protons. At even lower temperature the first atomic nuclei


are formed. They are those of the light elements, mainly helium.
It is the epoch of nucleosynthesis. We are at one minute after the
Big Bang.
Time goes on and the temperature drops further. Between
three and four hundred thousand years later it has fallen to a level
such that thermal agitation is no longer capable of keeping the
electrons free. They bind to the nuclei and form the first atoms.
Matter becomes electrically neutral and the photons, which
interact only with electrically charged particles, can propag-
ate freely. We call this the moment of decoupling, which means that
the photons are no longer coupled to matter. The latter is an
amorphous gas filling all space. It consists of atoms of light ele-
ments, mainly hydrogen and helium. Very slowly, under the
action of gravitation, gaseous masses start concentrating. It is a
very slow process; it takes several hundred million years for the
first celestial bodies—we call them proto-galaxies—to start form-
ing, and even more for the first sufficiently massive bodies to
appear. Under their gravitational attraction nuclear reactions
become possible in their interior and they start radiating. They
are the first shining stars . . . and there was light. The Universe
starts taking its familiar form.
These sketchy notes do not pretend to give a precise account
of the evolution of the Cosmos. There exist many specialised
books which the reader may consult. In fact, the Universe has
gone through several other phase transitions which we have not
presented. Some are shown in Figure 2.2. If today we ask the ques-
tion: what is the Universe made of, which are the constituents
of its total energy content, we shall get a surprising answer. The
visible mass, i.e. stars, galaxies, interstellar gas, everything we can
see, counts for less than 5% of the total. A much larger part, of the
order of 25%, seems to be composed of some sort of matter which
does not interact at all with light and, therefore is not visible. We
call it dark matter and we believe that it contains unknown neutral
particles. We hope that LHC will discover them. The remaining
A Brief History of Cosmology 17

part, of the order of 70% of the total energy balance, is not made
out of any form of matter. It is a kind of diffuse energy density
which manifests itself by causing an acceleration of the expansion
of the Universe shown in Figure 2.2. We call it dark energy and its
precise nature is unknown. It may be related to the cosmological
constant Einstein introduced in 1917 on the basis of wrong reason-
ing. It is responsible for the acceleration of the expansion which
has been observed in recent times and is shown in the figure.
At first sight this story resembles the scenario of a science-
fiction movie and we want to show here that it is real science and
not fiction. We said earlier that our views on cosmology are based
on the results of observations. Their detailed presentation goes
beyond the scope of this book, but we want, at least, to explain
their nature.
First question: by looking at stars and galaxies, how far back can
we go towards the early Universe? The answer is simple: as far
back as there were luminous objects to send us light. With our
present instruments we can observe the earliest proto-galaxies,
the most distant celestial bodies. Figure 2.4 shows a picture taken
by the satellite Hubble, launched in 1990. It is not obvious how
to interpret this, but astronomers have been able to extract, out
of the background of known stars, the signal coming from the
proto-galaxies. Among its numerous instruments, Hubble had an
ultra sensitive CCD (charge-coupled device) camera with a field of vis-
ion as narrow as a few fractions of a degree. It could point to
star-free parts of the sky and take long-exposure pictures. With its
high sensitivity it discovered these objects which are the first light
sources in the history of the Universe. They are almost 13 billion
years older than our time and represent the last step in the history
of the Cosmos we have presented (see Figure 2.4).
Second question: can we ‘see’ anything older? Answer: no, if we
search only for luminous objects, because before that time there
were none. We should change our instruments and try to detect
the diffuse radiation which existed in the Universe before the cre-
ation of any celestial body, when all matter was a hot soup of
18 The Origin of Mass

Figure 2.4 A picture taken by the space telescope Hubble. It shows the
‘young’ Universe at age one billion years.

atoms and photons. This was at the moment of the decoupling


which we placed at a few hundred thousand years after the Big
Bang. These photons travel freely and we can try to detect them.
Because of the Doppler–Fizeau frequency shift, we expect to see
them in the domain of microwaves.
With the development of high performance detectors over past
decades we have been able to obtain very precise measurements of
this radiation.11 Its fundamental importance is due to the fact that
11 The first observation of this cosmic radiation, of the utmost importance

to cosmology, was accidental. In 1964, Arno Allan Penzias (1933–, Nobel Prize
1978) and Robert Woodrow Wilson (1936–, Nobel Prize 1978), from the Bell labs,
A Brief History of Cosmology 19

Figure 2.5 The Universe seen by Planck.

these photons have remained free from interactions throughout


the long evolution, from three to four hundred thousand years
after the Big Bang, to our era. Consequently, they bring us some
of the most ancient information we can get. With them we have a
‘picture’ of the Cosmos long before the formation of any celestial
body, when the Universe was amorphous with no well-defined
structure.
Figure 2.5 shows the latest measurements obtained by the
European space mission Planck. Launched in May 2009 it gave us
the most precise map of this radiation. The average temperature
corresponds to 2.7 K or, approximately, –270 ◦ C. It is remarkably
homogeneous with variations of the order of one part in 105 . In
the picture these variations are shown using a chromatic code,
red for the ‘hot’ regions and blue for the ‘cold’. Astrophysicists

had constructed an ultra sensitive (for the time) antenna for radio-astronomy
research. To their great surprise, they found that they were hindered by a dif-
fuse background radiation which appeared to come from everywhere. They
thought that it was the result of interferences from terrestrial sources and they
tried, with no success, to eliminate it. It was Robert Henry Dicke (1916–97), an
astronomer from Princeton, who first understood (i) that it was the cosmolo-
gical microwave radiation predicted by the Big Bang theory, and (ii) that it was
the most important discovery in cosmology since that of the expansion of the
Universe.
20 The Origin of Mass

are able to analyse these results and extract information on the


conditions prevailing at the primordial Universe. They provide
the most severe constraints on our cosmological models. In par-
ticular, all models without expansion, like Hoyle’s steady state
theory which we mentioned earlier, are essentially eliminated see
Figure 2.5.
This is the most ancient signal that has been directly observed.
As we have said already, before this era matter was ionised, elec-
trons were not bound to protons to form atoms, and photons
could not propagate freely. Thus, we cannot observe them today.
For even more ancient times we have only indirect indications.
For example: the number and nature of the various elementary
particles we observe today, tells us something about the condi-
tions prevailing at their creation, (the first fractions of a second
after the Big Bang) or, the relative abundances of light elements,
do the same for the moment of nucleosynthesis.
Could we ever observe the ‘first moments’ of creation? It is
humanity’s old dream, to see the ‘beginning’ of the world! This
is not absolutely excluded, but we must resort to different kinds
of ‘messengers’, other than photons. The problem is that light
cannot propagate through a dense, or ionised medium. We can-
not see behind a wall, the same way we cannot see the interior of
a massive star, or the first moments of the Big Bang. Are there any
messengers capable of doing so? We know of two, neutrinos and
gravitational waves.
The neutrinos are elementary particles among those we find in
Table A1.3. They are produced during nuclear reactions and they
are among the first particles to be created in the early Universe.
They interact very weakly with matter and they can go almost
unhindered through a massive star. Using solar neutrinos we
study the interior of the sun.
Gravitational waves are emitted from the acceleration of
massive bodies, something analogous to the electromagnetic
waves which are emitted from the acceleration of electrically
charged particles. Until recently we have only had indirect
A Brief History of Cosmology 21

evidence for their existence, through looking at binary stars, but


in September 2015 the first direct observation of gravitational
waves produced by the collision of two massive black holes has
been reported.
In principle, either of these messengers, neutrinos or gravita-
tional waves, could bring us direct information about the first
moments of creation. Unfortunately, in both cases we must
increase our detection capabilities by many orders of magnitude
before we can achieve this goal. An open window on the Big Bang
is still science-fiction.
3
Symmetries

In human history the concept of symmetry antedates that of


science. Its importance transcends the scientific domain and
extends to fields such as art and philosophy. Even a superficial
account of its historical development goes beyond the scope of
this book, as well as the competence of the author. We shall limit
ourselves to a brief exposition of our present ideas on symmetry
in microscopic physics.
In everyday language, symmetry usually invokes ideas from
classical art (see Figure 3.1); likewise in science. There is an old
theoretical prejudice, according to which, the ‘best’ theory is the
most ‘symmetric’ one. We shall see that this prejudice has often
guided us in the search for the theories of nature.
We start this chapter with an abstract idea which we shall illus-
trate with some simple examples. In physics the notion of a symmetry
follows from the assumption that some variable is not observable. Therefore, no
physical quantity can depend on it.
Let us consider an example: the most symmetric solid is a
sphere. We can use this property to give an abstract definition
of the sphere as the solid which looks the same no matter from
which angle one looks at it. Consequently, the equations which
define the sphere should not depend on the angles. Indeed, the
equations, which take the form

x2 + y2 + z2 = R2 , (3.1)

are independent of the angles. It is easy to show that this property


of angle independence is equivalent to equation (3.1).
Symmetries 23

Figure 3.1 The concept of symmetry is present in almost all artistic cre-
ations. However, rare are the works which present a perfect symmetry.
In most cases it gives a general impression, but in the details it is broken.
Here we see a reproduction of the Parthenon’s western pediment.

We can generalise this idea. A physical system is symmetric if


the equations that describe it remain invariant when we change
the value of some variables. There is a profound theorem, proven
by Amalie Emmy Noether,1 which states that such an invari-
ance implies a conservation law. These conserved quantities play
1 Amalie Emmy Noether (1882–1935; Figure 3.2) is the first great female
name in modern mathematics. Daughter of the German mathematician Max
Noether, she studied mathematics and theoretical physics at the University of
Erlangen, at a time when women did not often follow University studies. In
spite of her immense talent and the famous results which bear her name, she
did not have an easy life. She was invited by David Hilbert and Felix Klein to
join the Mathematics Department of the University of Göttingen, probably
the most famous mathematics department of the times, but her appointment
was refused by the conservative majority of the Faculty who could not admit a
woman to the rank of Professor. She later obtained tenure at Göttingen where
she stayed until 1933, when she was forced to leave because of her jewish origins.
She emigrated to the United States where she died following surgery in 1935.
24 The Origin of Mass

Figure 3.2 Amalie Emmy Noether (1882–1935).

an important role in our understanding of physical phenomena.


Some examples follow.

3.1 Space symmetries


A preliminary notion. In order to specify a point in space we use
a coordinate system. Often it is a system of three orthogonal axes, like
those shown in Figure 3.4. The position of a point is given by three
numbers (x, y, and z), which denote the distances between the ori-
gin of the coordinate system and the projections of the point’s
position on the three axes.
By ‘space symmetries’ we mean the changes of the coordinate
system which leave the dynamical equations invariant. They are
She was 53 years old. Among her students in Göttingen we find well-known
mathematicians, such as the Dutchman, Bartel Leendert van der Waerden. The
best scientists among her contemporaries, both mathematicians and physicists,
had a great respect for Noether’s work. Einstein wrote that she was ‘. . . the
most creative mathematical genius produced since women had access to higher
education.’
Symmetries 25

Figure 3.3 Euclid: The founder of geometry.

r = (x, y, z)

y Y
O

Figure 3.4 A coordinate system.

the simplest symmetries to understand intuitively. We shall


present here examples of translations and rotations. We find them
implicitly in the work of all geometers of antiquity, but the per-
son who formulated them clearly was Euclid (see Figure 3.3) in
his famous book The Elements, in which he gives the first axiomatic
26 The Origin of Mass

definition of geometry. Euclid wanted to define the notion of


equality between two geometric figures and set the axiom:

‘The superposable objects (figures) are equal’.2

Let us consider the example of two triangles. Euclid tells us that


they are equal if we can superpose one on the other. To proceed
towards this comparison we may have to perform two geomet-
rical operations. The first is a translation, which means that we may
have to move one of the triangles in order to bring its centre to
coincide with that of the other. The second is a rotation in order to
check whether the two are exactly superposable. Obviously, we
must assume that by applying either one of these two operations,
the triangle does not change. Euclid understood that this is an
intrinsic property of the space, independent of the other axioms;
if we want it, we must postulate it separately. We thus arrive at
the notions of translation, or rotation symmetry. We shall give a concrete
example presently.

3.1.1 Translation symmetry


Let us first consider translations. Let us assume that the space
is homogeneous and the absolute position of the origin of the
coordinate system is not a measurable quantity.3 It follows that
the dynamical equations should remain invariant under the
transformation consisting of displacing the coordinate system
by a constant vector: x → x + a (Figure 3.5). We call this trans-
formation space translation. We can show, using Noether’s theorem,
that the invariance of the dynamical equations under these trans-
lations implies the conservation of momentum.4 In Figure 3.5
we illustrate this hypothesis for the case of a free particle. Its
trajectory is a straight line represented by A. Under the effect of
2Kαι τα εϕαρμóζoντα επ άλληλα ίσα αλλήλoις εστίν.
3This corresponds to the intuitive idea that in a perfectly homogeneous
space there is no way to specify a position in the absence of any fixed object.
4 In classical physics the momentum of a particle is defined as the product

of its mass times its velocity.


Symmetries 27

x' = x + a

z
A

a
y

A'
x

Figure 3.5 By space translation the image of a free particle trajectory A


is the straight line A .

a translation by a constant vector a, the line A becomes A . The


symmetry tells us that A is also the trajectory of a free particle.

3.1.2 Rotation symmetry


The same reasoning can be applied to the symmetry under rota-
tions. The physical assumption is that space is isotropic and has no
privileged direction. As a result, the particular orientation of the
coordinate system is not measurable and we can perform rota-
tions which will not affect any physical quantity. We say that the
latter are rotationally invariant. Noether’s theorem implies that to this
invariance corresponds the conservation of angular momentum.
A remark is necessary here. It must be clear that the hypothesis
of space homogeneity, or isotropy, applies to empty space. The
presence of fixed bodies affects these properties. For example, near
the earth, space is not translationally invariant: we do not breath
as easily at the summit of Mont Blanc as we do in the plain. The
same for rotations: a stone always falls vertically, which means
that the presence of the earth induces a privileged direction, the
one pointing to its centre.

3.1.3 Space inversion


Before closing this section on space symmetries we want to intro-
duce a third transformation of the coordinate system which will
be useful later: it is the space inversion x → –x, or, in coordinates,
28 The Origin of Mass

−x

−y y

−z

Figure 3.6 The mirror image of the coordinate system (x, y, z) is the sys-
tem (x, y, –z). It is obtained by an inversion followed by an 180◦ rotation
around the z axis.

(x, y, z) → (–x, –y, –z). In physicists’ jargon this transformation


is often called parity.5 It would have been more appropriate to
call it mirror symmetry, because the image of the coordinate system
(x, y, z), seen in a mirror placed, for example, on the (x, y) plane,
is the (x, y, –z), which is obtained by inversion followed by an
180 ◦ degree rotation around the z axis; see Figure 3.6. The same
figure also shows another aspect of space inversion: it is equi-
valent to the transformation left ↔ right. Indeed, the mirror
image of a left hand appears as a right hand. We must also stress
here that a space inversion transformation cannot be reproduced
by a sequence of rotations. Parity is an independent transforma-
tion.6 The equations of classical physics, like Newton’s equation
for mechanics, or Maxwell’s equations for electrodynamics, are
5A technical remark: contrary to translations, or rotations, we cannot visu-
alise inversion as a sequence of small transformations. In mathematics we
call translations or rotations continuous transformations, while inversion is a discrete
transformation. This technical difference has a physical consequence: Noether’s
theorem does not apply and there is no associated conserved quantity.
6 It seems that it was Lev Davidovich Landau (Nobel Prize 1962) who stated

the famous aphorism: ‘An acrobat may jump in the air and turn around as
many times as he wants, his heart will always remain at his left!’
Symmetries 29

invariant under parity transformations, so physicists were con-


vinced that this invariance was an exact law of nature. Therefore,
it came as a great surprise when an experiment performed by
Chien Shiung Wu (1912–97) of Columbia University, showed that
this invariance was violated by the weak interactions of nuclear
physics (see Appendix 1).

3.2 Time symmetries


As we explained previously, in order to specify a point in space we
must give three numbers, its three coordinates. In order to spe-
cify an event we need four: the first three will tell us where and
the fourth when the event took place. Therefore, already in clas-
sical physics, time enters as a ‘fourth dimension’ in our equations.
Time is measured starting from an arbitrarily chosen origin. As
we did for space, we assume that this choice is unimportant; phys-
ics should not depend on the way we measure time, be it since
the birth of Christ, or the foundation of Rome. Consequently
our equations should be invariant under time translations of the
form: t → t + τ , where τ denotes an arbitrary time interval.
Noether’s theorem tells us that to this invariance corresponds the
conservation of energy.
We can also consider the discrete transformation of time reversal
t → –t. Applied to a system of moving particles this transform-
ation results in the reversal of all velocities, since the speed of a
particle, which equals the distance covered in unit time, changes
sign under this transformation. As for space inversion, invariance
under this discrete transformation is not related to a conserva-
tion law. Classical physics is invariant under time inversion,7 but
very precise experiments had already shown, in 1964, that the
7 This statement requires an explanation. Newton’s equation, which
describes the motion of a particle, is indeed invariant under t → –t, but every-
day experience shows that very often the evolution of a macroscopic physical
system is time irreversible. Air moves from a high pressure region to one of
lower pressure, but not the other way around. We get older, but not younger.
The emergence of this macroscopic irreversibility out of microscopic equations
30 The Origin of Mass

interactions among elementary particles present a slight violation


of this invariance.
We have just seen the role of time as a ‘fourth dimension’ in
classical physics. In the framework of Einstein’s theory of spe-
cial relativity this notion acquires a more profound significance.
In a relativistic theory we can talk about the combined space-time
which admits transformations that mix time with the space com-
ponents. These transformations generalise in a non-trivial way
the ones we have presented so far and are part of our microscopic
theories. However, their detailed study goes beyond the scope of
this book.

3.3 Internal symmetries


All the symmetries we have seen so far concern transformations
of space and time. They are geometric transformations, in the
real sense of the word, easy to visualise and understand intuit-
ively. We shall need a certain degree of abstraction in order to
imagine transformations which do not affect the coordinate sys-
tem of space and time, but do change the dynamical variables of
the problem we are studying. We shall call the resulting sym-
metries internal symmetries. A characteristic example is provided by
Heisenberg’s theory of isotopic spin, first introduced in 1932, which
in today’s language, can be expressed as follows.
In Appendix 1 we point out that elementary particles often pos-
sess a proper angular momentum which we call spin. As is the case
with many quantities in quantum mechanics, spin can take only
discrete values, integer, or half-integer. For example, the spin of
the electrons equals 1/2. Like angular momentum, spin is a vector
in space and we can look at its projection on one of the axes of the
coordinate system, for example the z axis. Thus, the component
sz of the electron spin can take two values sz = +1/2 and sz = –1/2.

which are time reversible is a complex question which we shall not address in
this book.
Symmetries 31

electron with spin = +1/2

electron with spin = –1/2

Proton = (nucleon with isospin = +1/2)

neutron = (nucleon with isospin = –1/2)

Figure 3.7 A space rotation exchanges electrons with opposite spin. A


rotation in isotopic spin space exchanges protons with neutrons.

Strictly speaking we should say that we have two kinds of elec-


trons, those whose spin points upwards and those downwards, as
indicated in Figure 3.7. The reason why this distinction is useless
is the existence of transformations, to wit 180◦ rotations around
either the x or the y axis, which are symmetries of the theory and
transform an electron with sz = +1/2 to one with sz = –1/2, and
vice versa.
It is this spin analogy that Heisenberg used in his famous 1932
article on the nuclear forces. The constituents of nuclei are pro-
tons and neutrons.8 The former are electrically charged, the latter
are neutral. If we neglect this difference, the experimental res-
ults show that, as far as the nuclear forces are concerned, protons
and neutrons play very similar roles: their masses differ very little
and the nuclear energy levels do not change significantly if we
interchange one for the other. It is this exchange symmetry that
Heisenberg generalised to continuous transformations.9 He pos-
tulated that the system proton–neutron behaves as if there were
8 It is worth noticing that Heisenberg wrote this article in the same year that

the neutron was discovered.


9 In Heisenberg’s 1932 paper the symmetry under these transformations was

not complete. It was extended and completed in 1938 by Nicholas Kemmer.


32 The Origin of Mass

only one particle, which we shall call a nucleon, which can appear in
two states, either as a proton, or as a neutron. In this work Heisenberg
introduced for the first time a physical quantity which is not con-
nected at all with space-time. Today we call it isotopic spin, or isospin
and we attribute to it, formally, the same properties as those
of ordinary spin: it can take only integer, or half-integer values
and Heisenberg assumed that the nucleon has isotopic spin equal
to 1/2. Like spin, isotopic spin is assumed to be a vector. However,
it is not a vector of ordinary space, but of another, abstract space,
the isospace. In this abstract space we can consider a coordinate sys-
tem where the component tz of the nucleon’s isospin could take
two values, tz = +1/2, or tz = –1/2. Heisenberg identified the first
state with a proton and the second with a neutron. As we did with
the electron spin, we can avoid talking about two distinct particles
if we assume that the theory is invariant under rotations in this
new space. An 180◦ rotation around the x axis in isospace trans-
forms a proton into a neutron (see Figure 3.7). This is the concept
of an internal symmetry. The isotopic spin symmetry is approximate
because the previous reasoning neglected the effects due to the
electric charge of the proton. But, as far as the nuclear forces are
concerned, it is a good approximation. For the first time in physics
we have considered coordinate transformations other than those
of our familiar space-time.
From a purely mathematical point of view, if we have a sys-
tem described by dynamical equations, it is natural to associate
with it a space; it is the space in which all the transformations
act, leaving the equations invariant; in other words all the sym-
metries of the system. It follows that, for nuclear physics, the
‘space’ has seven dimensions, four of our familiar space and time,
and three for isospace. Indeed, this last one is three-dimensional,
isomorphic to our familiar space. However, the idea was later
generalised to more complicated spaces, as we were discovering
larger internal symmetries. The notion of space became abstract,
the space of microscopic physics became a mathematical multi-
dimensional object with a complicated topology, of which only
Symmetries 33

a part, the space of our everyday experience, is accessible to


our senses. We shall often use the term internal space to denote
this abstract space in which the transformations of an internal
symmetry act.

3.4 Local or gauge symmetries


With the exception of space or time inversions, all the transform-
ations we have introduced so far depend continuously on one
or more parameters. The space translations depend on the three
components of the vector a, those of time on τ , the rotations
in either ordinary space or isospace on the three angles of
rotation etc. These parameters are independent of the space-time
point (x, t).
We shall now introduce a second abstraction of the concept of
space by considering transformations depending on parameters
which are themselves arbitrary functions of (x, t). Such trans-
formations are called local, or gauge transformations.10 The physical
motivation to introduce this abstract notion is not obvious, but

Figure 3.8 Abstract symmetries: Werner Heisenberg (1901–76, Nobel


Prize 1932); Chen Ning Yang (1922–, Nobel Prize 1957); Robert Mills
(1927–99). Heisenberg : Max Planck Institute, Munich archives.
10 The origin of the term gauge goes back to the work of Hermann Klaus

Hugo Weyl (1885–1955) in 1918 in which he attempted to construct a the-


ory invariant under ‘scale’ transformations. He wanted to have a theory in
which the unit of length, ‘the gauge’, would change from one point of space to
34 The Origin of Mass

it has a mathematical, or even an aesthetic origin; the equations


we obtain by imposing this invariance have a richer mathem-
atical content and a higher predictive power. It is through its
consequences that the requirement for gauge invariance will be
justified.
Independently of one’s motivations, the first question we
should ask is: can gauge transformations be symmetries of a phys-
ical theory, in other words, can they leave its dynamical equations
invariant? At first sight the answer seems to be no!

3.4.1 Local translations


Let us look at space translations. The local ones can be described
by x → x + a(x, t). If the three components of a(x, t) are func-
tions of the space-time point (x, t), the image of the straight line A
which corresponds to the trajectory of a free particle, would be a
curve A (see Figure 3.9). No free particle would follow such a tra-
jectory. Indeed, the equations of motion, which in this case take

x" = x + a(x, t)

z
A

a(x, t)
y

A"
x

Figure 3.9 Local space translations.

another. It was natural to call such a theory gauge invariant. Even if this motiv-
ation was dropped later, Weyl continued to use the word ‘gauge’ for every
transformation whose parameters depended on the space-time point and this
terminology remained in the scientific literature. In this book we shall use the
terms ‘local’ or ‘gauge’ transformations interchangeably.
Symmetries 35

the simple form: acceleration = x¨ = 0,11 do not remain invariant


under the translation x → x + a(x, t). A particle will not follow a
trajectory like A , unless it is subject to specific forces.
Can we determine these forces? In other words, can we find the
dynamics which remains invariant under local translations? The
question seems to be only geometric, with no obvious physical
significance. Therefore we would expect to find an answer inter-
esting only in mathematics, but not in physics. Nevertheless, the
surprising result is that the dynamics we find does not describe
some obscure force, but one of the fundamental forces of nature,
the force of gravitation. The equations which remain invariant
under local translations are the equations of general relativity.
Einstein’s initial reasoning, which established general relativity by
trying to enlarge the applicability of the principle of equivalence,
uses precisely the same concepts.
We shall not prove this result here; we shall limit ourselves to
explaining an underlying physical principle. A simple dynamical
postulate states that the trajectory of a free particle between two
points in space is given by the shortest path that joins these points.
For the example of the particles of Figure 3.5, they are the straight
lines A and A . In mathematics, the shortest path between two
points is called a geodesic and it is a concept which characterises the
geometry of space. For example, on the surface of a sphere the
geodesic between two points is the arc of the circle which goes
through these two points and whose centre coincides with the
centre of the sphere. Let us come now to Figure 3.9. In order for a
free particle to follow the curve A , the latter should be a geodesic
of the ambient space. It is precisely what general relativity pre-
dicts: the presence of massive bodies deforms the geometry of the
space and straight lines are no longer geodesics. The conclusion is
that gravitation has a geometric origin.
Is this relation between geometry and dynamics accidental? Is it
limited to gravitation, or is it of more general validity?
11 A free particle is one which is subject to no force and therefore, according
to Newton’s equation, its acceleration vanishes.
36 The Origin of Mass

3.4.2 Internal gauge symmetries


In 1926, just two months after Erwin Schrödinger (see
Figure A1.5) had published the equation which bears his
name in quantum mechanics, the Russian physicist Vladimir
Aleksandrovich Fock (1898–1974) made a fundamental contribu-
tion which has not been fully appreciated, even by physicists. The
basic quantity which enters Schrödinger’s equation for a particle is
its wave function (r, t) (see Box 3.1). It is a complex valued function
of the space pointr and time t, and the square of its modulus gives
the probability density of finding the particle at the pointr at time
t. It follows that for a single particle, only the modulus of its wave
function has a physical meaning, the phase is not a measurable
quantity. In mathematical terms a change of phase of the wave
function is written as: (r, t) → eiθ (r, t), where θ is a constant
phase (see Box 3.1). Under this transformation Schrödinger’s
equation remains invariant and, following the discussion at
the beginning of this chapter, the phase must correspond to a
symmetry. It is an internal symmetry, because the transformation
does not affect the space-time coordinate system. In fact, this is
the first example of an internal symmetry in quantum theory.

Box 3.1: The wave function in quantum mechanics


The history of quantum mechanics is quite complex and there
exist several specialised books on the subject. Its origins go back
to the end of the nineteenth century with experiments on
the spectrum of black body radiation and the first results on
atomic spectroscopy. The first theoretical work which broke
away from classical ideas was that of Planck (see Figure A1.3)
who, in 1900, postulated that the energy exchanges between
matter and radiation are quantised, with quanta proportional to
the frequency of radiation. The proportionality constant is the
famous Planck’s constant, h. In 1905 Einstein followed this principle
and offered the first explanation of the photoelectric effect
in quantum terms. This period ends sometime between 1913
Symmetries 37

and 1916 with the formulation by Niels Bohr (see Figure A1.3)
and Arnold Sommerfelda of the quantisation rules for the
electronic orbits in atoms. This is the old quantum theory.
What was missing was a dynamical foundation. The rules
were postulated but they did not follow from any precise
equations. This need was fulfilled during the 1920s by Werner
Heisenberg (see Figure 3.8), who proposed a system of equa-
tions in matrix form, and by Erwin Schrödinger with his
differential equation. It is Dirac (see Figure A1.5) who showed
that the two formulations were, in fact, equivalent.
In 1923, Louis de Broglie (see Figure A1.3) proposed a cor-
respondence law between a particle and a wave. This is the
famous particle–wave duality law. Schrödinger, probably try-
ing to answer a question by Peter Debye,b made this corres-
pondence precise by writing a wave equation for de Broglie’s
matter waves. This is the famous Schrödinger equation.
Consider a particle of mass m in a potential V(r).
Schrödinger’s equation involves the wave function (r, t),
which satisfies:
∂ p2
ih̄ =  + V(r), (3.2)
∂t 2m
where p, with components px , py , and pz , is the momentum of
the particle which, in Schrödinger’s formulation, is given by

∂ ∂ ∂
px = –ih̄ ; py = –ih̄ ; pz = –ih̄ . (3.3)
∂x ∂y ∂z

The factor i = –1 which multiplies the left-hand side of
Schrödinger’s equation shows that the wave function  takes
a Arnold Johannes Wilhelm Sommerfeld (1868–1951).
b Petrus Josephus Wilhelmus Debije (1884–1966, Chemistry Nobel Prize

1936).
38 The Origin of Mass

complex values. Like every complex number, it can be written


in terms of two real functions, R (r, t) and I (r, t),

(r, t) = R (r, t) + iI (r, t), (3.4)

which we call the ‘real’ and the ‘imaginary parts’ of , respect-


ively. Only the modulus of , defined by

|(r, t)|2 = (R (r, t))2 + (I (r, t))2 (3.5)

has a physical significance. It gives the probability density of


finding the particle at the point r at time t. From a mathem-
atical point of view, it is clear that, if a function 1 (r, t) is a
solution of Schrödinger’s equation, so is any other function
of the form 2 (r, t) = C1 (r, t), with C an arbitrary complex
number. We say that Schrödinger’s equation determines the
wave function only up to a multiplicative constant. This arbit-
rariness is restricted by the probabilistic interpretation of .
The particle must be somewhere with certainty at time t,
therefore the wave function must satisfy the normalisation condition

|(r, t)|2 dr = 1. (3.6)

It follows that the modulus of the constant C must be equal to


one: |C|2 = 1. It is easy to see that a constant which satisfies
this unimodularity condition is given by a phase,

C = eiθ = cosθ + isinθ, (3.7)

with θ an arbitrary angle –π < θ ≤ π. Consequently, the


wave function is determined up to a phase.

Fock searched for the forces necessary to make this invariance


local. The answer, which is rather easy to obtain, is that they are
Symmetries 39

the electromagnetic forces. Schrödinger’s equation with a local


phase invariance, i.e. an invariance in which the phase of the wave
function θ is not a constant, but an arbitrary function of (r, t),
describes the motion of a charged particle in an electromagnetic
field. Electromagnetism also has a geometric origin, although
here the geometry is not that of ordinary space.
After this first success, we would expect that physicists would
have followed the same method to construct the gauge invariant
theory corresponding to isotopic spin symmetry, immediately
after the latter was introduced by Heisenberg in 1932. But here
history took an unexpected turn. The fascination that general
relativity had exerted on that generation of physicists was such
that, for many years, physicists were unable to conceive local
transformations of an internal symmetry space without, at the
same time, also introducing those of ordinary space.12 It was only
in 1954 that Chen Ning Yang and Robert Mills (see Figure 3.8)
wrote the theory known under their name and which is the gauge
theory invariant under local isotopic spin transformations. But
then we had to face a new problem, that of mass.

12 Attempts in this direction were made by the Swede, Oscar Benjamin Klein

(1894–1977) in 1937 and the Austrian Wolfgang Ernst Pauli (see Figure A1.7) in
1953.
4
A Problem of Mass

The concept of mass in elementary particle physics, introduced


in Appendix 1, is not different from the one we know in classical
physics. We have the inertial mass, which is the parameter entering
Newton’s equations, and the gravitational mass, which determines
the coupling of the particle with the gravitational field. The prin-
ciple of equivalence tells us that the two are equal. All known
particles have non-zero masses, with the exception of the photon
and, we believe, the graviton. But, as we have said in Chapter 2,
we have good reasons to believe that this has not always been the
case in the early Universe. There was a time in which most of the
elementary particles had zero mass. In this chapter we want to
explain the reasons.

4.1 Mass and the range of interactions


One of the basic principles of quantum theory is the duality
between a field and a particle. It goes back to de Broglie’s found-
ing work in which he postulated that for every particle there is
a corresponding wave. This was further enriched by the inter-
pretation of Niels Bohr and the Copenhagen School, according
to which these two aspects are dual to each other: for every phe-
nomenon there are experiments which illustrate the corpuscular
aspects and others which do the same for the wave aspects.
Let us take the example of electromagnetic forces. In classical
physics, the interaction between two charged particles at a dis-
tance R apart is described by the electromagnetic field created by
either of them at the position of the other. In quantum physics,
A Problem of Mass 41

to this picture we superpose ‘a dual one’, according to which the


interaction is the result of the exchange of one, or more, photons,
the particle that corresponds to the electromagnetic field. Figure
4.2(a) shows a graphic representation for the case of the interac-
tion of an electron and a proton: one of the two emits a photon
which is absorbed by the other. We say that the photon is the
‘mediator’ of electromagnetic interactions.
In 1935 Hideki Yukawa (Figure 4.1) extended this concept of the
forces generated by particle exchange, to nuclear forces. He pos-
tulated the existence of a new particle, the π meson, or pion, which
would be the mediator of the nuclear forces. Figure 4.2(b) shows
the one pion exchange process between a proton and a neutron.
The experimental discovery of the pion in cosmic rays, in 1947,
confirmed this idea which became one of the cornerstones in
the physics of elementary particles. As we explain in Appendix 1,
at present we have identified experimentally the mediators of
three out of the four fundamental interactions among element-
ary particles, to wit the strong, the electromagnetic, and the weak

Figure 4.1 Hideki Yukawa (1907–81, Nobel Prize 1949).


42 The Origin of Mass

e e n n

γ π

p p p p
(a) (b)

Figure 4.2 Forces produced by particle exchange: (a) electromagnetic


forces between two charged particles; (b) the nuclear forces between a
proton and a neutron.

interactions. The existence of a mediator for the gravitational


interactions, the ‘graviton’, remains as conjecture.
What is the meaning of such exchanges? Under which form are
these particles exchanged? Let us look at Figure 4.2(b). A proton
emits a pion which is absorbed by a neutron. The emission process
is p → p + π. In the initial proton’s rest frame the energy of the
state equals mp , the proton mass.1 The energy of the final state is
larger than mp + mπ , the sum of the masses of the two particles.
It follows that such a process is forbidden by energy conservation.
We often say that it is a virtual process, although the term is not very
meaningful. It becomes a bit more precise in quantum mechanics
because of a relation we can write as Et ≥ 1/2, which means
that the energy of a state is defined up to an uncertainty E
which becomes larger when the time of measurement becomes
shorter.2 We say that for a sufficiently short time the quantum

1We recall that we are using a system of units in which the speed of light c
and Planck’s constant h devided by 2π are both equal to one c = h̄ = h/(2π) = 1.
2 We often call this relation the ‘time–energy uncertainty relation’, but this

is very misleading. This terminology is reminiscent of the position–momentum


uncertainty relations xp ≥ h̄/2, which are fundamental in quantum mech-
anics. There is nothing comparable for E and t. Time is not an observable
associated with a particle. The meaning of the time–energy relation depends on
the kind of measurement we have in mind. For energy we often make frequency
measurements of the associated wave packet and, in classical wave mechanics,
we have the relation νt ≥ 1/2. Quantum mechanics enters via the fact that
A Problem of Mass 43

fluctuations of the energy become important and the emission


process becomes possible.
These remarks indicate that there exists a relation between the
range of an interaction and the mass of the exchanged particle.
For a large mass, E is large and therefore t must be short. A
short time implies a short range for the interaction. Although
the above arguments were hand-waving, it is straightforward to
derive a rigorous and precise relation, first obtained by Yukawa:
the exchange of a particle with mass m produces a potential, the
Yukawa potential, given by
e–mr
V(r) = , (4.1)
r
where V is the potential and r the distance. We see that V decreases
exponentially fast with distance with a characteristic constant
equal to 1/m. We also notice that, for m = 0, this constant goes to
infinity and we recover the Coulomb potential 1/r. The long range
of electromagnetic interactions is related to the zero mass of their
mediator, the photon.3

4.2 Gauge interactions


In Chapter 3 we introduced the notion of a transformation in
the form of a change of the coordinate system. This refers to
either the space-time coordinates, or to those of an internal
space. If the laws of physics remain invariant under such a trans-
formation, we say that we have a symmetry. We also introduced
the energy is proportional to frequency using Planck’s constant. It would have
been more correct to write this relation as E T ≥ h̄/2, where T represents the
time interval during which the energy of the system varies significantly. For
example, for an atom in its ground state, T can be as large as we wish and the
energy can be determined with arbitrary precision.
3 This result, as well as many others we shall derive in this book, is specific

to a three-dimensional space. For example, the electrostatic field created by


two opposite charges +q and –q, in a one-dimensional space remains constant,
independent of the distance.
44 The Origin of Mass

the notion of a global transformation, which is the same for all


points of space-time, as well as that of a local, or gauge transform-
ation. A gauge transformation is a generalisation of a global one
in which the parameters of the transformation depend on the
point of space-time. Figure 3.9 shows the case of space translations
r → r + α (r, t). The parameter is the vector α which determines
the translation and, for a gauge transformation, it is an arbitrary
function of the space-time point (r, t). We explained that invari-
ance under local transformations implies the presence of forces.
We indicated, without proving it, that the forces corresponding to
local translations are those of gravitation. Here we want to pursue
this idea a bit further; staying with Figure 3.9, let us consider two
points in space P1 and P2 , at a distance R apart. We can imagine
a function α (r, t) which vanishes everywhere except at the vicin-
ity of each one of the two points. The translation of the straight
line A of Figure 3.9 will give the curve à of Figure 4.3. We see that
the gravitational forces must have a range at least as large as the
distance R between the points P1 and P2 . Since this distance can
be chosen as large as we wish, we conclude that the invariance
under local translations implies a long range for the gravitational
forces. By the previous argument this means that their assumed
mediator, the graviton, must have a zero mass.
It is easy to generalise this argument to all forces produced
by gauge symmetries. They are all produced by the exchange of
mediators which in Appendix 1 we called gauge bosons. Since by
the previous argument the forces must have a long range, the
gauge bosons must be massless. In fact we have as many such

P1 R P2
A

Figure 4.3 Under a space translation with a vector α which is different


from zero only at the vicinity of the points P1 and P2 , the image of a free
particle trajectory A is the curve Ã.
A Problem of Mass 45

massless particles as independent gauge transformations which


are symmetries of our system. It is a direct consequence of the
geometric properties of gauge theories which establish correla-
tions between points at arbitrarily large distances. This result is
in perfect agreement with experiment for the two interactions of
classical physics, gravitation and electromagnetism, which have,
indeed, a long range. Contrary to this, it is in violent disagree-
ment with the other two interactions of microscopic physics, to
wit the strong and the weak interactions, whose ranges are meas-
ured to be 10–15 and 10–18 m, respectively. This is the first, and
the most fundamental problem of mass which has haunted gauge
theories since their formulation by Yang and Mills in 1954 up
to their first successful application by Steven Weinberg in 1967.

4.3 The masses of matter constituents:


quarks and leptons
In Appendix 1 we introduced two kinds of elementary particles:
force mediators and matter constituents. We also indicated that
particles with integer spin are bosons and those with half-integer
spin are fermions. The force mediators are bosons and we have just
seen that gauge invariance seems to force them to be massless.
We want to show here that the same result applies, albeit for
different reasons, to the matter constituents, quarks and leptons,
which are fermions.

4.3.1 Chirality
The constituents of matter are all particles with spin equal to 1/2
(see Appendix 1). We have already noticed that spin is a vector.
In classical mechanics the projection of a vector on an axis equals
the vector’s modulus times the cosine of the angle between the
vector and the axis. A remarkable result of quantum mechanics is
that the projection of a spin can take only discrete values: for the
46 The Origin of Mass

s = +1/2 v

s = −1/2 v

Figure 4.4 The two chirality states.

electron they can be +1/2 or –1/2, independent of which axis we


choose.4
For a spin 1/2 particle, such as an electron or a quark, moving
with a speed v, it is instructive to choose as a projection axis the
direction ofv. We call the resulting projection chirality. In Figure 4.4
we present the two possible values: +1/2 when the spin projection
is parallel to v and –1/2 when it is opposite.
In the physics jargon these two states are called right chiral-
ity and left chirality,5 respectively. Chirality is preserved by rota-
tions, because both vectors, spin and speed, turn the same way.
Conversely, the two states are exchanged by space inversion,
or parity, which we introduced in Section 3.1.3. We can under-
stand this by recalling that spin has the properties of angular
momentum, which in classical mechanics is proportional to the
vector product of the position and the speed vectors L ∼ r × v.
Under parity both change sign, so the angular momentum does
not. Since speed changes sign, we conclude that parity changes an
electron with right chirality to one with left.
These two states can also be connected with a second set of
transformations. Consider an electron with speed v and chirality
+1/2. We can perform a Lorentz transformation and go to the rest
frame of the electron, i.e. the frame in which its speed vanishes.
4 For a spin s, integer or half-integer, the possible values of the projection are

s, s – 1, s – 2, . . . , –s.
5 The terminology comes from the two states of circular polarisation of

light, although the term ‘chirality’ is used only for fermions.


A Problem of Mass 47

χ = +1/2 v

m=0

v χ = –1/2

Figure 4.5 For a massive particle the two chirality states are connected
by Lorentz transformations.

The spin projection remains unchanged. With a second Lorentz


transformation we can go to the system in which the speed of the
electron equals –v (see Figure 4.5). The combination of these two
transformations allows us to change a particle with right chiral-
ity to one with left chirality, under one condition: the mass of
the particle must be non-zero. Indeed, a massless particle always
travels with the speed of light and there exists no transformation
which can bring it to rest.

4.3.2 Chirality and weak interactions


In this section we will talk only about weak interactions. We
recall that they are responsible, in particular, for β-decay of the
neutron of the form

n → p + e + ν̄(e) , (4.2)

where n denotes the neutron, p the proton, e the electron, and


ν̄(e) the anti-particle of the neutrino associated with the electron.6
We find the same reaction in nuclear physics where the neut-
ron and the proton are bound inside the nuclei. In the inverse
direction, this gives rise to the fusion of two protons (hydrogen
6 In Appendix 1 we indicate that we have three species of neutrinos, each

being associated with one of the three known leptons, the electron (e), the
muon (μ), and the tau (τ ).
48 The Origin of Mass

nuclei), giving a nucleus of deuterium, D, which is a bound state


of a proton and a neutron,

p + p → D(p + n) + e+ + ν(e) , (4.3)

where e+ is the positron, the anti-particle of the electron. This


reaction is the starting point of the process of nucleosynthesis in
the interior of stars. They are reactions which generate all heavy
elements and are at the origin of stellar energy. We can rewrite
these reactions in terms of quarks by replacing the proton and
neutron by quarks u and d respectively. We can also generalise
them with the quarks of other families and describe the decays
of several unstable particles.
In this section we want to present some properties of weak
interactions which distinguish them in a very significant way
from all other interactions.
Among all interactions of elementary particles, the only ones
that do not respect the invariance under space inversions, are the
weak interactions. This is an experimental fact, not a theoretical
result.
A second property of these interactions, also established by
experiment, is that they involve primarily fermions, quarks or
leptons, of left chirality. The states with right chirality seem to be
insensitive to weak interactions. We shall come back to this point
in Section 6.4.
These two properties make weak interactions unique. We see
that, in order to describe them correctly we must assume (i) the
violation of parity invariance and (ii) a zero mass for all con-
stituents of matter, leptons and quarks. If (i) is rather easy to
implement, (ii) is in contradiction with all experimental results
which show that all these particles have non-zero mass. It is the
second problem of mass that we encounter. It appears to be less ‘fun-
damental’ than the first, which refers to the gauge bosons which
transmit interactions. The first was a consequence of the geo-
metric property of gauge interactions, while the second seems to
A Problem of Mass 49

have a phenomenological origin. Nevertheless, they are both very


important and they call for profound modification of our ideas on
symmetries and the four interactions we introduce in Appendix 1.
It is this part of the story that we shall present in the following
chapters.
5
Spontaneously Broken Symmetries

5.1 Curie’s theorem


In this chapter we shall present a phenomenon which seems to
contradict our physical intuition, although, as we shall see, we
encounter it in everyday life. It is the phenomenon of spontaneous
symmetry breaking.
Usually, for a symmetric problem, we are looking for symmet-
ric solutions. We cite Pierre Curie:

‘When certain causes produce certain effects, the symmetry


elements of the causes must be present in the produced effects’

Talking about spontaneous symmetry breaking seems to con-


tradict this assertion. Of course, Curie knew about these
phenomena.
Let us take an example. Consider a sphere of radius r uniformly
charged with an electric charge Q. We want to compute the elec-
trostatic field produced at a point P at a distance R > r from the
centre of the sphere (see Figure 5.1).
In order to solve this problem it is sufficient to notice that
‘the cause’, i.e. the charged sphere, has a spherical symmetry.
Consequently, the same will be true for ‘the effect’, in other
words, the field E will be radial because it is the only direction
which respects the spherical symmetry. Furthermore, for the
same symmetry reasons, the field must have the same absolute
value at any point on the spherical surface going through P with
centre at the centre of the sphere. It is now straightforward to
Spontaneously Broken Symmetries 51

EA
| EA | = | EB |

Q
EB
B

Figure 5.1 The electric field produced by a uniformly charged sphere


at a distance R.

compute E by applying a theorem by Carl Friedrich Gauss which


gives: E = (Q/4π 0 R2 )r̂, where r̂ is the unit vector in the radial
direction and 0 is the dielectric constant.1
In practice a sphere is never perfect and the symmetry could
only be approximate. Nevertheless, we still apply the above reas-
oning assuming implicitly that ‘small’ violations of the symmetry
of the cause will induce only ‘small’ deviations from the sym-
metric solution. But this is a much stronger assumption going
beyond Curie’s theorem. Indeed, the latter shows only the exist-
ence of a symmetric solution, while in practice we may need in
addition an assumption on its stability. In the case of spontaneous
symmetry breaking we shall study here, it is this last assumption
which will fail.

1 This is a special case of the general Gauss theorem, which we shall not

prove here. Consider an arbitrary distribution of static electric charges in a finite


region of space and a closed surface which surrounds them. The theorem states
that the flux of the electric field through the surface, i.e. the integral over the sur-
face of the perpendicular component of the field, equals the sum of the charges
divided by 0 .
52 The Origin of Mass

5.2 Spontaneous symmetry breaking


in classical physics
A simple example is shown in Figure 5.2. A cylindrical rod of
length l is loaded with a load F along the z axis. The problem
is symmetric under rotations around this axis, so we expect the
final state to be a compressed but straight rod. This is in fact
the answer when F is small. But we know from experience that
when F increases, the form of the rod changes: it bends in an
arbitrary direction. This is the phenomenon of buckling. The bent
rod no longer has the rotational symmetry because the bending
has introduced a spacial direction in the (x, y) plane. This class of
phenomena we call spontaneous symmetry breaking.

y
x −F

Figure 5.2 The rod bends under the action of the load F.
Spontaneously Broken Symmetries 53

Question: what happened to the original symmetry? Answer:


it is still there, but it is hidden. We cannot predict the direction
of the bending. They are all equally probable; in other words, we
have an infinite number of possible solutions, all connected via
the transformations of the original symmetry, i.e. the rotations
around the z axis.
The mathematical analysis of this example is quite simple
and we present it in Box 5.1. This is required for quantitative
computations, but it is not necessary for a qualitative under-
standing of the phenomenon. We can easily guess the essential
features:

1. For small values of the load the rod stays straight but it
starts being compressed. This compression costs a certain
elastic energy which increases with the compression rate,
therefore, with F.
2. There exists a critical value of the load, Fcr , above which it
costs less in energy to the rod to bend and decrease its com-
pression rate (a bent rod has a greater effective length and,
consequently, a smaller compression rate). The analytical
computation we present in Box 5.1 gives the value of Fcr pre-
cisely, in terms of the rod’s parameters for the configuration
in Figure 5.2.
3. Theoretically, there always exists the possibility of having a
straight but very compressed rod, but, since the associated
elastic energy is very large, this is an unstable state.
4. It is convenient to introduce a parameter δ.  This is a vec-
tor in the horizontal (x, y) plane and its components δx
and δy give the displacement of the centre of the rod from
the symmetric point. We can compute the elastic energy
as a function of δx and δy , and we obtain Figure 5.3. For
F < Fcr , we obtain the system on the left with a single min-
imum energy configuration. It is the symmetric solution
with δ = 0. For F > Fcr , we obtain the system on the right.
The symmetric solution is always present, which means that
54 The Origin of Mass

Figure 5.3 The energy of the system as a function of the order para-
meter, for F < Fc (left) and F > Fc (right). Dr A. Hoecker, CERN.

Curie’s theorem is verified, but it corresponds to a local


maximum of the energy and it is, therefore, unstable.
5. We have an infinity of stable solutions. The energy depends
only on the modulus of δ. Starting from any one of these
solutions we obtain any one of the others by applying a rota-
tion around the z axis. In the physics jargon, a state of a
system with the minimum energy is called a ground state. Here
we have an infinite number of ground states. We say that the
ground state is infinitely degenerate.

Box 5.1: The phenomenon of buckling


We present here a simplified version of the mathematical ana-
lysis of buckling without entering into all the details of the
theory of elasticity. The reader who is not interested in these
technical points can go directly to Section 5.3.
Let us call X(z) and Y(z) the x and y displacements of the
point z on the axis of the rod from its symmetric position. z var-
ies from 0 to l. In terms of these variables the order parameter
is given by
Spontaneously Broken Symmetries 55

δx = X(z = l/2); δy = Y(z = l/2). (5.1)

In the symmetric phase we have X(z) = Y(z) = 0. The gen-


eral equations of elasticity are non-linear differential equations
which we can only solve numerically. However, we can sim-
plify them in the neighbourhood of the critical point, keeping
only first-order terms in X and Y. For the rod of Figure 5.2,
which is articulated at both ends, they take the form,
d2 X d2 Y
IM + FX = 0; IM + FY = 0, (5.2)
dz2 dz2
where I = πR4 /4 is the moment of inertia of the rod, R is the
radius of its section, and M a parameter which characterises its
elastic properties and is called the Young modulus. This system of
equations is invariant under rotations around the z axis, with
an angle θ under which the variables X and Y transform as
X → Xcos θ + Ysin θ; Y → –Xsin θ + Ycos θ. (5.3)

We want to solve system (5.2) with the following boundary


conditions at z = 0 and z = l: X = Y = 0, which impose on
the extremities to remain fixed, and X = Y  = 0 which cor-
respond to an articulated rod. It is clear that the symmetric
solution X = Y = 0 is always present. However, we have also
asymmetric solutions of the form X = Csinkz with k2 = F/MI,
provided kl = nπ; n = 1, 2, . . . The first of these solutions with
n = 1 appears as soon as F takes the critical value
π 2 MI
Fcr = . (5.4)
l2
Starting from this solution we can obtain an infinity of others
by applying the rotation (5.3).
The appearance of these asymmetric solutions at the vicinity
(i.e. at first order in the displacements X and Y) of the symmet-
ric solution is a sign of the latter’s instability. We can verify this
56 The Origin of Mass

by an explicit computation of the corresponding elastic energy,


which we shall not present here. However, it will be instructive
to guess its form as a function of the order parameter.
Rotational invariance imposes a dependance on δ · δ = ρ 2 .
At the vicinity of the critical point δ is small and we can write
an expansion of the form E = C0 + C1 δ · δ + C2 (δ · δ)
 2 + ...
= C0 + C1 ρ 2 + C2 ρ 4 + . . . , neglecting higher-order terms.
The Cs are constants depending on the parameters of the rod
and the load F. The ground state of the system is determined
by the minimum of the energy, which gives, in terms of the
variables defined in (5.8),

dE
(ρ = v) = 0 ⇒ v(C1 + 2C2 v2 ) = 0 (5.5)

This equation has two solutions: the symmetric one, v = 0


and a second one, v2 = –C1 /2C2 . ρ being a real number, this
second solution is acceptable only if the ratio C1 /C2 is negat-
ive. It is the solution with spontaneous symmetry breaking. If
C2 is negative there is no minimum energy solution; for large
ρ, E → –∞. It follows that the existence of a ground state
imposes C2 > 0, in which case C1 must vanish at the vicinity of
the critical point and we can write it as C1 = Ĉ1 (Fcr – F), with
Ĉ1 positive. The phase with spontaneous symmetry breaking
is obtained for F > Fcr and, in this phase, we can write the
energy as

E = C0 + Ĉ1 (Fcr – F)δ · δ + C2 (δ · δ)


2
(ρ 2 – v2 )2
= Ĉ1 (F – Fcr ) , (5.6)
2v2

where v is given by the non-zero solution of equation (5.5). The


energy being defined up to an arbitrary additive constant, we
have determined C0 by the condition of having the ground
Spontaneously Broken Symmetries 57

state (ρ = v) energy vanishing. In the phase with spontan-


eous symmetry breaking, the energy of the symmetric solution
ρ = 0 is positive and given by
v2
E0 = Ĉ1 (F – Fcr ) . (5.7)
2
This expression gives the form of Figure 5.3. For F < Fcr we
have a single minimum with δ = 0 (Figure 5.3 left), and for
F > Fcr we obtain the circle of Figure 5.3 right.

We just described an example of a class of phenomena known


as phase transitions. The stressed rod is a physical system which may
exist in two different phases: the phase of the straight rod and
that of the bent rod. We shall call the first one the symmetric phase
and the second the broken symmetry phase. There exists a parameter,
external to the system, which determines which one of the two
phases will be chosen by the system. In the case of the rod it is
the load F. We call this the control quantity. The two phases differ by
the value of another parameter, for the rod the vector δ,  which
we call the order parameter. Its value is zero in the first phase and
different from zero in the second. The transition from one phase
to the other is accompanied by a change of the symmetry of
the system. In the case of the rod it is the rotational symmetry
around the z axis which seems to be absent in the second phase.
As we have pointed out, the symmetry is hidden because the
ground state is degenerate.
The order parameter will play an important role in our discus-
sion of spontaneous symmetry breaking. It will be convenient to
parametrise δ by its modulus ρ and a phase θ, rather than its com-
ponents δx and δy . We introduce a complex number δ = δx + iδy
with
δx = ρ cosθ; δy = ρ sinθ; δ = ρ eiθ . (5.8)
With these variables the points on the minimum energy circle
correspond to ρ constant and arbitrary θ, between –π and π.
58 The Origin of Mass

We have many examples of spontaneous symmetry break-


ing in classical physics, such as crystallisation, turbulence, many
problems of erosion, etc.

5.3 Spontaneous symmetry breaking


in quantum physics
Ferromagnetism, i.e. iron magnetic properties, offers an example
of spontaneous symmetry breaking in quantum physics. The phe-
nomenon is as follows: looking at the magnetic properties of
an iron rich metal we often find two phases. We shall describe
them using the terminology we introduced in the case of the
bent rod.

1. The symmetry is that of three-dimensional rotations.


2. The control quantity is the temperature T. There exists a crit-
ical value Tc , called the Curie point. For T > Tc we are in the
symmetric phase, and for T < Tc in that with spontaneous
symmetry breaking. For iron the Curie point is on the order
of 770 ◦ C.
3. The order parameter is the magnetisation. It is a macroscopic
magnetic moment which determines the interaction of
the sample with a magnetic field. The simplest compass
is precisely a needle made out of magnetised iron which
points parallel to the earth’s magnetic field. Above the
Curie temperature the magnetisation vanishes. At T < Tc
a magnetisation appears spontaneously.
4. In the high temperature phase there is no magnetisation
(para-magnetic phase). There is no privileged direction in space
and we have the full three-dimensional rotation symmetry.
This is the symmetric phase. In the low temperature phase
we have a spontaneous magnetisation (ferro-magnetic phase).
It defines a privileged direction in space. The symmetry is
reduced to rotations only around this axis. This is the broken
symmetry phase.
Spontaneously Broken Symmetries 59

5. In the ferro-magnetic phase the ground state of the system is


degenerate. All orientations of the magnetisation are a priori
equivalent.

W. Heisenberg proposed a simple model which, in a series


of approximations, eliminates all inessential complications and
captures the main features of the phenomenon.
The first approximation concerns the crystal structure. We
assume a regular lattice and we ignore all possible defects. In order
to simplify the presentation we take a cubic lattice, although the
reasoning also applies to other regular crystals.
The iron atoms occupy the lattice sites. In principle we must
take into account all atomic degrees of freedom, but, and this
is the second approximation, we assume that, for the magnetic
properties of the crystal, only the spin degrees of freedom are
important. We consider a spin in every lattice site, for example a
spin equal to 1/2, whose projection on the z axis can take the two
values ±1/2.
The spins interact among themselves and, following
Heisenberg, we assume the interactions to be of short range
so that only pairs of nearest neighbours are coupled. For a three-
dimensional cubic lattice, each spin has six nearest neighbours.
The last assumption concerns the form of the interaction. We
assume rotational invariance and, the simplest form of coupling
yields an energy given by
1 
E=– J Si · Sj , (5.9)
2 i,j

where Si is the spin on the ith site and the sum runs over all
pairs of nearest neighbours. J is a constant, which we assume to
be positive.2 Rotational invariance is ensured by the form of the
interaction which is a scalar product of two vectors.
2 The case with J < 0 is also interesting and describes a phenomenon called

anti-ferromagnetism. It has been studied by Louis Eugène Félix Néel (1904–2000,


Nobel Prize 1970), but we shall not present it here.
60 The Origin of Mass

We can solve this model numerically and in some simple cases,


for one- or two-dimensional crystals, analytically, but the physics
of the phenomenon is obvious by inspection.
At very high temperature we are in the symmetric phase.
Thermal fluctuations are very important and, at any particular
moment, the spins are randomly oriented. We obtain the con-
figuration of Figure 5.4 (a) with vanishing mean value of the
magnetisation. This is a disordered phase.
At very low temperature, when thermal fluctuations can be
neglected, the ground state of the system is obtained by the
configuration which minimises the energy. For positive J this cor-
responds to maximising the scalar product, which means making
the spins parallel. Each spin tends to orient its nearest neigh-
bours in its own direction, so that, from nearest neighbour to
nearest neighbour, we obtain the configuration of Figure 5.4 (b),
yielding a non-zero magnetisation M.  This is the ordered phase
in which the original three-dimensional rotation symmetry is
spontaneously broken.3

T > Tc T < Tc

(a) (b)

Figure 5.4 (a) At T > Tc the spins are oriented randomly due to the
thermal fluctuations. The average magnetisation vanishes. (b) : At
T < Tc the interaction orients the spins and we have a spontaneous
magnetisation.
3
What happens in a real ferromagnet is more complex. This process of spin
orientations starts independently from various points of the solid and the image
shown in Figure 5.4 (b) is valid only locally in a small domain. A sample of
Spontaneously Broken Symmetries 61

To summarise, at every temperature, we have two compet-


ing effects: the thermal fluctuations favour disorder but their
strength decreases with temperature; the interaction favours
order. We can understand that there exists a temperature below
which order wins. This is the Curie temperature.4
A remark before closing this section: the order parameter of
ferromagnetism is the vector of magnetisation M.  In the high

temperature phase M = 0 and the system is invariant under the
full set of three-dimensional rotations. This is the symmetric
phase. An arbitrary rotation can be parametrised by the three
Euler angles, i.e. two angles to define the direction of the axis
around which we perform the rotation, and a third which gives
the rotation angle. In the low temperature phase the three-
dimensional rotational symmetry is not completely broken. A
part, corresponding to the rotations around the M  axis, remains
exact. Consequently, we need two angles to label the states of
minimum energy. In the notation we used in the case of the bent
rod, we write

Mx = ρ sinθ cosφ; My = ρ sinθ sinφ; Mz = ρ cosθ. (5.10)

We see that in the low temperature phase, (i) the symmetry


is only partly broken and (ii) the set of ground states form the
surface of a sphere. In Chapter 6 we shall encounter cases in which
macroscopic size contains a large number of domains, each with a magnetisa-
tion pointing in a random direction. This is the reason why a piece of iron does
not always appear to be magnetised. Strictly speaking, this domain structure
does not minimise the energy because in the boundaries between domains the
spins are not parallel. Nevertheless, transition to the real ground state would
require a simultaneous change of orientation of the spins in every domain and
the probability of such an event goes to zero for a large sample. In order to
obtain a magnetised piece of iron we must place the sample in an external
magnetic field during the phase transition.
4 Another result which depends on the number of space dimensions. We can

find a qualitative argument which explains this dependence: the interaction,


which favours order, increases with the number of nearest neighbour pairs, i.e.
the number of space dimensions. For example, we can show that a linear chain
of spins, in which every spin has only two nearest neighbours, has no phase
transition for any finite value of the temperature.
62 The Origin of Mass

this set forms a multi-dimensional hyper-surface. Notice also that


in the low temperature phase, which is the ordered phase, the
energy of the symmetric solution is given by the analogue of for-
mula (5.7), but multiplied by a factor proportional to the total
number of spins. This goes to infinity for an infinite system.

5.4 Goldstone theorem


In this section we want to present a very important consequence
of spontaneous symmetry breaking which, in addition, will bring
us closer to the subject of this book, namely the Brout-Englert-
Higgs particle. This is a quantum effect but, as we have done up to
now, we shall try to find classical analogies, with all the dangers of
offering misleading explanations inherent in such an approach.
Let us look at Figure 5.4 (b). At low temperature the system
is in the ordered phase, all the spins are parallel, and we have
long range correlations. In Chapter 4 we showed that in quantum
physics and in a three-dimensional space, such correlations imply
long range interactions and, consequently, massless particles. This
is Goldstone’s theorem:5 spontaneous symmetry breaking6 results in the appearance
of massless particles in the spectrum of the theory.
We shall not give here a mathematical proof of this theorem;
we shall try instead to extract its physical consequences.
The first image is the degeneracy of the ground state of a
system in the phase with spontaneous symmetry breaking. It is
represented classically by the right-hand side of Figure 5.3. All
points in the circle ρ = |δ| = v, where v is the radius of the
circle, correspond to the same value of the energy. We want to
find the corresponding representation in quantum physics. Two
simple principles will help us:

1. For every configuration of classical variables there is a cor-


responding state in quantum physics. This is a direct con-
sequence of the axioms of quantum mechanics.
5 It is also called the Nambu–Goldstone theorem in honour of Yoichiro

Nambu and Jeffrey Goldstone (see Figure 5.5).


6 The theorem applies only to continuous symmetries.
Spontaneously Broken Symmetries 63

Figure 5.5 Spontaneous symmetry breaking: Pierre Curie (1859–1906,


Nobel Prize 1903); Yoichiro Nambu (1921–, Nobel Prize 2008); Jeffrey
Goldstone (1933–). Goldstone : J. Goldstone archives.

2. In quantum physics of elementary particles, each state is


characterised by the number and the properties of the
particles it contains.

With the help of these two principles we want to find the clas-
sical analogue of Goldstone’s theorem. We have already alluded
to the field–particle duality in quantum physics. Therefore we
expect to find the fields which correspond to the Goldstone
particles and, in the classical limit, to identify the corresponding
classical variables.
Let us start with Figure 5.3. There are two relevant variables,
ρ and θ. A classical configuration is described by precise values of
these two variables. According to the aforementioned first prin-
ciple, we also have a quantum state. On the other hand, going
from classical to quantum physics, classical variables become
quantum fields which correspond to particles. We conclude that
the quantum description of spontaneous symmetry breaking will
involve two kinds of particles, those which correspond to ρ and
those to θ. According to the second principle, each state contains
a different number of these particles.
Consider two quantum states, each corresponding to a differ-
ent point in the circle of minima in Figure 5.3. In classical physics
they differ only by the value of θ, therefore, each one of the two
64 The Origin of Mass

quantum states has the same number of ρ particles, but a different


number of θ particles. Let us call this difference .7 The two states
must have the same energy. It follows that we can add a number
 of θ particles in a state without changing its energy. In a relativ-
istic theory the energy of a particle is always larger, or equal, to
that given by its mass. Therefore, the mass of the θ particles must
be equal to zero. These are the Goldstone particles. We can now
state the classical version of Goldstone’s theorem:

Spontaneous breaking of a continuous symmetry implies the degeneracy of the


minimum energy state (ground state) of a physical system.

This degeneracy is translated, in the quantum system, into the


appearance of zero mass particles.
In fact, our analysis also yields a second conclusion. We have
seen that the quantum description of spontaneous symmetry
breaking involves two kinds of particles, the θs, which are the
massless Goldstone particles, and the ρs which are massive.8 In
the symmetric phase the ground state is the one which cor-
responds to ρ = 0. In the phase with spontaneous symmetry
breaking we have an infinite number of ground states, all corres-
ponding to ρ = v. They differ from the symmetric state by the
number of ρ particles. In our jargon we often call the ground
state vacuum. Therefore, the phase transition implies a change in
the vacuum state of the theory. In Section 5.3 we pointed out
that, for a large system, the energy difference between these two
states goes to infinity. The two ‘vacua’, the symmetric one and
the one with spontaneous symmetry breaking, not only are they
not ‘empty’, but, in addition, they differ by an infinite number

7From the mathematical point of view, this is not a rigorous argument. It


turns out that each one of the quantum states which corresponds to these clas-
sical configurations, contains an infinite number of particles. As a result  is
not well defined.
8 We recall that we may have several kinds of ‘θ ’ particles. In the example of

ferro-magnetism we had two: the θ s and the φs.


Spontaneously Broken Symmetries 65

of ρ particles.9 In Section 5.5 we shall establish that they are the


Brout–Englert–Higgs particles.
A last remark: in Appendix 1 we argue that the mass of a
particle may depend on the interactions in which it participates.
Consider a particle which interacts with ρ. We expect to find out
that its mass may be different in each of the two phases of the sys-
tem. If we call m0 its mass in the symmetric phase which has no
ρ particles, in the phase with spontaneous symmetry breaking it
will be given by

m = m0 + δm, (5.11)

where δm is the change in the mass that the particle gets from its
interactions with ρ.

5.5 Spontaneous symmetry breaking


in the presence of gauge interactions
This is the most important section of this book. It is here that the
significance of the CERN discovery will be presented.
Let us summarise the results we obtained previously on mass-
less particles. In Chapter 4 we showed that the presence of inter-
actions with local symmetries, i.e. gauge interactions, seems to
make the bosons which mediate these interactions massless. This
was the first mass problem. Next we discovered a second problem,
namely, that the weak interactions also make the constituents
of matter, leptons and quarks, massless. In this chapter we have
proved Goldstone’s theorem which sounds like a third mass
problem: the phenomenon of spontaneous symmetry breaking
implies the presence of new massless particles, those we called
Goldstone particles. However, looking at the table of elementary
particles in Appendix 1, we see that all particles with the exception
of the photon and, possibly the graviton, are massive. We could
9 We also see that the bosonic character of the ρ particles is essential. Only

bosons can accumulate in large numbers in the same state and it is this property
that makes this phenomenon of phase transition possible.
66 The Origin of Mass

conclude that each of these problems, taken separately, indicates


that gauge theories, as well as the phenomenon of spontaneous
symmetry breaking, are in contradiction with the experimental
results. In this section we will show that, in fact, the opposite
is true: if we consider the three problems together, each will
solve the other. In a gauge theory with spontaneous symmetry
breaking, all particles can be massive. It was the discovery of this
property, we could call it ‘miraculous’, which paved the way for
spectacular progress in our understanding of the fundamental
interactions.
In this section we shall try to explain this phenomenon.
Unfortunately, it is a complex and purely quantum phe-
nomenon, for which we have no classical analogue. As we just
said, we have to face three mass problems: (i) the masses of the
matter constituents, (ii) those of the gauge bosons, and (iii) the
mass of the Goldstone particles, which we previously called ‘θ
particles’.
The simplest to start with is probably the first one. We have
already referred to it at the end of Section 5.4 by writing equation
(5.11). We noticed in Chapter 4 that the weak interaction sym-
metries force quarks and leptons to be massless. As a result, in the
symmetric phase we must have m0 = 0. Spontaneous symmetry
breaking allows these particles to develop a δm and acquire a mass.
This happens through their interactions with the ρ particles
which populate the ground state in the broken symmetry phase.
We expect δm to be proportional to the strength of the interac-
tion between the particle and ρ and this allows us to fit the large
spectrum of masses we observe among the matter constituents.
But, beware: fit is not synonymous with understand. Table 5.1 gives
the experimentally determined masses of quarks and leptons. We
see that the ratio between the mass of the quark t (the heaviest)
and that of the electron (the lightest)10 is of order 350,000. By

10 In this book we shall not discuss the problem of the neutrino masses. We

know, by indirect measurements, that they are different from zero, but they
Spontaneously Broken Symmetries 67

Table 5.1 The masses of quarks and leptons. The unit is the megaelec-
tronvolt (MeV). 1 MeV ∼ 1.7810–30 kg. The values shown are approxim-
ate. Quark masses are not directly measurable (see Chapter 6) and those
of the neutrinos are too small and beyond our measuring capabilities.

Masses of matter constituents


Quarks u d c s t b
2.3 MeV 5.0 MeV 1275 MeV 95 MeV 173,000 MeV 4180 MeV
Leptons ν(e) e ν(μ) μ ν(τ ) τ
?? 0.51 MeV ?? 106 MeV ?? 1777 MeV

adjusting the coupling strength of each of these particles with ρ


we can reproduce the values we see in Table 5.1, but we do not
understand the origin of such a dispersion. This makes us believe
that an important part of the mass creation mechanism is still
beyond our comprehension.
Let us now come to the second problem, namely the masses of
the gauge bosons which transmit weak interactions. In Table A1.3
these are indicated as W ± and Z0 . Their masses have been meas-
ured and found to be 80,385 and 91,188 MeV, respectively. We
could be tempted to use the previous argument and attribute
their masses in the phase with spontaneous symmetry breaking to
their interactions with ρ. This is not totally wrong but it describes
only one aspect of the phenomenon. In fact, the real story is more
complex.
These gauge bosons have spin equal to one. This follows from a
mathematical theorem, which we have not proven here; but it is
also an experimental result. In quantum mechanics we show that,
for a particle with spin equal to s, the spin projection on any axis
can take 2s + 1 values, to wit +s, +s – 1, . . . , –s. Therefore, for
s = 1, we have three possible values, +1, 0, and –1. We say that a
spin 1 particle is described by three degrees of freedom. This rule applies
are so small that we have not been able to measure them directly. Present limits
indicate that they are at least 500000 times lighter than the electron.
68 The Origin of Mass

to massive particles. For massless particles the rule is different: for


all s different from zero we have only two degrees of freedom and
the spin projection on the direction of motion can only take the
values +s and –s.
These properties result from the relativistic invariance of the
theory and we cannot demonstrate them here. We can only
give a classical analogue. It is known that electromagnetic waves
can be polarised, but only in a transverse direction with respect
to their propagation. We never observe them polarised in the
longitudinal direction. In the equations of classical electrodynam-
ics, the electromagnetic field has two independent components,
corresponding to the two transverse polarisations.11
This is the complication we mentioned previously: for the
gauge bosons it is not enough to look for interactions to generate
mass, we must also find an extra degree of freedom to allow for
the transition between a massless and a massive spin one particle.
Who supplies this degree of freedom?
In order to answer this question we must come back to the reas-
ons why we have massless particles in the first place, in both gauge
theories and the Goldstone phenomenon. We have explained that
in both cases this was the result of long range correlations. The
gauge interactions impose a long range because the transforma-
tions depend, in an arbitrary way, on the space-time point. The
spontaneous symmetry breaking imposes a different long range
order, as is indicated in the spin example of Figure 5.4(b). It is
easy to imagine that the two may not be compatible with each
other. For example, it is easy to see in this figure that, if we
allow ourselves to turn the spins independently at every point,
we can destroy the order imposed by the spontaneous symmetry

11 There is an exception to this rule: the electromagnetic field propagating

in a superconductor has, in addition, a longitudinal component. It is the prop-


erty which led Philip Anderson to describe, for the first time, the phenomenon
we are in the process of studying here, namely that of spontaneous symmetry
breaking in the presence of electromagnetic interactions, in a non-relativistic
context.
Spontaneously Broken Symmetries 69

breaking.12 The absence of long range order implies the absence


of massless particles. Therefore we expect that in the phase with
spontaneous symmetry breaking, all particles are massive.
Let us see the consequences of this analysis, first on a simple sys-
tem with a single gauge boson. In the symmetric phase we have
the massless gauge boson described by the two degrees of free-
dom corresponding to its transverse polarisations. In addition, we
have two massive spin zero particles, those we called θ and ρ.
Altogether four degrees of freedom. In the phase with spontan-
eous symmetry breaking the same physical system is described
by a massive spin one particle (which makes three degrees of
freedom corresponding to its transverse and longitudinal polar-
isations), and a massive spin zero particle, the ρ. Again, a total
of four degrees of freedom. The θ particle, which would have
been massless as a consequence of Goldstone’s theorem, is absent.
Its degree of freedom has been used in order to allow the gauge
boson to acquire a mass. We can say that the θ particle is the
component of longitudinal polarisation of the massive spin one
boson.
This picture is immediately generalised to a more complex sys-
tem, like that of ferromagnetic spins in the presence of gauge
symmetry. The symmetry is that of three-dimensional rotations.
We have three independent transformations and this implies that,
in the high temperature phase, we have three massless gauge
bosons. This gives us six degrees of freedom. We add the three
massive, spinless particles corresponding to ρ, θ, and φ. In the
low temperature phase we have a partial spontaneous symmetry
breaking. The symmetry of rotations around the axis defined by
the direction of M, remains intact. Only the other two are spon-
taneously broken. It follows that only two among the three gauge
bosons acquire a mass; the third remains massless. In this process,
the first two gauge bosons absorb the degrees of freedom of the
12 Again an argument which we should consider as an indication and not

as a proof. It is easy to visualise and understand it on a discrete lattice, but its


applicability in a continuous space is much more subtle.
70 The Origin of Mass

particles θ and φ. So, in this phase we have: (i) a massless gauge


boson with two degrees of freedom, (ii) two massive spin one
bosons each having three degrees of freedom, and (iii) the spin
zero boson ρ. The sum of degrees of freedom is again nine, the
same as in the symmetric phase.
The trace of this phase transition is the presence of the
ρ particle. We believe it is the particle discovered at CERN.
Therefore this discovery not only adds a new entry to our table of
elementary particles, but also, it opens a window into one of the
most extraordinary phenomena in the history of the Universe.
6
The Standard Theory

6.1 Introduction
In Appendix 1 we see that all experimental results show the
presence of four fundamental interactions among elementary
particles: gravitational interactions, weak interactions, electromagnetic inter-
actions, and strong interactions. In this chapter we want to use the
mathematical and conceptual tools we have developed so far in
order to build theoretical models describing these interactions.
The ingredients of these models are:

1. Gauge symmetries. These are generated by transforma-


tions whose parameters depend on the space-time point. In
Chapter 3 our motivation for introducing them was essen-
tially aesthetic, but, in fact, they are absolutely necessary.
We can prove that the only theories which are mathem-
atically consistent and have the predictive power to cor-
rectly describe fundamental interactions, are those which
are invariant under local, i.e. gauge, transformations. We
have distinguished two kinds of gauge transformations:
those which act on the coordinate system of space-time, on
the one hand, and those on the coordinate system of an
internal space, on the other. Both are used in fundamental
physics. (i) Invariance under space-time gauge transform-
ations gives us general relativity, the classical theory of
gravitational interactions. (ii) Invariance under gauge trans-
formations of internal spaces will give us the theoretical
framework to describe the other three interactions. In this
72 The Origin of Mass

book we shall only mention gravitational interactions very


briefly in the last chapter; we shall concentrate instead on
the other three, the weak, the electromagnetic, and the
strong interactions.
2. Spontaneous symmetry breaking for some of the internal
symmetries. The necessity for this step is obvious for weak
interactions whose intermediaries, the W ± and the Z0 gauge
bosons, are massive. In contradistinction, the photon, which
is the gauge boson of the electromagnetic interactions, is
massless and the corresponding gauge symmetry should
remain unbroken. We shall discuss the situation concerning
strong interactions later.

6.2 The electromagnetic and weak


interactions
A priori it is not at all obvious why these two interactions should be
studied together. At first sight, they have very little in common,
except for the fact that they both have spin one intermediaries.
But this is true for every gauge theory of an internal symmetry.
On the contrary, almost everything seems to separate them:

1. Weak interactions violate parity; the electromagnetic inter-


actions conserve parity.
2. Weak interactions seem to involve only quarks and
leptons of left chirality; electromagnetic interactions
involve both chiralities and do not distinguish right from
left. Consequently, as we explained in Chapter 4, without
spontaneous symmetry breaking, weak interactions cause
quarks and leptons to be massless.
3. The weak interaction intermediate gauge bosons are
massive; the photon is massless.
4. Electromagnetic interactions are mediated by a single gauge
boson, the photon. Weak interactions involve three: W ± and
Z0 . In Chapter 4, we saw that we have one gauge boson for
The Standard Theory 73

each independent gauge transformation. It follows that the


gauge symmetry of weak interactions is larger than that of
electromagnetic interactions. For the first we should use the
full Yang–Mills theory.

Because of all these differences, early attempts to apply the


Yang–Mills theory only addressed weak interactions. It was the
American physicist Sheldon Lee Glashow who first understood,
in 1960, that we obtain a richer theory by considering simul-
taneously weak and electromagnetic interactions in a unified
framework. His work was done before that of Brout–Englert–
Higgs (1964) and consequently, he did not deal with a mechanism
to give masses to W ± and Z0 . The synthesis of all these ideas
was due to Steven Weinberg in 1967 and Abdus Salam in 1968
(see Figure 6.1). Their model only applied to leptons and the
extension to quarks came later. We are not going to follow the
historical order, naturally longer to explain; we present directly
the final theory. We are going to construct it deductively by tak-
ing into account all experimental results known up to today. The
construction goes through three steps:

1. The choice of gauge symmetry. We must choose the trans-


formations which are supposed to leave invariant the
dynamical equations of the theory. In Chapter 4 we saw

Figure 6.1 The Standard Model: Sheldon Lee Glashow (1932–, Nobel
Prize 1979); Steven Weinberg (1933–, Nobel Prize 1979); Abdus Salam
(1926–1996, Nobel Prize 1979); Glashow : S.L. Glashow archives;
Weinberg : S. Weinberg archives; Salam : ICTP archives.
74 The Origin of Mass

that there exists an exact correspondence between the num-


ber of independent gauge symmetries and that of the gauge
bosons which transmit the interactions. In Table A1.3 of the
elementary particles we see that, experimentally, we have
four intermediate gauge bosons for the weak and electro-
magnetic interactions: the photon (γ ) and the three bosons
of weak interactions (W + , W – , and Z0 ), where +, –, and 0
indicate their electric charges. It follows that we must pos-
tulate the invariance of the theory under four independent
gauge transformations. We recall that these transformations
act on the coordinate system of an internal space and do not
affect at all that of our space-time.
Only knowing the number of independent transformations
which leave the theory invariant does not sound like a seri-
ous constraint, but in fact, the opposite is true. We can show
that this number largely determines the geometrical prop-
erties of the theory and, after the work of Yang and Mills, an
important part of the dynamical equations. The underlying
mathematical theory had been established during the late
part of the nineteenth century by the Norwegian mathem-
atician Sophus Lie, followed by the work of the Frenchman,
Élie Cartan. It is the theory of Lie groups and Lie algebras. It
is a very beautiful mathematical theory, too technical to
present here. We give a brief summary of the results which
will be of interest to us in Appendix 2. The conclusion is that,
with four independent symmetry transformations, there
exists only one non-trivial solution which can be visualised
as follows: three rotations in a three-dimensional internal
space and an additional rotation around a fourth fixed axis.
Thus, the geometry of the internal space is determined. In
Appendix 2 we show a more abstract and more convenient
way to describe these transformations.
2. In the symmetric phase all four gauge bosons are massless.
Therefore, the second step consists in creating the necessary
conditions for spontaneous symmetry breaking. In practice
The Standard Theory 75

this implies the introduction of particles, and thus of fields,


of type ρ and θ, which we presented in Chapter 5. The goal
is to give masses to three gauge bosons and leave the fourth
massless. The latter will be identified as being the photon.
We shall have a partial symmetry breaking, like the one
described in the ferromagnetic example. One out of the four
symmetry transformations will remain exact and the other
three will be spontaneously broken. The counting of degrees
of freedom we presented in Section 5.5 shows that we need
three particles of θ type, one for every gauge boson which
becomes massive. If we add at least one particle of ρ type,
we obtain a minimum of four zero spin degrees of freedom.
It is important to realise that this reasoning gives the min-
imum number of such particles. If we add more, let us say
N > 4, we shall end up having, in the phase with spon-
taneously broken symmetry, N – 3 physical particles. One
of the items in the LHC agenda is, precisely, the search for
possible additional BEH particles.
Let us try to establish a correspondence between the gauge
bosons in the two phases. In the symmetric phase, in which
they are all massless, we have three which correspond
to the rotations in the three-dimensional internal space
and one which corresponds to the rotations around the
fourth axis. The naive solution would be to say that this
last one is the photon and remains massless and the other
three become massive and transmit the weak interactions.
Glashow remarked1 that there exists a more general solu-
tion, in which the transformation that remains exact is a
rotation around an axis making an angle with respect to the
initial ones. In this picture, the distinction between the elec-
tromagnetic and weak interactions is the result of spontan-
eous symmetry breaking. Physicists have introduced a new

1 For the understanding of the structure of the electroweak theory he was

awarded, together with Weinberg and Salam, the 1979 Physics Nobel Prize.
76 The Origin of Mass

term to describe this phenomenon and talk about electroweak


interactions, in order to emphasise the common origin of both.
The phenomenon has precise physical consequences, which
have been verified experimentally. We shall present them at
the end of this chapter.
3. The third step involves the introduction of the particles
which are the matter constituents. There is a lot of arbit-
rariness in this step. Firstly, we must decide which are the
elementary constituents of matter. Even if today there is
a consensus in favour of quarks and leptons, we should
recall that this is a phenomenological result which could
change if tomorrow we discover that quarks and/or leptons
are themselves bound states of other, more ‘elementary’
constituents. Secondly, we do not have the slightest the-
oretical idea concerning the total number of these quarks
and leptons. As we have pointed out already, this is one
of the profound questions for which the Standard Theory
offers no answer. In addition, we observe in nature that all
these particles are grouped together to form three ‘famil-
ies’, but we do not know the precise reasons that dictate
such an organisation. One point seems to be important (see
Appendix 1): the algebraic sum of the electric charges of
all the particles in a family must vanish. This is a necessary
condition for the mathematical consistency of the theory,
but we shall not give the proof here. In the symmetric
phase all these particles are massless and they acquire their
masses in the phase with spontaneous symmetry breaking
through their interactions with the BEH particles. Although
we are able to adjust the strength of these interactions in
order to reproduce the observed mass spectrum, we have
no explanation of their precise values.

Through these three steps the theory of electroweak interac-


tions, often called the Standard Model, is complete. Its agreement
The Standard Theory 77

with experiment is spectacular. We shall present the principal


results shortly.

6.3 The strong interactions


Although the study of the strong interactions is outside the scope
of this book, we would like to present their main properties in
this section. As we shall see, they involve new and unexpec-
ted concepts whose deep understanding remains a challenge for
physicists.
Strong interactions entered the scene along with the discov-
ery of nuclear structure. We know that nuclei are composed of
protons and neutrons, which we call, collectively, nucleons. The
observed stability of nuclear matter shows the existence of attract-
ive forces among the nucleons which, at least at distances as large
as a typical nuclear size (∼10–15 m), must be stronger than the
electrostatic repulsion among protons. It is this observation, com-
bined with the estimate on the range of nuclear forces, which led
Yukawa in 1935 to predict the existence of π mesons, as medi-
ators of these new forces. The discovery of these particles, the
pions, in cosmic rays in 1947 confirmed the prediction and opened
a new discipline of fundamental physics, the study of strong
interactions.
Construction of the first powerful accelerators in the 1950s
brought into evidence a large number of new particles subject
to the strong interactions.2 Today we know around a hundred of
these, so the question of which particles are ‘elementary’ became
almost meaningless. The quark model, considered initially as a math-
ematical model of classification, was proposed precisely in order
to bring some order to this chaos. All these hadrons were assumed
to be bound states of a small number of elementary constituents
which Murray Gell-Mann called quarks.3
2 We called them hadrons (see Appendix 1).
3 This word comes from a rather obscure verse in James Joyce’s Finnegan’s
Wake.
78 The Origin of Mass

Table 6.1 Examples of the quark composition of various hadrons.


Baryons are bound states of three quarks and mesons of a quark–
antiquark pair. We also include the example of a baryon, called 0 , in
whose composition enters a quark of the second family, the strange quark s.

From quarks to hadrons


Hadrons Proton Neutron Meson π + Meson π – Meson π 0 0
Quarks uud udd ud̄ ūd ūu and d̄d uds

We are not going to present here the historical evolution of


this concept but today the physical ‘existence’ of quarks as basic
constituents of all hadrons is no longer a matter of controversy.
It is now firmly established through a large number of exper-
iments which have observed the presence of hard grains in the
interior of hadrons, the high energy analogue of Rutherford’s
experiment which in 1911 discovered the nuclei inside atoms.
Table 6.1 presents some examples of the quark composition of
various hadrons.
In the table of elementary particles (Table A1.3 in Appendix 1)
we see that we know today six species of quarks: u, d, c, s, t, and b.
Table 5.1 shows that their masses span a wide spectrum of val-
ues, from 2.3 MeV for the lightest quark u, to 173,000 MeV for the
heaviest, t. The quark t is the heaviest elementary particle known
today.
With the arrival of quarks the very nature of strong interac-
tions changed radically. They were no longer interactions among
nucleons, but at a deeper level, those of quarks. Nuclear forces
would be derivable from these fundamental interactions in the
same way that forces among atoms and molecules (called van der
Waals forces4 ) are derivable from the fundamental electromagnetic

4 Johannes Diderik van der Waals from the Netherlands (1837–1923, Nobel

Prize 1910).
The Standard Theory 79

interactions among charged particles. It was soon realised that the


forces among quarks presented several very peculiar features:

1. They are weak at very short distances but get stronger with
increasing distance. Today we have measurements which
cover almost four orders of magnitude, from 10–15 m, the
typical size of a nucleon, to 10–19 m, the LHC resolution.
In Section 6.4 we present the experimentally observed vari-
ation and compare it with the theoretical prediction.
2. The quarks do not appear as free particles. In spite of all
experimental efforts, we have not succeeded in ‘breaking’
a hadron and liberating its constituent quarks. We call this
property confinement. Quarks seem to be permanently con-
fined inside the hadrons.
3. If we compare the proton mass (∼938.3 MeV) with the sum
of the masses of its constituents (∼9.6 MeV), we find that
they differ by a factor of order 100. This means that the main
part of a proton’s mass is not due to quark masses. This raises
two questions: the first concerns the origin of the additional
mass, the second concerns the proton’s stability. The pro-
ton, as a bound state of three quarks, is much heavier than the
quarks. Usually it is the other way around. For example,
the mass of the deuteron, a proton–neutron bound state,
is equal to ∼1875.6 MeV, therefore it is smaller than the sum
of the mass of a proton (∼938.3 MeV) and that of a neutron
(∼939.6 MeV). The difference is called the binding energy and
explains the deuteron’s stability. But then, what prevents
the disintegration of a proton into three quarks?

At first sight it looks as though a theory fitting all these experi-


mental constraints must have almost miraculous properties. And
yet, the miracle does occur. There exists one theory, and only one,
which seems to satisfy all three properties. Without entering into
the technical details, we shall try to obtain it by following these
properties.
80 The Origin of Mass

1. Concerning the first, let us point out that the variation of


the effective strength of an interaction with the distance is
a general feature of all quantum field theories, so, by itself,
this is not a problem. What is peculiar with the interaction
among quarks is that this variation appears to be counterin-
tuitive. We expect the intensity to decrease with the distance
and go to zero at infinity, while experiments involving
quarks show the opposite. Using standard techniques of
quantum field theory we can study this dependence and we
find that, indeed, practically all theories exhibit the beha-
viour expected intuitively: negligible at very large distances,
their strength increases with decreasing distance. All, with
one exception, Yang–Mills theories in the symmetric phase,
have counterintuitive behaviour. When we compute the
effective force Feff (R) produced by such an interaction as
a function of the distance R, we find an increasing function
which goes to zero when R → 0. We call this property asymp-
totic freedom and we can formulate it as a theorem: the only
asymptotically free theories are Yang–Mills theories in the symmetric
phase.
2. The theorem of asymptotic freedom tells us that, in order
to satisfy the first property of strong interactions, we must
assume that they are described by a gauge invariant inter-
action with several independent transformations, like those
proposed by Yang and Mills.5 Furthermore, we must assume
that the theory is in the symmetric phase, because the
property of asymptotic freedom is lost under spontaneous
symmetry breaking. But this creates a new problem: in the
symmetric phase the gauge bosons are massless and the
interactions are long ranged. Such a behaviour seems to be
in violent contradiction with the observed short range of
nuclear forces, which are supposed to be derivable from the
fundamental interaction between quarks.
5The condition of having a Yang–Mills theory is essential. We can prove that
gauge invariant theories with only one mediator, like quantum electrodynam-
ics, are not asymptotically free.
The Standard Theory 81

3. The answer to this question brings us to the second property


of quark interactions, that of confinement. We can visualise
this as follows: consider the example of the hydrogen atom.
It is a bound state of a proton and an electron. The binding
force is electrostatic attraction. If we introduce an external
electrostatic field, the electron will be pulled to one side and
the proton to the other (see Figure 6.2(a)). If the external
field is sufficiently strong, we can beat the attractive force
and separate the two constituents. We say that the atom is
ionised. Consider now the case of a π 0 meson which, as we
see in Table 6.1, is a quark–antiquark pair. These particles
have opposite charges so, as happened with the electron–
proton system, they will be pulled in opposite directions
by the external field. However, experiments show that we
cannot separate them, no matter how strong the external
field. It is the property of confinement (see Figure 6.2(b)).
Phrased more generally, confinement postulates that all
the ‘elementary particles’ in a Yang–Mills theory in the
symmetric phase, to wit quarks, antiquarks, as well as the
gauge bosons which mediate the interaction, are perman-
ently confined inside the hadrons. We cannot observe them
as free particles.

FC ∼1/R2 FQCD ∼?

p e− q −
q
(a) (b)

Figure 6.2 The binding forces. In (a) a proton and an electron are bound
by the Coulomb force which behaves like 1/R2 . The binding can be
broken and the atom ionised. In (b) a quark–antiquark pair is bound
by the strong Yang–Mills force. We do not know its precise behaviour
at large distances but, if it does not decrease fast enough, the quark and
the antiquark will remain confined. In (b) we represent this binding by
a spring. At short distances the spring is loose and the resulting force is
weak. At large distances the spring is tight and the pull increases.
82 The Origin of Mass

We have good reasons to believe that all Yang–Mills theor-


ies exhibit this property of confinement but so far we have
no rigorous proof.6 These reasons are of two types. For the
first, let us come back to Figure 6.2. The possibility of atom
ionisation is a consequence of the large distance behaviour
of the force Fef (R). The energy needed to break the atom is
given by the work provided by the force in order to separate
the electron from the proton and bring them far apart from
each other. In mathematical
 terms this work is expressed
by the integral Feff (R)dR from the size of the atom to
infinity. The Coulomb force decreases as 1/R2 and the integ-
ral converges. If the force Feff (R) which binds the quark–
antiquark pair had a much slower decrease,7 the integral
would diverge and we would need an infinite amount of
energy to separate the pair. Figure 6.2(b) shows an artist’s
view of this phenomenon. Unfortunately, we are only able
to compute the function Feff (R) when R is very small. We
do find that it is an increasing function of R, but we do not
know whether this behaviour persists at large distances. The
second indication comes from numerical simulations which
we perform by approximating the continuum of space-time
by the points of a discrete lattice. We can prove the prop-
erty of confinement on the lattice, but we cannot control
the limit in which the distance between two lattice points
goes to zero in order to recover the continuous space.

We shall follow the same method we used for the theory


of electroweak interactions and determine the number and the
properties of the transformations which will be the symmetries
of the strong interactions. We shall try the most ‘natural’ choice

6 The rigorous proof of this property constitutes one of the seven problems in
mathematics of the 21st century, proclaimed in the year 2000 by the Clay Mathematical
Institute. Each is endowed with a one million dollar prize.
7 Any behaviour, such as a slow decrease (e.g. 1/R), a constant, or an increase

in R, would make the integral divergent.


The Standard Theory 83

and we shall verify it a posteriori, by comparing its predictions with


the experimental results.
Let us consider the table of masses of quarks and leptons
(Table 5.1). We know of six species u, d, c, s, t, and b with their
masses, as shown in the table. In Appendix 1 we remark that
each quark comes in three different types, which we call ‘colours’.
Therefore, it is more precise to write the quarks as ui , di , etc, with
the index i taking three values, 1, 2, or 3. It is convenient to group
them together in a lattice form with three lines and six columns:

u1 d1 c1 s1 t1 b1
u2 d2 c2 s2 t2 b2 (6.1)
u3 d3 c3 s3 t3 b3 .

Looking at this we observe that it admits two sets of ‘natural’


transformations: (i) those which mix the columns and leave the
lines untouched and (ii) those which do the opposite, mix the
lines and leave the columns. The first cannot form an exact sym-
metry because the quark masses change considerably from one
column to the other. So, let us try the second. These are trans-
formations which mix the three colours in the lattice 6.1. We may
be tempted to view them as acting on the coordinate system in a
three-dimensional space. However, there is a complication: in this
space the coordinates are the wave functions of the three quarks.
We have noticed previously that wave functions in quantum
mechanics take complex values, so we must consider transform-
ations in a complex three-dimensional space. The counting of
these transformations is not difficult, but we can also refer to
the results presented in Appendix 2 on the number of transform-
ations in the Cartan classification. We find that we obtain nine
independent transformations which we may split into 8 + 1. We
will assume that they all correspond to exact symmetries of the
theory. Let us look separately at the eight and the ninth.
The latter acts on every quark. Following Noether’s theorem,
this implies the existence of a conserved quantity which we can
identify with the number of quarks (minus that of antiquarks).
84 The Origin of Mass

This number must remain conserved during a reaction. Taking


into account the fact that quarks are confined inside hadrons,
this conserved quantum number must correspond to the baryon
number we introduce in Appendix 1. With the reservations we
have already expressed, all present experiments show that it is
indeed a conserved quantum number. Can we conclude that this
gives a gauge symmetry as that associated with electric charge
conservation? The answer is no, because in the table of element-
ary particles there is no massless particle, the analogue of the
photon, for this quantum number. It follows that this is a global
symmetry, which does not generate an interaction.
The remaining eight transformations form the gauge sym-
metry of strong interactions. We have eight massless gauge
bosons, the eight ‘gluons’ of Table A1.3. Like the quarks, they
are also confined inside the hadrons. This property of confine-
ment explains why nuclear forces are not of long range, in spite
of gluons’ zero mass. The range of nuclear forces is limited by
the characteristic size of confinement, which, experimentally, is
of order 1 fermi, or 10–15 m.
The symmetry laws determine the dynamics of interactions
among quarks. By analogy with electromagnetism, we called

Figure 6.3 Quantum chromodynamics: David Gross (1941–, Nobel


Prize 2004); David Politzer (1949–, Nobel Prize 2004); Frank Wilczek
(1951–, Nobel Prize 2004). Gross : D. Gross archives; Politzer : D.
Politzer archives; Wilczek : Photo 2007 by Kenneth C. Zirkel, Wikimedia
Commons.
The Standard Theory 85

this theory quantum chromodynamics.8 It is an asymptotically free


Yang–Mills theory. In Section 6.4 we present the evolution of the
effective interaction strength with distance and we compare this
with the experimental results (see Figure 6.5). We see that the
interaction becomes very strong at distances of order 1 fermi, in
agreement with the range of nuclear forces.
A last remark before closing this section. We have just seen that
quantum chromodynamics contains a characteristic distance of
the order of 1 fermi. In our system of units in which h̄ = c = 1,
this distance corresponds to an energy of the order of 200 MeV,
much larger than the masses of the quarks u and d which form
the nucleons. We believe that this energy scale determines a
new phase transition which generates the nucleon masses. This
explains why, as pointed out previously, protons and neutrons are
much heavier than their constituents, the u and d quarks. Again,
we have no rigorous proof of this result, only indirect evidence,
both phenomenological and numerical.

6.4 The Standard Theory and experiment


In earlier sections we have developed a theoretical framework in
order to describe weak, electromagnetic, and strong interactions.
The essential ingredient is the invariance under local transform-
ations, which we called gauge transformations. Altogether we found
12 independent transformations, four for the electroweak theory
and eight for quantum chromodynamics. As a result, we have 12
gauge bosons which transmit these interactions. The symmetry
is partly broken through the Brout–Englert–Higgs mechanism
and three among these bosons, W ± and Z0 , become massive. The
other nine, the photon and the eight gluons, remain massless.
This theory has revolutionised our ideas concerning the funda-
mental interactions. We often say, and it is usually true, that pro-
gress in physics occurs when an unexpected experimental result
8 From the greek word χρώμα which means ‘colour’. This theory was

formulated by David Gross, David Politzer and Franck Wilczek (see Figure 6.3).
86 The Origin of Mass

contradicts the current theoretical beliefs. This forces physicists


to abandon old concepts and to imagine new ones. However, this
time that is not the way it happened. There was no disagreement
with any experiment. Gauge theories, which introduced geo-
metry into physics, were invented and studied for their intrinsic
beauty and mathematical coherence, not for experimental reas-
ons. The proof of this coherence was first obtained by two Dutch
physicists, Gerardus ’t Hooft and Martinus J.G. Veltman, who
shared the 1999 Nobel Prize. A remarkable feature is that the the-
ory has often been ahead of the experiments and confirmation
of the theoretical predictions was not always immediate. We saw
that in the case of the Brout–Englert–Higgs particle this confirm-
ation took almost 50 years to come. In this section we shall give a
brief account of the great experimental discoveries which consol-
idated the theoretical edifice known for a long time as the Standard
Model and which promoted it to the Standard Theory of Fundamental
Interactions.

1. In 1967, when the formulation of the electroweak model


was presented, we only knew the weak interaction pro-
cesses mediated by the exchange of charged gauge bosons.
A typical example is neutron decay, or the decay of the μ
lepton:

n → p + e– + ν̄(e) ; μ– → ν(μ) + e– + ν̄(e) . (6.2)

In a gauge theory these two reactions are produced by the


exchange of a charged boson W – ; see Figure 6.4(a) and (b).
The reactions mediated by the neutral gauge boson Z0 had
not yet been observed. An example is given by neutrino–
proton elastic scattering,

ν(μ) + p → ν(μ) + p, (6.3)

represented in Figure 6.4(c). This reaction is difficult to


observe experimentally because the neutrinos are neutral;
they interact very weakly and this makes their detection
The Standard Theory 87

ν(μ) p
ν(μ) ν(μ)

μ− ν(e) n ν(e) Z0
W− W−
p p

e− e−
(a) (b) (c)

Figure 6.4 Three weak interaction processes: μ lepton decay (a) ; neut-
ron decay (b); and elastic neutrino–proton scattering (c). The first two
are produced by the exchange of a charged W, the third by that of the
neutral boson Z.

very difficult. The existence of these reactions was first estab-


lished at CERN in 1972, giving concrete evidence that with
the gauge theories we were on the right track. The theory
gives very precise predictions concerning their properties. In
particular, these reactions often involve quarks and leptons
with both right and left chiralities in well-defined propor-
tions. These properties have been verified experimentally.
2. In 1967 only the first three quarks were known, u, d, and s.
On the other hand we knew the four leptons of the first
two families, namely the electron (e– ), its associated neut-
rino (ν(e) ), the muon (μ– ), and its neutrino (ν(μ) ). Looking
at Table A1.3, we see that the second family was not com-
plete. The quark c was missing. This is the reason why the
1967 model proposed by Weinberg applied only to leptons.
Its extension to hadrons led to the prediction of the fourth
quark c (c standing for ‘charm’) and the new hadrons in the
composition of which this quark participates. The discovery
of these ‘charmed’ particles between 1974 and 1976 was the
second great success of this theory.
3. The most characteristic prediction of gauge invariant the-
ories is the existence of gauge bosons which transmit these
interactions. As we have already pointed out, their number
and their properties are fixed by the theory. For electroweak
88 The Origin of Mass

interactions, spontaneous symmetry breaking also determ-


ines their masses. Their actual discovery was a great exper-
imental challenge, because the predicted masses were put-
ting them out of reach of the 1970 accelerators. A new form
of proton–antiproton collider was invented in order to pro-
duce and identify them. This discovery, made at CERN in
1983, gave the 1984 Nobel Prize to Carlo Rubbia, the exper-
imentalist who proposed and supported the project, and to
Simon van der Meer, the engineer who designed and built
the essential element of the accelerator. Later, in the 1990s,
the properties of the W and Z bosons were studied in detail
and all theoretical predictions precisely confirmed.
4. We point out in Appendix 1 that the mathematical consist-
ency of the electroweak theory requires the families to be
complete. We cannot have a doublet of leptons without the
corresponding quarks. More precisely the theorem states
that the algebraic sum of the electric charges of all particles
in a family (quarks + leptons) must vanish. The proof is
quite technical and we shall not present it here. But this
theorem yields precise predictions. The discovery of a new
lepton, the τ , together with its associated neutrino, at the
Stanford Linear Accelerator Centre (SLAC) by Martin Lewis
Perl9 was interpreted as the opening of a new family and the
prediction of two new quarks. Indeed, the quark b was dis-
covered by Leon Lederman at FermiLab, near Chicago, in
1977, and the quark t a little later, first indirectly at CERN
and then directly at FermiLab.
5. The success of quantum chromodynamics is based on the
property of asymptotic freedom which predicts a weak
coupling at short distances. Figure 6.5 shows a theoretical
prediction of the variation of the effective coupling with
the distance, together with the experimental points. The
agreement is impressive.

9 1995 Nobel Prize.


The Standard Theory 89

αs(Q) sept. 2013


τ decays (N3LO)
Lattice QCD (NNLO)
DIS jets (NLO)
0.3 Heavy Quarkonia (NLO)
e+e– jets & shapes (res. NNLO)
Z pole fit (N3LO)
(-)
PP-> jets (NLO)
0.2

0.1
≡ QCDαs(MZ) = 0.1185 ± 0.0006

1 10 100 1000
Q [GeV]

Figure 6.5 The variation of strong interaction effective strength with


the energy scale. The points with error bars are the experimental, or
numerical, measurements. The width of the curve represents the the-
oretical uncertainties. Notice that, in our system of units, distance is the
inverse of energy. CERN, CMS collaboration.

6. According to quantum chromodynamics, the strongly


interacting ‘elementary particles’ are the quarks and the
gluons. The latter are those which transmit the interaction.
Therefore, we must find these inside the hadrons, together
with the quarks. It is more difficult to detect them because
we use mostly electromagnetic interactions to probe the
interior of hadrons, and the gluons are electrically neutral.
Nevertheless, they have been detected and identified at the
DESY research centre in Hamburg.
7. The curve of Figure 6.5 shows that at distances of the order
of a few fermi (or, approximatively, 1 GeV) the interaction
becomes very strong. It is the region in which the hadronic
bound states, such as the proton, the neutron, the mesons
etc., are formed. In Section 6.3 we explained that a phase
transition occurs which generates the largest part of the
90 The Origin of Mass

*
*
B * , B ,B s * , B
s
*
D, D D s D s BC B C
2500

2000
B mesons offset by –4000 MeV

1500
(MeV)

1000

500

0
π ρ K K* η η' 𝛚 ɸ N Λ Σ Ξ Δ Σ* Ξ* Ω

Figure 6.6 The mass spectrum of the light hadrons computed using
numerical simulations of quantum chromodynamics on a space-time
lattice. The masses of hadrons with a b quark are displaced by 4 GeV. The
figure also shows the uncertainties, both theoretical and experimental.
The agreement is impressive. A. Kornfeld, ArXiV 1203.1204.

mass of these hadrons. Our ability to perform analytical cal-


culations is limited to the weak coupling regime, so, in order
to study these strong interaction phenomena, we appeal
to numerical simulations. Figure 6.6 shows the results of
such simulations for the lightest hadronic states, together
with the experimental points. Another confirmation of the
Standard Theory.
8. Last but not least, a brilliant confirmation of the theory is
the recent discovery of the Brout–Englert–Higgs particle,
predicted half a century ago.
7
Epilogue

With the discovery of the Brout–Englert–Higgs particle, the trace


of mass generation, the Standard Theory is complete. All its pre-
dictions have been brilliantly confirmed by experiment. Is this the
end of the story? In this chapter we shall argue that the answer is
no. We shall indicate possible avenues to go beyond the Standard
Theory towards a richer and more coherent model.
The guiding principle is that the precise knowledge of physics
at a given energy scale allows us to guess the possible presence of
new physics at a higher scale. An analogy: imagine that extrater-
restrials are observing us from outer space with telescopes whose
resolution power is limited to 10 m. They can distinguish our
constructions and large pieces of equipment of all kinds, build-
ings, bridges, trains, ships, planes, etc., but they cannot see and
study us. However, based on their observations, they can easily
deduce the presence of living creatures on this planet and even
guess some of our properties: our approximate size must be of the
order of 1 m, (all these constructions are not the work of ants),
that we live in open air, etc. They can therefore conclude that,
by increasing the resolution power of their telescopes, they may
make a discovery, that of new inhabitants of the Universe.
In physics we often find ourselves in a similar situation. The
knowledge of the range of nuclear forces led Yukawa to predict
the existence of the π meson. The properties of weak interactions
suggested the existence of a new class of hadrons, the charmed
particles. We have good reasons to expect that the story will
repeat itself and the detailed study of the Standard Theory will
reveal to us the existence of new physics, hopefully accessible to
92 The Origin of Mass

LHC. We want to present here very briefly some of these reasons.


They are of two sorts:

1. We have already mentioned several questions to which the


Standard Theory offers no answer. It does not explain why
quarks and leptons appear in three families, neither why
their masses are spread over at least 11 orders of magnitude.1
The very origin of the neutrino masses remains a mystery.
In the same spirit, in the Standard Theory the observed elec-
tric charge quantisation, i.e. the fact that electric charges of
all particles appear to be integer multiplets of an elementary
charge, is a mere coincidence.2 A truly fundamental theory
should be able to answer these types of questions.
The most important limitation of the Standard Theory is
the fact that it ignores gravitational interactions. This does
not affect its agreement with experiment because gravita-
tional effects are completely negligible in particle physics
experiments. However, this omission shows that the theor-
etical model is incomplete. This is even more so, since, as
we remarked in Chapter 3, the gravitational interactions are
described classically by general relativity, which is the gauge
theory of space-time symmetries. In spite of all the theorists’
efforts and ingenuity, we have not been able to incorporate
gravitation into the Standard Theory. In fact we are essen-
tially convinced that quantum field theory may not be the
right framework to describe quantum gravity. During the
past 30 years efforts have moved away from the concepts
of point particles and local fields towards the more gen-
eral ones of extended objects, strings, or membranes. From
the theoretical point of view, this approach has produced

1
We recall that the top quark mass equals 173 GeV (1 GeV = 109 eV) and the
neutrino masses are smaller than 1 eV.
2 A large piece of matter, containing a huge number of protons and elec-

trons, always appears electrically neutral. This puts a very stringent limit on a
possible difference between the absolute values of the electric charges on these
two particles. The limit can be written as: |QP /Qe | = 1 ± with < 10–20 .
Epilogue 93

models of great mathematical depth. It offers the only con-


sistent framework to bring together the two major discov-
eries of the early twentieth century, general relativity and
quantum mechanics. On the other hand, from the physical
point of view, we are still far from obtaining models with
sufficient predictive power in order to compare them with
experiment.
2. One could object that all the limitations mentioned earlier
only show that the Standard Theory is not The final theory
of all interactions, hardly a surprising conclusion from the
epistemological point of view. They predict the existence
of new physics, but don’t allow us to estimate the energy
scale of its appearance. In the absence of such an estimation,
even as an order of magnitude, the prediction is not very
interesting. In other words, a prediction must be quantitat-
ive and not merely qualitative. We would like to argue here
that the Standard Theory is sufficiently precise to allow for
quantitative predictions.
Let us come back to the Standard Theory as exposed in
Chapter 6. It is a gauge invariant theory. In the low tem-
perature phase, the phase of our present Universe, the
symmetry is partly broken and the gauge bosons W ± and Z0
which mediate the weak interactions, as well as the fermions
which are the building blocks of matter, are massive. They
acquired their masses through the Brout–Englert–Higgs
mechanism. In the high temperature phase, the symmetric
phase, all these particles are massless; all but the four BEH
spin zero bosons. They are the only particles which do not
get their masses through the BEH mechanism. The mass of
these bosons in the symmetric phase is the only dimension-
ful parameter of the theory, the one which determines the
mass scale. We have seen that, experimentally, this scale is
given by the BEH boson mass, which equals 126 GeV.
The theory contains in addition other dimensionful para-
meters at higher scales. For example, the gravitational
interactions introduce Newton’s constant, which, expressed
94 The Origin of Mass

in GeV, corresponds to a huge scale of order 1019 GeV, often


called Planck’s mass, MPl .
The simultaneous presence of two so widely separated mass
scales is not natural in a fundamental theory. It implies the
introduction of a dimensionless constant, their ratio, which
is of order 10–17 .3 We cannot believe that a fundamental
theory contains parameters of that order. In fact, the situ-
ation is even more serious: quantum field theory allows us
to estimate the corrections to the BEH mass induced by the
presence of the Planck mass. Not surprisingly, these correc-
tions are proportional to MPl2 . In other words, the theory is
not able to naturally sustain two mass scales, so far away
from each other. Strictly speaking the problem is aesthetic:
we do not like the presence of parameters equal to 10–34
in equations of a fundamental theory. Nevertheless, exper-
ience has taught us that aesthetic criteria are often good
guides for deciphering nature’s secrets.
Is there a solution to this problem? The answer is yes, and
even more than one, but they all imply new physics, often
at a scale immediately above that of the BEH particle, of the
order of 1000 Gev. The details of this new physics are model
dependent and cannot be stated precisely. We are in the
place of the extraterrestrials who are looking at our planet.
They know that it is inhabited, but they cannot guess the
details of our appearance. They do not know whether we
are four-legged, two-legged, or three-legged. The new phys-
ics we speculate about, often contains new ‘elementary’
particles with masses of order 1000 GeV, but their nature and
their properties depend on the particular assumptions we
make.
A model which has been analysed in detail postulates the
existence of a new symmetry which relates fermions and

3 In fact, the parameters which enter in the fundamental equations are the

squares of the masses, so the dimensionless parameter is of order 10–34 .


Epilogue 95

bosons. We call it super-symmetry and it has remarkable math-


ematical properties. We also find it as an important ingredi-
ent of string theories which attempt to build a consistent
theory of quantum gravity. Supersymmetry predicts the
existence of a whole series of new particles, the supersym-
metric partners of all existing particles.
There are even more exotic ideas according to which, at a
scale of 1000 GeV, the number of space dimensions changes
and we may discover hidden ones. An example: imagine
a cylinder of radius r and length l, much larger than r. If
we look at it with a spatial resolution worse than r we will
think we are looking at an one-dimensional line. With bet-
ter resolution we will find out it is in fact a two-dimensional
cylindrical surface.

LHC has already given us the BEH particle. In 2015 it started


operating again at higher energies. Naturally, the first item on its
agenda is the detailed study of the new boson properties. We have
very precise theoretical predictions about these. For example,
its interactions with quarks and leptons must depend crucially
on their masses, since it is through these interactions that these
particles become massive. All this must be measured very accur-
ately if we want to confirm all aspects of the BEH mechanism.
But, in addition to this, already very rich programme, it is the
search for new physics which fascinates physicists. The two ideas
we have mentioned, supersymmetry and hidden dimensions, are
two examples among many others. Nature may be hiding fur-
ther surprises. We are reasonably confident that new physics lies
at higher energies and that LHC will be able to find it. Great dis-
coveries, which seem to mark the end of a story, mark in fact the
beginning of a new, more extraordinary one.
AP P E N D I X 1

The Elementary Particles

A1.1 Introduction
If we divide a drop of water into two parts we obtain two drops. They
are smaller but the substance is the same. If we repeat the exercise and
divide each one into two, we obtain four. Carrying on we obtain suc-
cessively eight, sixteen, etc. droplets. How many times can we continue?
Can we envisage that this process will lead us eventually to the smallest
possible drop of water, the elementary drop, or, contrary to this, that it is an
endless process? In other words, is the structure of matter continuous,
or discontinuous?
It seems that it was Democritus from Abdera, Greek philosopher
from the end of the fifth century BC,1 who first gave the right answer:
discontinuous (see Figure A1.1). Although we do not know any details of his
reasoning, we do know that his answer was correct. There does exist an
‘elementary drop’ of water, which we call a water molecule, and we know
very well its properties and its chemical composition. Democritus con-
sidered these elementary constituents of matter as ‘unbreakables’ (he
called them ‘atoms’), but here he was only half right. We can break a
water molecule, but the pieces are no longer water.
Today, matter composition is no longer a philosophical question, but
a domain of intense experimental research. This evolution is due to the
technical progress which has made possible the design, construction,

1 An almost legendary figure, Democritus was born probably in Abdera of


Thrace at around 460 BC and died in 370. Although he was essentially a con-
temporary of Socrates, he is considered among the pre-Socratic philosophers.
According to ancient sources, his work had been immense, but only small frag-
ments of it are known to us. On the other hand numerous apocryphal stories
are attributed to him. Together with his master Leucippus, he is considered to
be the father of atomic theory. According to Democritus, matter is composed
of atoms, (‘impossible to break’, from the Greek word τέμνω, which means ‘to
break’), and the vacuum which fills the space between the atoms.
98 Appendix 1: The Elementary Particles

Figure A1.1 The ‘fathers’ of the structure of matter: Democritus (460–


370 BC), the inventor of the atomic concept; Ernest Rutherford, 1st
Baron of Nelson (1871–1937, Nobel Prize in Chemistry 1908) the first
atomic model; Murray Gell-Mann (1929–, Nobel Prize 1969), the quarks.

Table A1.1 The resolution power of the principal microscopes.

Microscopes
Naked eye 10–4 –10–5 m
Optical microscopes 10–7 m
Electronic microscopes 10–10 m
X rays 10–11 m
α rays 10–13 m
Accelerators ∼ 100 MeV 10–14 –10–15 m
Accelerators ∼ 10 GeV 10–16 –10–17 m
L.E.P., Tevatron 10–18 m
LHC 10–19 m

and operation of more and more powerful microscopes. Table A1.1 gives
a list of the most important types.
Some words of caution concerning this table. The indicated resolu-
tion powers are only approximate because, in fact, the table contains
instruments which are not directly comparable. The resolution of the
human eye is limited by physiological factors, that of optical micro-
scopes by the wavelength of visible light, of order 5 × 10–7 m. A first
revolution came in 1895 with Wilhelm Röntgen, (1845–1923, first Physics
Nobel Prize in 1901), who discovered X rays with wavelength 10–8 m and
Appendix 1: The Elementary Particles 99

obtained his famous pictures. The motivation was not so much resolu-
tion, but rather penetration. With X rays we can ‘see’ bones and other
internal organs. Since they are not in the domain of visible light, it is
not with our eyes that we see the image, but on a photographic plate,
or a computer screen. Today the X ray technique reaches wavelengths
of order 10–11 m or shorter, for the study of structures in materials, or
biological macro-molecules. Figure A1.2 shows the spectrum of elec-
tromagnetic radiation. The separations between γ rays, X rays, etc. are
conventional. Visible light occupies only a small part of the spectrum,
between 400 and 750 nm.
A second revolution was due to Ernest Rutherford, 1st Baron of
Nelson, who, in 1909, had the idea to use α particles. Along with Hans
Geiger and Ernest Marsden, he studied the scattering of α particles, in
fact helium nuclei, from a thin foil of gold. The results were aston-
ishing. Most α particles passed through unaffected, but, occasionally,
some were deflected at large angles. Rutherford interpreted these res-
ults as meaning that the space occupied by the atoms was mostly empty
with some hard grains inside. He proposed a classical atomic struc-
ture following the stellar model: a massive positively charged nucleus
in the middle with electrons, with light masses and negative electric

Increasing freqency (ν)

24 22 20 18 16 14 12 10 8 6 4 2 0
10 10 10 10 10 10 10 10 10 10 10 10 10 v (Hz)

γ rays Χ rays UV IR Microwave FM AM Long radio waves


Radio waves

-16 -14 -12 -10 -8 -6 -4 -2 0 2 4 6 8


λ (m)
10 10 10 10 10 10 10 10 10 10 10 10 10
Increasing wavelength (λ)
Visible spectrum

400 500 600 700


Increasing wavelength (λ) in nm

Figure A1.2 The spectrum of electromagnetic radiation. The part con-


taining visible light is shown expanded. The figure shows also some of
the principal applications of electromagnetic waves according to their
wavelength.
100 Appendix 1: The Elementary Particles

Figure A1.3 The first fathers of quantum theory: Max Karl Ernst
Ludwig Planck (1858–1947, Nobel Prize 1918) the quantum concept;
Niels Henrik David Bohr (1885–1962, Nobel Prize 1922) the old quantum
theory; Prince, later Duke Louis-Victor Pierre Raymond de Broglie
(1892–1987, Nobel Prize 1929) the wave–particle duality. We could add
Albert Einstein for the photoelectric effect; see Figure 2.1.

charge, gravitating around it. Today we know that this simple model
is very naive: a classical electron orbiting around the nucleus would
lose energy by radiation and would fall rapidly into the centre. It was
the Dane, Niels Bohr (Figure A1.3) who completed the picture with
the introduction of quantisation rules which later gave rise to quantum
mechanics.
Rutherford, by introducing with his model the concept of the atomic
nucleus, became the founder of nuclear physics, but in addition, with
his revolutionary experimental method, the father of a new generation
of ‘microscopes’, which use energetic particles to probe the structure
of matter. Table A1.1 shows the evolution of this method. The most
powerful microscopes are, in fact, the particle accelerators. The nat-
ural quantity to label their power is the final energy of the accelerated
particles. We express this in electronvolts (eV), but we often use mul-
tiples of this unit: 1 MeV = 106 eV and 1 GeV = 109 eV. We recall
that in our system of units in which c = h̄ = 1, energy has dimensions:
[energy] = [distance]–1 . Substituting the numerical values of c and h̄, we
find formula (1.1), which gives 1 f = 10–15 m = [200 MeV]–1 . This allows
us to find an approximate relation between the energy of an acceler-
ator and its spatial resolution. Table A1.1 shows that in the course of the
Appendix 1: The Elementary Particles 101

twentieth century we gained a factor of 1 million in resolution. Using


these instruments the following were discovered successively:

molecules → atoms → nuclei + electrons → protons + neutrons + electrons → quarks


+ electrons → ???

Figure A1.4 presents ‘an artist’s view’ of this process towards the
infinitely small.
There is no reason to think that the series ends somewhere, and even
less, that we have already reached this end.

CELL MOLECULE ATOM

10–10 m

10–5 m 10–9 m

NUCLEUS

10–14 m

QUARK
NUCLEON

? 10–15 m

Figure A1.4 Towards the infinitely small. The dimensions are only
indicative, especially for cells and biological macromolecules. In fact
these dimensions vary considerably according to the cell, or the
molecule. For the quarks we know only that they are smaller than
∼10–18 m–10–19 m, which is the LHC resolution.
102 Appendix 1: The Elementary Particles

A1.2 The four interactions


Nature is characterised by an incredible diversity. The variety of forms,
structures, colours, and properties which are offered to our senses
seems to be infinite. Nevertheless, we know today that this is only
apparent; nature is indeed plethoric in forms and properties, but it is
extremely economical in building blocks and fundamental forces. If, as
we have just seen, our ideas on the elementary constituents of matter
have profoundly evolved during the last century, those on the funda-
mental forces have remained remarkably stable. No matter which scale
in matter composition we are looking at, from the smallest particles we
produce with our accelerators, to the largest clusters of galaxies in the
Universe, as well as all degrees of complexity, from the simplest hydro-
gen atom to the most complex biological molecule, all owe their struc-
ture to four fundamental forces. In increasing strength, they are the
following:

1. Gravitational interactions. These are well known from our everyday


experience. They are responsible for falling apples, for the motion
of the planets around the sun, and the structure of the Universe.
However, at the microscopic scale, their intensity is negligible,
much weaker than anything we have been able to measure up to
now.
2. Weak interactions. These are responsible for most radioactive decays
and they initiate the fusion processes at the origin of stellar
energy.
3. Electromagnetic interactions. These are also well known in macroscopic
physics. They are repulsive between same charged particles and
attractive between oppositely charged ones. They have a long
range and they are responsible for the structure of atoms and
molecules, as well as many properties of condensed matter.
4. Strong interactions. These are responsible for nuclear structure. They
are attractive between the constituents of nuclei (protons and
neutrons) and stronger than the electrostatic repulsion between
protons. They have a short range of order 1 f = 10–15 m.

A large part of this book is devoted to the theoretical description


of these interactions. From a historical point of view, it is a process
which started long ago, but the modern version of theoretical particle
Appendix 1: The Elementary Particles 103

physics has a precise birth date: 2 June 1947, the date of the Shelter
Island Conference. The most important contributions presented in
this conference were not brilliant theoretical breakthroughs, but two
experimental results. Willis Eugene Lamb (1913–2008, Nobel Prize 1955),
of Columbia University, announced the measurement of an energy
shift between the levels 22 S1/2 and 22 P1/2 of the hydrogen atom and
Isidor Isaac Rabi (1898–1988, Nobel Prize 1944), that of an ‘anomal-
ous’ magnetic moment of the electron. The interest in these results
was that, for the first time, they were in clear contradiction with the
Dirac (Figure A1.5) theory, which was ‘the bible’ of theoretical phys-
ics up to that time. They forced theorists to elaborate a more general
and mathematically coherent theory which revolutionised our funda-
mental physical ideas and gave rise to the theory we describe in this
book.
This theory is called quantum field theory and it is the quantum mech-
anics of a system with a very large, in fact infinite, number of degrees
of freedom. Attempts to formulate such a theory go back to 1925, with
the work of many physicists among whom we find Born, Heisenberg,
Jordan, Dirac, Pauli, Fermi, etc., but, under its modern form, it was
developed by Richard Phillips Feynman, Julian Seymour Schwinger,

Figure A1.5 Together with W. Heisenberg, see Figure 3.8, they are
the fathers of quantum mechanics: Erwin Rudolf Josef Alexander
Schrödinger (1887–1961, Nobel Prize 1933) the first wave equation in
quantum mechanics; Paul Adrien Maurice Dirac (1902–84, Nobel Prize
1933) the relativistic wave equation and the first quantum field theory
computation.
104 Appendix 1: The Elementary Particles

Figure A1.6 The founders of ‘modern’ quantum electrodynamics:


Richard Philips Feynman (1918–88, Nobel Prize 1965); Julian Seymour
Schwinger (1918–94, Nobel Prize 1965); Sin Itiro Tomonaga (1906–79,
Nobel Prize 1965).

and Sin Itiro Tomonaga (Figure A1.6) in 1947 and completed the follow-
ing year by Freeman John Dyson. The first application is named quantum
electrodynamics and describes with high precision the interaction of elec-
trons with an electromagnetic field. An example of the extraordinary
agreement between this theory and experiment is given by the electron
anomalous magnetic moment measured by Rabi.2 The most recent values are:

ae (exp) = 0.00115965218076(27)
ae (th) = 0.00115965218178(77), (A1.1)
2 The magnetic moment is the quantity which determines the interaction of
a particle with a magnetic field. That of the electron, denoted as μe , is usually
expressed in units of ‘an elementary magnetic moment’, the Bohr magneton, given
by μB = eh̄/2me , where e and me are the electron charge and mass, respectively.
The term anomalous has a purely historical origin. Before the advent of quantum
electrodynamics, physicists were computing the electron magnetic moment
using Dirac’s equation, which is the relativistic generalisation of the quantum
mechanical Schrödinger equation for the electron. The Dirac equation predicts
a value for the ratio ge = μe /μB , called the gyromagnetic ratio, equal to 2. The early
experiments, which were not very precise, were compatible with this predic-
tion and provided one of the successes of the Dirac theory. Therefore, a value
different from 2, like that announced by Rabi, was at first considered to be
‘anomalous’. Of course, today we know that there is nothing anomalous about
such a value, but the term has survived and the results are always presented for
the quantity ae = (ge – 2)/2.
Appendix 1: The Elementary Particles 105

where we have indicated the theoretical prediction of quantum


electrodynamics. The numbers in parentheses correspond to the
uncertainties, experimental or theoretical. The agreement is impress-
ive. In fact, the precision of this measurement, combined with the
theoretical calculations, yield today the best determination of the value
of the electron charge.
A remarkable feature of quantum electrodynamics is that it asso-
ciates a particle with the electromagnetic field, the ‘quantum’ of the
field, which we call a photon.3 The interaction between charged particles
is described by the exchange of one or more photons. We say that
the photon is the messenger of the electromagnetic interactions, a notion
which, as we saw in Chapter 4, is generalised to all other interactions. By
analogy, we shall call the messengers of all interactions radiation quanta.

A1.3 Some basic notions


Before presenting the table of elementary particles, we want to intro-
duce a dictionary of some concepts and the associated terminology,
which will be useful throughout this book. They are the various prop-
erties which we associate with elementary particles.

1. Mass. This is the familiar concept which we know from clas-


sical mechanics: under the influence of a force F, the particle
acquires an acceleration a, given by Newton’s formula F = ma. A
first abstraction comes from special relativity, which tells us that
the value of the mass that enters Newton’s formula depends on
the speed of the particle and is given by m = m0 / 1 – v2 /c2 , where
m0 is the mass of the particle in the coordinate system in which
it is at rest and c is the speed of light in vacuum. It is the mass
m and not m0 which we use in Einstein’s formula E = mc2 . It fol-
lows that a massive particle can never reach the speed of light,
because the resulting energy would be infinite. On the other hand
we can have particles whose ‘rest mass’ is equal to zero but which
have non-zero energy because their speed equals the speed of light

3 The concept of the photon as the mediator of an electromagnetic inter-

action had been introduced earlier, but it is in the framework of quantum


electrodynamics that this concept acquires a precise meaning.
106 Appendix 1: The Elementary Particles

in all reference frames. Such particles exist in nature, the well-


known example being the photon, the particle associated with an
electromagnetic field.
An old theoretical idea is that particles acquire their mass, at
least partly, through their interactions. Already by the end of the
nineteenth century Lorentz had attempted to compute the ‘self
energy’ of the electron, i.e. the energy coming from the electron’s
interaction with its own electromagnetic field. Since the electro-
static potential of a point charge at a distance r is proportional
to 1/r, this energy, computed at the point r = 0 of the position
of the particle, is, in fact, infinite. This problem haunted theor-
etical physics throughout the first half of the twentieth century
and it was only with the advent of quantum field theory that we
understood that this particular problem is due to the fact that the
question is not well defined. In the electron example, we can only
measure its physical mass and not that which it would have had in
the absence of its own field. However, the question, appropriately
rephrased, is still relevant and people have tried, though unsuc-
cessfully, to attribute the small proton–neutron mass difference
to the proton’s electric charge.
2. Quantum numbers. This concept generalises that of electric charge. A
particle’s charge can be positive, negative, or zero. We know that
electric charge is a conserved quantity: in a reaction, the algebraic
sums of all electric charges in the initial and the final states are
equal. Charge conservation has an immediate consequence: the
electron, which is the lightest charged particle, is stable. All obser-
vations reveal another remarkable property of electric charge: all
charges are integer multiplets of an elementary unit of charge. We
do not have a deep understanding of this property and this shows
that our theoretical ideas are still incomplete.
Experiments indicate the existence of additional quantities which
appear also to be conserved. We call these ‘quantum numbers’
and they also take discrete values. The most common example
is the ‘baryon number’4 which equals 1 for the constituents of
nuclei, the proton and the neutron,5 and zero for electrons. The
study of nuclear reactions shows that, although protons and
neutrons may transform into each other, their absolute number
4 From the Greek word βαρύς, which means ‘heavy’.
5 We call them collectively nucleons.
Appendix 1: The Elementary Particles 107

remains constant. We have never observed nucleons appearing, or


disappearing spontaneously. If baryon number is conserved, the
proton, which is the lightest baryon, must be stable and this would
explain the great stability of nuclear matter. On the other hand, if
this conservation is exact, we cannot explain the creation of bary-
ons just after the Big Bang. For this reason physicists consider that
baryon number can be only conserved approximately and con-
siderable experimental effort aims at detecting a possible proton
decay.
Another remark concerning quantum numbers. Relativistic
quantum mechanics predicts, and experiments confirm, that for
every particle there is an associated anti-particle, which has exactly
the same mass as the particle, but carries the opposites of all
quantum numbers. To the electron there corresponds the positron,
which has a positive electric charge,6 to the proton, the antiproton
with negative electric charge and baryon number, to the neut-
ron, the antineutron with zero electric charge and baryon number
equal to –1, etc. There exist particles, like the photon, which carry
no quantum number different from zero. It follows that they are
indistinguishable from their anti-particles. The ‘antiphotons’ are
identical to the photons.
3. Spin. In classical mechanics this term denotes the motion of a spin-
ning top and it is a special case of the more general concept of
angular momentum. In quantum mechanics we define a particle’s spin
as its intrinsic angular momentum, but we must abandon the
image of a spinning top, which is meaningless for a point particle.
Like many other quantum mechanical quantities, the spin can
take only discrete values, which, in the appropriate units, are
positive numbers, integers, or half-integers. Particles with integer
spin (0, 1, 2, . . . ) are called bosons, in honour of the Indian physi-
cist Satyendra Nath Bose (1894–1974) and those with half-integer
spin (1/2, 3/2, . . . ) fermions, in honour of the Italian physicist Enrico
Fermi.7 The photon, which has spin equal to 1, and the BEH

6 The existence of an electron’s anti-particle was predicted by Dirac as a con-


sequence of the equation of relativistic quantum mechanics which bears his
name. This prediction was brilliantly confirmed in 1932 by Carl David Anderson
(1905–91, Nobel Prize 1936), who discovered the positron in cosmic rays.
7 To be precise, this result, along with many others we present in this book,

applies to particles in our three-dimensional space. This precision is necessary


108 Appendix 1: The Elementary Particles

particle, which has spin equal to 0, are examples of bosons,8 while


the electron, the proton, the neutron, etc., which have spin equal
to 1/2, are examples of fermions.
In our three-dimensional space there is a profound difference in
the behaviour of these two kinds of particles. Let us consider two
identical particles, one at point x and the other at point y. In
Box 3.1 we explained that in quantum mechanics, the state of a
physical system is described by a complex valued function which
we call a ‘wave function’. We also explained that the phase of this
function has no physical significance. It follows that the state of
these two particles will be described by a wave function (x, y, t)
defined up to a phase. Let us imagine now that we interchange the
two particles and we bring the first one to point y and the second
to x. Since the particles are assumed to be identical, we expect the
new system to be described by the same wave function, up to a
phase:

(x, y, t) = C(y, x, t), (A1.2)

where C is a complex number of modulus 1. If we repeat the


exchange operation we bring the system back to its original state,
but the wave function is multiplied by C2 ,

(x, y, t) = C(y, x, t) = C2 (x, y, t). (A1.3)

We conclude that C2 = 1 and, consequently, C = ±1. The wave


function of two identical particles is either symmetric, or anti-
symmetric, under their exchange. It turns out that a deep the-
orem of quantum field theory proves, and experiment confirms,

because the properties of a physical system often depend on the number of


dimensions of the ambient space. In particular, the spin concept, as well as the
distinction between fermions and bosons, change if we consider a system on
a two-dimensional surface, or a one-dimensional line. This is easy to under-
stand. We defined spin as an angular momentum, therefore it is related to the
properties of spatial rotations. These properties depend on the number of space-
dimensions. In a three-dimensional space we can perform three independent
rotations, on a plane we have only one, and in a line the very notion of rotations
is lost.
8 We often call it the ‘BEH boson’.
Appendix 1: The Elementary Particles 109

that under such an exchange, the wave function is symmetric,


(x, y, t) = (y, x, t), if the particles have integer spin (bosons),
and anti-symmetric, (x, y, t) = –(y, x, t), if they have half-integer
spin (fermions). This implies in particular that for the fermions,
the wave function must vanish if x = y; in other words, two fer-
mions cannot occupy the same place.9 This difference, which
sounds purely technical, is at the origin of the partition of elec-
trons in atomic orbits giving rise to the structure of matter. Let
us take, for example, an atom with N electrons. Because of Pauli’s
principle, they cannot all sit at the lowest energy state and they
are forced to occupy successive atomic shells, whose number
increases with N. This is the explanation for Mendeleev’s periodic
table of elements. In contradistinction, there is no exclusion prin-
ciple for bosons and nothing prevents an arbitrarily large number
of them agglomerating in the same quantum state. We call this
phenomenon Bose–Einstein condensation and we expect to find it at
low temperature where thermal agitation is small. Many phys-
ical phenomena with important technological applications result
from this symmetry property of bosons. To name but a few, we
have the laser, and also superconductivity, superfluidity, etc. In
Chapter 5 we indicated that the electroweak phase transition, the
subject of this book, which is responsible for the creation of most
particle masses, is precisely due to the bosonic character of the
Brout–Englert–Higgs particle.
4. Leptons, hadrons. In Section A1.2 we introduced the four funda-
mental interactions between elementary particles. Experiments
show that some particles are not subject to all these interactions.
In particular, some fermions, like the electron, are insensitive to
strong interactions and we call them leptons.10 Others, such as the
nucleons, protons, and neutrons, are sensitive and we call them
hadrons.11 In the way we introduced the baryon quantum number,
we can also introduce a leptonic quantum number, or lepton number. This

9 This property was first postulated by W. Pauli for electrons and it is known
as Pauli’s exclusion principle. To be precise, it applies under the complete exchange
of the two particles, including their position in space as well as other variables
which characterise their state, such as their spin orientations.
10 From the Greek word λεπτóς, which means ‘fine’ or ‘thin’.
11 From the Greek word αδρóς, which means ‘strong’.
110 Appendix 1: The Elementary Particles

equals 0 for the hadrons and 1 for the leptons; the electron has
leptonic number equal to +1 and its anti-particle, the positron, –1.

A1.4 The neutrino saga


The neutrino is the most elusive and probably the most fascinating
of all elementary particles. Its story is an extraordinary and instruct-
ive chapter in the history of modern physics. It is intimately related
to all developments of the twentieth century, it involves remarkable
experiments and brilliant theoretical ideas, but also it provides for an
exemplary illustration of the scientific process. We shall present here a
very short account of the early part of the story.
It begins with the discovery of nuclear β-decay by Henri Becquerel12
in 1896 and the identification of the emitted particles as electrons in 1902.
Today we know that the reaction is, in fact, the decay of a neutron giving
a proton, an electron, and an anti-neutrino. For a neutron bound to a
nucleus, this reaction appears as a nucleus giving a second nucleus, an
electron, and an anti-neutrino:

n → p + e– + ν̄ ⇒ N1 → N2 + e– + ν̄. (A1.4)

For the physicists of the early twentieth century this reaction was a
seemingly endless source of questions:

1. They knew neither the neutron nor the neutrino. In any case,
with the technology available at the time, the latter was undetect-
able. So, they were seeing only

N1 → N2 + e– . (A1.5)

2. Since they could not imagine that an electron could come out of
a nucleus without being there in the first place (the quantum field
theory for electrons which describes how particles can be created
by the interaction was first formulated by E. Fermi in 1933), they
had a nuclear model according to which nuclei were bound states
of protons and electrons.

12 Antoine Henri Becquerel, 1852–1908, Nobel Prize 1903.


Appendix 1: The Elementary Particles 111

But the serious problems started with the study of the emitted elec-
tron energy spectrum. Let us begin with a little kinematical exercise.
Let a particle A decay into two other particles B and C : A → B + C.
The energy E of a particle of mass m and momentum p is given by:
E = m2 + |p|2 . In the rest frame of particle A, momentum conser-
vation gives pB + pC = 0. Energy conservation gives mA = EB + EC . It
follows that
m2A + m2B – m2C m2 + m2C – m2B
EB = ; EC = A ; (A1.6)
2mA 2mA
in other words, kinematics uniquely determines the energy of each one
of the final particles. This reasoning, applied to reaction (A1.5) which
was believed to describe β-decay, implies that the emitted electrons ought
to be mono-energetic.
Indeed, the first experiments, between 1906 and 1914, were roughly
in agreement with this result. These came mostly from Berlin, the
chemistry lab of Otto Hahn (1879–1968, Chemistry Nobel Prize 1944)
and Lise Meitner,13 in which the electron energy was estimated by its
penetration length. Poor precision allowed them to interpret the res-
ults as a series of mono-energetic rays and to attribute the absence of a
single ray to impurities in the radioactive source.
The situation changed in 1914 thanks to James Chadwick, former
student of Rutherford in Manchester (see Figure A1.7). Chadwick too
arrived in Berlin to work with Geiger, Rutherford’s former assistant in
the α-scattering experiment. Chadwick used a magnetic field to meas-
ure the electron energy and a system of counters for their detection.
The result was unambiguous: no unique ray. The spectrum appeared to
be continuous14 with the electron energy between zero and a few MeV.
This result is in violent contradiction with equation (A1.6) which, as we
13 A ‘grande dame’ of nuclear physics (1878–1968). Born in Vienna, she stud-
ied with Boltzmann and came to Berlin to work with Max Planck in 1907.
Together with Otto Hahn, she played a very important role in all experiments
on radioactive elements. The story goes that they had to work in an abandoned
warehouse because the regulations, in effect up to 1909, did not allow women
to access the main laboratory. She left Berlin in 1938, just in time to escape Nazi
persecution and found shelter in Sweden.
14 This experiment offers several fascinating aspects. Firstly the result,

which, as we have noted, was against all established ideas. Secondly the novel
experimental method. Finally, the absence of any apparent contact, let alone
112 Appendix 1: The Elementary Particles

Figure A1.7 The neutrino saga: James Chadwick (1891–1974, Nobel


Prize 1935); Wolfgang Pauli (1900–58, Nobel Prize 1945); Enrico Fermi
(1901–54, Nobel Prize 1938); Pauli : CERN Pauli archives; Fermi : Univ.
Roma I, Amaldi archives.

explained, is a straightforward consequence of energy and momentum


conservation. This was the first energy crisis.
The result was confirmed by an experiment in 1916, but apart from
that, nothing happened until the beginning of the 1920s; there was a
war going on. The following years were those of great confusion. Several
experiments were reported among which we find the name of the
young Ellis. We shall come back to him. Hahn and Meitner performed a
new series of measurements, always insisting on their interpretation of
mono-energetic rays. They proposed an ingenious theoretical scheme
which was roughly the following. The decay produces electrons with
energy given by formula (A1.6). But we should recall that the electrons
collaboration, between Hahn’s team in chemistry and that of Geiger in phys-
ics. This is surprising since they were all in Berlin, they were studying the same
decays and, even more so, they were all coming from Rutherford’s lab. But
what follows is also interesting. Chadwick found himself in Germany when
the war started. He was arrested and interned at a prison camp in Ruhleben
for the duration of the war. The surprising thing is that Chadwick found the
means to continue his experiments in the camp, kept his correspondence with
Rutherford, and continued to receive scientific publications as well as visits
from German colleagues, such as Geiger and Otto Frisch. In the camp he met
Charles Drummond Ellis, an officer cadet of the British army who was on hol-
iday in Germany and was also interned in Ruhleben. Apparently, Chadwick’s
enthusiasm for science was such that Ellis, four years younger, decided to give
up his plans for a military career and study physics. We shall have the occasion
to present his work shortly.
Appendix 1: The Elementary Particles 113

were supposed to be among the nucleon’s constituents. Therefore, an


electron ejected from the interior of the nucleus, has a ‘primary’ energy
given by (A1.6), but it loses part of this by rescattering before getting
out. Thus, the continuous spectrum. Apparently unanswerable. Other
explanations of a similar nature were also proposed.
At this point Ellis arrives. After the war he studied physics at
Cambridge and joined Rutherford’s team at Cavendish. He published
some work on nuclear spectra, which showed already his great skill in
experimental techniques, but his masterpiece is a revolutionary method
for resolving experimentally the controversy on β-decays.
How can we measure the primary energy of an electron, if we only
have access to its final energy? Ellis and William Alfred Wooster15 found
the answer and in doing so, established a new experimental technique,
calorimetry.16 Their reasoning was as follows. Experiments show a con-
tinuous spectrum for β-decay electrons. It is comprised between zero
and a maximum value Emax . But we do not know whether it is the
‘primary’ energy, or that we observe after rescattering. Let us consider
a radioactive substance. We can measure the number N of decays in a
time interval T. If we manage to measure all the energy Etot released
during the time T, we can have the answer without knowing the path
of each individual electron. Because, if Hahn and Meitner are right, we
must find Etot = NEmax , while, if Chadwick is right and the primary
electrons are indeed released with a continuous spectrum, the result
will be Etot = NEav , where Eav is the average value of the energy. Therefore,
we need a calorimeter and not a spectrometer.
It took Ellis and Wooster two years to build and operate their calor-
imeter. The increase in temperature they measured was of order 10–3
degrees Celsius. The total energy deposited was proportional to the
average value of order 0.34 MeV, far below the maximum value, which
was larger than 1 MeV. This measurement put an end to all attempts to
explain the spectrum by secondary effects. The year 1927 was that of the
second, and most serious, energy crisis.
What was the reaction of the scientific community to this new
measurement? Firstly, Lise Meitner declared she had felt a great shock.
15 A student at Cambridge (1903–84), Wooster followed a scientific career in
crystallography.
16 Today calorimeters are essential components of any modern detector.

The BEH boson identification at CERN was largely based on data produced by
electromagnetic calorimeters.
114 Appendix 1: The Elementary Particles

As a good experimentalist, she repeated the experiment and confirmed


the result. In her 1929 article she correctly acknowledges the work
of Ellis and Wooster, but she proposes no explanation. Resolving the
energy crisis was now left to the theorists.
No lesser man than Niels Bohr, in collaboration with Hendrik
Anthony Kramers from Holland and John Clarke Slater from the USA,
proposed in 1924 a theory in which all conservation laws, including that
of energy, held only ‘on average’ and not for every individual process.
These were the first years of quantum mechanics and Bohr was willing
to upset any established principle. Einstein also envisaged such solutions
for a while, but he soon gave up. Pauli remained very critical towards all
these speculations. The energy crisis was the principal problem but, in
fact, there were also others which increased the general confusion. The
first spin measurements of various nuclei often gave results in apparent
contradiction with angular momentum conservation. Pauli’s exclusion
principle, which explained atomic spectra very well, seemed to fail in
nuclear physics. Niels Bohr, at a conference in London in 1930 (‘The
Faraday Conference’) presented a good summary of all the problems
encountered in nuclear physics. Today we know that they were due in
large part to the nuclear model itself, which assumed the nuclei to be
bound states of protons and electrons but, at that time, physicists had
no solid basis for questioning this.
This brings us to December 1930. A conference was organised in
Tübingen to discuss all problems related to what was termed at that
time ‘radioactivity’. Pauli was invited but decided not to attend. He sent
a letter instead, dated December 4, written in his own light, inimit-
able style, in which he launches one of the most speculative, but also
profound ideas. The letter starts with ‘Dear radioactive ladies and gen-
tlemen’. A little further down Pauli declares that he regrets not being
able to participate in the conference, but his presence at a ball organ-
ised by the Italian Students Association in Zürich was indispensable.
In the main part of the letter he writes ‘. . . I found a desperate way
to solve the problem of statistics . . . as well as that of the continuous
spectrum in β-decay. . . . the possibility of the existence inside the nuc-
leus of neutral particles with spin 1/2. Thus the continuous spectrum
could be explained by the hypothesis that one of these particles is emit-
ted together with the electron in the decay . . . Your humble servant, W.
Pauli’.
Appendix 1: The Elementary Particles 115

Pauli calls this new particle a neutron but the properties he attributes
to it, in particular its mass,17 are not those of the future neutron. It is
also worth noticing that Pauli does not challenge the prevailing hypo-
thesis according to which everything that comes out of a nucleus exists
already inside it.
Pauli never published this proposal. He had probably submitted it to
his friend Heisenberg, because the latter, in a letter dated December 1st,
refers to ‘your neutrons’. Pauli presented it to various conferences and
in June 1931, the news about this new particle made headlines in the
New York Times.
In January 1932, Chadwick discovered a new particle which he also
called a ‘neutron’. We thus had two particles with the same name. Fermi
coined the term ‘neutrino’, meaning ‘little neutron’ in italian. The same
year Heisenberg proposed the theory of isotopic spin, which we have
presented in Chapter 3. Soon Chadwick’s neutron became the accepted
ingredient of nuclear matter, next to the proton, and nuclear physics
took its present form. The following year Fermi arrived at a general syn-
thesis with the theory of β-decay. It was the first theory formulated in
the language of quantum field theory. It also established the idea that
particles are created or annihilated by the interactions. Electrons and
neutrinos don’t ‘come out’ of a nucleus, they are created at the moment
of decay.
Even if the controversy concerning possible violations of the conser-
vation laws continued for some time,18 the year 1933 marks the end of
the first act of the neutrino saga. As we have noted already, it is a perfect
example of the scientific process:

A new phenomenon is discovered and physicists launch a series of


experiments which often yield incompatible results.
New experimental techniques are invented in order to obtain reliable
measurements.
As far as possible, people try to interpret the results in the framework of
existing theories.
When this is proven impossible, all avenues are explored, even the most
speculative.

17 Pauli estimates that the mass of this particle should be less than one

hundredth of that of the proton.


18 We find contributions by Bohr and even Dirac, as late as 1936.
116 Appendix 1: The Elementary Particles

At the end a new idea, often simple and elegant, emerges and imposes
itself. The neutrino’s existence ceased very soon to be a subject of
controversy. In Fermi’s theory the neutrino is a ‘particle’ like any
other.

It is the end of the first act, but not the end of the story. Neutrinos
revealed many more surprises throughout the twentieth century. We
shall not tell the entire story here, we will just mark the most important
milestones.

1. 1956: First direct observation of a neutrino. Frederick Reines


and Clyde Lorrain Cowan19 working in the Los Alamos
nuclear reactor, observed a reaction produced by an
(anti-)neutrino. In fact neutrinos have neither strong nor
electromagnetic interactions. As a result, they interact with
matter very weakly. For a neutrino with an energy of a few
MeV the probability of having an interaction with a nucleus
is extremely small. Hence the great difficulty in observing it.
Nuclear reactors produce, by fission, a large number of unstable
nuclei whose decay products contain neutrinos. Reines and
Cowan succeeded in detecting the inverse of the reaction (A1.4):
ν̄ + N1 → N2 + e+ , with the detection of the recoil of the
nucleus N2 and the positron, in coincidence.
2. 1957: Maurice Goldhaber measured the polarisation of a neutrino
produced in β-decay. It was an incredible ‘tour de force’: measur-
ing the polarisation of a particle we cannot see. One of the most
beautiful experiments in the history of elementary particles.
3. 1962: The second neutrino. The arrival of large accelerators
marked the era of a new source of neutrinos. They are those
found among the decay products of the π mesons, predicted
by Yukawa and presented in Chapter 4. Their main decay mode
is π → μ + ν, where μ is a new lepton, like the electron, but
much heavier (see Section A1.5.2). So the question arose whether
these neutrinos were the same as those produced in β-decay. In
1962 Leon Max Lederman, Melvin Schwartz, and Jack Steinberger,
using the first neutrino beam built in the Brookhaven acceler-
ator, showed that the answer was negative: there indeed exist two
19 Clyde Lorrain Cowan Jr, born in 1919, died in 1974. For the first neutrino

observation his collaborator Frederick Reines received the Nobel Prize in 1995.
Appendix 1: The Elementary Particles 117

distinct kinds of neutrinos.20 Later we discovered a third.21 We


present these in the particle table at the end of this appendix.
We have introduced the notation ν(e) , ν(μ) , and ν(τ ) in order to
distinguish between them.
4. 1972: Neutral currents. In Section 6.4 we explained the import-
ance of reactions like the elastic scattering of neutrinos on nuclei,
which we called ‘neutral current reactions’. These were first
observed at CERN by the ‘Gargamelle’ collaboration.
5. 1998: First observation in Japan, by the ‘Super-Kamiokande’ col-
laboration of a strange phenomenon named ‘neutrino oscilla-
tions.’22 We have just explained that there exist three species of
neutrinos. The aforementioned experiments showed that a neut-
rino produced in one of these species has a certain probability of
manifesting itself later as belonging to another species. In other
words, during their propagation, neutrinos oscillate among the
three species. The first observation concerned solar neutrinos, but
the phenomenon was confirmed for all neutrinos, irrespective of
their origin. A beautiful illustration of the fundamental laws of
quantum mechanics.

A1.5 The table of elementary particles


A1.5.1 The elementary particles in 1932: the world is simple
Before presenting the table of elementary particles known today, we
start with the much simpler one of 1932. The year is not chosen ran-
domly. 1932, the year the neutron was discovered, marks the beginning
of elementary particle physics, the year in which our ideas on the
structure of matter started to take their present form.
In 1932 we knew only one ‘quantum of radiation’, the photon, the
quantum associated with an electromagnetic field. With regard to the
constituents of matter, we knew, of course, the electron, we had just
20 For this discovery Lederman, Schwartz, and Steinberger shared the 1988
Nobel Prize.
21 The Large Electron–Positron (LEP) collider at CERN has shown that there

are no further light neutrino species.


22 For this observation Raymond Davis Jr. and Masatoshi Koshiba received

the 2002 Nobel Prize. In fact, the 2015 Nobel Prize was attributed to Arthur B.
McDonald and Takaaki Kajita, who studied this phenomenon further.
118 Appendix 1: The Elementary Particles

Table A1.2 The table of elementary particles in 1932.

Table of elementary particles


Radiation quanta
Photon (γ )
Matter particles
Leptons Hadrons
νe , e– p, n

completed the doublet of nucleons with the discovery of the neutron,


and two years earlier Pauli had postulated the existence of the neutrino
in order to explain the electron spectra in β-decay (see Table A1.2).
This completes the presentation of the 1932 table of elementary
particles. By looking again we can make the following remarks:

1. All matter particles have spin 1/2. The quantum of radiation has
spin 1.
2. The algebraic sum of electric charges of all matter particles is equal
to zero. At this stage this property appears to be a coincidence but,
as explained in Section 6.4, it is important.
3. Every entry in Table A1.2 plays an important and well understood
role in the structure of matter. This is obvious for the electron, the
proton, and the neutron, which are the constituents of atoms and
molecules. It is also obvious for the photon, which transmits elec-
tromagnetic interactions and allows for the formation of atoms.
But it is also true for the neutrino, whose role is not visible at
the classical level. In fact, neutrinos appear in neutron decay and
make nuclei with a large neutron excess unstable. On the other
hand, the same reaction is at the origin of stellar evolution and,
indirectly, energy production in stars. The cycle of nuclear reac-
tions in the interior of stars, such as that of the sun, starts with
the fusion of two protons to give a deuteron, a neutron–proton
bound state:

p + p → d(p, n) + e+ + ν(e) . (A1.7)

In addition, the great transparency of matter to neutrino radi-


ation, makes the latter essentially the only means of a massive star
Appendix 1: The Elementary Particles 119

radiating away energy and lowering its temperature. Thus, neut-


rinos too, have a decisive role in the structure of the Universe.

A1.5.2 The elementary particles today


The disorder
This simple picture with a small number of elementary particles did
not last for long. In 1937 a new particle was discovered in cosmic rays,
which, after a few episodes, turned out to be a new lepton, called a muon
(μ). Later, in 1962, it was established that it is accompanied by its own
neutrino and it carries a new quantum number. In order to distinguish
among the various neutrinos, we have put the lepton with which each
one is associated as an index. Thus we write ν(e) , ν(μ) , etc. The muon is
200 times heavier than the electron but otherwise, the doublet (ν(μ) , μ)
seems to have the same properties as the more familiar (ν(e) , e). In ret-
rospect, muon discovery marked a turning point in our concepts of
elementary particle physics: it is the first particle whose specific role in
the structure of matter still remains unknown.23
After the Second World War the proliferation of ‘elementary’ particles
increased at a fast pace. Looking for the mediator of nuclear forces,
whose existence was predicted in 1935 by Hideki Yukawa, physi-
cists discovered a large number of hadrons. The first was precisely
Yukawa’s π meson24 which was discovered in cosmic rays in 1947. With
the operation of the large accelerators the discovery rate increased
exponentially and today’s Particle Data Book has more than one
hundred entries. Even more importantly, all simple rules deduced
from Table A1.2 were violated: among the new hadrons some had
integer and some half-integer spin and the distinction between mat-
ter particles and radiation quanta was impossible. As for the role of
each one of those particles in the structure of the world, physicists
did not even dare ask the question. Complete disorder appeared to
reign.
23 It seems that it was I. Rabi who, after discovery of the muon, asked the
question: ‘Who ordered that?’
24 From the Greek word μέσoν, which means ‘middle’. In fact, the first

mesons to be discovered had masses intermediate between that of the nucleons


and of the electrons.
120 Appendix 1: The Elementary Particles

The quarks
By the early 1960s we had accumulated a considerable number of had-
rons and their properties had been sufficiently studied for physicists to
be able to attempt to apply some order to this chaos.25 Past experience
had taught us that the enormous variety of atoms and molecules which
surround us result from a surprisingly small number of constituents. It
was thus natural to try to interpret all these hadrons as bound states
of a few, more ‘elementary’ particles. There had been several attempts
in this direction, but the model which was finally confirmed by exper-
iment is the one proposed independently by Murray Gell-Mann and
George Zweig in 1964. Its basic postulate is that all hadrons are made
out of spin 1/2 building blocks, the quarks.26 In 1964 we thought there
were three types of quarks, today we know there are six. In Table A1.3
we write them as (u, d, c, s, t, b). Baryons are formed out of three quarks
and mesons out of a quark–antiquark pair. For example, the proton is a
bound state of two quarks u and one quark d, the neutron of one quark u
and two quarks d. Table 6.1 contains more examples of hadron compos-
ition in terms of quarks. These examples show that the electric charge
of the quark u equals 2/3 and that of the quark d is –1/3. Therefore, the
‘elementary charge’ in the hadronic world equals 1/3 times the elec-
tron charge. Similarly, the other four quarks, c, s, t, and b, form hadrons
which are produced in our accelerators and are unstable. We can find
them in the Particle Data Book, which is updated regularly. We shall
not need them in this book.
A last remark concerning the number of quarks. We are in fact obliged
to complicate this simple picture of six quarks, sole constituents of
all hadrons. We have seen that quarks have spin 1/2, they are there-
fore fermions. In Section A1.3 we noticed that identical fermions have
the strange property of anti-symmetry under the exchange of any pair
among them. Let us consider the example of a hadron called N ∗++ ,

25 In a review article in 1957 on elementary particles, Murray Gell-Mann and


Edward P. Rosenbaum wrote: ‘. . . At present, our level of understanding is about
that of Mendeleyev, who discovered only that certain regularities in the prop-
erties of the elements existed. What we aim for is the kind of understanding
achieved by Pauli, whose exclusion principle showed why these regularities were
there, and by the inventors of quantum mechanics, who made possible exact
and detailed predictions about atomic systems . . . ’
26 This is the name given by Gell-Mann. Zweig called them ‘aces’.
Appendix 1: The Elementary Particles 121

which was discovered in the 1950s. It has a spin equal to 3/2 and an elec-
tric charge equal to 2. It is formed as a bound state of three quarks u.
Its wave function must be anti-symmetric under the exchange of any
pair of these. On the other hand we can convince ourselves of the fol-
lowing: (i) the spins of the three quarks must be parallel in order to
build a total spin equal to 3/2, therefore the spin part of the wave func-
tion is symmetric; (ii) the space part must also be symmetric, because all
theoretical computations show that the minimum energy bound state
is obtained when the wave function is symmetric under the exchange
x ↔ y. We conclude that the anti-symmetry cannot come from either
the position or the spin of the quarks and therefore we must have
another variable associated with them. This reasoning led to the intro-
duction of the concept of colour. This assumes that each one of the six
quarks (u, d, c, s, t, b) may exist in three different types. We call them col-
ours, but we want to stress that there is no relation with the ordinary
sense of the word. It simply implies that we should attribute to each
quark an extra index ui , di , . . . , where i takes the values 1, 2, and 3.
It is with respect to this index that the N ∗++ wave function is anti-
symmetric. This sounds like an artifact, just another way of saying
that we have in fact 18 quarks. Although the counting is correct, the
physical interpretation is not. We must supplement the model with an
assumption, fully confirmed by experiment, that every hadron con-
tains all three colours in equal proportions and matter appears, in fact,
colourless.
When this idea was first proposed in 196427 it did not raise a great deal
of enthusiasm. It was gradually imposed in the 1970s when it was found
that its predictions were verified by experiment and it could provide the
basis for a fundamental theory of strong interactions, the one presented
in Section 6.3. A basic difficulty, not easy to explain, was the fact that
states with a single colour never appear in nature.
This brings us to the other strange property of quarks, which we also
presented in Section 6.3. The presence of quarks inside hadrons has been
verified experimentally. The first results came from the scattering of
high energy electrons off nucleons. It is the analogue of Rutherford’s
experiment which established the existence of the atomic nucleus. Here

27 The first proposal, in a slightly different form, was made by O. W.

Greenberg. It was followed a year later by a similar one by M. Y. Han and Y.


Nambu for integer charge quarks.
122 Appendix 1: The Elementary Particles

Table A1.3 The table of elementary particles today. This table reflects
our present ideas on the structure of matter. Quarks and gluons do not
appear as physical particles and the graviton has not been observed.

Table of elementary particles

Radiation quanta
Strong interactions Eight gluons
Electromagnetic interactions Photon (γ )
Weak interactions Bosons W + , W – , Z0
Gravitational interactions Graviton (?)
Matter particles
Leptons Quarks
1st family ν(e) , e– ui , di , i = 1, 2, 3
2nd family ν(μ) , μ– ci , si , i = 1, 2, 3
3rd family ν(τ ) , τ – ti , bi , i = 1, 2, 3
BEH boson

the experiments showed the presence of hard grains inside protons and
neutrons. Nevertheless, we have never succeeded in breaking a nuc-
leon and taking out a quark. As we explain in Section 6.3, we have
good reasons, experimental, numerical, and theoretical, to believe that
this is due to a property of the interaction which binds the quarks
inside the hadrons. The force between the quarks increases with the
distance; weak at short distances, it becomes very strong, possibly infin-
ite, when the separation becomes macroscopic and tends to infinity. We
called this property confinement and its rigorous derivation from first prin-
ciples is always a challenge for theorists. We present it schematically in
Section 6.3.

The table today


With the arrival of quarks the table of elementary particles again took a
rather simple form (see Table A1.3).
The table presents the elementary particles, or rather those assumed
to be, in three sectors: the radiation quanta, the matter particles, and, finally,
the famous BEH boson.

1. The radiation quanta. We have seen the role of the photon as the mes-
senger of electromagnetic interactions. In Chapter 4 we saw that
Appendix 1: The Elementary Particles 123

this concept of messenger is generalised to all interactions. The


particles which appear in this first sector of Table A1.3, are pre-
cisely the messengers of the four interactions presented in Section
A1.2. The geometrical theory we have developed in this book
determines uniquely their number and their properties. There
are eight gluons for strong interactions; the photon for electromagnetic
interactions; three intermediate bosons, denoted by W + , W – , and Z0 , for
weak interactions, and the graviton for gravitational interactions. All
these particles, with the exception of the graviton, have been iden-
tified experimentally. They are all bosons. The gluons, the photon,
and the intermediate bosons of the weak interactions have spin
equal to 1. We shall call them collectively gauge bosons. The spin of
the graviton is expected to be 2.
2. The matter particles. Here we find the particles which, according to
our present understanding, are the elementary constituents of
matter. They are spin 1/2 fermions and in Section A1.4 we saw that
the fermionic character is essential for the formation of nuclei and
atoms. In the table these particles are classified into three groups
which we call families. Let us start by looking at the first one: it con-
tains the electron e and its neutrino ν(e) , together with the pair
of quarks u and d. The index i which takes three values denotes
the three colours. This first family resembles the matter particles
of Table A1.2 which we considered elementary in 1932, with the
quarks u and d replacing the nucleons, proton, and neutron. All
macroscopic matter is made out of constituents belonging to this
first family.
The other two families appear as copies of the first. Each one con-
tains a doublet of leptons, with one charged lepton, μ and τ , and
one neutrino, as well as a doublet of quarks in three colours. These
heavy quarks are also confined and produce new hadrons. The
leptons μ and τ , as well as the new hadrons, are all unstable and
decay into particles of the first family.
3. The BEH boson. This is the subject of this book. It is electrically
neutral and its spin is equal to zero. According to our present
understanding, it is an independent component in the world of
elementary particles. In Chapter 7 we presented some theoretical
speculations attempting to guess a new physics beyond that of the
Standard Theory. In such a framework this boson could be linked
with the other particles of Table A1.3.
124 Appendix 1: The Elementary Particles

We end with the three rules which we formulated after the 1932 table.
The first two remain valid. Matter particles have spin 1/2 and all radi-
ation quanta which have been identified so far, have spin 1. The sum of
electric charges in each family is still equal to zero. It is worth noticing
that, in order to verify this rule, we need the hypothesis of the three col-
ours for the quarks. Indeed the electric charge of each lepton is –1. The
quarks of each family contribute 2/3 – 1/3 = 1/3 and we must multiply
this result by 3 in order to obtain the value +1. However, the third rule
is no longer valid. If we understand very well the role of the first family
particles in the structure of matter, we know of no good reason for the
existence of the other two. Why do we have three families, since just one
would have been sufficient? Why does nature reproduce what seems to
be three copies of the same thing? Such questions show the limits of our
understanding of the fundamental physical laws.
AP P E N D I X 2

From Sophus Lie to Élie Cartan

Introduction of the group concept is relatively recent in mathematics.


It appears for the first time, although only implicitly, in the work of
Leonhard Euler in the eighteenth century. The term ‘group’ was first
used by Évariste Galois in around 1830. As is often the case in mathem-
atics, this initial concept was enriched with time, becoming more and
more abstract. Since it underlies all ideas on symmetries that we use in
this book, we shall present in this appendix a few elementary notions.
Even in a very simplified form, they remain quite technical and require
some familiarity with the language used in mathematics. Some addi-
tional simple concepts will be explained as we go along. The aim of this
appendix is to help the reader who wishes to follow the reasoning which
led to the physical theories we have presented in this book, but it is by no
means necessary to understand the results. We shall mainly talk about
the work of two great mathematicians, the Norwegian Marius Sophus
Lie and the Frenchman Élie Joseph Cartan (see Figure A2.1).1

1 Marius Sophus Lie, one of the great figures of nineteenth-century math-


ematics, was born in Nordfjordeide, Norway in 1842. He wanted to follow a
military career, but he was forced to give up because of poor vision. He studied
at the University of Christiania (today Oslo) but it was only in 1867, at the age of
25, that he decided to major in mathematics. He travelled a lot to meet the great
mathematicians of his time. It seems that it was in Paris, in contact with the
French mathematician Camille Jordan, that he discovered the importance of
group theory in geometry. At the out break of the 1870 war he decided to leave
Paris on foot to go to Italy, but he was arrested as a German spy and imprisoned
in Fontainebleau. He was liberated only after the intervention of Jean Gaston
Darboux and returned to Norway. He occupied the chairs of Mathematics at the
Universities of Christiania and Leipzig and, between 1888 and 1893, published a
three volume monumental work under the title The Theory of Group Transformations.
His reasoning follows a strong geometrical intuition, not always appreciated
by his fellow mathematicians at the time. In 1874, he married Anna Birch, a
great-niece of Niels Henrik Abel, another famous Norwegian mathematician
126 Appendix 2: From Sophus Lie to Élie Cartan

Figure A2.1 Marius Sophus Lie (1842–99); Élie Joseph Cartan (1869–
1951).

Rather than introducing the abstract and precise definition of the


general group concept, we shall limit the discussion to a particular case,

who died in 1829 at the age of 27. Lie had worked in editing Abel’s collected
works. In poor health himself, he died in Norway in 1899.
Élie Joseph Cartan, born in 1869 at Dolomieu, is considered to be one of
the greatest French mathematicians from the turn of the nineteenth cen-
tury. He entered the École Normale Supeure (class 1888), where he followed
the courses of famous mathematicians like Goursat, Hermite, Darboux, and
Poincaré. Between 1892 and 1894 he studied at the Collège de France with a
scholarship from the Pécot Foundation. It was during this period that he corres-
ponded with Sophus Lie. In 1894 he published his first results on the complete
classification of all finite dimensional Lie algebras, one of the most remarkable
mathematical results of the end of the nineteenth century. He was 25 years old.
In 1910, after a study of the representations of the three-dimensional rotation
group, he introduced the notion of spinors. Notice that in physics, the concept
of spin was only introduced in 1925 by the Dutchmen Samuel Goudsmit and
George Uhlenbeck. Cartan made many other important contributions which
turned out to be interesting for physics, such as the classification of symmet-
ric spaces and a reformulation of differential geometry for general relativity.
An outstanding teacher, he taught mathematics at the University, the École
Normale Supeure, and the École de Physique-Chimie. He is considered to be
one of the main founders of the modern school of French mathematics. He
died in Paris in 1951.
Appendix 2: From Sophus Lie to Élie Cartan 127
which happens to be the only one we have discussed in this book. This
is the concept of transformations of the coordinate system in a given
space.2
Definition: Let G be a set of transformations which act on the coordinate system of some
space. We denote a particular transformation by g. We shall say that G forms a group if
the following conditions are fulfilled.

1. If we apply to the coordinate system two successive transformations g1 and g2


belonging to G , the result is equivalent to applying to the coordinate system a
transformation g3 which also belongs to G . We write this composition property
as g1 g2 → g3 .
2. G contains the trivial transformation g0 which leaves the coordinate system
unchanged. We call it ‘identity’.
3. If G contains the transformation g, it contains also the inverse transformation,
denoted g–1 , such that the successive application of g and g–1 , taken at any
order, brings the coordinate system to its initial state. We write this property as
gg–1 = g–1 g → g0 .

It is obvious that all the transformations we consider in this book,


translations, rotations, inversions, etc. form groups. We can distinguish
between finite groups, which contain only a finite number of transforma-
tions, and infinite groups which contain an infinite number. For example,
space inversion is a group with only two elements, to wit the identity
x → x and the proper inversion x → –x. The same with time inver-
sion. The symmetry groups of regular crystals are also examples of finite
groups. Contrary to this, rotations or translations are infinite groups.
Another notion which is intuitively obvious is that of a sub-group.
The rotations around one axis is a sub-group of the three-dimensional
rotation group.
In this book we have also used a property of group transformations
which is familiar from rotations. Consider the rotations around an axis.
Each one is characterised by an angle. It is easy to write the composi-
tion law: the result of two successive rotations, one with angle θ1 and a
second one with θ2 , is equivalent to a rotation with angle θ3 = θ1 + θ2 .
It is also obvious that we obtain the same result if we change the order
and apply the rotation with angle θ2 first and θ1 second. We say that

2 In fact, this is not far from Lie’s geometrical ideas. A mathematician would

say today that we study some particular group representations.


128 Appendix 2: From Sophus Lie to Élie Cartan
these transformations commute and we call the group commutative. We also
call it abelian, in honour of the Norwegian mathematician Niels Henrik
Abel (1802–29).
We have also encountered non-commutative, also called non-abelian,
groups. The simplest example is the group of rotations in a three-
dimensional space. It is easy to check that the result of the composition
of a rotation around the x axis and one around the y axis, depends on
the order in which we perform them. From the mathematical point
of view, non-abelian groups are more interesting because they have a
richer structure. The Yang–Mills gauge theories we have presented in
this book are based precisely on non-abelian groups.
Were we to give free rein to our imagination, we could define a
large variety of groups, including groups of transformations, because
the definition we have given is very general. It was the great merit of
Sophus Lie that he chose a particular class among infinite groups3 which
is large enough to cover most of the interesting cases and precise enough
to allow for detailed study. We shall not follow his introduction, which
today represents only a historical interest, but our approach, which
follows a physicist’s motivations, respects his own geometrical spirit.
We have seen that the elements of the transformation groups we
have been considering depend on one or several parameters. Rotations
on angles, translations on vectors, etc. Lie understood that an essential
notion which will characterise the properties of these groups is that of
continuity, therefore we can restrict ourselves to the study of groups whose
elements depend continuously on the parameters. Even if the notion of continuity
sounds intuitively obvious, its precise definition is quite technical and
we are not going to present it here. On the other hand it is clear that
this choice allows us to apply to groups a good part of the theory of
continuous functions, and we can convince ourselves that these groups,
which we call Lie groups, have rich and multiple properties which make
them interesting to both mathematicians and physicists.4

3 The study and classification of finite groups turned out to be a very inter-
esting chapter of modern mathematics and it was completed only during the
second half of the twentieth century. We shall not have occasion to use it in this
book.
4 Rather than engaging in lengthy and abstract definitions, it may be

instructive to give an example of an infinite group of transformations which is


not a Lie group. Let us take our familiar example of the group of rotations around
an axis. Its elements depend on the rotation angle θ which we may decide to
Appendix 2: From Sophus Lie to Élie Cartan 129
A consequence of this property of continuity, which was already
noticed by Lie, is that, for a given group, we can study its structure
by limiting ourselves to small transformations in the neighbourhood
of the identity element. For example, if we want to study the group of
rotations, it is sufficient to consider rotations with infinitesimal angles.
The property of continuity will allow us as a next step to reconstruct
large angle rotations by a succession of ‘small’ rotations. At first sight
this remark sounds rather trivial, but Lie understood that the small
transformations in the neighbourhood of the identity give rise to a new
structure which encodes all the essential properties of the group. Today
we call this structure Lie algebra, but Lie called it an infinitesimal group. Once
more, we shall not give the precise definition but we shall show its
importance with some particular examples.
Consider again the example of the group of rotations around an axis.
We have seen already that the composition law is given by the sum-
mation over the angles: θ3 = θ1 + θ2 . θ is an angle, therefore it takes
values between –π and +π. We can visualise the transformations as
the motion on a circle. If we limit ourselves to infinitesimal angles, θ
also measures the length of the arc on the circle, therefore the same
law also describes the group of translations along an axis. We have just
seen the interest of this analysis: two groups, which are globally very
different, motion on a circle from –π to +π and on a line from –∞ to
+∞, are locally the same, or, using the technical term, they have the same
Lie algebra.
We can object that this example is trivial and one could have guessed
the result by inspection. This objection is valid, but we can give more
sophisticated examples for which the tools developed for the study of
Lie algebras will be essential. To do so, we need the work of Cartan.
We have already introduced the number of independent transforma-
tions we can perform in the space we are considering. This number will

measure, for example, in degrees. Let us now consider a particular sub-group


formed by the rotations whose angle, expressed in degrees, is given by a rational
number (a number which we can write as a ratio of two integers). It is obvi-
ously a group because the sum of two rational numbers is a rational number
and the other two conditions are also fulfilled. It is not a Lie group because we
do not have the property of continuity; between two rational numbers there is
an infinity of irrational numbers. So this group is not covered by Lie’s definition
and it is also clear that it is not particularly interesting.
130 Appendix 2: From Sophus Lie to Élie Cartan
play an important role in the discussion that follows, so we shall give
it a name: the dimension of the Lie algebra. We have seen that for rotations
in a three-dimensional space, this number is equal to three. It corres-
ponds to the three independent rotations we have learnt in geometry.
A warning: this equality between the number of space dimensions and
that of the independent rotations is a numerical accident, valid only in
three dimensions. For example, if we consider a plane (two dimensions),
we have only one rotation around the axis perpendicular to the plane.
With a bit of imagination we can convince ourselves that the number
of independent rotations we can perform in a four-dimensional space
equals six.5
Up to now we have considered only spaces with real coordinates. But
already in Section 3.3, when we were discussing isotopic spin symmetry,
we remarked that in an internal symmetry space the basic vectors of
the coordinate system may be the particle fields which, like the wave
functions, can take complex values. Therefore we must define trans-
formations in such spaces. We did it implicitly in Section 6.3 when we
talked about quantum chromodynamics as the theory invariant under
transformations which mix the quark fields with different colours. It
is obvious that the further we go in this direction, the more abstract
the discussion becomes and it loses a simple geometrical significance.
Cartan’s results allow us to put a definite order to this labyrinth of
spaces and transformations.
In 1894 Cartan obtained the complete classification of all finite-
dimensional Lie algebras,6 a result of capital importance for both physics
and mathematics. He showed that there exist five classes of Lie algebras,
four of them form infinite series and the fifth contains five exceptional
cases. For each algebra he gave also its dimension, i.e. the number of
independent transformations. His notation is not the one we use in
physics, but we can summarise his results as follows.

5 The counting for any dimension is easy: we know that in a plane we have
only one possible rotation. So, let us count planes rather than axes. A plane is
defined by two axes, therefore we must compute the independent pairs of axes.
In n dimensions each axis can pair with n – 1 of the remaining ones. This gives
us n(n – 1). But this counts every pair twice, so the final result is n(n – 1)/2. For
n = 2 we get 1, for n = 3 we get 3, for n = 4 we get 6, etc.
6 If we want to be more precise we must add ‘...for semi-simple groups on

complex numbers’, but these properties will not be needed in our discussion.
Appendix 2: From Sophus Lie to Élie Cartan 131
1. For the rotations in an n-dimensional real space we write the
group as O(n), where O is an abbreviation for ‘Orthogonal’.7 As
we have seen, the dimension of the corresponding algebra equals
n(n – 1)/2.
2. For the transformations in an n-dimensional complex space we
write the group as SU(n). ‘S’ stands for ‘special’ and means that we
do not consider a group of transformations which simultaneously
change the phase of all vectors. ‘U’ stands for ‘Unitary’ which
means that the transformations do not change the modulus of
the vectors.8 The dimension of the algebra equals n2 – 1.9

We can search in this list the groups we have used for the Standard
Theory.

(i) The group of rotations around a single axis is not in the Cartan
list. It is an abelian group with a one-dimensional algebra which
mathematicians consider to be trivial. We denote this by U(1).
(ii) In Cartan’s classification the algebra of the group O(3), i.e. the
rotations in a three-dimensional real space, is identical to that
of SU(2). We used this result implicitly when we said that the
Heisenberg group of isotopic spin, whose transformations act on
the two complex vectors given by the proton and neutron fields,
is equivalent to the group of three-dimensional real rotations;
see Chapter 3. The dimension of the algebra equals 3, hence the
three weak interaction gauge bosons W + , W – , and Z0 .
7 In Cartan’s notation the algebras of these groups form two of his four infin-

ite series, which he denotes as Bn n ≥ 2 for the groups O(2n + 1) and Dn n ≥ 4 for
the groups O(2n).
8 Cartan notes this series as A n ≥ 1 for the group SU(n + 1).
n
9 The counting is also relatively simple. A general transformation applied to

a set of n complex vectors depends on n2 complex numbers, which give 2n2 real
numbers. It is convenient to write them as a tableau with n lines and n columns.
We call these tableaux matrices. We must now count the constraints implied by
the unitarity condition, i.e. the conservation of vector modulus. In the matrix
notation this condition can be written as U(n)U ∗ (n) = 1, where 1 is the unit
matrix whose only non-zero elements are on the diagonal and are all equal to 1.
U ∗ (n) represents the generalisation of the concept of complex conjugation for
matrices. This condition gives n2 constraints. A further constraint comes from
not counting the transformation of a common change of phase for all vectors.
The result is n2 – 1.
132 Appendix 2: From Sophus Lie to Élie Cartan
(iii) The algebra of the group SU(3), which is the group of unit-
ary transformations acting on the three-dimensional complex
space of the three coloured quark fields, has dimension 32 –
1 = 8. Indeed, the number of gluons, the gauge bosons of
quantum chromodynamics, equals 8.
Index

O(n) 131 broken symmetry phase 57, 58, 66, 69,


SU(n) 131 75, 76
α particle 99 Brookhaven 116
β-decay 47, 110, 113, 115 Brout, Robert (1928–2011) 15
π meson 41, 77, 91, 116 Brout-Englert-Higgs 73
ρ 57 buckling 52
ρ particle 63, 66, 70, 75
θ 57 C
θ particle 63, 66, 70, 75 calorimeter 113
’t Hooft, Gerardus (1946–) 86 Cartan classification 130
Cartan, Élie Joseph (1869–1951) 74, 83,
A 125
Abel, Niels Henrik (1802–1829) 128 CERN 1, 13, 65, 70, 87, 88, 113, 117
Anderson, Carl David (1905–1991) 107 Chadwick, James (1891–1974) 111, 115
Anderson, Philip Warren (1923–) 68 charm 87, 91
angular momentum 27, 30, 46, 107 chirality 45, 87
anti-ferromagnetism 59 left 46, 48
anti-particle 107 right 46
antineutron 107 colour 83, 121, 130
antiphoton 107 confinement 79, 81, 84, 122
antiproton 107 conservation
asymptotic freedom 80, 88 angular momentum 27, 114
energy 29
B momentum 26
baryon 120 conservation law 23
baryon number 84, 106, 109 constituents of matter 45, 48, 65, 76,
baryon number conservation 107 118, 122, 123
Becquerel, Antoine Henri continuity 128
(1852–1908) 110 continuous spectrum 111
BEH boson 108, 109, 113, 122, 123 control quantity 57, 58
BEH mechanism 85, 93, 95 coordinate system 24, 27, 28, 30, 32, 36,
BEH particle 15, 62, 65, 75, 76, 86, 90, 43, 71, 83, 105, 127
94, 95 cosmological constant 7, 17
big bang 9 cosmology 3, 6, 10
Bohr magneton 104 Coulomb potential 43
Bohr, Niels Henrik David Cowan, Clyde Lorrain (1919–1974) 116
(1885–1962) 37, 40, 100, critical point 55
114 critical value 53, 55, 58
Bose, Satyendra Nath (1894–1974) 107 Curie point 58
Bose-Einstein condensation 109 Curie’s theorem 50
boson 45, 65, 107, 109 Curie, Pierre (1859–1906) 50, 63
134 Index

D energy crisis 112, 113


dark energy 17 Englert, François (1932–) 15
dark matter 16 Euclid, (∼300 BC) 25
Davis, Raymond Jr (1914–2006) 117
de Broglie, Louis Victor (1892–1987) 37, F
40, 100 family 76, 88, 92, 123
de Sitter, Willem (1872–1934) 7 fermi 5, 85
Debije, Petrus Josephus Wilhelmus Fermi, Enrico (1901–1954) 107, 110, 111,
(1884–1966) 37 115
decoupling 16 FermiLab 88
degenerate ground state 54, 62 fermion 45, 93, 107, 109, 120, 123
degree of freedom 67, 75 ferromagnetism 58
Democritus, (∼460–370 BC) 97 Feynman, Richard Phillips
DESY 89 (1918–1988) 103
Dicke, Robert Henry (1916–1997) 19 Fizeau, Armand Hippolyte Louis
Dirac equation 104, 107 (1819–1896) 8
Dirac, Paul Adrien Maurice Fock, Vladimir Aleksandrovich
(1902–1984) 37, 103 (1898–1974) 36
disordered phase 60 Friedmann, Alexandr Alexandrovich
domain 60 (1888–1925) 7
Doppler, Christian Andreas
(1803–1853) 8 G
Doppler-Fizeau effect 8 Gargamelle collaboration 117
duality gauge boson 44, 48, 66, 67, 69, 72, 74, 75,
field-particle 40, 63 84, 85, 87, 93, 123, 131
particle-wave 37 charged 86
Dyson, Freeman John (1923–) 104 neutral 86
gauge interaction 43, 65
E gauge invariance 34
Eddington, Arthur Stanley gauge symmetry 33, 44, 71, 73, 84
(1882–1944) 8 internal 36
Einstein, Albert (1879–1955) 6, 9, 24, 30, gauge theory 86, 92, 128
35, 36 Gauss’ theorem 51
electric charge 84, 106 Geiger, Johannes Wilhelm
electric charge conservation 106 (1882–1945) 99, 111
electric charge quantisation 92 Gell-Mann, Murray (1929–) 97, 120
electromagnetic interactions 41, 43, 68, geodesic 35
71, 72, 79, 89, 102, 105, 118, 123 geometry 4, 35, 74, 86
electron-Volt, eV 100 GeV 100
electroweak 14 Glashow, Sheldon Lee (1932–) 73, 75
electroweak interactions 76 gluon 84, 89, 123, 132
elementary particle 1, 97 Goldhaber, Maurice (1911–2011) 116
Elementary Particles Goldstone particle 64, 65
Table 117, 122 Goldstone theorem 62, 64
Ellis, Charles Drummond Goldstone, Jeffrey (1933–) 63
(1895–1980) 112 gravitational interactions 71, 92, 102
Index 135

gravitational waves 20 Koshiba, Masatoshi (1926–) 117


graviton 123 Kramers, Hendrik Anthony
Gross, David (1941–) 84 (1894–1952) 114
ground state 54, 56, 61, 66
group 125, 127 L
abelian 128 Lamb, Willis Eugene (1913–2008) 103
commutative 128 Landau, Lev Davidovich (1908–1968) 28
finite 127 lattice simulations 82, 90
infinite 127 Lederman, Leon Max (1922–) 88, 116
non-abelian 128 Lemaître, George (1894–1966) 7, 9
non-commutative 128 LEP 117
group of transformations 127 lepton 45, 65, 76, 87, 109
lepton number 109
H LHC 13, 16, 75, 79, 92, 95
hadron 77, 109, 119, 120 Lie algebra 74, 129
Hahn, Otto (1879–1968) 111 dimension 130
Hale, George Ellery (1868–1938) 7 Lie group 74, 128
Heisenberg, Werner Karl Lie, Marius Sophus (1842–1899) 74, 125
(1901–1976) 30, 37, 39, 59 local phase invariance 39
higgs particle 15 long range correlation 62, 68
Higgs, Peter (1929–) 15 long range force 43, 44
high temperature phase 58, 61, 93 Lorentz transformation 46
Hilbert, David (1862–1943) 23 Lorentz, Hendrik Antoon
Hoyle, Fred (1915–2001) 11, 20 (1853–1928) 46
Hubble space telescope 17 low temperature phase 61, 69, 93
Hubble’s law 9 Lowell Observatory 8
Hubble, Edwin Powell (1889–1953)
6, 9 M
magnetic moment 104
I anomalous 103, 104
inflation 14 magnetisation 58, 60, 61
intermediate boson 72, 74, 123 Marsden, Ernest (1889–1970) 99
internal space 33, 43, 71, 74 mass 40, 105
internal symmetry 32, 36, 39, 72, 130 mass generation 15
invariance 23, 26, 27, 29, 30, 34, 38, 44, Maxwell’s equations 28
48, 56, 59, 68, 71, 74, 85 McDonald, Arthur Bruce (1943–) 117
isospace 32 mediator 41, 43–45, 77, 80, 105, 119
isotopic spin 30, 39, 115, 130, 131 Meitner, Lise (1878–1968) 111
Merleau-Ponty, Jacques (1916–2002) 10
K meson 81, 89, 119, 120
Kajita, Takaaki (1959–) 117 MeV 100
Kamiokande 117 microwave cosmic radiation 18
Kant, Immanuel (1724–1804) 6 Mills, Robert (1927–1999) 33, 39
Kemmer, Nicholas (1911–1998) 31 mirror symmetry 28
Klein, Felix Christian (1849–1925) 23 moment of inertia 55
Klein, Oscar Benjamin (1894–1977) 39 momentum 26
136 Index

Mount Palomar Telescope 7 proto-galaxies 16


Mount Wilson Observatory 7 proton decay 107
muon 119
Q
N quantum chromodynamics 85, 88, 130
Néel, Louis Eugène Félix (1904–2000) 59 quantum electrodynamics 104
Nambu, Yoichiro (1921–) 63 quantum field theory 80, 103
neutral current 117 Quantum Mechanics 36, 93, 114
neutrino 20, 47, 110, 115 quantum number 84, 106, 119
neutrino mass 92 conserved 84
neutrino oscillations 117 quantum theory
neutrinos old 37
three species 117 quark 15, 45, 65, 76–78, 83, 87, 89, 120
Newton’s equation 28, 29, 35
Noether’s theorem 23, 26, 27, 29, 83 R
Noether, Amalie Emmy (1882–1935) 23 Röntgen, Wilhelm Conrad
normalisation condition 38 (1845–1923) 98
nucleon 32, 77, 106 Rabi, Isaac (1898–1988) 103, 119
nucleosynthesis 16, 48 radiation quantum 105, 118, 122
range of interactions 43
O Reines, Frederick (1918–1998) 116
order parameter 54, 56, 57, 58, 61 relativity
ordered phase 60, 62 general 7, 9, 35, 39, 71, 92
special 30, 68
P rest frame 46
parity 28, 46, 48 rotation 25, 27
parity violation 29 Rubbia, Carlo (1934–) 88
Particle Data Book 120 Rutherford, Ernest, 1st Baron of Nelson
Pauli’s exclusion principle 109, 114 (1871–1937) 78, 97, 111
Pauli, Wolfgang Ernst (1900–1958) 39,
111, 114 S
Penzias, Arno Allan (1933–) 18 Salam, Abdus (1926–1996) 73
Perl, Martin Lewis (1927–2014) 88 Schrödinger equation 36, 104
phase transition 14, 57, 61, 64, 85, 89 Schrödinger, Erwin Rudolf Josef
photon 41, 105, 106, 123 Alexander (1887–1961) 36, 103
mediator of the electromagnetic Schwartz, Melvin (1932–2006) 116
interaction 105 Schwinger, Julian Seymour
pion 41 (1918–1994) 103
Planck space mission 19 Shelter Island Conference 103
Planck’s constant 36 short range force 43
Planck, Max Karl Ernst Ludwig SLAC 88
(1858–1947) 36, 100 Slater, John Clarke (1900–1976) 114
polarisation 68 Slipher, Vesto Melvin (1875–1969) 8
Politzer, David (1949–) 84 Sommerfeld, Arnold Johannes
positron 107 Wilhelm (1868–1951) 37
problem of mass 45, 48, 66 space inversion 27, 46, 48
Index 137

space symmetry 24 V
space-time 30, 71 vacuum 64
spin 30, 45, 59, 67, 107, 114 van der Meer, Simon (1925–2011) 88
spontaneous symmetry breaking 14, van der Waals, Johannes Diderik
52, 62, 74, 80 (1837–1923) 79
gauge interactions 65 van der Waerden, Bartel Leendert
spontaneously broken symmetry 50 (1903–1996) 24
stability 51 Veltman, Martinus Justinus
Standard Model 4, 76, 86, 91 Godefriedus (1931–) 86
steady state theory 11, 20 visible mass 16
Steinberger, Hans Jacob (1921–) 116 Vogel, Hermann Carl (1841–1907) 9
stelar energy 48
strong interactions 45, 71, 77, 102 W
subgroup 127 wall of ignorance 13
supersymmetry 95 wave function 36, 36, 83, 108
symmetric phase 57, 58, 69, 74, 76, 80, 93
antisymmetric 108, 121
symmetric solution 53 symmetric 108
symmetry 4, 22, 43, 50 weak interactions 29, 45, 47, 65, 71, 72,
gauge 33 102
global 84 Weinberg, Steven (1933–) 45, 73
internal 30 Weyl, Hermann Klaus Hugo
local 33, 38 (1885–1955) 33
Wilczek, Frank (1951–) 84
T Wilson, Robert Woodrow (1936–) 18
thermal fluctuations 60
Wooster, William Alfred
time reversal 29
(1903–1984) 113
time symmetry 29
Wu, Chien Shiung (1912–1997) 29
Tomonaga, Sin Itiro (1906–1979) 103,
104
transformation X
continuous 28 X rays 98
discrete 28
gauge 34, 44, 74, 85 Y
global 44 Yang, Chen Ning (1922–) 33, 39
local 34, 44 Yang-Mills theory 39, 73, 80, 85,
translation 25 128
local 34, 35 Young modulus 55
space 26, 44 Yukawa potential 43
time 29 Yukawa, Hideki (1907–1981) 41,
77, 91
U
uncertainty relation 42
units 4 Z
unstable state 53 Zweig, George (1937–) 120

You might also like